Microfluidic Approaches for Probing Protein Phosphorylation in Cells

Jingren Deng

Dissertation submitted to the faculty of the Virginia Polytechnic Institute and State

University in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

In

Biological Sciences

Committee Members

Iulia M. Lazar, Chair

Deborah F. Kelly

Florian Schubot

Mark A. Stremler

May 24, 2018

Blacksburg, Virginia

Keywords: microfluidics, fast cell processing, protein phosphorylation, mass

spectrometry

Microfluidic Approaches for Probing Protein Phosphorylation in Cells

Jingren Deng

ABSTRACT

Protein phosphorylation plays critical roles in diverse cellular functions, including cell

cycle, growth, differentiation, and apoptosis. Deregulated phospho-signaling is often

associated with many human diseases and cancers. Despite tremendous efforts to

investigate the molecular mechanisms that control the functionality of phospho-signaling

pathways, only limited progress has been made on describing the temporal and spatial

profiles of cellular protein phosphorylation. The main challenges associated with the

study of phospho-signaling processes in cells are related to the short time-scale of certain

phosphorylation and dephosphorylation events, the low abundance of the phosphorylated

protein forms as compared to their non-phosphorylated counterparts, the complicated and

time-consuming sample preparation methods that are accompanying such type of work,

and, last, the performance of the detection methods that are suitable for assessing protein

phosphorylation. To tackle the challenges associated with the investigation of protein

phosphorylation in cells, our objective was to develop a combined mass spectrometry

(MS) and microfluidics strategy that enables fast sampling and sensitive detection of key

signaling phosphoproteins in complex biological samples. MS is the most widely used

analytical tool in the field of proteomics due to its high sensitivity, specificity, and

throughput. Microfluidics has been proven as a suitable platform for handling small

volumes of scarce samples, being also amenable to automation, integration, and

multiplexing. To achieve our objective, this study was conducted in multiple steps: (1)

We performed a comprehensive analysis of the factors that affect the performance of

mass spectrometry detection (i.e., sensitivity, reproducibility, ability to accurately

identify a large number of proteins from complex samples), when used in conjunction

with technologies that are conducted in a non-standard fashion, on short time-scales; (2)

We developed and evaluated a miniaturized strategy for rapid proteolytic digestion and

phosphopeptide enrichment; (3) We demonstrated sensitive detection and quantification

of phosphopeptides from complex biological samples using multiple reaction monitoring

mass spectrometry (MRM-MS) and microfluidic sample processing; and (4) We

developed a microfluidic platform for handling and processing cells that enables the

investigation of biological processes that occur on short time-scales, and that can be

integrated with the devices developed for the analysis of phospho-proteins. SKBR3 cells

were used as a model system for developing and demonstrating the microfluidic chips.

The detection and quantification of phospho-proteins involved in MAPK (mitogen

activated protein kinase) signaling pathways was achieved at the low nM level. Overall,

this study demonstrates proof-of-concept applicability of a microfluidics-MS strategy for

monitoring phosphorylation processes in signaling networks.

Microfluidic Approaches for Probing Protein Phosphorylation in Cells

Jingren Deng

GENERAL AUDIENCE ABSTRACT

Cellular protein phosphorylation plays critical roles in cellular functions, and deregulated

phosphorylation is often associated with many human diseases and cancers. Despite

tremendous efforts to investigate the molecular mechanisms that control cellular protein

phosphorylation events, limited progress has been made on describing the temporal and

spatial profiles. The main challenges are related to the short time-scale of certain

phosphorylation and dephosphorylation events, the low abundance of the phosphorylated

protein forms as compared to their non-phosphorylated counterparts, the complicated and

time-consuming sample preparation methods that are accompanying such type of work,

and, last, the performance of the detection methods that are suitable for assessing protein

phosphorylation. To address the issues involved in the investigation of protein

phosphorylation in cells, we developed a novel strategy using mass spectrometry (MS)

and microfluidics. This study was conducted in multiple steps: (1) We performed a

comprehensive analysis of the factors that affect the performance of mass spectrometry

detection; (2) We developed and evaluated a miniaturized strategy for rapid proteolytic

digestion and phosphopeptide enrichment; (3) We demonstrated sensitive detection and

quantification of phosphopeptides from complex biological samples; and (4) We

developed a microfluidic platform for handling and processing cells that enables the

investigation of biological processes that occur on short time-scales, and that can be

integrated with the devices developed for the analysis of phospho-proteins.

v

ACKNOWLEDGEMENTS

The work that led to the completion of this dissertation would not be possible without the

support of many people. I would like to acknowledge all of them for how they have

influenced my life in so many positive ways and made me the person I am today.

First and foremost, I would like to thank my advisor, Dr. Iuliana M. Lazar, for her patient

guidance and support throughout the years. I truly appreciate all her contribution of time,

caring, and financial support to make my Ph.D. journey a rich experience. I am grateful

to have had this opportunity to work with her. I would also like to thank all my

committee members, Dr. Deborah F. Kelly, Dr. Florian Schubot, and Dr. Mark A.

Stremler for their insightful suggestions, advice, and encouragement on my research and

career choices.

I am fortunate to have had a wonderful support network of colleagues and friends during

my time at Blacksburg. Thanks to all members of the Lazar group, especially Wooram

Lee, Fumio Ikenishi, and Shreya Ahuja, for their support and assistance.

This work was supported by an NSF grant number DBI-1255991, and my academic

training was also partially covered by the BIOTRANS, formerly the MULTISTEPS, at

Virginia Tech.

vi

Above all, I want to thank my parents, Shihao Deng and Baoyu Dai, and my girlfriend,

Yi Liu, for their unconditional love and support.

vii

TABLE OF CONTENTS

ABSTRACT ................................................................................................................................... ii

ACKNOWLEDGMENTS ............................................................................................................ iv

TABLE OF CONTENTS ............................................................................................................. vi

LIST OF FIGURES ...................................................................................................................... ix

LIST OF TABLES ........................................................................................................................ xi

LIST OF ABBREVIATIONS ..................................................................................................... xii

CHAPTER 1: Purpose of Study .................................................................................................. 1

CHAPTER 2: Introduction ......................................................................................................... 3

2.1 Cancer ...................................................................................................................................... 3

2.2 Cell signaling and protein phosphorylation ......................................................................... 5

2.3 Mass spectrometry and proteomics ...................................................................................... 8

2.4 Microfluidics ......................................................................................................................... 17

2.5 References ............................................................................................................................. 18

CHAPTER 3: Materials and Methods ..................................................................................... 23

3.1 Chemicals and materials ...................................................................................................... 23

3.2 SKBR3 cell culture ............................................................................................................... 24

3.3 Fluorescence-activated cell sorting ..................................................................................... 24

3.4 Cell processing ...................................................................................................................... 25

3.5 Conventional proteolytic digestion and phosphopeptide enrichment .............................. 26

3.6 MS analysis and data processing ......................................................................................... 27

3.7 Microfluidic chip fabrication ............................................................................................... 29

CHAPTER 4: Fast Enzymatic Digestion: Counts, Coverage, Reproducibility ..................... 31

4.1 Introduction .......................................................................................................................... 33

4.2 Material and methods .......................................................................................................... 36

4.3 Results and discussion .......................................................................................................... 40

Peptide identification and protein sequence coverage ...................................................... 40

Reproducibility ..................................................................................................................... 43

Detection of low-abundance peptides ................................................................................. 44

Missed cleavage sites ............................................................................................................ 49

Peptide size and quality of MS identifications ................................................................... 51

Gravy index ........................................................................................................................... 56

4.4 Conclusions ........................................................................................................................... 57

viii

4.5 References .............................................................................................................................. 58

CHAPTER 5: Proteolytic Digestion and TiO2 Phosphopeptide Enrichment Microreactor for

Fast MS Identification of Proteins ............................................................................................ 60

5.1 Introduction .......................................................................................................................... 62

5.2 Material and methods .......................................................................................................... 66

5.3 Results and discussion .......................................................................................................... 69

Microreactor fabrication ..................................................................................................... 69

Microreactor for proteolytic digestion ............................................................................... 70

Microreactor for phosphopeptide enrichment ................................................................... 81

Microreactor for proteolytic digestion and phosphopeptide enrichment ....................... 86

Analysis of cell extracts ........................................................................................................ 90

5.4 Conclusions ........................................................................................................................... 91

5.5 References .............................................................................................................................. 92

CHAPTER 6: Streamlined Microfluidic Analysis of Phosphopeptides Using Stable Isotope-

Labeled Synthetic Peptides and MRM-MS Detection ............................................................. 98

6.1 Introduction ........................................................................................................................ 100

6.2 Material and methods ........................................................................................................ 102

6.3 Results and discussion ........................................................................................................ 108

Selection of MRM transitions ............................................................................................ 108

Microfluidic chip design for the detection of phosphopeptides ...................................... 112

Effect of high pH on the detection of phosphopeptides ................................................... 116

Two-step sequential elution of hydrophilic and hydrophobic phosphopeptides ........... 124

MRM analysis of phosphopeptides from SKBR3 breast cancer cell extracts ............... 125

6.4 Conclusions ......................................................................................................................... 128

6.5 References ............................................................................................................................ 129

CHAPTER 7: Microfluidic Reactors for Cell Stimulation, Lysis and Sample Collection for

MS Analysis ............................................................................................................................... 135

7.1 Introduction ........................................................................................................................ 136

7.2 Material and methods ........................................................................................................ 139

7.3 Results and discussion ........................................................................................................ 143

Design of the devices ........................................................................................................... 143

COMSOL simulation of EGF infusion .............................................................................. 149

Flow visualization ............................................................................................................... 150

Sample processing-cell loading, stimulation, and lysis .................................................... 155

HPLC-MS analysis and data interpretation .................................................................... 158

ix

7.4 Conclusions ......................................................................................................................... 158

7.5 References ........................................................................................................................... 159

CHAPTER 8: Conclusions and Future Work ....................................................................... 162

x

LIST OF FIGURES

Figure 4.1 Effect of enzymatic reaction time on the identification of peptides and proteins

generated from SKBR3 cell extracts .......................................................................................... 42

Figure 4.2 Venn diagrams of unique peptide (A-F) and protein (G-I) overlaps from the 7

min-18 h time-point experiments ............................................................................................... 46

Figure 4.3 The percetage change in peptides with various numbers of missed cleavage sites

by the progression of enzymatic digestion ................................................................................. 47

Figure 4.4 Trends in missed cleavage sites and peptide length as a function of enzymatic

reaction time ................................................................................................................................ 47

Figure 4.5 Stacked column chart illustrating amino acid frequency distributions ............... 48

Figure 4.6 Xcorr score distributions as a function of m/z ........................................................ 54

Figure 4.7 Gravy index histograms for peptides with 0 to 4 missed cleavages ...................... 55

Figure 5.1 Protein and peptide identifications using various conditions for enzymatic

protein digestion .......................................................................................................................... 73

Figure 5.2 Venn diagram comparisons for unique peptide identifications in proteolytic

digestion reactions performed with the microreactor and overnight protocols .................... 78

Figure 5.3 Full mass scans acquired during the analysis of standard protein mixture digests

....................................................................................................................................................... 84

Figure 5.4 Comparison of the performance of the microreactor and conventional approach

....................................................................................................................................................... 89

Figure 6.1 Schematic diagrams of the phosphoprotein analysis microfluidic chips ............ 113

Figure 6.2 Full mass scans acquired during the infusion of a mixture of three stable isotope-

labeled synthetic phosphopeptides ........................................................................................... 117

xi

Figure 6.3 Microfluidic chip MRM-MS analysis of an SKBR3 cell extract digest spiked with

three heavy standards ............................................................................................................... 118

Figure 6.4 Retention of stable isotope-labeled synthetic phosphopeptides on C18 particles

using loading solutions of different pH .................................................................................... 121

Figure 6.5 Elution of stable isotope-labeled synthetic phosphopeptides from TiO2 particles

using eluents of different pH .................................................................................................... 122

Figure 6.6 Two-step sequential elution of on-column enriched stable isotope-labeled

synthetic phosphopeptides ........................................................................................................ 123

Figure 6.7 Two-step sequential elution of phosphopeptides from an on-chip enriched

SKBR3 cell extract spiked with stable isotope-labeled synthetic peptides ........................... 127

Figure 7.1 Schematic diagram of the first chip design ........................................................... 146

Figure 7.2 Schematic diagram and simulation of the second chip design ............................ 147

Figure 7.3 COMSOL simulation of the on-chip EGF infusing process ................................ 148

Figure 7.4 Visualization of flow in an empty chip .................................................................. 152

Figure 7.5 Visualization of flow in a chip fully packed with latex beads of 11.6 µm in

diameter ...................................................................................................................................... 153

Figure 7.6 On-chip processing of HBEC5i cells ...................................................................... 154

Figure 7.7 Sample processing on a chip of the second design ................................................ 157

xii

LIST OF TABLES

Table 5.1 Analytical processing steps for the microreactors ................................................... 72

Table 5.2 Comparison of results generated with the overnight and microreactor digestion

protocols ....................................................................................................................................... 77

Table 5.3 Selectivity of the phosphopeptide enrichment process for the spin tip and

microreactor ................................................................................................................................. 83

Table 5.4 Phosphopeptide sequences identified in a mixture of bovine proteins by using

conventional approach and the microreactors ......................................................................... 83

Table 6.1 Proteins mapped to ERBB2/MAPK pathways in EGF stimulated SKBR3 cells 110

Table 6.2 Phosphopeptides selected for MRM-MS analysis of EGF-stimulated SKBR3 cell

extracts ....................................................................................................................................... 111

Table 6.3 Steps for the phosphopeptide enrichment and analysis using the microfluidic chip

..................................................................................................................................................... 114

Table 6.4 Quantitative measurement of heavy and light phosphopeptides using the chip-

MRM/MS approach .................................................................................................................. 114

xiii

LIST OF ABBREVIATIONS

SYMBOL DESCRIPTION

µ-TAS Micro-total analysis system

2DE Two-dimensional gel electrophoresis

BOE Buffered oxide etch

CE Capillary electrophoresis

CID Collision-induced dissociation

DAVID Database for annotation, visualization and integrated discovery

DDA Data-dependent analysis

DI water Deionized water

DMEM Dulbecco's modified eagle medium

DTT Dithiothreitol

ECD Electron capture dissociation

ECGS Endothelial cell growth supplement

EDTA Ethylenediaminetetraacetic acid

EGF Epidermal growth factor

EGFR Epidermal growth factor

ELISA Enzyme-linked immunosorbent assay

EOF Electroosmotic flow

ER Estrogen receptor

ERLIC Electrostatic repulsion-hydrophilic interaction chromatography

ESI Electrospray ionization

ETD Electron transfer dissociation

FACS Fluorescence-activated cell sorting

FBS Fetal bovine serum

FDR False discovery rate

FTICR Fourier-transform ion cyclotron resonance

GRAVY Grand average of hydropathy

HER-1/2/3/4 Human epidermal growth factor receptor 1/2/3/4

HILIC Hydrophilic interaction chromatography

HPLC High-performance liquid chromatography

IHC Immunohistochemistry

IMAC Immobilized metal affinity chromatography

xiv

IMER Immobilized enzymatic reactor

K Lysine

LC Liquid chromatography

LIT Linear ion trap

MALDI Matrix-assisted laser desorption/ionization

MAP2K Mitogen-activated protein kinase kinase

MAP3K Mitogen-activated protein kinase kinase kinase

MAPK Mitogen-activated protein kinase

MOAC Oxide affinity chromatography

MRM Multiple reaction monitoring

MS Mass spectrometry

MS/MS Tandem MS

MW Molecular weight

NL Neutral loss

PBS Phosphate-buffered saline

PCR Polymerase chain reaction

PDMS Polydimethylsiloxane

PI Propidium iodide

PMA Protein array

PMF Peptide mass fingerprinting

PMMA Polymethyl methacrylate

PSM Peptide spectrum match

PTM Posttranslational modifications

Q Quadrupole

QIT Quadrupole ion trap

R Arginine

RNA Ribonucleic acid

SCX Strong cation exchange chromatography

STRING Search tool for the retrieval of interactive genes

TBP Tributylphosphine

TCEP-HCl Tris (2-carboxyethyl) phosphine hydrochloride

TFA Trifluoroacetic acid

TMA Tissue microarray

TOF Time-of-flight

1

CHAPTER 1: Purpose of Study

The purpose of this study was to develop and optimize microfluidic and mass

spectrometry (MS) technologies that enable the observation, sampling and analysis of fast

biological processes such as phospho-signaling events in cells. Reversible protein

phosphorylation is a critical regulatory mechanism that controls diverse cellular processes,

and deregulated phospho-signaling pathways are frequently observed in many cancers.

However, as phosphorylation and dephosphorylation are often transient processes,

accurate time-dependent profiling of protein phosphorylation and signaling pathways is

challenging, if not impossible to conduct. It is envisioned that the development of

combined microfluidics-mass spectrometry technologies that advance the ability to

identify and quantify events that occur on short time-scales will help elucidate the

properties of dysfunctional signal transduction pathways, to lead to a better understanding

of the mechanisms that drive aberrant cancer cell behavior.

To address the challenges that are associated with the study of fast cellular processes, the

objectives of this study were: (1) To identify proteins involved in early MAPK (mitogen

activated protein kinase) signaling pathways, possibly aberrantly phosphorylated, that can

be mapped by mass spectrometry and used as a test-model for the development of

microfluidic chips that enable the study of cell behavior; to address this objective, we

performed global proteomic and phospho-proteomic profiling of SKBR3/HER-2+ breast

cancer cells in multiple stages of the cell cycle, and identified MAPK proteins that

carried phosphorylated sites of relevance to proliferative signaling in cancer cellss; (2) To

2

design and develop microfluidic devices that enable the characterization of biological

processes that unfold within a short time-frame, with focus on capturing phospho-

signaling events; to achive this aim, we designed and evaluated microfluidic devices that

enable cell capture, stimulation, lysis, rapid tryptic digestion, phosphopeptide enrichment,

and electrospray ionization, and integrated various combinations of these processes

within one microfluidic platform; and (3) To demonstrate, as proof of principle, the

functionality of the microfluidic platforms that have been developed; to this end, we

performed a variety of optimization experiments with both benchtop instrumentation and

microfluidic devices; we tested various chip configurations to assess optimal

performance; in addition, we developed targeted multiple reaction monitoring (MRM)-

MS methods for the identification of phosphorylated peptides representative of MAPK

proteins at concentration levels commensurate with what would be expected during

cellular signaling processes; last, we determined the phosphorylation level of proteins in

cell extracts that enable MS detection from microfluidic chips and profiling of

phosphorylation processes. The results of this work are captured in four chapters that

describe the critical steps necessary for the implementation of this technology in the

biological laboratory.

3

CHAPTER 2: Introduction

2.1 Cancer

2.1.1 Cancer overview

Cancer cells display unrestrained proliferation, a process that may be induced by the

external environment or inherited genetic factors. In addition, cancer cells have the ability

to cause primary tumors, which, in turn, can shed cells to spread the disease through the

circulatory system or by direct organ invasion. According to the National Cancer Institute

(NCI), cancer is comprised of a group of more than 100 diseases (NCI 2018). It is the

second largest non-communicable disease and can affect any part of the body. According

to a study of the American Cancer Society (ACS), half of all men and one-third of all

women in the United States will be affected by some form of cancer during their lifetime.

On a global scale, lung, breast, and colorectal cancers have the highest incidence, lung,

stomach, and liver cancers having relatively higher death rates than others (NCI 2018).

Although the incidence of cancer continues to rise worldwide, the death rates are

declining due to the development of early diagnosis and advanced treatment methods.

Based on the origin of the tumor cells, cancers can be classified into at least the following

five major groups: carcinoma, sarcomas, hematopoietic cancers, germ cell cancers, and

blastoma. Among them, the most common type of cancer are carcinomas, which start in

epithelial tissues and account for more than 80% of all human cancers1.

4

2.1.2 Breast cancer

Breast cancer is a malignant or cancerous tumor which begins from the cells of the breast

tissue, and occurs primarily in women, although men may also be affected. According to

ACS, breast cancer is the most frequently diagnosed cancer among women in the United

States, and is also the second leading cause of cancer death, after lung cancer. On average,

U.S. women have a one-in-eight chance of developing invasive breast cancer over the

course of their lifetime (up to age 85), and about 1 in 33 women will die from breast

cancer (ACS 2018). Some women may develop breast lumps which are benign or

noncancerous abnormal growths that do not spread outside of the breast and are not life-

threatening, but these growths can place a person at increased risk for developing cancer.

It has become widely accepted that breast cancer should be viewed as a group of different

diseases, which are differing in histological, biological, and immunological

characteristics. Survival rates for patients now largely depend on the intrinsic growth rate

of the tumor, age at the time of diagnosis, and numerous biological parameters defining

the natural history of the disease. Detecting a malignancy before clinical appearance is

the goal of early diagnosis and treatment of cancer. Patients who are diagnosed with stage

0 or stage 1 breast cancers have a five-year relative survival rate close to 100%2.

However, less than 20% of patients who are diagnosed with distally metastasized breast

cancer can survive five years3. The diagnosis and prognosis of breast cancer is a

complicated, yet vital, component of reducing the death rate, and, therefore, further

research is needed to improve the survival rates.

5

Other than conventional screening methods, the diagnosis of breast cancer can also be

achieved by careful examination of the relative levels of some specific biomarkers.

Biomarkers are biochemical substances that exist in body fluids, tissues or blood, and can

be used as indicators of the presence of specific diseases. Cancer biomarkers are widely

studied at various levels, including in the context of tumor morphology and the

molecular-level genomics, epigenomics, and proteomics profiles. Routinely utilized

markers in the prediction of breast cancer include: growth factor receptors (e.g., the

epidermal growth factor receptors), tumor-suppressor genes (p53), proteolytic enzymes

that may facilitate the invasion of disease and metastasis (cathepsin D), and metastasis-

suppressor genes (nm23). Of these molecular makers, HER-2 is probably the most

commonly used breast cancer marker. HER-2 (also known as Human Epidermal Growth

Factor Receptor 2/EGFR-2 or ERBB-2) is part of the HER family of receptors that

consists of four members with similar structure, i.e., HER-1, HER-2, HER-3, and HER-44.

Among them, HER-2 is vital in controlling cell growth, survival, and differentiation of

normal cells5. The overexpression and deregulation of HER-2 receptors are often related

to carcinogenesis and metastasis. SKBR3 is a commonly used HER-2 positive human

breast cancer cell line, and it is also the model system used in this study.

2.2 Cell signaling and protein phosphorylation

2.2.1 Cell signaling overview

Cell signaling is a process involved in the communication between and within cells in an

organism. Communication between cells can occur through both electrical and chemical

signals. Although electrical signals deliver the messages much faster, the dominant form

6

of information transduction in cells is chemical signaling. Most human cells are enclosed

within a phospholipid bilayer, which represents a physical barrier to most signals outside

the cell. For some hydrophobic molecules, such as steroid hormones, the plasma

membrane can be easily crossed. However, for hydrophilic molecules, more elaborate

mechanisms are required for passing the barrier. One of the most critical steps in cell

signaling is receiving the external stimulus and then transferring this information into the

cell to activate downstream sensors and effectors that perform specific functions. When

chemical signals arrive at the outer surface of the cell, they can be received and

recognized by specific cell-surface receptors. As a result, a variety of transducers and

amplifiers will be activated to act as intracellular messengers, and ensure the transfer of

the information into the cell. These messengers stimulate sensors and effectors that are

responsible for activating various cellular responses. This mechanism is called “ON

mechanism,” as it is responsible for transmitting information into the cell. There is also

an opposite “OFF mechanism” that is responsible for the deactivation of cellular

responses. Cell signaling is involved in the regulation of diverse cellular activities,

including cell growth, proliferation, differentiation, metabolism, and apoptosis6.

2.2.2 Protein phosphorylation and the ERBB pathway

Protein phosphorylation, principally on serine, threonine or tyrosine residues, is one of

the most common and well-studied posttranslational modifications (PTM). It is believed

that around 70% of all human proteins are phosphorylated, with phosphorylation playing

critical roles in the regulation of virtually all cellular activities in eukaryotic cells,

including cell cycle progression, apoptosis, and signal transduction7. Protein

7

phosphorylation and dephosphorylation in cells are highly regulated events through the

activities of kinases and phosphatases, and abnormal phosphorylation/dephosphorylation

is often associated with many human diseases, including cancers. Phosphorylation

signaling events occur in cells on a large scale, many of these events unfolding on very

short time-scales, ranging from seconds to minutes. This leads to a relatively poor ability

to monitor these events, and inadequate understanding of the temporal and spatial profiles

of protein phosphorylation.

In this study, a human breast cancer cell line, SKBR3, which overexpresses ERBB-2

protein (also known as HER-2), was used as the model system. ERBB-2 is a member of

the ERBB family, which includes three other receptor tyrosine kinases (RTKs):

epidermal growth factor receptor (EGFR), ERBB-3, and ERBB-4. All the four members

are typical RTKs with an extracellular ligand binding region, a transmembrane region,

and an intracellular kinase containing domain8. Once bound to ligands, receptors form

homodimer or heterodimer and activate the intracellular kinase domain by

transphosphorylation of the tyrosine on the tail of each receptor. The activated receptor

can initiate a number of downstream signaling events by recruiting and phosphorylating

specific signaling molecules. It needs to be pointed out that ERBB-3 has defeated kinase

activity and ERBB-2 does not bind any of the ERBB ligands9,10. However, both of them

can be activated via heterodimerization with other ligand-bound receptors. Some large-

scale phosphoproteomic studies have revealed that ErbB receptors potentially bind to

more than 100 proteins11,12. Consistent with their critical roles in various cellular

activities, including cell proliferation, differentiation, and survival, the ERBB signaling

8

pathways have been implicated in the pathogenesis of a variety of human disorders

including cancer, some chronic and genetic disease13. For example, the overexpression of

ERBB-2 was discovered three decades ago and has been consistently observed in breast

cancer and approximately 20% of sporadic tumors. Given that the ERBB signaling

pathways function differently in a pathological situation, understanding the disease-

specific regulatory mechanisms of ERBB signaling pathways might provide an

opportunity for the discovery of new drugs for human diseases.

2.3 Mass spectrometry and proteomics

2.3.1 Mass spectrometry overview

Mass spectrometry is a technique that relies on the ionization of molecules to measure

their mass-to-charge ratio. Over the past decades, MS has evolved into the most powerful

platform for investigating the proteomic profiles of biological samples14. The applications

of this technology include but are not limited to biological analysis, diagnostics, and drug

discovery. In general, mass spectrometers have three primary components, i.e., ionization

source, the mass analyzer, and the detector. MS has the ability to handle the complexity

associated with whole-cell proteomes, when other techniques such as two-dimensional

gel electrophoresis (2DE) and protein microarrays fail to provide the same sensitivity and

accuracy. Three primary applications of proteomics that are pursued by MS include

identifying protein expression, understanding protein interactions, and locating the sites

of protein modifications15.

9

2.3.2 Ion source and mass analyzer

The ion source is an essential component of the mass spectrometer which is used to

ionize the sample molecules. Several ionization methods have been developed as

different analytes can have distinct characteristics. In the early stages of mass

spectrometry, the availability of ionization methods was limited mainly to electron

impact. Two technical breakthroughs introduced in the late 1980s, electrospray ionization

(ESI)16 and matrix-assisted laser desorption/ionization (MALDI)17, solved the critical

problem of generating ions from large, nonvolatile samples such as peptides or even

proteins. In both methods, the analyte molecules are ionized with minimal fragmentation.

ESI is considered to be the most “soft” ionization method. The energy placed in the

sample molecules during ionization is typically less than that by the MALDI method.

Under optimized conditions, even non-covalent bonds, such as those found in protein-

protein interactions, can be conserved in the transition from liquid phase to the gas phase

during ionization. Another reason for the rapid development of ESI relates to the ease

with which it can be interfaced with liquid chromatography (LC)18 and capillary

electrophoresis (CE) separation methods19 for enhancing the effectiveness of mass

analysis. LC-ESI coupling is currently one of the most popular configurations for high-

throughput, automated protein identification and peptide sequencing. Furthermore, due to

the propensity of ESI to produce multiply charged ions, simple quadrupole (Q) and ion

trap instruments with limited m/z range can be used to detect analytes with masses over

the normal m/z range of the analyzer.

10

In MALDI, the target analytes are co-crystallized with a volatile matrix. Pulses of laser

light are then utilized to vaporize the matrix and generate analyte ions for mass analysis.

MALDI coupling to time-of-flight (TOF)-MS has been extensively used for protein

identification due to the robustness and simplicity of the technique20. The “peptide mass

fingerprinting” method is a good example of MALDI-TOF applications. MALDI mass

spectra are easy to interpret, due to the propensity of the technique to generate

predominantly singly charged ions. The method is relatively resistant to interference with

matrices commonly used in protein chemistry.

Once leaving the ion source, the ionized molecules are directed to the mass analyzer

where they are separated based on their mass-to-charge ratios. The most widely used

mass analyzers in proteomics include the quadrupole (Q), quadrupole ion trap (QIT),

linear ion trap (LIT), TOF and Fourier-transform ion cyclotron resonance (FTICR)21.

These analyzers can be grouped into two categories: scanning analyzers, in which only

specific m/z are transmitted at any time, and non-scanning, in which all ions are

transmitted and detected at the same time. A number of parameters, such as sensitivity,

transmission efficiency, resolution, mass accuracy, dynamic range and scan rate, should

be considered to find the suitable mass analyzer for a specific application.

2.3.3 Mass spectrometric analysis

Mass spectrometry has been widely used as a primary method for protein identification

from complex biological samples. Two methods play the most significant roles: peptide

mass fingerprinting (PMF) and tandem MS (or MS/MS)22. In the PMF method, a sample

11

is subjected to digestion with a specific proteolytic enzyme, usually trypsin, then a mass

spectrum is generated which collects the information regarding the masses of all peptides.

These masses are utilized as fingerprints of corresponding peptides that match a protein

in a database. Some algorithms are used to calculate the number of matches and matching

scores such that the hits for each protein are ranked. In contrast, a different principle is

followed in protein identification by tandem MS sequencing. Typically, two mass

analyzers are constructed in sequence with a collision cell in-between, where the ions

collide with an inert gas, fragment and produce daughter ions. For proteomic analyses,

cell extracts are typically digested with trypsin, the samples are processed to remove the

contaminants, and the peptides are separated by HPLC prior to MS detection. In the MS

analyzer, the peptides are fragmented by a method known as collision-induced

dissociation (CID)23. Thousands of MS/MS spectra are acquired for each sample. Each

MS/MS spectrum pertains to a specific parent ion which fragments in a rather predictable

fashion. The masses of parents and daughter ions are compared with known masses of

peptides matched to a protein in a database. By matching multiple discontinuous peptides

from one protein, one can identify the protein of interest.

As a popular method for protein identification, PMF has some unique and significant

advantages over others24. The first benefit comes with its easiness on both operation and

sample preparation. Compared to the tandem MS experiments requiring an MS expert,

PMF can be performed by a moderately skilled operator. Most importantly, PMF is

inherently amenable to analysis by using high-throughput robotics, which theoretically

allows up to 500 samples to be analyzed in a day25. The process is also less time-

12

consuming than other methods. For example, PMF data can be acquired within one

minute, whereas LC-MS/MS data acquisition takes more than one hour. The database

search takes only several minutes to provide straightforward results. In contrast, within

each LC-MS/MS experiment, thousands of spectra are produced, and the subsequent

database search may take up to an hour.

Tandem MS is becoming, however, a superior platform for protein identifications, and is

gradually replacing PMF26. This revolution was mainly caused by two characteristics of

MS/MS: sensitivity and reliability. Two factors are contributing to the high sensitivity.

First, chromatography is introduced prior to MS/MS for sample separation and

concentration. Unlike gel-based separations that are commonly used prior to PMF, high

performance liquid chromatography (HPLC) methods, which are preferred prior to

MS/MS, present high separation efficiencies and amenability for easy automation. HPLC

also ensures peptide pre-concentration that ultimately leads to a corresponding increase in

the MS signal. Moreover, MS/MS can provide accurate sequence-specific data.

Therefore, a lower number of peptides is needed for the confident identification of the

protein. For example, due to the rapid growth of database sizes, 5-10 peptides are

required in PMF for reliable protein identification. However, even only one peptide can

be sufficient for MS/MS identification. Since the observed signals are proportional to the

concentration of the sample, lower requirements on the observed number of peptides

provide a significant gain in sensitivity. The high reliability of MS/MS is also achieved

by two factors. First, PMF can only provide a list of ranked candidates with

corresponding scores. Various factors contribute to the calculation of the score, which

13

may lead to false positive hits. However, with tandem mass spectra, one can easily

achieve >90% confidence levels with a single peptide if the peptide provides an

informative fragmentation pattern. In practice, multiple peptides are used for the

identification of each protein so that high confidence levels can be achieved. Second,

PMF may provide erroneous results when the gel spot or sample contains more than one

protein, since the reliability of identification decreases dramatically. Unfortunately, the

majority of samples contain more than one protein. In contrast, efficient on-line

separations prior to MS/MS enable the analysis of very complex samples.

2.3.4 Proteomics overview

The term “proteomics” was invented by merging “protein” and “genomics” in the

1990s27. Proteomics is a relatively new field of study that focuses on the identification

and quantitation of all the proteins of a proteome, including expression, post-translational

modifications (PTMs), interactions, and cellular localization13. According to Human

Proteome Map statistics (http://www.humanproteomemap.org), as of May 13, 2018, more

than 30,000 proteins derived from over 17,000 human genes have been identified. The

rapid progress of proteomics has been driven by the development of technologies and

tools for sensitive MS analysis, efficient peptide/protein separation, sophisticated

bioinformatic software for data analysis. In the past two decades, mass spectrometry has

emerged as the most powerful analytical tool for large-scale protein analysis due to the

advance in the resolution, mass accuracy, sensitivity, and scan rate. There are three major

proteomic strategies: a) the bottom-up strategy analyzes fully digested peptides, usually

14

with trypsin; b)the top-down approach analyzes intact proteins directly; c) the middle-

down method analyzes relatively larger peptides resulting from restricted proteolysis29.

2.3.5 Mass spectrometry-based phosphoproteomics

More than 2,000 human genes guide the synthesis of protein kinases, which constitute the

largest single enzyme family in humans30. A vast majority of biological processes are

controlled by the reversible phosphorylation of proteins, specifically on serine, threonine

and tyrosine residues31. Around 70% of all cellular proteins are believed to contain

covalently bound phosphate groups. Almost all protein phosphorylations are transient and

reversible, and the protein phosphorylation state converts significantly over time, not

only correlated to cell cycle but also affected by the life stages of the whole organism.

Despite the significance and widespread occurrence of this posttranslational modification

(PTM), our ability to explain complex phosphorylation events encounters many

challenges. Therefore, strategies for quantitative phosphoprotein analysis are needed to

adequately assess the function of specific protein phosphorylation sites, and accurate MS-

based phosphoproteomic methods are becoming increasingly crucial for the systematic

analysis of this PTM. Global MS-based phosphoproteomic profiling provides a strategy

for the systematic study of protein phosphorylation in cell signaling networks and has

been widely used for the identification and quantification of dynamic temporal and

spatial changes of this PTM. Current phosphoproteomics research offers a robust

platform that empowers researchers with vital information regarding the regulatory

phosphorylation-mediated events in a cell.

15

A general phosphoproteomics workflow includes several steps: protein extraction and

digestion, enrichment in phosphopeptides, cleanup, HPLC-MS analysis, and data

interpretation32. Once the proteins are extracted from cells, they are treated with

proteolytic enzymes to generate peptides that are amenable to MS detection. In

comparison to regular proteomic workflows, phosphoproteomic analyses requires

specialized techniques for enrichment in phosphopeptides, detection, sequencing, and

quantification33. The reasons for such specific needs include: (a) the abundance of

signaling phosphoproteins is low within cells; (b) the time duration of phospho-signaling

events can be often short, as only a small proportion of phosphorylated kinases are active

at any specific time point; (c) the time adaptation over signaling pathways is an essential

factor for kinase phosphorylation; (d) phosphopeptides generate low signal intensities in

MS.

As a result, various phosphopeptide enrichment methods have been established and

successfully applied to reveal the presence of phosphorylation in complex biological

samples. These methods include immobilized metal affinity chromatography (IMAC),

TiO2 or other metal oxides, ion exchange chromatography, phosphorylation-specific

antibodies, sequential elution from IMAC (SIMAC), hydrophilic interaction

chromatography (HILIC), and electrostatic repulsion-hydrophilic interaction

chromatography (ERLIC) 34,35.

To fully understand the molecular mechanisms involved in phospho-signaling, it is

critical to localize the presence of the phosphate group on a specific amino acid. In

16

collision-induced dissociation (CID), peptide ions are accelerated in the mass analyzer

and then collided with neutral gas molecules for producing a series of y-type and b-type

peptide fragments. During CID of phosphopeptides, fragmentation occurs mainly at the

O-phosphoric acid bond, as this is much weaker than the peptide bond. Therefore, the

most common result of MS/MS fragmentation is the neutral loss (NL) of a phosphate

group (i.e., the loss of H3PO4, ~98 Da) plus a water molecule. Further fragmentation of

the peptide is forfeited, and the tandem mass spectrum cannot be interpreted. The

phosphate group on serine and threonine residues is more labile than the one on tyrosine

residues. Therefore, a neutral loss of 98 is mainly observable for phosphoserine and

phosphothreonine-containing peptides; phosphotyrosine-containing peptides are resistant

to this loss, but generate mostly intense signals at [M+H]+-80 Da (loss of HPO3) in

MALDI-MS. Besides the CID described above, alternative methods that preserve the

presence of the phosphate group on a peptide, such as electron transfer dissociation

(ETD)36 and electron capture dissociation (ECD)37, have been described. For assigning

unambiguous phosphorylation sites, a data-dependent neutral loss-triggered MS3

(MS/MS/MS) analysis has also been established. The general principle involves the

followings: a peptide is chosen for MS/MS fragmentation, and when an NL of 98 Da

(phosphoric acid) is observed in the tandem mass spectrum, the ion generated by the NL

will automatically be selected and isolated for further fragmentation by MS/MS/MS.

Using this method, the site of phosphorylation can be located more easily and reliably.

17

2.4 Microfluidics

2.4.1 Microfluidics overview

Microfluidics is an emerging technology focused on the study of miniaturized systems

that use fluids as a working medium and integrate different functionalities on a small

scale, typically with dimensions in the range of tens to hundreds of micrometers.

Microfluidics is a relatively new field of science and technology in which considerable

progress has been made in the past two decades. The most important application areas of

microfluidics include sample amplification, 1D/2D separations, immunoassays,

biomedical diagnostics, and biosensing38. The basic idea of microfluidic biochips is to

integrate operations such as sample pre-treatment, separation, amplification, and

detection on a miniaturized platform39. The reason why we consider it as a new branch is

not only the fact that it can carry out complex microfluidic protocols, but also because of

the different physical events that occur in the micro-domain environment. In conventional

branches, such as chemical process technology, fluids are contained, transported, and

processed in macro-scale devices. The fundamental laws or equations describing the

physics and chemistry of such processes are the same as in microfluidics; however, some

effects that are significant on a macroscopic scale became unimportant on small scales,

while other phenomena that are less important macroscopically turn out to be prevalent in

microfluidics.

Compared to traditional methods, microfluidics has the potential to revolutionize the way

modern biological research is performed. Firstly, the analysis times and reagent costs can

be reduced dramatically due to miniaturization. Besides, the overall small surface areas

18

(despite increased surface-to-volume ratios) reduce adsorption-related sample losses, and

thus offer the ability to work with scarce samples. Furthermore, microfluidics holds the

promise to integrate an entire laboratory onto a single chip, which could lead to high-

speed and high-throughput analysis. Currently, various techniques are used for the

fabrication of microfluidic devices including micromachining, photolithography, soft

lithography, embossing, in situ construction, injection molding, and laser ablation. Each

of these techniques has its advantages and disadvantages, and the choice of the

fabrication method often depends on the specific application of the device. Microfluidic

devices are mainly made of glass, quartz, silicon, polydimethylsiloxane (PDMS), and

poly(methyl methacrylate) (PMMA). A number of factors, such as the use of the chip, the

physical and chemical surface characteristics, and the cost of fabrication, are considered

when choosing a particular material for chip fabrication40.

2.5 References

1. S. Rogalla, C. H. Contag. Early cancer detection at the epithelial surface. Cancer J.

(United States) 2015, 21, 179.

2. Breast Cancer Survival Rates & Statistics, https://www.cancer.org/cancer/breast-

cancer/understanding-a-breast-cancer-diagnosis/breast-cancer-survival-rates.html,

accessed in June 2018.

3. S. Ely, A. N. Vioral. Breast cancer overview. Plast. Surg. Nurs. 2007, 27, 128.

4. D. C. Koboldt, R. S. Fulton, M. D. McLellan, H. Schmidt, J. Kalicki-Veizer, J. F.

McMichael, L. L. Fulton, D. J. Dooling, L. Ding, E. R. Mardis, R. K. Wilson, A. Ally, M.

19

Balasundaram, Y. S. N. Butterfield, et al. Comprehensive molecular portraits of human

breast tumours. Nature 2012, 490, 61.

5. C. Yeung, J. Hilton, M. Clemons, S. Mazzarello, B. Hutton, F. Haggar, C. L. Addison,

I. Kuchuk, X. Zhu, K. Gelmon, A. Arnaout. Estrogen, progesterone, and HER2/neu

receptor discordance between primary and metastatic breast tumours—a review. Cancer

Metastasis Rev. 2016, 35, 427.

6. M. A. Lemmon, J. Schlessinger. Cell signaling by receptor tyrosine kinases. Cell 2010,

141, 1117.

7. S. Wang. Protein Phosphorylation in Human Health. 2012.

8. Z. Wang, ErbB Receptors and Cancer, in ErbB Receptor Signaling: Methods and

Protocols. 2017, pp. 3–35.

9. C. Kiel, E. Verschueren, J. S. Yang, L. Serrano. Integration of protein abundance and

structure data reveals competition in the ErbB signaling network. Sci. Signal. 2013, 6,

DOI 10.1126/scisignal.2004560.

10. M. a Olayioye, R. M. Neve, H. a Lane, N. E. Hynes. The ErbB signaling network:

receptor heterodimerization in development and cancer. EMBO J. 2000, 19, 3159.

11. W. X. Schulze, L. Deng, M. Mann. Phosphotyrosine interactome of the ErbB-receptor

kinase family. Mol. Syst. Biol. 2005, 1, E1.

12. J. V. Olsen, B. Blagoev, F. Gnad, B. Macek, C. Kumar, P. Mortensen, M. Mann.

Global, In Vivo, and Site-Specific Phosphorylation Dynamics in Signaling Networks.

Cell 2006, 127, 635.

13. N. E. Hynes, G. MacDonald. ErbB receptors and signaling pathways in cancer. Curr.

Opin. Cell Biol. 2009, 21, 177.

20

14. R. Aebersold, M. Mann. Mass spectrometry-based proteomics. Nature 2003, 422,

198.

15. A. Bensimon, A. Heck, R. Aebersold. Mass Spectrometry–Based Proteomics and

Network Biology. Mass Spectrom. Proteomics Netw. Biol. 2012, 81, 379.

16. J. B. Fenn, M. Mann, C. K. Meng, S. F. Wong, C. M. Whitehouse. Electrospray

ionization for mass spectrometry of large biomolecules. Science (80-. ). 1989, 246, 64.

17. F. Hillenkamp, M. Karas, R. C. Beavis, B. T. Chait. Matrix-Assisted Laser

Desorption/Ionization Mass Spectrometry of Biopolymers. Anal. Chem. 1991, 63, DOI

10.1021/ac00024a716.

18. M. S. Lee, E. H. Kerns. LC/MS applications in drug development. Mass Spectrom.

Rev. 1999, 18, 187.

19. R. P. Oda, J. P. Landers, Introduction to Capillary Electrophoresis, in Handbook of

Capillary Electrophoresis. 1996, p. 894.

20. P. of MALDI. Principles of MALDI-TOF Mass Spectrometry. can be found under

http://www.shimadzu.com/an/lifescience/maldi/princpl1.html, 2012.

21. E. de Hoffmann, V. Stroobant. Mass Analyzers. Mass Spectrom. Princ. Appl. 2007,

85.

22. J. D. Wulfkuhle, K. C. McLean, C. P. Paweletz, D. C. Sgroi, B. J. Trock, P. S. Steeg,

E. F. 3rd Petricoin. New approaches to proteomic analysis of breast cancer. Proteomics

2001, 1, 1205.

23. R. N. Hayes, M. L. Gross. Collision-induced dissociation. Methods Enzymol. 1990,

193, 237.

24. X. J. Qin, B. X. Ling. Proteomic studies in breast cancer (Review). Oncol. Lett. 2012,

21

3, 735.

25. B. Thiede, W. Höhenwarter, A. Krah, J. Mattow, M. Schmid, F. Schmidt, P. R.

Jungblut. Peptide mass fingerprinting. Methods 2005, 35, 237.

26. G. Hudelist, C. F. Singer, K. I. D. Pischinger, K. Kaserer, M. Manavi, E. Kubista, K. F.

Czerwenka. Proteomic analysis in human breast cancer: Identification of a characteristic

protein expression profile of malignant breast epithelium. Proteomics 2006, 6, 1989.

27. M. R. Wilkins, C. Pasquali, R. D. Appel, K. Ou, O. Golaz, J.-C. Sanchez, J. X. Yan, A.

A. Gooley, G. Hughes, I. Humphery-Smith, K. L. Williams, D. F. Hochstrasser. From

Proteins to Proteomes: Large Scale Protein Identification by Two-Dimensional

Electrophoresis and Arnino Acid Analysis. Nat. Biotechnol. 1996, 14, 61.

28. Y. Zhang, B. R. Fonslow, B. Shan, M. C. Baek, J. R. Yates. Protein analysis by

shotgun/bottom-up proteomics. Chem. Rev. 2013, 113, 2343.

29. C. Wu, J. C. Tran, L. Zamdborg, K. R. Durbin, M. Li, D. R. Ahlf, B. P. Early, P. M.

Thomas, J. V Sweedler, N. L. Kelleher. A protease for’middle-down’proteomics. Nat.

Methods 2012, 9, 6.

30. S. B. Ficarro, M. L. McCleland, P. T. Stukenberg, D. J. Burke, M. M. Ross, J.

Shabanowitz, D. F. Hunt, F. M. White. Phosphoproteome analysis by mass spectrometry

and its application to Saccharomyces cerevisiae. Nat. Biotechnol. 2002, 20, 301.

31. D. Miranda-Saavedra, G. J. Barton. Classification and functional annotation of

eukaryotic protein kinases. Proteins 2007, 68, 893.

32. N. Ozlu, B. Akten, W. Timm, N. Haseley, H. Steen, J. A. J. Steen. Phosphoproteomics.

Wiley Interdiscip. Rev. Syst. Biol. Med. 2010, 2, 255.

33. E. López, X. Wang, L. Madero, J. López-Pascual, M. Latterich. Functional

22

phosphoproteomic mass spectrometry-based approaches. Clin. Transl. Med. 2012, 1,

34. L. Beltran, P. R. Cutillas. Advances in phosphopeptide enrichment techniques for

phosphoproteomics. Amino Acids 2012, 43, 1009.

35. X. S. Li, B. F. Yuan, Y. Q. Feng. Recent advances in phosphopeptide enrichment:

Strategies and techniques. TrAC - Trends Anal. Chem. 2016, 78, 70.

36. M.-S. Kim, A. Pandey. Electron transfer dissociation mass spectrometry in

proteomics. Proteomics 2012, 12, 530.

37. R. A. Zubarev. Electron-capture dissociation tandem mass spectrometry. Curr. Opin.

Biotechnol. 2004, 15, 12.

38. Lii, J., W.-J. Hsu, S. P. Lee, S. K. Sia. Microfluidics. Kirk-Othmer Encycl. Chem.

Technol. 2006, DOI 10.1002/0471238961.micrsia.a01.

39. I. M. Lazar, J. Grym, F. Foret. Microfabricated devices: A new sample introduction

approach to mass spectrometry. Mass Spectrom. Rev. 2006, 25, 573.

40. T. A. Duncombe, A. M. Tentori, A. E. Herr. Microfluidics: Reframing biological

enquiry. Nat. Rev. Mol. Cell Biol. 2015, 16, 554.

23

CHAPTER 3: Materials and Methods

3.1 Chemicals and materials

All bovine protein standards (α-casein, β-casein, carbonic anhydrase, hemoglobin α/β, α-

lactalbumin, fetuin, and cytochrome c), protease inhibitor cocktail P8340, phosphatase

inhibitor cocktail 2 and 3, Cell Lytic™ NuCLEAR™, urea, dithiothreitol (DTT), Tris,

NaCl, acetone, acetic acid, trifluoroacetic acid (TFA), ammonium bicarbonate

(NH4HCO3), hydrogen peroxide, and ammonium hydroxide were obtained from Sigma-

Aldrich (St. Louis, MO). SKBR3 breast cancer cells, 0.25% trypsin/0.53mM EDTA, and

phosphate-buffered saline (PBS) were purchased from ATCC (Manassas, VA). McCoy’s

5A medium was obtained from Life Technologies (Carlsbad, CA), human epidermal

growth factor (hEGF) from PeproTech (Rocky Hill, NJ), fetal bovine serum (FBS) from

Gemini-Bio Products (West Sacramento, CA), and sequencing grade modified trypsin

from Promega Corporation (Madison, WI). SPEC-PT-SCX and SPEC-PT-C18 solid

extraction pipette tips, and Zorbax SB-C18/5 μm particles were purchased from Agilent

Technologies (Santa Clara, CA). GL-Tip SDB, GL-Tip GC, Titansphere Phos-TiO/10 μm

particles, and lactic acid were purchased from GL Sciences (Torrance, CA). HPLC-grade

acetonitrile and methanol were purchased from Fisher Scientific (Fair Lawn, NJ), and DI

water was from a MilliQ Ultrapure water system (Millipore, Bedford, MA). Buffered

oxide etch (BOE) and chrome etchant were purchased from Transene Co. (Danvers,

MA). MF-319 developer was from Rohm and Haas (Philadelphia, PA). Sulfuric acid was

obtained from Mallinckrodt (St. Louis, MO). The heavy forms of the selected

24

phosphopeptides were synthesized by New England Peptide (Gardner, MA). The stable

isotope-label (13C, 15N) was incorporated at leucine, resulting in a mass shift of +7 Da.

3.2 SKBR3 cell culture

SKBR3 cells were cultured using McCoy’s 5A medium supplemented with 10% FBS and

incubated at 37 °C with 5% CO2 atmosphere. Typically, the cells were cultured without

other any treatment until full confluence, and harvested using 0.25% trypsin/0.53 mM

EDTA, then stored at -80 °C until use. For the study of cell cycle-specific proteins, cell

cycle arrest and release was performed. After reaching ~60-70% confluence, the cells

were arrested in the G1-phase by serum-deprivation for 48 h to synchronize the cells. The

cells were released into the S-phase by 36 h incubation in medium with 100 nM EGF and

10% FBS. Cell harvesting was accomplished by rinsing cells the with PBS (pH 7.4), then

incubating with 0.25% trypsin/0.53 mM EDTA for 15 min. The cells were harvested in

two different phases, G1 and S, and stored at -80 °C until use.

3.3 Fluorescence-activated cell sorting

A small fraction of cells (~ 106 - 107) was collected separately during cell harvesting to

perform cell cycle analysis using fluorescence-activated cell sorting (FACS). Cells were

resuspended in 5 mL ice-cold PBS followed by centrifugation at 1000 rpm for 5 min. The

supernatant was removed, and the cells were suspended again in 500 µL ice-cold PBS.

The cells were fixed by adding 4.5 mL cold 70% ethanol, and stored at -20 °C until use.

Before FACS analysis, the cell suspension was centrifuged at 1000 rpm for 5 min. The

supernatant was removed, and the cells were resuspended in 5 mL PBS. The suspension

25

was centrifuged again at 1000 rpm for 5 min, and the supernatant was removed. The cell

pellet was resuspended and stained in 1mL of fresh propidium iodide (PI)/Triton X-100

staining solution (20 µg/mL propidium iodide, 200 µg/mL DNase-free RNase A, 0.1%

(v/v) Triton X-100 dissolved in PBS). Cells were kept in a dark cold room for at least 30

minutes before analysis. FACS was performed using a Coulter EPICS XL-MCL benchtop

analyzer with a 488 nm excitation source.

3.4 Cell processing

The frozen cells were thawed at room temperature and lysed using two methods for

different applications. The first method was for the separation of nuclear and cytoplasmic

fractions by using the Cell LyticTM NuCLEARTM extraction kit. The cells were incubated

on ice for 15 min in 5X volume of a hypotonic buffer composed of 10 mM HEPES, 1.5

mM MgCl2, 10 mM KCl, 0.01 M DTT, protease inhibitor cocktail and phosphatase

inhibitors. After incubation, 10% IGEPAL CA-630 was added to the cell suspension to a

final concentration of 0.6%, and the mixture was vortexed vigorously for 10-15 seconds.

The lysate was centrifuged at 10,500 g for 30 seconds, and the supernatant was collected

as the cytoplasmic fraction. The cell pellet was resuspended in 2/3X volume of a

hypertonic buffer composed of 20 mM HEPES, 1.5 mM MgCl2, 0.42 M NaCl, 0.2 mM

EDTA, 0.01 M DTT, 25% glycerol (v/v), protease inhibitor cocktail and phosphatase

inhibitors. The mixture was vortexed at medium speed for 45 min at 4 °C, and then

centrifuged at 20,500 g for 5 min. The supernatant was collected as the nuclear fraction.

Both fractions were stored at -80 °C until use.

26

The second method was for the direct lysis without separating cytoplasmic and nuclear

fraction. Cells were mixed with a lysis buffer composed of Tris (50 mM), NaCl (75 mM),

urea (8 M), DTT (1 mM), protease inhibitors cocktail (1%), phosphatase inhibitor

cocktail 2 (2%) and 3 (2%), and lysed by sonication for 1 min cycles repeated 5 times in

ice-cold water. After sonication, the mixture was centrifuged at 10,500 g for 5 min, and

the supernatant was collected and stored at -80 °C until use.

3.5 Conventional proteolytic digestion and phosphopeptide enrichment

The protein concentration was measured using a SmartSpec Plus spectrophotometer and

the Bradford assay (Bio-Rad, Hercules, CA). The samples were aliquoted into 500 µg

each. Protein denaturation and disulfide reduction were conducted by incubating the

sample with 8 M urea, 4.5 mM DTT, and 50 mM NH4HCO3 at 58 °C for 1 hour. The

mixture was diluted 10X with 50 mM NH4HCO3 obtaining a final protein concentration

of 0.5 mg/mL. A mixture of 8 standard bovine proteins (5 µM) was spiked into the

sample, and the sample was digested with sequencing grade modified trypsin at a ratio of

50:1 (substrate:enzyme) for 15-20 hours at 37 °C, and quenched with glacial CH3COOH

(1%).

For regular proteomic analysis, the sample was cleaned-up using SPEC-PT-C18 and

SPEC-PT-SCX tips and then dissolved in H2O/CH3CN/TFA 98:2:0.01 to a concentration

of 2 μg/μL. For the analysis of phosphopeptides using the benchtop method, the samples

were subjected to C18 cleanup, and phosphopeptide enrichment using TiO2, styrene

divinylbenzene (SDB) co-polymer and graphitic carbon (GC) tips. The final sample was

27

prepared in H2O/CH3CN/TFA 98:2:0.01 to a concentration of 2 μg/μL. For microfluidics-

based phosphopeptide enrichment and analysis, the sample was subjected to a

conventional clean-up method using C18 tips, and then dissolved in H2O/CH3CN/Lactic

Acid/TFA 18:72:10:0.4 to a concentration of 1 μg/μL.

3.6 MS analysis and data processing

For the global proteomic study of SKBR3, five technical replicates were conducted for

each sample using an Agilent 1100 micro-LC separation system (Agilent) coupled with

an LTQ or an LTQ-XL mass spectrometer purchased from Thermo Electron (San Jose,

CA). A nano-separation column (100 μm i.d. × ~12 cm) was prepared in-house using 5

μm Zorbax SB-C18 particles (Agilent), and a ~1 cm long fused silica capillary was

inserted into the open end as a nanospray emitter. The nano-separation column was

connected to the HPLC system and operated at a flow rate of ~180 nL/min during

analysis. The MS was set in the positive-ion mode with a capillary voltage of ~2 kV. The

separation gradient from 100% mobile phase A (H2O/CH3CN/TFA 96:4:0.01, v/v) to

100% mobile phase B (H2O/CH3CN/TFA 20:80:0.01, v/v) was 3-hour long. The sample

was analyzed via a data-dependent method by performing Zoom/MS2 scans on the top

three most intense peaks in each MS scan (5 microscans averaged), and the data were

acquired over a range of 500-2000 m/z. The isolation width, normalized collision energy,

activation Q, and activation time were set at 3 m/z, 35%, 0.25, and 30 ms for collision-

induced dissociation (CID). To facilitate the identification of phosphopeptides, data-

dependent neutral loss MS3 was also used for some samples. The neutral loss was

activated for the loss of 98, 49, 32.7, and 24.5, which represent singly, doubly, triply, and

28

quadruply charged phosphopeptides, respectively. The raw data files were searched

against a minimally redundant Homo sapiens protein database from UniProt using

Discoverer 1.4 software package (Thermo Electron). A mass range of 500 – 5000 Da and

up to 4 missed cleavages for each fully tryptic fragment were allowed in the search.

Maximum mass tolerance for precursor ion and product ion were set to 2 Da and 1 Da,

respectively. The maximum false discovery rate was set at 3% at the peptide level.

Phosphorylation was the only posttranslational modification allowed, and the dynamic

modification mode was enabled for Ser, Thr, and Tyr. To analyze the data in a biological

context, multiple bioinformatics tools including GoMiner, DAVID (Database for

Annotation, Visualization and Integrated Discovery), and STRING (Search Tool for the

Retrieval of Interactive Genes), were used.

For microfluidics-based analysis, the sample was analyzed using an LTQ mass

spectrometer with all parameters, unless otherwise stated, set the same as described

above. A flow rate of 300 nL/min was achieved using a syringe pump purchased from

Harvard Apparatus (Holliston, MA). For direct infusion ESI-MS analysis, the sample was

analyzed via a data-dependent method by performing Zoom/MS2 scans on the top 10

most intense peaks in each MS scan. Under multiple reaction monitoring (MRM) mode,

each precursor ion was assigned 5-7 fragment ions. The m/z ratio tolerances were set to

+/- 1 and +/- 0.6 for precursor and fragment ions, respectively. MRM/MS analysis was

also performed using HPLC with a concentration gradient of 4-hour to facilitate the

selection of MRM targets based on the quality of spectra. The MRM results were

processed and analyzed using Skyline, an open source software developed by MacCoss

29

Lab. At peptide level, the following parameters were used for the search: minimum

length of 5, maximum length of 25, and maximum missed cleavages of 2. At transition

level, only b and y ions with a charge state of 1 or 2 can be selected, the ion match

tolerance was 1 m/z, and only the top 10 most intense product ions were picked.

3.7 Microfluidic chip fabrication

Microfluidic chip layouts were designed using AutoCAD software (Autodesk, San

Rafael, CA) and the photomasks were prepared by HTA Photomask (San Jose, CA).

Chips were fabricated from 1.5 mm thick white crown glass coated with chrome and

photoresist (Nanofilm, Shelton, CA) using standard photolithography and wet chemical

etching techniques. The UV light source was purchased from OAI (San Jose, CA). Glass

substrates were etched with buffered oxide etch solution, and the channel depths were

measured using a Dektak profilometer (Veeco, Plainview, NY). Holes of ~1 mm i.d. were

drilled in substrates using a rotary tool (Dremel, Racine, WI), to access the channels. The

photoresist and chrome layers were removed with methanol/acetone and chrome etchant,

respectively. The substrate plates were carefully cleaned with sulfuric acid, ammonium

hydroxide, acetone, methanol, and distilled DI water, sequentially, and bonded by gradual

heating from room temperature to 550 °C. Fine fused-silica capillaries (20 μm i.d. × 90

µm o.d. × ~ 2 cm long) were inserted into one end of the microchip and secured in place

with a removable glue, to work as the electrospray ionization (ESI) emitter. The two

reactor chambers (50 μm deep × 500 µm wide × 5 mm long) in each setup were fully

packed with C18 (5 μm) and TiO2 (10 μm) particles. Alternatively, in some designs

featuring only one reactor chamber, C18 and TiO2 particles were loaded in the chamber

30

and the main channel (50 μm deep × 120 µm wide × 10 mm long), respectively.

Isopropanol is the commonly used solvent to prepare the C18 slurry. However, for the

larger and heavier TiO2 particles, a more viscous solvent was needed to prevent particle

deposition prior to finalizing the loading process. Lactic acid is a very viscous solution

and often used as an additive to improve the selectivity of TiO2-based phosphopeptide

enrichment. Therefore, a mix of isopropanol/lactic acid (80:20 v/v) was used to enable

trouble-free loading of TiO2 particles. All loading processes were performed manually

using a 250 μL gas-tight syringe (Hamilton, Reno, MA), and the packed chambers were

sealed with a two-part epoxy glue and cured at 95 °C for 30 min to retain the particles.

The ports on chips were prepared from unions (Valco Instruments, Houston, TX) evenly

cut into two pieces.

31

CHAPTER 4: Fast Enzymatic Digestion: Counts, Coverage, and Reproducibility

Jingren Deng, Morgan H. Julian and Iulia M. Lazar*

Department of Biological Sciences, Virginia Tech

1981 Kraft Drive, Blacksburg, VA 24061, USA

*Corresponding author: malazar@vt.edu

Author contributions

Designed the experiments: JD IML

Performed the experiments: JD MHJ

Analyzed the data: JD MHJ IML

Wrote the paper: JD IML

Conceived and coordinated the study: IML

Keywords: proteolytic digestion, fast analysis, proteomics, mass spectrometry

32

Abstract

Biological studies are conducted at ever increasing rates by relying on proteomics

workflows that provide information about the presence, function and interaction of

proteins from a global perspective. Although MS data acquisition is highly automated

and rapid, sample preparation continues to be the bottleneck of developing high-

throughput workflows. Enzymatic protein processing, in particular, involves time-

consuming protocols that can extend from one day to another. To accelerate the

conversion of proteins into peptides, either enzymatic reactors of improved design, or

physical stresses that facilitate more effective proteolytic reactions, have been explored.

However, no comprehensive study has been conducted, so far, for assessing the quality of

peptides generated through rather simple, but rapid, in-solution enzymatic digestion

processes. To address this gap, we developed and evaluated the performance of a fast

enzymatic reaction (7 min-60 min), and compared it to that of a routine approach,

conducted for 18 h. The methodology was applied to the digestion of proteins originating

from SKBR3 cancer cell extracts. The reaction products were analyzed by nano-

HPLC/MS/MS, and the quality of the process was assessed in terms of peptide/protein

identifications, sequence coverage, peptide length, missed-cleavage sites, cross-

correlation scores, and peptide hydrophilic/hydrophobic properties. The results

demonstrate that rapid enzymatic processes lead to a large number of peptide fragments

that improve protein sequence and proteome coverage, that the tandem mass spectra

produced from these peptides are of high quality for enabling reliable protein

identifications, and that the features and physical properties of peptides are prone to

supporting the development of alternative multi-dimensional separations and middle-

33

down proteomics analysis strategies. As the reproducibility of generating the same

peptides within a few minutes of enzymatic digestion was remarkably close to that

obtained from 18 h long reactions, the data support the implementation of proteomic

protocols that could substantially streamline the analysis of biological samples.

4.1 Introduction

Enzymatic digestion of cellular proteins is conducted according to protocols that seek

complete and reproducible protein cleavage at particular amino acid sites. This is

typically accomplished by first denaturing the protein samples with chaotropic agents

(e.g., urea, guanidinium chloride]), breaking the disulfide bridges with a reducing agent

[e.g., 1,4-dithiothreitol (DTT), Tris (2-carboxyethyl) phosphine hydrochloride (TCEP-

HCl), tributylphosphine (TBP), or 2-mercaptoethanol], alkylating the products with an

agent that blocks the released cysteine residues (e.g., iodoacetamide), and digesting the

proteins with enzymes that produce peptides of optimal length for MS analysis (e.g.,

trypsin, LysC). Trypsin, a commonly used enzyme in proteomics experiments, will

cleave the proteins at the carboxyl terminus of Lys (K) and Arg (R). Typically, the

enzymatic digestion process is allowed to unfold overnight in basic buffer solutions

(pH~7-8), at a substrate:enzyme ratio of (50-100):1, but 5-6 h long exposure of the

proteins to the enzyme has been shown to produce satisfactory results. After the removal

of salts, additives and cell lysing detergents, the generated peptides are separated by a

well-controlled HPLC gradient prior to subsequent electrospray ionization (ESI)-MS

analysis. For fast screening applications that enable high-throughput analysis, the entire

process must be streamlined and reduced in length. This can be accomplished most

34

effectively by shortening the proteolytic digestion process, or, if possible, by eliminating

certain sample preparation steps.

A variety of approaches have been proposed for accelerating the enzymatic digestion of

proteins, most importantly, by immobilizing the enzyme on diverse substrates, by

designing efficient microreactors that facilitate a close interaction between the substrate

and the enzyme, and by assisting the process with various physical stresses such as heat,

ultrasound, pressure, or electrical fields [1-13]. For example, Yin et al. developed an

approach to synthesize an enzyme-inorganic hybrid nanoflower that was able to perform

rapid proteolytic digestion of bovine and human serum albumins within 2 min with

greater than 40 % sequence coverages [4]. Yuan et al. reported an immobilized trypsin-

based strategy that enabled not only simultaneous rapid digestion and 18O labeling, but

also online integration with nano-HPLC-ESI-MS/MS. A sequence coverage of 60 % and

labeling efficiency of 98.5 % were obtained for bovine serum albumin within a reaction

time of 2.5 min [5]. Ning et al. proposed a rapid digestion method by integrating a

protease-modified membrane at the end of pipette tips, and demonstrated 100 % peptide

coverage, within 30 s, for the monoclonal antibody Herceptin [6]. Although most of such

immobilized trypsin reactors are suitable for rapid digestion, their practical applications

continue to be limited primarily because of time-consuming fabrication protocols and the

possibility of introducing contaminants in the reaction products. Nevertheless, other

methods that utilize physical stresses, have been also developed for accelerating the

digestion process. For example, by using microwave irradiation and a stable, immobilized

enzyme, Kim et al. obtained 75 % sequence coverage for bovine serum albumin within

35

10 min digestion time [11]. Guo et al demonstrated that microwave- and ultrasound-

assisted digestion enable fast enzymatic reactions at the expense of higher missed-

cleavage rates [12], while Ge at al. reported a method that relied on trypsin-immobilized

miniature incandescent bulbs and infrared radiation to enabled the digestion of standard

proteins within 5 min, with a sequence coverages of ~50 %-90 % [13]. Although the

digestion process could be accelerated in all these procedures, the trypsin had to be

modified to shift its optimum operating temperature to 50-60 ℃ , and additional

equipment such as microwave ovens or sonicators were necessary for completing the

enzymatic reactions. Moreover, the applicability of many of the above described methods

was limited to the analysis of simple mixtures of proteins.

In prior work we described a capillary microreactor that enabled fast proteolytic digestion

reactions (1-10 min) by immobilizing not the enzyme, but the proteins, through

adsorption on reversed phase C18 particles, and flowing the enzyme solution over the

adsorbed proteins. The performance of the microreactor was characterized with standard

protein digests [14,15]. At the expense of a somewhat incomplete enzymatic reaction,

this simple procedure generated a larger number of tryptic peptides and better sequence

coverage than conventional overnight digestion protocols. Based on the outcome of our

recent work, and of others, and given the substantial reduction in analysis time enabled

by shortening the enzymatic processing step and the associated benefits of increased

protein sequence coverage, we reasoned that the protocol could perform well for the

characterization of whole, complex proteomes, not just of simple mixtures of standards.

As a result, the present study was aimed at conducting an in-depth evaluation of the

36

quality of peptides generated through fast enzymatic reactions, and the impact on MS

proteomic profiling of complex cellular extracts. Comparisons of proteolysis reactions

performed in less then one hour, to a reference overnight protocol, were performed. A

simple, in-solution enzymatic process was chosen for demonstration, that can find

applicability to any proteomic analysis that involves biological samples, and that can be

most easily implemented in high-throughput analysis work-flows.

4.2 Material and methods

Materials. SKBR3 breast cancer cells, HBEC-5i human brain endothelial cells, phosphate

buffered saline (PBS) and trypsin/EDTA were purchased from the American Tissue

Culture Collection (ATCC, Manassas, VA, USA). Fetal bovine serum (FBS) was from

Gemini-Bio Products (West Sacramento, CA, USA), McCoy’s 5A and DMEM)/F-12

(1:1) cell culture media from Life Technologies (Carlsbad, CA, USA), human epidermal

growth factor (hEGF) from PeproTech (Rocky Hill, NJ), Normocin from InvivoGen (San

Diego, CA, USA), and sequencing grade modified trypsin and LysC from Promega

Corporation (Madison, WI, USA). Endothelial cell growth supplement (ECGS), urea,

dithiothreitol (DTT), acetic acid, trifluoroacetic acid (TFA), ammonium bicarbonate,

sodium chloride, Trizma base and hydrochloride buffer, protease inhibitor solution and

phosphatase inhibitor cocktails 2 and 3 were obtained from Sigma-Aldrich (St. Louis,

MO, USA). Zorbax SB-C18/5 μm particles, SPEC-PTC18 and SPEC-PTSCX solid-phase

extraction pipette tips were purchased from Agilent Technologies (Santa Clara, CA,

USA), and fused silica capillary columns from Polymicro Technologies. HPLC-grade

acetonitrile and methanol were from Fisher Scientific (Fair Lawn, NJ, USA), and DI

37

water was prepared with a MilliQ Ultrapure water system (Millipore, Bedford, MA,

USA).

Cell culture and processing. SKBR3 breast cancer cells were cultured in McCoy’s 5A

cell culture medium, with 10 % FBS, by incubation at 37 °C with 5 % CO2. Some

cultures were arrested by serum deprivation for 48 h, and released with EGF (20 ng/mL)

for 10 min. HBEC-5i cells were cultured in DMEM/F12 supplemented with 10 % FBS

and ECGS (40 µg/mL), and arrested for 48 h by serum deprivation. Normocin (0.1

mg/mL) was added to the cell culture to protect against bacteria, fungi and mycoplasma.

At full confluence, the cells were harvested by trypsinization, washed with cold PBS, and

stored at -80 °C. For further processing, the cells were suspended in a lysis buffer

prepared from 50 mM Tris (pH~8), 75 mM NaCl, 8 M urea, 1-2 mM DTT, and protease

(lysis buffer:protease inhibitor solution 100:1 v/v) and phosphatase inhibitor cocktails 2

and 3 (lysis buffer:phosphatase inhibitor solution 50:1 v/v). Lysis was performed through

intermittent sonication for 10 min (5 x 1 min sonication bursts followed by 1 min pause)

in an ice-cooled sonic bath. The lysis buffer to packed cell volume ratio was 5:1.The

protein concentration was measured with the Bradford assay (SmartSpec Plus

spectrophotometer, Bio-Rad, Hercules, CA, USA).

Enzymatic digestion. The SKBR3 protein extracts were denatured and reduced at 56 °C

for 1 h in the lysis buffer that already contained denaturing and reducing agents, diluted

10-fold with 50 mM NH4HCO3, and digested with trypsin (50:1 substrate:enzyme ratio)

in solution at 37 °C, for various amounts of time (7 min, 15 min, 30 min, 60 min, 18 h).

38

Alkylation was not performed to avoid the presence of incomplete reactions and side-

products [16]. The protein digestion solution was quenched with glacial CH3COOH (10

µL per 1 mL protein digest), and sample cleanup was performed with SPEC C18/SCX

cartridges. The LysC proteolysis of SKBR3 cells was performed by following the same

procedure. HBEC-5i cells were processed according to a protocol that enabled separation

in nuclear and cytoplasmic extracts, described in detail in previous work [17]. All peptide

samples were brought to dryness in a vacuum centrifuge, dissolved in a solution of

H2O/CH3CN/TFA 98:2:0.01 v/v to a final concentration 2 μg/μL, and further analyzed by

nano-HPLC-MS.

LC-MS/MS analysis and data processing. LC-MS/MS analysis and data processing

protocols were described in detail in previous work [17]. In short, MS analysis was

performed with an LTQ-XL mass spectrometer (Thermo Electron, San Jose, CA, USA),

operated in positive-ion electrospray mode at ~2 kV. Nano-LC was performed with an

Agilent 1260 micro-LC separation system and in-house prepared separation columns

(100 μm i.d. × 360 µm o.d., 10-12 cm long, packed with 5 µm/C18 Zorbax particles, 300

Å pore size), with the eluent flow rate set at 180-200 nL/min. The flow was generated by

the micro-LC pumps at 10 µL/min, and split to the desired level with an in-house built

split/splitless injector. Sample injections (8 µL) were performed in splitless mode. The

eluent was prepared from H2O/CH3CN/TFA, and the concentration gradient was from

96:4:0.01 v/v (solvent A) to 10:90:0.01 v/v (solvent B). For global proteomic profiling of

cell extracts, a 4 h long HPLC gradient was used, and the samples were analyzed via a

data-dependent analysis (DDA) method by performing zoom/MS2 scans on the top 5

39

most intense peaks from each MS scan (produced by averaging 5 scans), with the data

being acquired over a mass range of 500-2000 m/z. The collision induced dissociation

(CID) parameters were 3 m/z ion isolation width, 21 % normalized collision energy, 0.25

activation Q, 30 ms activation time, and threshold for triggering MS2 scans 100 counts.

Conditions for data dependent analysis included: 5 m/z zoom scan width, 1.5 m/z

exclusion mass width, dynamic exclusion at repeat count 1, repeat duration of 30 s,

exclusion list size 200, and exclusion duration 60 s.

Bioinformatics. Raw data files were analyzed with the Proteome Discoverer 1.4 software

package, using the Sequest HT search engine (Thermo Electron) for performing searches

against a Homo sapiens protein database from UniProt (January 2015 download)

comprising 20,198 reviewed/non-redundant sequences (500-5000 mass range,

minimum/maximum peptide length of 6/144 amino acids, precursor ion tolerance 2 Da,

fragment ion tolerance 1 Da, b/y/a ion fragments only, fully tryptic fragments with up to

4 missed cleavages allowed, no PTMs). Three replicates of each sample were performed.

The peptide false discovery rate (FDR) settings were 3 % and 1 % for relaxed and

stringent database searches, respectively. FDRs were calculated with the Target Decoy

PSM Validator node based on Xcorr vs. charge state values. The GRAVY (grand average

of hydropathy) index for tryptic peptides was calculated with tools provided by

Bioinformatics.org [18].

40

4.3 Results and discussion

To assess the effectiveness of fast enzymatic reactions performed in solution, and their

value for MS proteomics experiments, SKBR3 breast cancer cell extracts were subjected

to enzymatic digestion with trypsin for 7 min, 15 min, 30 min, 60 min and 18 h. The

reaction performance and quality of products was evaluated by measuring the total

number of identifiable peptides and proteins, the protein sequence coverage, the

reproducibility of detection, the detection of low abundance proteins, the missed cleavage

sites, the quality of peptides as assessed by Xcorr scores, and the GRAVY index. The

results of the 18 h long enzymatic process were used as a reference. Each dataset was

produced by combining three LC-MS/MS analysis replicates of a sample. Additional

datasets produced either from SKBR3 cells subjected to various biological treatments (7

min digestion), or from HBEC-5i brain endothelial cells (18 h digestion), were used to

confirm the results of the time-point experiments.

Peptide identification and protein sequence coverage. An effective proteomics protocol

aimed at profiling complex cellular extracts leads to the identification of thousands of

proteins, matched, ideally, by multiple peptides per protein to ensure unambiguous

identifications. The proteomic profiling of the SKBR3 cell extracts subjected to short

proteolysis reaction times (7-60 min) led to very similar results in terms of protein IDs

(e.g., groups), with numbers ranging from 1,093 to 1,162, but with matching unique

peptides increasing progressively from 3,386 to 3,567 as the reaction time decreased from

60 min to 7 min (Figure 4.1). Both numbers exceeded the results for the 18 h reference

(1,023 protein groups matched by 2,622 peptides). However, while all protein IDs were

41

maintained within +/-10 % of a grand average, the unique peptide IDs increased by ~25-

30 % when tryptic digestion was performed on a short time scale. Overall, for this dataset,

the range of unique peptides and sequence coverage per protein dropped from 3-30 % to

0-15 %, and from 5-77 % to 0-45 %, respectively, when extending the digestion time

from 7 min to 18 h. The sequence coverage was less at 18 h due to the loss of short,

typically 2-6 amino acid long sequences, some with multiple K and R residues that

flanked longer sequences with no missed cleavage sites. Many of the short peptides fell

in a range of m/z<500 that was not included in the data acquisition process due to the

presence of background ions that interfered with, and reduced the overall effectiveness of

the DDA process. Being also more hydrophilic, such peptides can be lost during the

sample cleanup and enrichment steps. Essentially, none of the digestion products

contained peptides with <8 amino acid residues.

42

3567 3500 34963386

2622

1093 1122 1155 11621023

0

500

1000

1500

2000

2500

3000

3500

4000

7 min 15 min 30 min 60 min 18 h

Pro

tein

an

d p

epti

de

cou

nts

Enzymatic digestion time

Number of unique peptides Number of unique proteins

Figure 4.1 Effect of enzymatic reaction time on the identification of peptides and proteins generated

from SKBR3 cell extracts. Conditions: SKBR3 cells were lysed through sonication, digested with trypsin

for 7 min, 15 min, 30 min, 60 min or 18 h, subjected to C18/SCX cleanup, and analyzed by nano-HPLC-

MS/MS; sample concentration 2 μg/μL, HPLC injection volume 8 µL; nano-LC gradient 4 h long, from

4 % to 96 % CH3CN (TFA 0.01 %). The number of unique peptides and protein groups per time-point

represent each the combined results of 3 LC-MS/MS analyses.

43

Reproducibility. The ability to detect more peptides per protein at short enzymatic

digestion times has a number of benefits, most importantly, increased confidence in

protein identification and quantitation. Nevertheless, this is the result of an incomplete

digestion process, which may lead to inferior reproducibility in the detection of the same

peptides in replicate experiments. To explore the level of reproducibility that can be

achieved within a few minutes of enzymatic digestion, the overlap between unique

peptides generated at different time-points, and from different sample replicates, was

assessed. Between successive time-points, the reproducibility of detecting the same

peptides was consistently preserved at 71 % (RSD 2 %) (Figure 4.2A-4.2C), but it

dropped to ~35 % when comparing the 7 min to the 18 h experiment (Figure 4.2D). If

only peptides containing zero missed cleavages were compared, the reproducibility was

only marginally better, by ~5 %. Independent experiments conducted with different cell

batches confirmed that the % overlaps between unique peptides were similar for the 7

min (76 %, RSD 5 %) and the 18 h (80 %, RSD 3 %) time-point experiments

(representative examples shown in Figures 4.2E-4.2F). As expected, at the protein-,

relative to the peptide-, levels, the overlaps were higher, but followed the same trends

(Figures 4.2G-4.2I). Overall, at both levels, the reproducibility of detecting the same

components was not substantially different, when conducting the digestion process for 7

min or for 18 h. Interestingly, however, neither the peptides nor the proteins detected in

the 18 h experiment that led to a more complete digestion process, were a subset of the 7

min one, but the two sets were rather complementary (Figures 4.2D and 4.2G), with

larger differences than those that would have arisen from experiments conducted using

identical conditions (Figures 4.2E, 4.2F, 4.2H, 4.2I)). Therefore, an analysis strategy

44

that would combine the products of short and long proteolytic digestion times could be

used to increase the peptide and protein identification rates, to lead to a more complete

characterization of complex cellular extracts.

Detection of low-abundance peptides. Of concern was the possibility that peptides of

lower abundance would be lost in the early stages of the digestion process, being

overshadowed by the large number of peptides originating from abundant proteins. Such

peptides were matched by only a few spectral counts, being at the limit of detection under

the MS experimental condition used for data acquisition. These peptides belonged to a

range of proteins involved in cell cycle regulation, defense, stress response and various

metabolic processes. To investigate whether this was the case, the loss of peptides

matched by a low number of counts was assessed for samples digested for 18 h and 7 min.

Comparisons were made at both protein and peptide levels, as the variability of DDA

detection reaches its highest level for low abundance peptides, having, however, a lower

impact on protein detection. The variability induced by DDA detection alone was

evaluated by comparing two sets of results, consisting each of 3 combined LC-MS/MS

analyses of the exact same sample. The reproducibility of identifying the same peptides

and proteins, and the associated losses, were 78-80 % (~20 % loss) and 83-84 % (~16 %

loss), respectively (Figures 4.2F and 4.2I). The proteins that were lost were typically

matched by only one unique peptide and 1-2 peptide spectrum matches (PSMs). This

was considered as a reference, a minimal loss level, reflective of the variability

introduced by DDA/LC-MS/MS analysis. When proteolytic digest replicates of a sample

were compared (i.e., 7 min vs. 7 min), the total protein count loss increased to ~19-22 %,

45

indicating that the enzymatic reaction and sample clean-up steps contributed only

marginally, by an additional ~5 % to loss in reproducibility (Figures 4.2H). On the other

hand, when an 18 h product was compared to the 7 min product, the loss essentially

doubled to ~35 % (Figure 4.2G). Nevertheless, this loss was compensated by the

identification of a similar number of new peptides and proteins that were not detectable in

the 18 h experiment. Altogether, on a global level, the results indicate that the

detectability of proteins mapped by only a few peptides was not hampered by the short

enzymatic reaction, but rather diversified.

46

Figure 4.2 Venn diagrams of unique peptide (A-F) and protein (G-I) overlaps from the 7 min-18 h

time-point experiments. Numbers on the bottom indicate IDs in each replicate and the % overlap, while

numbers on the top indicate the combined IDs in both replicates. Conditions: (A-D) and (G) Venn diagrams

of peptide overlaps between the 7 min, 15 min, 30 min, 60 min and 18 h enzymatic digestion experiments;

proteins extracted from proliferating SKBR3 cells. (E) and (H) Venn diagrams of peptide and protein

overlaps from two tryptic digest replicates, each conducted for 7 min; proteins extracted from SKBR3 cells

arrested for 48 h in serum free medium, and released for 10 min with EGF (20 ng/mL). (F) and (I) Venn

diagrams of peptide and protein overlaps from 2 x 3 LC-MS/MS replicates of the same cell extract tryptic

digestion product; proteins extracted from HBEC-5i cells arrested for 48 in serum free medium.

47

Figure 4.3 The percetage change in peptides with various numbers of missed cleavage sites by the

progression of enzymatic digestion. Stacked column chart illustrating the % change in peptides with

various numbers of missed cleavage sites, as the enzymatic digestion progressed from 7 min to 18 h.

Conditions: the same as in Figure 1.

Figure 4.4 Trends in missed cleavage sites and peptide length as a function of enzymatic reaction

time. (A) Line chart illustrating the % change in missed K and R cleavage sites with the progression of the

proteolytic digestion reaction. (B) and (C) Pie charts illustrating the % change in peptide length, in terms of

number of amino acids, for enzymatic reactions conducted for 7 min and 18 h. Conditions: the same as in

Figure 1.

48

Figure 4.5 Stacked column chart illustrating amino acid frequency distributions. (A) Theoretical

distribution of amino acids in the Homo sapiens proteome. (B) Reference set of 34,388 peptides generated

from a combination of proteomic experiments conducted on various human cell lines. (C) Reference set

from above, including only 18,617 peptides with zero missed cleavage sites. (D) Full set of 2,622 peptides

from the 18 h experiment. (E) Sub-set of 2,187 peptides with zero missed cleavage sites from the 18 h

experiment. (F) Full set of 3,567 peptides from the 7 min experiment. (G) Sub-set of 1,767 peptides with

zero missed cleavage sites from the 7 min experiment. The lower bar indicates the theoretical frequency of

amino acids in the Homo sapiens proteome.

49

Missed cleavage sites. A count of the missed cleavages at internal K and R amino acid

residues provided very clear trends over the time period taken under study (Figure 4.3).

The peptides with no missed cleavage sites increased from ~ 50 % to 83 %, mainly at the

expense of peptides with 1 or 2 missed cleavages that dropped from 34 % to 13 %, and

from 12 % to 2 %, respectively, as the proteolysis reaction was increased from 7 min to

18 h. Peptides with one missed K experienced the largest change, followed by peptides

with one missed R and two missed K and R residues (Figure 4.4A). Peptides with two

missed R residues were generally low in number. Experiments conducted with cells

cultured under various other conditions confirmed these results, the fraction of peptides

with no missed cleavages being in the range of 42-50 % for the 7 min digestion reactions,

and 72-83 % for the 18-20 h digestions. The presence of various detergents used for cell

lysis and fractionation impacted, however, to a certain extent, the effectiveness of the

proteolysis reaction.

It is worth to note that the highest contribution to missed cleavages was provided by one

missed K, and this was much larger than that contributed by one missed R. Given that the

frequency of K and R in the Homo sapiens proteome is roughly the same, 5.6 % and

5.7 % [19], respectively, the observed results led to the conclusion that either there was a

bias in the detection of peptides with multiple missed R residues, or that the enzymatic

digestion was more effective at R than at K, or both. The frequencies of amino acid

detection in the 7 min and 18 h experiment were therefore compared to the theoretical

frequency of amino acids in the Homo sapiens proteome and to a reference set of 34,288

peptides generated from a variety of proteomic experiments in our laboratory from MCF7,

50

MCF10 and SKBR3 cell lines. Entire datasets, or subsets containing only peptides with

zero missed cleavages, were compared (Figure 4.5). Overall, for the entire set of 20

amino acids, the experimental results closely matched the theoretical predictions. The

largest discrepancies were found for K and R. For the full datasets, the frequency of K

detection (4.2-5.2 %) came close to the theoretical value (5.6 %), while the frequency of

R (3.0-3.4 %) was lower than the theoretical one (5.7 %). When only peptides with zero

missed cleavage sites were counted, the frequencies dropped even more, to 2.8-3.4 % for

K, and to 2.2-2.5 % for R (encircled areas in the bar graph from Figure 4.5).

These findings confirmed that the frequency of R detection is lower than optimal for all

datasets, but this is not a bias introduced by the fast digestion process. Moreover,

prolonged reaction times that led to a more complete enzymatic reaction, also led to the

preferential loss of both K and R residue containing peptides - an expected outcome, as

the short peptides that are cleaved by the enzyme always carry at least one K or R residue.

The analysis of the shortest detectable peptides from these datasets (i.e, peptides

containing 8-10 amino acids) confirmed this assumption, revealing that the frequencies of

both K and R in these peptides approached or even exceed the theoretical values (i.e.,

5.8 % K and 5.0 % R for 18 h, and 7.3 % K and 6.3 % R for 7 min reactions). Such short

peptide products that are missed in proteomics experiments lead therefore not just to

reduced protein sequence coverage, but are also a source of K/R loss in the full proteome

datasets. The presence of inhibitory effects of tryptic attack, induced by flanking or

neighboring D, E, or P residues, or of RK sequences, calls for additional scrutiny [20],

even though these did not appear to affect the main trends in these datasets. Peptide

51

sequences rich in basic amino acid residues are part of protein motifs and domains that

are involved in the regulation of a variety of cellular processes. Arginine-rich domains,

for example, have critical roles in RNA processing, maturation and ribonucleotide

assembly [21], while arginine methylation has been found to control the subcellular

localization of the oncoprotein splicing factor SF2/ASF [22]. Fine-tuning the enzymatic

reaction and further optimizing the sample processing and data acquisition process could

therefore enable a more accurate interpretation of the biological implications of

proteomic data.

Peptide size and quality of MS identifications. While some K/R residues from short

peptide sequences escaped detection, the presence of missed cleavage sites in the

products of the fast proteolysis reactions led to a larger proportion of long peptides

containing 20-50 amino acid residues (Figures 4B/C). The production of such peptides

could support applications that use middle-down proteomic sequencing and peptide

fragmentation techniques such as ETD (electron transfer dissociation). Middle–down

proteomics seeks the detection of peptides in the mass range of 3,000-15,000 Da. Such

peptides result in better sequence coverage and improved ability to detect protein variants,

as well as a more accurate assignment to gene products [23,24]. ETD, on the other hand,

is a fragmentation technique that has been developed for the analysis of large, multiply-

charged peptides, to capture posttranslational modifications (PTMs) and enable the

identification of the modified amino acid sites. It benefits, therefore, from the presence of

basic amino acid residues in the sequence of the peptide. A number of enzymes (Lys-C,

Lys-N, Arg-C, Asp-N, Glu-C, ompT) and chemically induced digestion protocols have

52

been explored for generating peptides in this mass range [23,24]. Generally, it was found

that the use of such enzymes led to only a modest increase in peptide identifications and

in average peptide length (e.g, 1.9 kDa for Glu-C and Asp-N peptides, relative to 1.5 kDa

for tryptic peptides) [24,25]. However, the benefit of alternative proteases manifested

itself in using their combined peptide results that led to improved sequence coverage, and

ability to detect low abundance proteins [25]. In comparison, the analysis of the 7 min

reaction products showed that the proportion of peptides with >20 amino acid residues

was markedly higher (43-47 %) than in the 18 h products (26-29 %) (Figures 4B and 5C).

We also note that the combination of 7 min and 18 h tryptic peptides, when compared to

all other combinations that were explored, led to the largest increase in both peptide and

protein IDs (94 % and 40 % increase in new peptide and protein IDs relative to the 18h

products) (Figures 4.2D and 4.2G). These values are much higher than reported by the

use of combined enzymes that resulted in an average increase of only 16 % of new

protein IDs per use of a new enzyme with different cleavage specificity [25]. A fast

proteolysis strategy could complement, therefore, either bottom-up or middle-down

approaches enabled by the use of a variety of enzymes.

To demonstrate the broader applicability of the procedure, a Lys-C digestion reaction was

conducted for 7 min. The results were similar to the tryptic experiments: peptides with

zero missed cleavage sites 53 %, peptides with >20 amino acid residues 46 %, K sites

7.4 %, and R sites 2.8 %, with K detectability higher, and R detectability similar to the

tryptic digests. The median and maximum peptide length for all 7 min tryptic and LysC

digestions was 19-20 and 47-51, respectively, higher than typically observed for

53

overnight digestions (median 16-17, maximum 46-48 amino acids), with the major

difference lying in the higher proportion of peptides with >20 amino acid residues (see

above). Most importantly, when combining two sets of peptides produced in the 7 min

LysC and tryptic enzymatic reactions, the increase in unique peptides reached a

maximum of 5717 IDs and sequence coverage.

To assess the quality of identification of higher mass peptides generated through fast

enzymatic reactions, the peptide pools from the 7 min and 18 h experiments were

compared based on the distribution of XCorr scores vs. m/z for charge states 1+, 2+ and

3+ (Figure 4.6). The Xcorr scores were very similar for the two experiments, distributed

in the ranges of 2.7-3.8 for singly charged peptides, 3.3-5.6 for double charged peptides,

and 3.8-5.5 for triply charged peptides, with slightly higher trends for the 7 min peptides.

A similar assessment was performed for peptides originating from LysC digests, as well

as for peptides containing only K or R residues from either the tryptic and LysC (7 min)

digests. The ranges of the Xcorr scores were the same as above, for LysC or K- or R-only

containing peptides, confirming that there was no bias in their detection through tandem

MS. While accurate characterization of higher charge states would require high mass

accuracy MS instruments, overall, the observed scores for the 7 min experiments

corroborated the quality of the generated peptides, supporting the applicability of fast

enzymatic processing for proteomic explorations of complex samples and integration in

workflows and platforms that seek rapid sample analysis [26].

54

Figure 4.6 Xcorr score distributions as a function of m/z. Box plots representing Xcorr score

distributions for (1+), (2+) and (3+) charged peptides, as a function of m/z for (A) 7 min and (B) 18 h

enzymatic digest products.

55

Figure 4.7 Gravy index histograms for peptides with 0 to 4 missed cleavages. Results were generated

from the 7 min (A-E) and 18 h (F-J) proteolytic digestion experiments. A reference bar is drawn for an

index value of zero, to help visualize the shift in GRAVY scores with an increased number of missed K/R

cleavage sites.

56

Gravy index. Longer peptide sequences are often associated with an increase in

hydrophobic properties that impact negatively their recovery from clean-up cartridges or

separation on C18/LC columns. To answer this concern, the distributions of GRAVY

scores for peptides generated from the 7 min and 18 h long enzymatic reactions, were

compared (Figure 4.7). The GRAVY index values represent the hydrophilic and

hydrophobic properties of an amino acid, and cover a range from (-4) to (+4), the more

negative values being characteristic of the more hydrophilic components [27]. The

GRAVY histograms were generated by calculating the GRAVY score for each peptide

and partitioning the values in bins of 0.1 width. Peptides with a different number of

missed cleavage sites were placed in different histograms. As observed from the figure,

for all peptide sub-sets at 7 min (4.7A-E) and 18 h (4.7F-J), the presence of a larger

number of missed cleavages resulted in a shift toward more negative GRAVY scores,

revealing that the longer peptides were, in fact, more hydrophilic, rather than

hydrophobic. The shift in properties was introduced by the presence of additional K and

R residues, which are the most hydrophilic amino acids with a hydropathy score of -4.5

(R) and -3.9 (K). This result brings an unintended advantage, as a broader distribution of

the hydrophobic/hydrophilic properties of peptides would allow for a superior refinement

of reversed phase LC separations, and a better utilization of the landscape of available

techniques (ion exchange, HILIC, or high-pH reversed phase LC) for devising multi-

dimensional peptide pre-fractionation strategies.

57

4.4 Conclusions

To streamline proteomic experiments, in this work, we propose the use of accelerated

enzymatic reactions for processing the protein complement of cells. The performance of

this approach was assessed by evaluating the impact of the enzymatic digestion time on

the quality of generated peptides. We demonstrate that a rapid digestion process can

produce results that are superior to established protocols in terms of achieving protein

and proteome coverage, with no sacrifice in ability to perform efficient tandem mass

spectrometric analysis. While the reaction products of short and long enzymatic reactions

differed considerably, the reproducibility of obtaining the same results from fast replicate

enzymatic processes was only marginally affected by the short time-scale. By properly

combining the reaction products generated through short and long enzymatic reactions,

possibly completed with two distinct enzymes, a substantial increase in the identification

of peptides and proteins can be achieved. Moreover, by fine-tuning the reaction

conditions, changes in the length and physical properties of peptides can be induced, to

support more effective separations and complementary analysis approaches in middle-

down proteomics experiments. Altogether, these results demonstrate the value of fast

enzymatic digestions and lay the necessary premise for catalyzing the development of

novel high-throughput proteomics workflows.

Acknowledgments. This work was supported by awards from the NSF (DBI-1255991)

and NIGMS (1R01GM121920-01A1) to IML. The authors thank Shreya Ahuja for

providing support with the HBEC-5i cell culture.

58

4.5 References

1. Hustoft, H. K.; Malerod, H.; Ray, S.; Reubsaet, L.; Lundanes, E.; Greibrokk, T.

Integr. Proteomics 2012, InTech.

2. Switzar, L.; Giera, M.; Niessen, W. M. A. J. Proteome Res. 2013, 1067–1077.

3. Safdar, M.; Sproß, J.; Jänis, J. J. Chromatogr. A 2014, 1324, 1-10.

4. Yin, Y.; Xiao, Y.; Lin, G.; Xiao, Q.; Lin, Z.; Cai, Z. J. Mater. Chem. B 2015, 3 (11),

2295–2300.

5. Yuan, H.; Zhang, S.; Zhao, B.; Weng, Y.; Zhu, X.; Li, S.; Zhang, L.; Zhang, Y. Anal.

Chem. 2017, 89 (12), 6324–6329.

6. Ning, W.; Bruening, M. L. Anal. Chem. 2015, 87 (24), 11984–11989.

7. Wu, S.; Zhang, L.; Yang, K.; Liang, Z.; Zhang, L.; Zhang, Y. Anal. Bioanal. Chem.

2012, 402(2), 703-710.

8. Yuan, H.; Zhang, L.; Zhang, Y. J. Chromatogr. A 2014, 1371, 48–57.

9. Liu, W.-L.; Lo, S.-H.; Singco, B.; Yang, C.-C.; Huang, H.-Y.; Lin, C.-H. J. Mater.

Chem. B 2013, 1 (7), 928.

10. Starke, S.; Went, M.; Prager, A.; Schulze, A. React. Funct. Polym. 2013, 73 (5), 698–

702.

11. Kim, H.; Kim, H. S.; Lee, D.; Shin, D.; Shin, D.; Kim, J.; Kim, J. Anal. Chem. 2017,

89 (20), 10655–10660.

12. Guo, Z.; Cheng, J.; Sun, H.; Sun, W. Rapid Commun. Mass Spectrom. 2017, 31 (16),

1353–1362.

13. Ge, H.; Bao, H.; Zhang, L.; Chen, G. Anal. Chim. Acta 2014, 845, 77–84.

14. Deng, J.; Lazar, I. M. J. Am. Soc. Mass Spectrom. 2016, 27 (4), 686–698.

59

15. Lazar, I. M.; Deng, J.; Smith, N. J. Vis. Exp. 2016, 110, e53564.

16. Guo, M.; Weng, G.; Yin, D.; Hu, X.; Han, J.; Du, Y.; Liu, Y.; Tang, D.; Pan, Y. RSC

Adv. 2015, 5(125), 103662-103668.

17. Lazar, I. M.; Hoeschele, I.; de Morais, J.; Tenga, M. J. Sci. Rep. 2017, 7 (1), 17989.

18. Sequence Manipulation Suite for Protein GRAVY.

http://www.bioinformatics.org/sms2/protein_gravy.html, version 2.

19. Pruess, M.; Apweiler, R. J. Biomed. Biotechnol. 2003, pp 231–236.

20. Gershon, P. D., J. Proteome Res. 2014, 13 (2), 702–709.

21. Godin, K. S.; Varani, G. RNA Biol. 2007, pp 69–75.

22. Sinha, R.; Allemand, E.; Zhang, Z.; Karni, R.; Myers, M. P.; Krainer, A. R. Mol. Cell.

Biol. 2010, 30 (11), 2762–2774.

23. Wu, C.; Tran, J. C.; Zamdborg, L.; Durbin, K. R.; Li, M.; Ahlf, D. R.; Early, B. P.;

Thomas, P. M.; Sweedler, J. V; Kelleher, N. L. Nat. Methods 2012, 9 (8), 6–10.

24. Cristobal, A.; Marino, F.; Post, H.; Van Den Toorn, H. W. P.; Mohammed, S.; Heck,

A. J. R. Anal. Chem. 2017, 89 (6), 3318–3325.

25. Swaney, D. L.; Wenger, C. D.; Coon, J. J. J. Proteome Res. 2010, 9 (3), 1323–1329.

26. Lazar, I. M.; Rockwood, A. L.; Lee, E. D.; Sin, J. C. H.; Lee, M. L. Anal. Chem.

1999, 71 (13), 2578–2581.

27. Kyte, J.; Doolittle, R. F. J. Mol. Biol. 1982, 157 (1), 105–132.

60

CHAPTER 5: Proteolytic Digestion and TiO2 Phosphopeptide Enrichment

Microreactor for Fast MS Identification of Proteins

Jingren Deng and Iulia M. Lazar*

Department of Biological Sciences, Virginia Tech

1981 Kraft Drive, Blacksburg, VA 24061, USA

*Corresponding author: malazar@vt.edu

Journal of the American Society for Mass Spectrometry

2016 Apr;27(4):686-98. doi: 10.1007/s13361-015-1332-6.

Author contributions

Designed the experiments: JD IML

Performed the experiments: JD

Analyzed the data: JD IML

Wrote the paper: JD IML

Conceived and coordinated the study: IML

Keywords: microreactor, proteolytic digestion, phosphopeptide enrichment, mass

spectrometry

61

Abstract

The characterization of phosphorylation state(s) of a protein is best accomplished by

using isolated or enriched phosphoprotein samples or their corresponding

phosphopeptides. The process is typically time-consuming as, often, a combination of

analytical approaches must be used. To facilitate throughput in the study of

phosphoproteins, a microreactor that enables a novel strategy for performing fast

proteolytic digestion and selective phosphopeptide enrichment was developed. The

microreactor was fabricated using 100 μm I.D. fused-silica capillaries packed with 1-2

mm beds of C18 and/or TiO2 particles. Proteolytic digestion-only, phosphopetide

enrichment-only and sequential proteolytic digestion/phosphopeptide enrichment

microreactors were developed and tested with standard protein mixtures. The protein

samples were adsorbed on the C18 particles, quickly digested with a proteolytic enzyme

infused over the adsorbed proteins, and further eluted onto the TiO2 microreactor for

enrichment in phosphopeptides. A number of parameters were optimized to speed up the

digestion and enrichments processes, including microreactor dimensions, sample

concentrations, digestion time, flow rates, buffer compositions and pH. The effective

time for the steps of proteolytic digestion and enrichment was less than 5 min. For simple

samples, such as standard protein mixtures, this approach provided equivalent or better

results than conventional bench-top methods, in terms of both enzymatic digestion and

selectivity. Analysis times and reagent costs were reduced ~10-15 fold. Preliminary

analysis of cell extracts indicated the feasibility of integration of these microreactors in

more advanced workflows amenable for handling complex biological samples.

62

5.1 Introduction

The analysis of phosphoproteins represents a topic of major interest to many biological

researchers due to the key roles that phosphorylation plays in the regulation of various

cellular activities, as well as the implications that aberrant phosphorylation has in the

development of a number of diseases such as cancer and neurodegenerative disorders. A

typical bottom-up mass spectrometry (MS)-based phosphoproteomic study involves a

number of steps that include cell processing/lysis, protein separation/purification,

proteolytic digestion, phosphopeptide enrichment, microanalytical separations and MS

detection. Alternatively, conventional biological experiments rely on the

isolation/purification of a protein of interest, followed by characterization by using a

variety of physico-chemical methods. Often, a combination of different analysis

strategies must be used to enable a comprehensive characterization of individual

phosphoproteins, or of the phosphoproteome, as a whole. The steps of proteolytic

digestion and phosphopeptide enrichment, as commonly practiced, are time consuming

and represent the main bottlenecks in achieving throughput. For example, in-solution

proteolytic digestion is usually performed overnight at substrate:enzyme ratios of 50-

100:1 (w/w), being often accompanied by the generation of undesired enzyme autolysis

products. Some of the most common techniques for phosphopeptide enrichment involve

the use of immobilized metal affinity chromatography (IMAC) and/or of TiO2 particles.

In the case of complex samples, prefractionation steps, using strong anion/cation

exchange or hydrophilic interaction chromatography, precede the phosphopeptide

enrichment process.1-3 In comparison to IMAC, the use of TiO2 particles has gained

particular momentum in the bioanalytical community due to the much improved

63

selectivities that can be achieved for enrichment in phosphopeptides. The sample

enrichment process on such particles, while very effective, tends to be also time-

consuming as it involves multiple conditioning, rinsing and centrifugations steps (1-2 h).

To accelerate the in-solution proteolytic digestion process, a number of approaches have

been developed.4,5 For example, microwave-assisted6 digestion, proposed by Pramanik et

al. in 2002,7 successfully reduced the digestion time from hours to 10 min. Turapov et al.

reported a fast digestion method using a PCR-type thermocycler,8 with the caveat that

many proteases cannot withstand the elevated temperatures. Ultrasound-assisted

digestion, first introduced in 2005,9 was carried out in less than 60 s. Alternative

techniques, making use of pressure,10 alternating electric fields11 and infrared radiation12

have also become powerful tools for accelerating the in-solution proteolytic digestion

process. To eliminate the need for additional instrumentation, reduce sample and reagent

consumption and enable process automation, a variety of enzymatic microreactors have

been developed as stand-alone devices or as part of micro-total analysis systems (µ-

TAS). The majority of these microreactors fall under the category of immobilized

enzymatic reactors (IMERs). By immobilizing the enzyme, high enzyme-to-substrate

ratios can be achieved, and undesired autolysis products are eliminated. Various methods

have been introduced for the enzyme-immobilization process.13-27 Landers and co-

workers introduced a digestion approach that uses electroosmotic flow (EOF) to pump

proteins through a proteolytic digestion system packed with immobilized trypsin gel

beads.17 Liu and co-workers immobilized trypsin on different substrates, including titania

and alumina sol-gel,22 gold nanoparticles,23 and nanozeolites.24 Wu et al. developed a

64

microdevice consisting of an immobilized proteolytic enzyme on the surface of acrylic

acid-grafted PDMS channels.25 To improve sensitivity and detection limits, Hustoft et al.

coupled for the first time an immobilized trypsin reactor on-line with open-tubular LC-

MS, to detect attomole amounts of isolated cancer proteins.26 Regnier and co-workers

proposed another innovative on-line system that integrates affinity selection, buffer

exchange, and a continuous flow (cf)-IMER, which was capable of converting native

proteins to peptides in a few minutes at elevated temperature with high recovery and

great reproducibility.27 Krenkova et al. introduced an enzyme reactor containing trypsin

and LysC immobilized on a porous polymer monolith, which was demonstrated for the

fast digestion (6 min) of high-molecular weight human immunoglobulin G.16 In regard to

phosphopeptide enrichment using IMAC, TiO2, immunoprecipitation or a combination of

any two strategies, only a limited number of microdevices have been reported. Yates and

co-workers developed a TiO2-based automated online multidimensional phosphopeptide

enrichment system coupled to ESI-MS. The study successfully demonstrated the superior

performance of TiO2 as a phosphopeptide enrichment resin, allowing high-throughput

analyses for assessing the phosphorylation states.28 Heck’s group, using the Agilent

microfabricated platform, described a TiO2-based chip coupled to Q-TOF/MS, which

identified 1012 unique phosphopeptides corresponding to 960 different phosphorylation

sites in human leukocytes.29,30

While most immobilized microfluidic reactors are very effective in achieving their

purpose, several key issues must be addressed before enabling trouble-free

implementation for routine lab operations. The fabrication of an enzyme-immobilized

65

reactor is time-consuming, complex and expensive. For example, the fabrication of a

monolithic structure inside a microchannel may require up to 10 hours, or even more, if

surface modifications are necessary. Commercial enzymatic cartridges are high-priced.

Covalent attachment or immobilization through non-covalent interactions or

encapsulation of enzymes has the potential to affect the enzyme activity. Sample

preparation and cleanup alone, such as for phosphopeptide enrichment with highly

selective TiO2 spin tips, requires several hours for completion. The undesired adsorption

of proteins on polymer-based microdevices fabricated, for example, from PDMS or

PMMA, poses additional concerns to achieving good detection limits. To circumvent

some of the above-described concerns, in this work, we propose a novel and very simple

strategy for the fabrication of a microreactor that enables both proteolytic digestion and

phosphopeptide enrichment. Commercially available C18 and TiO2 particles were packed

into fused silica capillaries to act as enzymatic digestion and enrichment microreactors.

Unlike in conventional immobilized enzyme systems, in this strategy, the protein samples

are the ones that are adsorbed on the C18 beads, and the proteolytic enzyme is delivered

through infusion. The newly generated peptides are then eluted onto the TiO2

microreactor for enrichment in phosphopeptides. This analysis approach enabled protein

digestion in 1-3 min, digestion/phosphopeptide enrichment in ~ 5 min, and complete

sample analysis in ~30-90 min. In comparison to conventional protocols this represents

~10-15 fold speed up of the analysis process.

66

5.2 Material and methods

Reagents. Hemoglobin α/β, bovine serum albumin, carbonic anhydrase, α-lactalbumin,

fetuin, α-casein, β-casein, cytochrome C, urea, dithiothreitol (DTT), acetic acid,

trifluoroacetic acid (TFA) and ammonium bicarbonate were purchased from Sigma-

Aldrich (St. Louis, MO). Sequencing-grade trypsin was acquired from Promega

Corporation (Madison, WI). Zorbax SB-C18/5 μm particles and SPEC-PTC18/SPEC-

PTSCX pipette tips were from Agilent Technologies (Santa Clara, CA), and Titansphere

Phos-TiO/10 μm particles, styrene divinylbenzene (SDB)/graphitic carbon (GC)

cartridges and lactic acid were products of GL Sciences (Torrance, CA). HPLC-grade

methanol and acetonitrile were purchased from Fisher Scientific (Fair Lawn, NI). DI

water was obtained from a MilliQ Ultrapure water system (Millipore, Bedford, MA).

SKBR3 breast cancer cells were purchased from the American Tissue Culture Collection

(ATCC, Manassas, VA), fetal bovine serum (FBS) from Gemini Bio Products (West

Sacramento, CA), and McCoy’s 5A cell culture medium from Life Technologies

(Carlsbad, CA).

Sample preparation and analysis using conventional proteolytic digestion and

phosphopeptide enrichment protocols. Standard bovine proteins, or mixtures of proteins,

1 μM each in NH4HCO3 (50 mM), were denatured with urea (8 M) in the presence of

DTT (5 mM) at 60 °C for 1 h, diluted 10-fold with NH4HCO3 (50 mM), digested

overnight with trypsin (50:1 substrate:enzyme ratio) at 37 °C, quenched with glacial

CH3COOH (1 % v/v final concentration), and then subjected to C18/SCX cleanup. In

preparation for nano-HPLC-MS/MS experiments, or for loading on the microreactor, the

67

samples were brought to dryness and dissolved in a solution of H2O/CH3CN/TFA

98:2:0.01 v/v to a final concentration of 0.5-1 μM in each protein. TiO2 phosphopeptide

enrichment was performed in the presence of lactic acid using the Titansphere Phos-TiO

Kit following the manufacturer’s protocol (adsorption at pH<3 in the presence of TFA

and lactic acid, and elution at pH>10 in a solution containing NH4OH). The

phosphopeptide enriched samples were subjected to SDB/GC cleanup, brought to dryness

in a vacuum centrifuge and dissolved in H2O/CH3CN/TFA 98:2:0.01 v/v to reach a final

concentration corresponding to the initial concentration of 1 μM in each protein. The

phosphopeptide samples were further analyzed by nano-HPLC-MS. For simple infusion-

MS analysis, all peptide samples were dissolved in a solution of CH3CN/H2O/TFA

50:50:0.01 v/v, phosphopeptide samples in CH3CN/H2O/28% NH4OH 50:40:10 v/v, and

infused through a 50 µm I.D. x 360 µm O.D. x 1 m long fused silica capillary at 300

nL/min.

SKBR3 cell processing. SKBR3 cells were cultured in McCoy’s 5A supplemented with

10 % FBS in an incubator (5 % CO2 and 37 °C), harvested when the cells reached full

confluence and stored at -80 °C. For processing, the frozen cells were thawed at room

temperature and lysed through sonication for 5 min at 4 ºC. The protein content was

measured with the Bradford assay (SmartSpec Plus spectrophotometer, Bio-Rad,

Hercules, CA). Conventional tryptic digestion of the cell extracts was performed

overnight at 37 °C, at substrate:enzyme ratios of ~50:1 (w/w). The digest was further

cleaned-up with C18/SCX pipette tips and prepared for MS analysis (1-2 μg/μL) in a

solution of H2O/CH3CN/TFA 98:2:0.01 v/v.31,32

68

Sample analysis. MS analysis was performed with an LTQ or LTQ-XL mass

spectrometer (Thermo Electron, San Jose, CA), in positive-ion mode, with the capillary

voltage set at 2.0-2.2 kV. The samples were either infused with a syringe pump 22

(Harvard Apparatus, Holliston, MA) at 300 nL/min through the microreactor or an empty

fused silica capillary, or analyzed with an Agilent 1100 micro-LC separation system,

using a nano-HPLC column operated at ~180 nL/min (100 µm I.D. x 360 µm O.D., 10

cm long, packed in-house with Zorbax C18/5 µm particles),32 with a 10 min

concentration gradient of H2O/CH3CN/TFA from 96:4:0.01 v/v (solvent A) to 10:90:0.01

v/v (solvent B). Fused silica capillaries were purchased from Polymicro Technologies

(Phoenix, AZ), and syringes (250 µL) from Hamilton (Reno, NV). The samples were

analyzed via a data-dependent method by performing Zoom/MS2 scans on the top 3 (fast

nano-HPLC-MS/MS) - top 10 (infusion MS/MS) most intense peaks in each MS scan (5

microscans averaged).32 Data were acquired over a mass range of 500–2000 m/z.

Collision induced dissociation (CID) was performed with isolation width 3 m/z,

normalized collision energy 35 %, activation Q 0.25, and activation time 30 ms. The raw

data files were searched against UniProt protein databases using the Proteome Discoverer

1.4 software package (Thermo Electron). Phosphorylation was enabled as a dynamic

modification on Ser, Thr and Tyr. Maximum missed cleavage allowance was set to 2,

precursor ion mass tolerance at 2 Da, fragment ion tolerance at 1 Da, and the peptide-

level FDR at <3 %. Each combination of treatments was replicated three times.

69

5.3 Results and discussion

The development of the proteolytic digestion and phosphopeptide enrichment protocols

included the fabrication and testing of microreactors of various dimensions, operated at

different flow rates under various sample loading and elution conditions, as well as

validation by performing comparisons with conventional sample preparation and analysis

protocols.

Microreactor fabrication. The microreactors were fabricated from fused-silica capillaries

(100 μm I.D.) that were packed with a slurry of C18/silica (5 μm) or TiO2 (10 μm)

particles delivered by a 250 μL gastight syringe. The particles were retained in the

microreactor by a 1 cm long capillary (20 μm I.D. x 90 μm O.D.) inserted in one end of

the microreactor, that also had the role of the electrospray ionization (ESI) emitter. The

choice of the solvent for the slurry preparation was critical to preventing particle

deposition prior to finalizing the microreactor fabrication process. Isopropanol proved to

be an adequate solvent for packing the C18/silica particles. For the heavier TiO2 particles,

a more viscous solution, prepared by adding lactic acid to isopropanol, had to be

prepared. Lactic acid is often used for improving the selectivity of the phosphopeptide

enrichment process, and is recommended in the phosphopeptide sample loading solvent

in concentrations as high as 25 % for the commercial TiO2 spin tips. Mixed with

isopropanol, it enabled trouble-free particle loading. Particle beds of 1-2 mm in length,

that contained enough particles for the retention of a sufficient amount of peptides to

achieve similar detection levels to conventional nano-LC separations interfaced to MS,

were prepared. As accurate control of a 1 mm long particle bed length was difficult to

70

achieve, most proteolytic digestion studies were performed with C18 microreactors of 2

mm in length. For phosphopeptide enrichment, the microreactor was first loaded with a

~1 mm long section of a C18 bed that acted as a filter for preventing the TiO2 particles

from blocking the ESI emitter, and then with a 2 mm long TiO2 bed. For performing

combined proteolytic digestion and phosphopeptide enrichment, microreactors

comprising three sections were prepared, i.e., ~1 mm C18 filter, 1-2 mm TiO2 for

phosphopeptide enrichment and 2 mm C18 particles for proteolytic digestion. All particle

loading operations were performed manually, under a microscope, to enable control of

the particle section lengths. The microreactor system was then washed at 2 µL/min with

isopropanol, CH3CN/H2O 80:20 v/v, and preconditioned with the same solution as used

for sample preparation, e.g., H2O/CH3CN 98:2 v/v at acidic or basic pH. Overall, the

preparation of the microreactors could be completed in 30-40 min.

Microreactor for proteolytic digestion. Microreactors that perform fast proteolytic

digestion protocols rely on the use of immobilized enzymes for ensuring high

enzyme:substrate ratios and preventing the generation of undesired autolysis products. In

this work, an alternative, easy-to-implement and cost-effective approach was developed,

that relies on the adsorption of the protein (i.e., the substrate) on hydrophobic C18/silica

particles, and the delivery of the enzyme through continuous flow to the microreactor.

The completion of the analysis involves the following steps: (a) adsorption of the protein

or protein mix on C18 particles through hydrophobic interactions, (b) infusion of the

enzyme over the adsorbed proteins in a solvent that preserves the retention of the newly

generated tryptic peptides, (c) removal of undesired buffer/contaminant components, and

71

(d) elution of the peptides from the microreactor in a solvent system compatible with

ESI-MS detection. The performance of the microreactor was evaluated for various

sample (0.1-1µM) and trypsin (0.1-5 µM) concentrations, incubation times (1-10 min)

and buffer compositions, and the results were compared to those obtained from

experiments in which the proteins were digested with a conventional overnight protocol.

Optimized microreactor operational conditions are provided in Table 5.1 A mixture of 10

proteins at concentrations typical for infusion experiments (e.g., 1 μM each), in a solvent

system which enables protein adsorption on C18 particles (H2O/CH3CN 98:2 v/v, pH~4

or pH~7.8), was infused for 5 min over the microreactor. Once the proteins were

adsorbed on the hydrophobic C18 particles, a trypsin solution at 5 times higher

concentration than the protein (e.g., 5 µM), in a basic buffer solution optimal for enabling

the proteolytic digestion process (H2O/CH3CN 98:2 v/v, 50 mM NH4HCO3, pH~7.8),

was pumped over the microreactor at 2 μL/min for a time ranging from 1 to 10 min (not

including the time necessary for rinsing the dead-volume associated with the transfer

capillary from the micropump to the microreactor, i.e., ~2 µL for a capillary of 50 µm

I.D. x 1 m length). To quench the enzymatic digestion process, a rinse step with acidic

solution was performed (H2O/CH3CN/TFA 98:2:0.01 v/v, 2 μL/min for 5 min).

Ultimately, the peptides were eluted in a high organic content solvent (H2O/CH3CN/TFA

50:50:0.01 v/v) at a flow rate of 300 nL/min for 25 minutes. While higher elution flow

rates would have resulted in overall shorter analysis times, the choice of 300 nL/min was

dictated by the maximum tolerable flow by the ESI emitter (20 µm I.D. x 90 µm O.D.)

that did not result in a loss of ESI efficiency.

72

Proteolytic digestion microreactor Solution composition Concentration Flow rate Infusion time

Microreactor preconditioning Acidic rinse solution: H2O/CH3CN/TFA

98:2:0.01 v/v 2 μL/min 5 min

Basic rinse solution: H2O/CH3CN/NH4HCO3

98:2 v/v, 50 mM NH4HCO3

2 μL/min 5 min

Sample loading Sample in acidic solution, or 1 μM 2 μL/min 5 min

Sample in basic solution 1 μM 2 μL/min 5 min

Enzymatic reaction Trypsin in basic solution 1.0-5.0 μM 2 μL/min 1-3 min

Tryptic reaction quenching Acidic rinse solution: H2O/CH3CN/TFA

98:2:0.01 v/v 2 μL/min 5 min

Sample elution for analysis Acidic elution solution: H2O/CH3CN/TFA

50:50:0.01 v/v 300 nL/min 25 min

Phosphopeptide enrichment microreactor

Solution composition Concentration Flow rate Infusion time

Microreactor preconditioning Acidic rinse solution A: H2O/CH3CN/TFA

20:80:0.4 v/v 2 μL/min 5 min

Equilibration with lactic acid Acidic equilibration solution B: solution A/lactic acid

97.5:2.5 v/v 2 μL/min 5 min

Sample loading Sample in acidic solution B 1 μM 1 μL/min 10 min

Lactic acid removal Acidic rinse solution A: H2O/CH3CN/TFA

20:80:0.4 v/v 1-2 μL/min 5 min

pH change Rinse solution C: H2O/CH3CN 50:50 v/v 1-2 μL/min 5 min

Sample elution for analysis Basic elution solution: H2O/CH3CN/28% NH4OH

40:50:10 v/v 300 nL/min 50 min

Microdigestion/phosphopeptide enrichment microreactor

Solution composition Concentration Flow rate Infusion time

Microreactor preconditioning Acidic rinse solution: H2O/CH3CN/TFA

98:2:0.01 v/v 2 μL/min 5 min

Sample loading Sample in acidic rinse solution 1 μM 2 μL/min 5 min

Enzymatic reaction Trypsin in basic solution 5 μM 2 μL/min 90 sec

Quenching tryptic reaction Acidic rinse solution: H2O/CH3CN/TFA

98:2:0.4 v/v 1-2 μL/min 5 min

Sample elution from C18/adsorption on TiO2

Rinse solution B: Solution A/Lactic Acid

97.5:2.5 v/v 1 μL/min 10-15 min

LA removal Rinse solution A: H2O/CH3CN/TFA

20:80:0.4 v/v 1-2 μL/min 5 min

pH change Rinse solution C: H2O/CH3CN 50:50 v/v 1-2 μL/min 5 min

Sample elution for analysis Elution solution: H2O/CH3CN/28% NH4OH

40:50:10 v/v 300 nL/min 25-50 min

Table 5.1 Analytical processing steps for the microreactors.

73

Figure 5.1 Protein and peptide identifications using various conditions for enzymatic protein

digestion. (A) Proteolytic digestion performed with the microreactor at various incubation times with

trypsin. Conditions: protein concentrations 1 µM, trypsin concentration 5 µM, sample loading/peptide

elution in acidic buffer (see Table 5.1). (B) Comparison of the microreactor performance with a

conventional overnight digestion protocol. The samples were either digested on the microreactor, or

digested overnight in a vial and loaded on the microreactor for further processing. Microreactor digestion

conditions: protein concentrations 1 µM, trypsin concentration 5 µM, sample loading in acidic or basic

buffer, peptide elution in acidic buffer (see Table 5.1). Overnight digestion conditions: protein denaturation

in urea/DTT, proteolytic digestion at substrate:enzyme ratio 50:1, buffer NH4HCO3 50 mM, pH~7.8,

C18/SCX cleanup of the tryptic peptide sample, peptide sample (1 µM) loading in acidic solution on the

microreactor, and rinsing/elution using the same conditions as for microcolumn digestions but without

trypsin infusion. See Table 5.1 and experimental section for details. Note: three replicates were conducted

for each experiment.

74

Figure 5.1A displays the results obtained for different trypsin infusion times over the

microreactor, i.e., 1, 1.5, 3, 5 and 10 min, respectively. A total of 8-9 proteins were

identifiable by 28-38 unique peptides, a maximum being observed at 1.5 min incubation

time, and a decrease in numbers after 3-5 min. A similar decrease in protein

identifications, albeit for experiments conducted on longer time-scales, was reported by

Klammer et al.33 A possible explanation for this observation takes into account the forces

that control the adsorption/desorption of proteins and peptides on hydrophobic surfaces.

Peptides and proteins are adsorbed on C18 particles as a result of hydrophobic

interactions. The overall outcome is controlled by the balancing of various forces which

arise as a result of intramolecular, solvation and surface effects. Initially, the sample

proteins are adsorbed on the C18 particles, and if the surface is saturated with sample

proteins, the adsorption/retention of trypsin is minimized. An increase in the duration of

trypsin infusion will increase the length of the enzymatic reaction and lead to a more

complete digestion process, a decreased number of missed K/R cleavage sites, and

eventually to an increased number of identifiable proteins. Nevertheless, at prolonged

reaction times, the larger molecular weight (MW) hydrophobic trypsin, and the buffer

solution itself, can also displace the smaller MW peptides, or even some proteins,

deteriorating the experimental outcome. This explanation was confirmed by experimental

evidence. While the large majority of peptides eluted from the microreactor with the high

organic content eluent, a few tryptic peptides were already observable at the start of the

elution process, at a stage when the organic solvent could not have reached yet the

microreactor, indicating that there were physical processes at work that prevented

complete retention of all peptides. For lower sample concentrations (e.g., 0.1 µM) the

75

concentration of trypsin had to be adjusted accordingly, best results being obtained for

infusions with 1 µM trypsin solutions. Due to low intensity peptide signals at these

concentrations, the results were somewhat less reproducible in terms of number of unique

peptide and protein identifications. The best conditions with the present experimental

setup were achieved for a proteolytic digestion process conducted for 1.5-3 min. A

substantial improvement in the average number of identified peptides (from 38 to 54) was

observed if the microreactor preconditioning and sample loading solution was changed

from an acidic (H2O/CH3CN/TFA 98:2:0.01 v/v, pH~4) to a basic buffer system such as

the one used for the digestion process (H2O/CH3CN 98:2 v/v, 50 mM NH4HCO3,

pH~7.8) (Figure 5.1B). This improvement resulted from the elimination of the

equilibration time needed for achieving the change from acidic to basic environmental

conditions which are favorable to the digestion process.

To evaluate the completeness of the digestion process, the proteolytic digestion results

obtained with the microreactor were compared to those of an overnight digestion

protocol. For this purpose, a sample containing a mixture of 10 proteins was denatured

and digested overnight with trypsin (substrate:enzyme ratio 50:1), cleaned-up with

C18/SCX cartridges, reconstituted to a final concentration of 1 µM, loaded on the

microreactor and processed using the same experimental conditions as for microreactor

digestion, but without the trypsin infusion step (Figure 5.1B). The mircoreactor

proteolytic digestion process performed within 1.5 min with basic buffer loading

conditions resulted in the identification of 9 proteins by an average of 54 unique peptides

per run and 83 unique peptides in combined 3 replicate experiments. The overnight

76

process returned all 10 protein I.D.s, matched by an average of 38 unique peptides per

run and 50 unique peptides in 3 replicate experiments. A larger number of missed tryptic

sites in the peptides generated with the microreactor accounted for the larger number of

unique peptides, the percent peptides with missed K/R cleavage sites being 53-59 % and

19-21 % for microreactor and overnight proteolysis protocols, respectively. The typical

range for missed K/R sites with the overnight digestion protocols was 20-30 %. While

this is an outcome of incomplete enzymatic digestion, the presence of such peptides

results in a more complete amino acid sequence coverage, and a more reliable

identification by MS. α-lactalbumin could not be identified in any of the experiments

conducted using the microreactor. Although it has a relatively small molecular weight

(~16 kDa), its globular structure and the lack of a protein denaturing step led, most likely

to resisting proteolytic digestion with trypsin.34,35 The performance of the microreactor,

as compared to overnight digestion in terms of unique peptides and % amino acid

sequence coverage, is provided in Table 5.2, and shown as a range for three replicate

experiments. Microreactor digestion resulted in each protein being identified by 3-9

unique peptides accounting for 7-75 % sequence coverage (~30 % average), except for α-

casein S2 which was not added to the mix but was present as a minor component of the

α-casein standard. Overnight digestion resulted in each protein being identified by 1-10

unique peptides accounting for 7-65 % sequence coverage (~26 % average). Table 5.1

provides an alignment of peptide identifications in all experiments. The data also

confirmed the results of earlier studies that evaluated the effectiveness of enzymatic

digestion of proteins adsorbed on various surfaces.36 These studies showed that the

enzymatic products for proteins adsorbed on surfaces differ from the ones proteolyzed in

77

solution, most likely due to the different protein sites that are exposed to the enzyme. The

same study showed that the attainable sequence coverage for cytochrome C adsorbed on

C18 beads can reach values as high as 89 %, but only after 2 h incubation time with C18

beads, or, after multiple protein passages through a C18-bead packed micro gel loader tip,

followed by 2h or 30 min proteolytic digestion, respectively.36 In the case of the

microreactor, the attainable sequence coverage for cytochrome was 43-57 % after 90 s of

proteolysis.

Protein name Amino acid coverage (%)

Unique peptides # AA

Overnight digest

Micro reactor

Overnight digest

Micro reactor

Hemo alpha 65 47-75 5 5 - 8 142 Hemo beta 47-59 56-61 5 - 7 8 - 9 145 Carbonic anhydrase 29-43 38-40 5 - 8 8 - 9 260 Alpha-2 HS-glycoprotein 17-26 11-15 3 - 6 3 - 5 359 Cytochrome c 17 44-57 1 - 2 8 - 9 105 Serum albumin 17-18 7-13 8 - 10 3 - 8 607 Alpha-S1-casein 12-26 28-41 2 - 4 6 - 9 214 Alpha-S2-casein 10-15 10-17 1 - 2 1 - 3 222 Beta-casein 8 15-18 1 3 224 Alpha-lactalbumin 7-19 0 1 - 2 0 142

Table 5.2 Comparison of results generated with the overnight and microreactor digestion protocols.

78

Figure 5.2 Venn diagram comparisons for unique peptide identifications in proteolytic digestion

reactions performed with the microreactor and overnight protocols. (A) Microreactor digestion (90 s)

with direct MS/MS analysis (1 µM sample, 10 µL loading); (B) Overnight digestion with tryptic peptide

loading on the microreactor and direct MS/MS analysis (1 µM sample, 10 µL loading) ; (C) Microreactor

digestion (3 min) with sample collection for nano-HPLC-MS/MS analysis (0.5 µM sample, 8 µL injection);

(D) Overnight digestion followed by nano-HPLC-MS/MS analysis (0.5 µM sample, 8 µL injection). See

Table 5.1 and experimental section for details. Note: for C and D, the count of unique peptides for each

experiment is the result of two combined nano-HPLC-MS/MS runs.

79

The reproducibility of peptide identifications in three replicate experiments was assessed

by comparing the microcolumn to overnight digestion by using two strategies: (a) direct

MS detection of peptides eluting from the microcolumn, with peptides being generated

either through digestion on the microcolumn, or overnight and loaded on the

microcolumn (as discussed above); and, (b) nano-HPLC-MS/MS analysis of peptides

generated and collected from the microcolumn or directly from overnight digestion. The

Venn diagrams in Figures 5.2A and 5.2B show that while a larger number of unique

peptides were identifiable from the microcolumn than the overnight enzymatic digestion

protocols (i.e., 83 vs. 50), the reproducibility of unique peptide identifications in 3

replicate runs dropped from 50 % to 35 % as changing from the overnight to the

microreactor digestion conditions. The larger number of peptides with missed cleavages

that contributed to a larger sequence coverage, also contributed to a somewhat lower

reproducibility in peptide identifications. For building the Venn diagrams for Figures

5.2C and 5.2D, the solution of tryptic peptides that eluted from the microcolumn had to

be diluted to provide for sufficient volume for an HPLC injection. The concentration of

all samples analyzed by HPLC was 0.5 µM and the sample consumed for analysis was 4

pmoles (assuming 100 % recovery from the digestion process). BSA was not part of this

protein mixture. To reduce the impact of peptides with missed K/R cleavages, proteolytic

digestion on the microcolumn was performed for 3 min. In addition, to avoid possible

bias induced by the data dependent data acquisition process, all samples were analyzed

twice by HPLC. The Venn diagrams from Figures 5.2C and 5.2D represent the overlap

of three replicates of two averages each, and confirmed a similar trend, 55 % vs. 39 %

overlap of peptide IDs with overnight vs. microreactor digestion. While these

80

experimental conditions resulted in a lower number of total peptide IDs (54 vs 83), a

better reproducibility, approaching that of the conventional protocol, was evident. The

percent of peptides with missed cleavages dropped, as well, from 53-59 % to 39-53 %.

Additional parameters that were evaluated for the microreactor included detection limit

and breakthrough volume (for conditions of sample loading at 2 µL/min, basic buffer, 1.5

min incubation with trypsin, and sample elution at 300 nL/min). Experiments performed

with a mixture of 9 proteins (no BSA) revealed that all proteins (except α-lactalbumin)

were detectable from 1 µM, 7 from 0.1 µM and 4 from 0.01 µM solutions, respectively.

Overall, this corresponded to a total sample consumption of 10, 1 and 0.1 pmoles in an

infusion experiment that eluted the great majority of peptides in a time-window of ~10-

15 min. It enlists the identifiable peptides for each protein. Sample breakthrough volumes

were determined with 1 µM protein mix solutions infused at 2 µL/min over microreactors

containing C18 packing of 5-12 mm in length, by connecting the outlet of the

microreactor to an Agilent HPLC-UV detector equipped with a microflow cell and

operated at 254 nm. The particle beds in this study were longer than the ones used for

proteolytic digestions (~2 mm) to minimize measurement errors that could be induced by

the variability in packing length and uniformity. After accounting for the dead volumes

associated with the transfer capillaries, sample breakthrough started to occur after ~ 7

min, 11 min and 13 min from microreactors containing 5 mm, 10 mm and 12 mm C18

packing, respectively. Breakthrough was considered to occur when the absorbance of the

solution emerging from the microreactor reached ~1 % of the absorption of the 1 µM

protein mix solution when infused directly through the UV detector without an inline

81

mounted proteolytic reactor. Therefore, we estimated that for short, 2 mm microreactors,

the optimal loading time of 1 µM solutions (total sample amount ~2 µg) should not

exceed 3-5 min, i.e., the time that maximizes the number of detectable peptides while

minimizing the protein losses due to breakthrough.

Microreactor for phosphopeptide enrichment. The performance of the TiO2 microreactor

was evaluated with peptides generated from the mixture of 10 proteins (1 µM each)

through an overnight digestion protocol. Among the 10 bovine proteins, only α-casein,

including subunits 1 and 2, β-casein and fetuin contained phosphorylated peptides.

Selective enrichment in phosphopeptides relies on their retention on TiO2 particles at

pH<3 (when the phosphate groups are still partially ionized while the acidic aspartic and

glutamic acid amino acids are not), and elution at pH>10 when buffer anions compete for

the Lewis acidic Ti4+ sites of TiO2 and liberate the phosphopeptides. The parameters that

were optimized for the phosphopeptide enrichment microreactor included the parameters

that affect selectivity, i.e., the composition and flow rate (1-2 µL/min) of the solutions

involved in sample loading and rinsing. The commercial TiO2 spin tips, that contain the

same type of particles as the microreactor, deliver selective enrichment in

phosphopeptides when conditioned properly in the presence of lactic acid (25 % in a

solution of CH3CN/H2O/TFA, 80:20:0.4 v/v). To minimize the interference of such a

high concentration of lactic acid with MS detection, the performance of the enrichment

process was evaluated at lower, 0.25 % and 2.5 %, lactic acid concentrations. While at

0.25 % the selectivity of commercial tips was deteriorated, at 2.5 % lactic acid

concentration the selectivity was the same as at 25 %. The spin tip enrichment in

82

phosphopeptides from a mixture of peptides generated through the overnight digestion of

a 10 protein mix, followed by nano-HPLC analysis with a 10 min gradient on a 10 cm x

100 µm ID capillary packed with C18 particles (5 µm), enabled the identification of all 4

phosphoproteins (α-casein S1, α-casein S2, β-casein and α-2HS-glycoprotein), typically

by 10-13 phosphopeptides variants corresponding to 6-8 unique peptide sequences

(Table 5.3). The selectivity in phosphopeptide enrichment was 100 %, with no non-

phosphorylated peptides detected. The same lactic acid concentration of 2.5 % was

further used for performing phosphopeptide enrichment with the microreactor. Two rinse

steps, one for removing the excess lactic acid, and one to facilitate the pH change to basic

conditions in preparation for phosphopeptide elution, were necessary for completing the

enrichment process. Long rinse times and high flow rates increased the selectivity in

phosphopeptide enrichment, but decreased the recovery rates, indicating that a proper

balance must be achieved to obtain both the desired selectivity and recovery rate. Lack of

adequate rinsing resulted in loss of selectivity. Two examples of optimized rinse-step

combinations that resulted in selective enrichment are provided in Table 5.3 (only

phosphopeptides, or 5-6 phosphopeptides out of a total of 6-7 peptides corresponding to

5-6 unique amino acid sequences, were detected). With few exceptions, the TiO2

microreactor enabled the identification of all 4 phosphoproteins with a selectivity of 95-

100 % in phosphopeptide enrichment.

83

Lactic acid

removal

H2O/CH3CN/TFA

20:80:0.4 v/v

pH-adjustment

H2O/CH3CN

50:50 v/v

Total

proteins

identified

P-

proteins

identified

Peptide

spectral

counts

P-peptide

spectral

counts

Total

unique

peptides

Unique

P-peptides

TiO2 Microreactor

2uL/min, 5min 1uL/min, 5 min 3-4 3-4 16-17 16-17 4-5 4-5 1uL/min, 5min 1uL/min, 5 min 4-5 3-4 22-23 21-22 6-7 5-6

TiO2 spin tip/ HPLC/MS

Centrifuging, 300 g

N/A 4 4 18-19 18-19 10-13 10-13

Table 5.3 Selectivity of the phosphopeptide enrichment process for the spin tip and microreactor.

Conditions: 10 protein mix overnight digest (1 µM), phosphopeptide enrichment in the presence of 2.5

lactic acid for both TiO2 spin tip and TiO2 microreactor.

Protein Overnight digestion/TiO2 spin tip/nano-HPLC-MS 8 unique peptide sequences 10-13 phosphopeptides per analysis 16 phosphopeptides per 3 replicates

Proteolytic digestion/TiO2 Microreactor 9 unique peptide sequences 7-10 phosphopeptides per analysis 14 phosphopeptides per 3 replicates

Alpha-S1-casein precursor DIGsESTEDQAMEDIK

Alpha-S1-casein precursor DIGSEsTEDQAMEDIK

Alpha-S1-casein precursor DIGsEsTEDQAMEDIK DIGsEsTEDQAMEDIK

Alpha-S1-casein precursor KYKVPQLEIVPNsAEER

Alpha-S1-casein precursor VNELsKDIGsEsTEDQAMEDIK

Alpha-S1-casein precursor VPQLEIVPNsAEER VPQLEIVPNsAEER

Alpha-S1-casein precursor YKVPQLEIVPNsAEER YKVPQLEIVPNsAEER

Alpha-S2-casein precursor TVDMEsTEVFTK TVDMEsTEVFTK

Alpha-S2-casein precursor TVDMEsTEVFTKK

Alpha-S2-casein precursor TVDMEStEVFTK

Alpha-S2-casein precursor KTVDMEsTEVFTK

Beta-casein precursor FQsEEQQQTEDELQDK FQsEEQQQTEDELQDK

Beta-casein precursor FQsEEQQQtEDELQDK

Beta-casein precursor IEKFQsEEQQQTEDELQDK

Alpha-2-HS-glycoprotein HTFSGVAsVESSSGEAFHVGK HTFSGVAsVESSSGEAFHVGK

Alpha-2-HS-glycoprotein HTFSGVASVEsSSGEAFHVGK

Alpha-2-HS-glycoprotein HTFSGVAsVEsSSGEAFHVGK HTFSGVAsVEsSSGEAFHVGK

Alpha-2-HS-glycoprotein HTFSGVASVESsSGEAFHVGK HTFSGVASVESsSGEAFHVGK

Alpha-2-HS-glycoprotein HTFSGVASVEssSGEAFHVGK

Alpha-2-HS-glycoprotein HTFSGVAsVEssSGEAFHVGK HTFSGVAsVEssSGEAFHVGK

Alpha-2-HS-glycoprotein HTFSGVASVESssGEAFHVGK

Table 5.4 Phosphopeptide sequences identified in a mixture of bovine proteins by using conventional

approach and the microreactors. Note: (a) combined results of 3 replicate analyses are shown; (b)

phosphorylated amino acid residues are shown in lower case letters.

84

Figure 5.3 Full mass scans acquired during the analysis of standard protein mixture digests. (A)

Capillary infusion of the whole protein mixture digest without phosphopeptide enrichment; (B) Capillary

infusion of the whole protein mixture digest after TiO2 spin tip phosphopeptide enrichment; (C) Infusion of

phosphopeptides enriched on the microreactor. Conditions: sample concentration 1 µM, infusion at 300

nL/min, MS data acquisition performed using data dependent analysis with all peptide identifications being

confirmed by tandem MS. See experimental section for details. Note: not all identifiable phosphopeptides

are shown in the mass spectra.

85

Ultimately, the optimal conditions for the operation of the microreactor included the use

of lactic acid at concentration of 2.5 % and two rinse-steps performed at 1-2 μL/min for 5

min. Before sample loading, the TiO2 microreactor was preconditioned with an

acidic/high organic content solution (CH3CN/H2O/TFA 80:20:0.4 v/v), and then with the

same solution containing also lactic acid (acidic buffer/lactic acid 97.5/2.5 v/v) at 2

μL/min for 5 min for each solution (Table 5.1). The protein mix digest was pumped next

through the microreactor at a flow rate of 1 µL/min for 10 min (a low flow rate was

chosen to facilitate the retention of phosphopeptides). The amount of protein loaded

through infusion on a microreactor with a 2 mm TiO2 bed was ~2 µg, i.e., at the same

level that was determined for the C18 microreactors, and at a sample/TiO2 packing load

recommended by the manufacturer for the spin tips. Following sample loading, the

microreactor was rinsed again with the acidic/organic solution and then just with an

organic solution (CH3CN/H2O 50:50 v/v) at 1-2 μL/min, for 5 min, to remove the lactic

acid and neutralize the environment, respectively. The enriched phosphopeptides were

eventually eluted using a basic/high organic content solution (CH3CN/H2O/NH4OH

50:40:10 v/v/v) at a flow rate of 300 nL/min for 50 min. As observed in Figure 5.3, both

conventional and microreactor enrichment protocols were highly selective. A full MS

scan for a sample without any enrichment identified only two phosphopeptides,

YKVPQLEIVPNsAEER and FQsEEQQQTEDELQDK, with low abundance in the mass

spectrum (Figure 5.3A). Figures 5.3B and 5.3C show the MS scans for the same protein

mix digest enriched in phosphopeptides by using a commercial TiO2 spin tip and the

microreactor, respectively. The majority of the intense ions in both MS scans could be

assigned to phosphopeptides. Peptide sequences and IDs were confirmed by tandem MS.

86

The back-pressure of both proteolytic and TiO2 enrichment microreactors was negligible.

When mounted on the outlet of an Agilent 1100 micro-HPLC pump, operated with

solvent A, the measured change in backpressure was only 3-4 bar for the C18 and 5-6 bar

for C18/TiO2/C18 microreactors, indicating that the use of simple, low-cost infusion

pumps and low-pressure connectors is adequate for the implementation of such

experiments in any bioanalytical laboratory. In terms of reusability, after thorough

flushing with high organic content solvents, the C18 microreactors were reusable when

the protein concentrations were ~1 µM. Carryover that would affect the measurements of

low protein concentration (<1 µM) was, however, observed. The TiO2 particles were not

reusable. Given the ease of preparation, the use of new microreactors for each experiment

is, therefore, recommended. As both types of particles are broadly used in

chromatographic separations and enrichment in phosphopeptides, respectively, and as

chemical modifications to the particles were not performed, problems related to stability

were not observed. Unspecific binding to the TiO2 particles was minimized with the

addition of the lactic acid prior and during sample loading, as described above.

Microreactor for proteolytic digestion and phosphopeptide enrichment. The rationale for

designing a microreactor for performing both proteolytic digestion and phosphopeptide

enrichment relies on the ability to elute peptides from hydrophobic media in acidic/high-

organic content solvents, and selectively retain only phosphopeptides in the same solution

on TiO2 particles. Using conditions that were optimized in the previous experiments, a

microreactor containing a C18 filter (1 mm), TiO2 bed for phosphopeptide enrichment (1-

87

2 mm) and C18/silica particles for proteolytic digestion (2 mm) was prepared and tested

with the same 10 standard protein mix (1 µM each). The sample was loaded on the

microreactor at 2 μL/min for 5 min, retained on the 2 mm C18 bed, and subjected to

proteolysis with a trypsin solution (5 μM) infused at 2 µL/min for 1.5 min. Following a

rinse-step (H2O/CH3CN/TFA 98:2:0.4 v/v, 2 µL/min, 5 min) for quenching the enzymatic

reaction, the peptides were eluted from the C18 onto the TiO2 microreactor with a

solution of (H2O/CH3CN/TFA 20:80:0.4)/lactic acid, 97.5:2.5 v/v (see Table 5.1). After

additional rinse-steps to remove the lactic acid and facilitate the change to high pH, the

sample was eluted with a solution of CH3CN/H2O/NH4OH 50:40:10 v/v, at 300 nL/min,

during a time-window of 25-50 min. Higher flow rates for the rinse-steps enabled a faster

switch to high pH eluting conditions and faster recovery of peptides from the

microreactor. MS analysis was performed under high pH conditions, which enables

efficient (+) ESI-MS detection of peptides due to the fact that roughly half the peptides

that result from a tryptic digestion process have a pI-value>8. The interference of

background ions is minimized at high pH, the overall outcome resulting in improved

signal/noise ratios.37,38

A comparison of the performance of the microdigestion/phosphopeptide enrichment

microreactor with that of a control experiment that used a conventional protocol

involving overnight protein digestion and phosphopeptide enrichment with TiO2 spin tips,

followed by nano-HPLC-MS/MS (10 min gradient), is provided in Figure 5.4. In both

experiments, only phosphopeptides were detected, indicating that a high selectivity,

approaching 100 % is achievable with the proposed microreactor. Nevertheless, the time

88

required for performing the analysis with the conventional protocol was roughly 15 h (12

h overnight digestion, 2 h phosphoenrichment, 1 h HPLC analysis-including column

equilibration and rinse-steps), while with the microreactor was only ~1.5 h. Table 5.4

enlists the phosphopeptide sequences identified by the two approaches, i.e., on-column

digestion/enrichment/infusion-MS and overnight digestion/spin tip enrichment/nano-

HPLC/MS analysis. Microreactor processing returned 7-10 phosphorylated variants of 9

unique peptide amino acid sequences per analysis, while conventional processing

returned 10-13 phosphorylated variants of 8 sequences. Table 5.3 provides an alignment

of the peptide identifications. Combined from three replicate analyses, 14 and 16

phosphorylated peptides were identifiable with the microreactor and conventional

protocols, respectively (Table 5.4), each method displaying a few unique sequences not

identified by the other. The peptide IDs were compared with what was previously

reported by Larsen et al.39 All peptides but one from α-S2, corresponding to six unique

sequences of α-S1, α-S2 and β-casein peptides, carrying one or two phosphorylation

groups were detectable with the microreactor setup, but none carrying 3-4

phosphorylation groups. The use of a different ionization method (ESI vs. MALDI) and a

mass spectrometer with a more limited mass range (ion trap vs. TOF) were the reason for

this outcome. Important also to note is that in the microreactor experiments, ESI-MS was

performed directly from the TiO2 elution buffer, at pH conditions that do not favor the

detection of multiple phosphorylated peptides (pH>10.5). The use of a different

enrichment protocol (lactic acid vs. 2,5-dihydroxybenzoic acid for improving selectivity)

could have been an additional contributor. Nonetheless, one β-casein (3+) peptide

carrying 4 phosphorylation sites, that was not identified by the Discoverer software

89

package, produced a tandem mass spectrum that enabled manual verification of 3

phosphate group losses and some additional fragment ions (Figures 5.2B and 5.2C).

Figure 5.4 Comparison of the performance of the microreactor and conventional approach. Note:

three replicates were conducted for each experiment. See Table 5.1 and experimental section for details.

90

Analysis of cell extracts. Without sample prefractionation and/or HPLC separation steps,

the proposed microreactor functions best for the analysis of simple mixtures of proteins,

as shown above, or of isolated proteins from cell extracts. To test, however, its capacity

for handling samples of biological origin, a cell extract obtained from SKBR3 breast

cancer cells was analyzed. Typically, such extracts prepared by denaturation/overnight

digestion and analyzed by a 4 h long nano-HPLC-MS/MS gradient yielded 700-1000

protein IDs. The length of the gradient was essential, a short 10 min gradient reducing the

numbers to only 65-70 protein IDs matched by 140-150 unique peptides. Without

denaturation, the number of identified proteins did not change, but the matching peptides

dropped by ~15 %. The proteolytic digestion of such a cell extract with the microreactor

followed by direct MS analysis yielded results that were commensurate with the fast

nano-HPLC gradients, i.e., 38-44 protein IDs, but a lower number of matching peptides,

i.e., 47-55. The microreactor contained a 5 mm long bed of C18 particles for enabling the

loading of larger sample amounts (10-20 µg), and peptide elution occurred in three steps

with progressively larger percentages of CH3CN in the eluent (10 %, 25 % and 50 %).

Table 5.4 enlists the proteins identified in three microreactor digestion/direct infusion

MS and three overnight digestion/with and without denaturation/nano-HPLC-MS/MS (10

min gradient) analyses. Mainly abundant proteins could be identified (keratins, actins,

GAPDH, fatty acid synthase, tubulin, etc.), but also a few proteins of interest such as

nucleophosmin and histone proteins. A cytoplasmic SKBR3 cell extract spiked with

bovine standards was also loaded on a digestion/phosphoenrichment microreactor,

returning 11 protein IDs, matched by 12 peptides of which 9 were phosphopeptides,

preponderantly from the bovine casein proteins (75 % selectivity). While additional work

91

is necessary for optimizing the microreactor for a specific biological application,

altogether, the preliminary results indicate the ability of the microreactor to handle

semicomplex samples and expand its potential utility for the analysis of affinity pulldown

biological experiments.

5.4 Conclusions

A microreactor that enables fast proteolytic digestion and selective enrichment in

phosphopeptides, followed by on-line ESI-MS detection, was described in this work.

Compared to conventional sample preparation protocols, the microreactor has multiple

advantages, including: (a) enables a novel, inversed strategy for performing enzymatic

digestion of proteins, i.e., with the sample adsorbed on the surface of C18 particles and

with a high concentration trypsin solution flowing over the adsorbed substrate; (b)

obviates the need for high-cost immobilized enzyme particles, while simultaneously

eliminating problems associated with altered activity of the enzyme due to covalent

attachment; (c) warrants fast, straightforward and cost-effective fabrication requiring only

common reagents, commercial C18 and TiO2 particles, and a simple particle loading

process in a fused silica capillary; overall preparation of the microreactor can be

accomplished in a few min and can be implemented in any bioanalytical laboratory; (d)

facilitates high performance enzymatic digestion of proteins in <3 min, and selective

enrichment and identification of phosphopeptides; the overall sample analysis process

can be completed in 30-90 min, i.e., 10-15 faster than commonly used bench-top

protocols that require 1-2 days for completion; (e) minimizes concerns related to the

generation of autolytic enzyme fragments due to the short analysis times; (f) minimizes

92

concerns related to the presence of high concentration enzymes in solution as a result of

continuous removal of the enzyme from the system; and (g) ratifies broad applicability

for the analysis of simple protein mixtures and potential for integration in more

sophisticated, multiplexed workflows. To the best of our knowledge, such a combined

proteolytic digestion/phosphopeptide enrichment microreactor is reported for the first

time. It is envisioned that the integration of the microreactor with high-performance

chromatographic separations, and/or within microfluidic devices with high-throughput

and multifaceted functionality will facilitate the fast analysis of complex samples of

biological origin.

Acknowledgments. This work was supported by the National Science Foundation, award

DBI-1255991 to IML. The authors thank Professor Daniel Capelluto from Virginia Tech

for providing the PKHF2 recombinant protein used for testing the effectiveness of the

microreactor.

5.5 References

1. Kettenbach, A.N., Gerber, S.A.: Rapid and reproducible single-stage phosphopeptide

enrichment of complex peptide mixtures: application to general and phosphotyrosine-

specific phosphoproteomics experiments. Anal. Chem. 83, 7635-7644 (2011)

2. Dunn, J.D., Reid, G.E., Bruening, M.L.: Techniques for phosphopeptide enrichment

prior to analysis by mass spectrometry. Mass Spectrom. Rev. 29, 29-54 (2010)

3. Fíla, J., Honys, D.: Enrichment techniques employed in phosphoproteomics. Amino

Acids. 43, 1025-1047 (2012)

93

4. Switzar, L., Giera, M., Niessen, W.M.: Protein digestion: an overview of the available

techniques and recent developments. J. Proteome Res. 12, 1067-1077 (2013)

5. Capelo, J., Carreira, R., Diniz, M., Fernandes, L., Galesio, M., Lodeiro, C., et al.:

Overview on modern approaches to speed up protein identification workflows relying

on enzymatic cleavage and mass spectrometry-based techniques. Anal. Chim. Acta.

650, 151-159 (2009)

6. Kuhnert, N.: Microwave‐assisted reactions in organic synthesis—Are there any

nonthermal microwave effects? Angew. Chem. Int. Ed. 41, 1863-1866 (2002)

7. Pramanik, B.N., Mirza, U.A., Hain, Y., Liu, Y.H., Bartner, P.L., Weber, P.C., Bose,

A.K.: Microwave-enhanced enzyme reaction for protein mapping by mass

spectrometry: A new approach to protein digestion in minutes. Protein Sci. 11, 2676-

2687 (2002)

8. Turapov, O.A., Mukamolova, G.V., Bottrill, A.R., Pangburn, M.K.: Digestion of

native proteins for proteomics using a thermocycler. Anal. Chem. 80, 6093-6099

(2008)

9. Lopez-Ferrer, D., Capelo, J., Vazquez, J.: Ultra fast trypsin digestion of proteins by

high intensity focused ultrasound. J. Proteome Res. 4, 1569-1574 (2005)

10. López-Ferrer, D., Petritis, K., Hixson, K.K., Heibeck, T.H., Moore, R.J., Belov, M.E.,

Camp, D.G., Smith, R.D.: Application of pressurized solvents for ultrafast trypsin

hydrolysis in proteomics: proteomics on the fly. J. Proteome Res. 7, 3276-3281 (2008)

11. Wang, S., Bao, H., Liu, T., Zhang, L., Yang, P., Chen, G.: Accelerated proteolysis in

alternating electric fields for peptide mapping. Rapid. Commun. Mass Spec. 22,

3225-3232 (2008)

94

12. Wang, S., Zhang, L., Yang, P., Chen, G.: Infrared-assisted tryptic proteolysis for

peptide mapping. Proteomics. 8, 2579-2582 (2008)

13. Asanomi, Y., Yamaguchi, H., Miyazaki, M., Maeda, H.: Enzyme-immobilized

microfluidic process reactors. Molecules. 16, 6041-6059 (2011)

14. Le Nel, A., Minc, N., Smadja, C., Slovakova, M., Bilkova, Z., Peyrin, J.-M., Viovy,

J.-L., Taverna, M.: Controlled proteolysis of normal and pathological prion protein in

a microfluidic chip. Lab Chip 8, 294-301 (2008)

15. Al-Lawati, H., Watts, P., Welham, K.J.: Efficient protein digestion with peptide

separation in a micro-device interfaced to electrospray mass spectrometry. Analyst.

131, 656-663 (2006)

16. Krenkova, J., Lacher, N.A., Svec, F.: Highly efficient enzyme reactors containing

trypsin and endoproteinase LysC immobilized on porous polymer monolith coupled

to MS suitable for analysis of antibodies. Anal. Chem. 81, 2004-2012 (2009)

17. Jin, L.J., Ferrance, J., Sanders, J.C., Landers, J.P.: A microchip-based proteolytic

digestion system driven by electroosmotic pumping. Lab Chip 3, 11-18 (2003)

18. Kim, J., Kim, B.C., Lopez‐Ferrer, D., Petritis, K., Smith, R.D.: Nanobiocatalysis for

protein digestion in proteomic analysis. Proteomics. 10, 687-699 (2010)

19. Fan, J., Shui, W., Yang, P., Wang, X., Xu, Y., Wang, H., Chen, X., Zhao, D.:

Mesoporous silica nanoreactors for highly efficient proteolysis. Chem. Eur. J. 11,

5391-5396 (2005)

20. Zuo, C., Yu, W., Zhou, X., Zhao, D., Yang, P.: Highly efficient enrichment and

subsequent digestion of proteins in the mesoporous molecular sieve silicate SBA‐15

for matrix‐assisted laser desorption/ionization mass spectrometry with time-of-

95

flight/time-of-flight analyzer peptide mapping. Rapid. Commun. Mass Spec. 20,

3139-3144 (2006)

21. Shui, W., Fan, J., Yang, P., Liu, C., Zhai, J., Lei, J., Yan, Y., Zhao, D., Chen, X.:

Nanopore-based proteolytic reactor for sensitive and comprehensive proteomic

analyses. Anal. Chem. 78, 4811-4819 (2006)

22. Wu, H., Tian, Y., Liu, B., Lu, H., Wang, X., Zhai, J., Jin, H., Yang, P., Xu, Y., Wang,

H.: Titania and alumina sol-gel-derived microfluidics enzymatic-reactors for peptide

mapping: design, characterization, and performance. J. Proteome Res. 3, 1201-1209

(2004)

23. Liu, Y., Xue, Y., Ji, J., Chen, X., Kong, J., Yang, P., Girault, H.H., Liu, B.: Gold

nanoparticle assembly microfluidic reactor for efficient on-line proteolysis. Mol. Cell.

Proteomics 6, 1428-1436 (2007)

24. Ji, J., Zhang, Y., Zhou, X., Kong, J., Tang, Y., Liu, B.: Enhanced protein digestion

through the confinement of nanozeolite-assembled microchip reactors. Anal. Chem.

80, 2457-2463 (2008)

25. Wu, H., Zhai, J., Tian, Y., Lu, H., Wang, X., Jia, W., Liu, B., Yang, P., Xu, Y., Wang,

H.: Microfluidic enzymatic-reactors for peptide mapping: strategy, characterization,

and performance. Lab Chip 4, 588-597 (2004)

26. Hustoft, H.K., Brandtzaeg, O.K., Rogeberg, M., Misaghian, D., Torsetnes, S.B.,

Greibrokk, T., Reubsaet, L., Wilson, S.R., Lundanes, E.: Integrated enzyme reactor

and high resolving chromatography in "sub-chip" dimensions for sensitive protein

mass spectrometry. Sci. Rep. 3, 3511 (2013)

27. Kim, J.-H., Inerowicz, D., Hedrick, V., Regnier, F.: Integrated sample preparation

96

methodology for proteomics: Analysis of native proteins. Anal. Chem. 85, 8039-8045

(2013)

28. Cantin, G.T., Shock, T.R., Park, S.K., Madhani, H.D., Yates, J.R.: Optimizing TiO2-

based phosphopeptide enrichment for automated multidimensional liquid

chromatography coupled to tandem mass spectrometry. Anal. Chem. 79, 4666-4673

(2007)

29. Mohammed, S., Kraiczek, K., Pinkse, M.W., Lemeer, S., Benschop, J.J., Heck, A.J.:

Chip-Based Enrichment and NanoLC-MS/MS Analysis of Phosphopeptides from

Whole Lysates. J. Proteome Res. 7, 1565-1571 (2008)

30. Raijmakers, R., Kraiczek, K., de Jong, A.P., Mohammed, S., Heck, A.J.: Exploring

the Human Leukocyte Phosphoproteome Using a Microfluidic Reversed-Phase-TiO2-

Reversed-Phase High-Performance Liquid Chromatography Phosphochip Coupled to

a Quadrupole Time-of-Flight Mass Spectrometer. Anal. Chem. 82, 824-832 (2010)

31. Tenga, M.J., Lazar, I.M.: Proteomic snapshot of breast cancer cell cycle: G1/S

transition point. Proteomics 13, 48-60 (2013)

32. Sarvaiya, H.A., Yoon, J.H., Lazar, I.M.: Proteome profile of the MCF7 cancer cell

line: a mass spectrometric evaluation. Rapid. Commun. Mass Spec. 20, 3039-3055

(2006)

33. Klammer, A.A., MacCoss, M.J.: Effects of modified digestion schemes on the

identification of proteins from complex mixtures. J. Proteome Res. 5, 695-700 (2006)

34. Kamau, S.M., Cheison, S.C., Chen, W., Liu, X.M., Lu, R.R.: Alpha-Lactalbumin: Its

Production Technologies and Bioactive Peptides. Compr. Rev. Food Sci. Food Safety

9, 197-212 (2010)

97

35. Almaas, H., Berner, V., Holm, H., Langsrud, T., Vegarud, G.: Degradation of whey

from caprine milk by human proteolytic enzymes, and the resulting antibacterial

effect against Listeria monocytogenes. Small Rumin. Res. 79, 11-15 (2008)

36. Doucette, A., Craft, D., Li., L.: Mass Spectrometric Study of the Effects of

Hydrophobic Surface Chemistry and Morphology on the Digestion of Surface-Bound

Proteins. J. Am. Soc. Mass Spectrom. 14, 203-214 (2003)

37. Mansoori, B.A., Volmer, D.A., Boyd, R.K.: "Wrong-way-round" Electrospray

Ionization of Amino Acids. Rapid. Commun. Mass Spec. 11, 1120-1130 (1997)

38. Lazar, I.M., Ramsey, R.S., Ramsey, J.M.: On-chip proteolytic digestion and analysis

using “wrong-way-round” electrospray time-of-flight mass spectrometry. Anal. Chem.

73, 1733-1739 (2001)

39. Larsen, M.R., Thingholm, T.E., Jensen, O.E., Roepstorff, P., Jorgensen, T.J.D.: Highly

selective enrichment of phosphorylated peptides from peptide mixtures using titanium

dioxide microcolumns. Mol. Cell. Proteomics 4, 873-886 (2005)

98

CHAPTER 6: Streamlined Microfluidic Analysis of Phosphopeptides Using Stable

Isotope-Labeled Synthetic Peptides and MRM-MS Detection

Jingren Deng, Fumio Ikenishi, Nicole Smith and Iulia M. Lazar*

Department of Biological Sciences, Virginia Tech

1981 Kraft Drive, Blacksburg, VA 24061, USA

*Corresponding author: malazar@vt.edu

Author contributions

Designed the experiments: JD IML

Performed the experiments: JD FI NS

Analyzed the data: JD IML

Wrote the paper: JD IML

Conceived and coordinated the study: IML

Key words: microfluidics, mass spectrometry, multiple reaction monitoring,

phosphopeptides, fast analysis

99

Abstract

The detection and analysis of phosphopeptides from complex biological samples

represents a challenge for proteomics studies, mainly because of the dynamic nature of

protein phosphorylation and sub-stoichiometric occupancy of phosphorylation sites. In an

attempt to address these issues, a number of miniaturized devices and approaches have

been developed in recent years. The proposed methods typically rely on the use of

complicated and time-consuming operations. In this work, we describe the development

of an integrated microfluidic chip that can perform targeted, quantitative analysis of

phosphopeptides involved in cancer-relevant signaling pathways, by using multiple

reaction monitoring (MRM)-mass spectrometry (MS) detection. The microfluidic chip

comprises microreactors packed with C18 and TiO2 particles that enable on-chip solid

phase extraction (SPE), phosphopeptide enrichment, and electrospray ionization (ESI).

The newly developed chips were demonstrated for the detection of three phosphopeptides

involved in ERBB2 signaling pathways. The peptides were selected from the outcome of

a proteomic study involving EGF stimulation of SKBR3/Her2+ breast cancer cells.

Stable isotope-labeled phosphopeptides were spiked into the SKBR3 protein extracts to

enable absolute quantification of the target peptides. A number of experimental

conditions were evaluated and optimized to increase the sensitivity of phosphopeptide

detection. The data demonstrate that the proposed microfluidic strategy can be used for

the MS quantification of phosphopeptides in the low nM range from cell lysates without

any prior sample pretreatment, fractionation or bioaffinity enrichment, and is generally

applicable to the analysis of any phosphopeptide targets.

100

6.1 Introduction

Phosphorylation is one of the most studied protein post-translational modifications that

plays essential roles in supporting signal transduction processes within a cell. Identifying

and quantifying the presence of phosphorylation in a protein is, however, very

challenging. Protein phosphorylation is transient, reversible, and often occurs at sub-

stoichiometry levels. A combination of efficient enrichment strategies and sensitive

detection methods, such as MS, are required for completing the task. Advances in

microfluidics over the past two decades have led to the development of many promising

platforms for bioanalytical applications, as well as for the study of a variety of biological

systems and processes. Comprehensive reviews describe the progress in detail.1-7 As a

critical step in phosphoproteomic experiments, phosphopeptide enrichment has drawn

much attention.8-20 Although a variety of bench-top technologies that can be used alone or

in combination have been advanced, including immobilized metal affinity

chromatography (IMAC), metal oxide affinity chromatography (MOAC), strong cation

exchange chromatography (SCX), and immunopurification, the vast majority of chip-

based strategies rely on the use TiO2 as an enrichment medium. For example, Min et al.

reported a gravity-driven microchip equipped with a tunable TiO2 nanotube array.10 The

chip was demonstrated for on-chip enrichment and isotope labeling of serum

phosphopeptides originating from healthy and cancer patients. Heck and coworkers

developed a phosphopeptide enrichment strategy relying on SCX pre-fractionation of cell

extracts in 24-32 fractions, and the use of the Agilent commercial chip connected to a

benchtop HPLC system for fluid delivery and post-enrichment nano-LC separations.11-13

Various pre- and post-TiO2 phosphopeptide trapping configurations were evaluated, the

101

optimized system enabling the detection of ~100 phosphopeptides from one SCX

fraction, with an upfront consumption of up to 1 mg cell extract. TiO2 was also deposited

as a film by flame aerosol deposition,14 sputtering, sputtering/annealing, and liquid

deposition,15 and demonstrated in these forms for the enrichment of standard

phosphopeptides.

Discovery proteomics is a powerful technology capable of reliably identifying a large

number of proteins (103 - 104) without any prerequisite information, but often comes at

the cost of low throughput and reduced speed of analysis. Large, MS-generated

phosphoproteomic maps, encompassing thousands of phosphoproteins and phosphosites,

are typically produced from mg-levels of cellular extracts by using protocols that involve

dozens of sample fractionation and preparation steps (SCX, IMAC, affinity enrichment)

that extend over hours or even days of work21. With the recent expanse of public data

repositories (PeptideAtlas22, GPM Proteomics Database23, PRIDE24, SRMAtlas25),

targeted MS methods, such as MRM with stable isotope-labeled internal standards26,

have become the method of choice for the quantitative monitoring of molecular species

of interest. While the need for extensive sample preparation continues to be a challenge,

the characteristic MRM-MS benefits of sensitivity and selectivity lead to shorter analysis

times and increased throughput, features that have been all exploited in cell signaling

research. For example, by using IMAC enrichment from 200 µg protein cell lysate and

nano-LC-MRM-MS detection, Kennedy et al. reported the quantification of 107

phosphopeptides involved in DNA damage response27. Zawadzka et al. employed an

MRM-based analytical workflow to assess changes in the phosphorylation of human

102

plasma proteins, and quantified 24 proteins of biological relevance to cellular

homeostasis, metabolism, immune response and signaling.28

To streamline the analysis of signaling processes in cells, in this study, we evaluated the

ability to detect phosphopeptides of interest, in a targeted mode of detection, from only

10 µg of protein cell extract. We developed and optimized a microfluidic approach that

enables enrichment and quantification of phosphopeptides from cell extract digests,

without any sample fractionation or depletion. Three phosphopeptides from three proteins

involved in MAPK/ERBB2 signaling pathways, targeted by 18 MRM transitions, were

used to demonstrate the feasibility of the approach. To our knowledge, this is the first

effort using microfluidic chips for on-line enrichment and direct MRM-MS detection for

conducting the quantification of phosphopeptides from cell extracts. The approach is

applicable to the analysis of any phosphopeptides, and has potential utility for studies that

aim at interrogating the dynamics of phosphorylation in signaling networks.

6.2 Material and methods

Chemicals and materials. McCoy’s 5A cell culture medium was purchased from Life

Technologies (Carlsbad, CA), fetal bovine serum (FBS) from Gemini-Bio Products (West

Sacramento, CA), human EGF from PeproTech (Rocky Hill, NJ), and Normocin from

InvivoGen (San Diego, CA). SKBR3 cells, trypsin/EDTA and phosphate buffered saline

(PBS) were from ATCC (Manassas, VA). Protease inhibitor solution, phosphatase

inhibitor cocktails 2 and 3, bovine α-casein, DTT, pyrrolidine, urea, Trizma base and

hdrochloride, acetic acid, trifluoroacetic acid (TFA), ammonium bicarbonate, sodium

103

chloride and hydrogen peroxide (35 %) were purchased from Sigma-Aldrich (St. Louis,

MO). Sequencing grade modified trypsin was obtained from Promega Corporation

(Madison, WI). SPEC-PT-C18 solid phase extraction pipette tips and Zorbax SB-C18/5

μm particles were purchased from Agilent Technologies (Santa Clara, CA). Phos-TiO (3

mg) phosphopeptide enrichment tips, Titansphere Phos-TiO bulk TiO2 particles (10 μm),

GL-SDB and GL-GC cleanup tips, and lactic acid were from GL Sciences (Torrance,

CA). Buffered oxide etch (BOE) and chrome etchant were purchased from Transene Co.

(Danvers, MA) and Microposit MF-319 developer from Rohm and Haas (Marlborough,

MA). Sulfuric acid (10N) was acquired from Mallinkrodt (St. Louis, MO) and

ammonium hydroxide 28-30 % from Spectrum Chemical (New Brunswick, NJ). HPLC-

grade acetonitrile, methanol and isopropanol were purchased from Fisher Scientific (Fair

Lawn, NJ), acetone from Acros (New Jersey), and DI water was generated in-house with

a MilliQ Ultrapure water system (Millipore, Bedford, MA). Heavy forms of selected

phosphopeptides were synthesized by New England Peptide (Gardner, MA), with a +7

Da stable isotope-label (13C, 15N) incorporated at Leu.

Sample preparation. SKBR3 cells were cultured in McCoy’s 5A medium supplemented

with 10 % FBS, arrested in serum-free medium for 24 h, and stimulated with FBS and

EGF (20 ng/mL for 1 min, 5 min, 15 min, 30 min, 60 min and 24 h) by incubation at 37

°C with 5 % CO2 atmosphere. Normocin (100 µg/mL) was added to all cell culture media

to prevent bacterial, fungal and mycoplasma contamination. The cells were harvested

when reached ~90 % confluence by trypsinization and stored at -80 °C. For protein

extraction, the frozen cells were thawed and lysed by sonication in a lysis buffer solution

104

containing Tris (50 mM, pH~8), NaCl (75 mM), and urea (8M), that was supplemented

with DTT (1 mM), protease inhibitors (lysis buffer:protease inhibitor solution 100:1 v/v),

and phosphatase inhibitor cocktails 2 and 3 (lysis buffer:phosphatase inhibitor solution

50:1 v/v). The cells were sonicated with the lysis buffer (packed cell volume:lysis buffer

1:5 v/v) for 5 min in ice-cold water, with intermittent interruptions to avoid solution

heating. The cell extract was centrifuged for 5 min at 16,000 g, the supernatant was

collected, and the protein concentration was measured with a SmartSpec Plus

spectrophotometer and the Bradford assay (Bio-Rad, Hercules, CA). The cell extracts

were denatured for 1 h at 50-60 °C, diluted 1:10 v/v with NH4HCO3 (50 mM), digested

with sequencing grade trypsin (substrate:enzyme ratio 50:1 w/w) at 37 °C overnight,

quenched with glacial CH3COOH (1 %), and stored at -80 °C until further use. For

phosphopeptide enrichment on the chip, the cell extracts were subjected to C18 cleanup

using SPEC-PT-C18 solid phase extraction pipette tips, and prepared for analysis by

dissolution in a solution of H2O/CH3CN/lactic acid/TFA (18:72:10:0.4 v/v) at a

concentration of 1 μg/μL. For global nano-LC-MS phosphoproteomic analysis, 500 µg of

cell extract was desalted with SPEC-PT-C18 pipette tips, enriched in phosphopeptides

according to the manufacturer’s protocol using Phos-TiO tips (3 mg), cleaned up with

SDB and GC tips, and dissolved in H2O/CH3CN/TFA (98:2:0.01 v/v) at a concentration

corresponding to 4 μg/μL initial whole protein extract.

MS analysis and data processing. The samples were analyzed using LTQ or LTQ-XL

mass spectrometers purchased from Thermo Electron (San Jose, CA). The MS was

operated in positive-ion mode with a capillary electrospray ionization (ESI) voltage of ~2

105

kV, and the data were acquired over a mass range of 500-2000 m/z. For chip infusion

ESI-MS experiments performed on the LTQ instrument, the phosphopeptide samples and

the eluents were delivered at a flow rate of 300 nL/min with a syringe pump purchased

from Harvard Apparatus (Holliston, MA). The samples were analyzed via a data-

dependent method by performing zoom/MS2 scans on the top 10 most intense peaks in

each MS scan (5 microscans averaged). For MRM-MS analysis, each precursor ion was

monitored by 5-7 fragment ions (transitions). The m/z ratio tolerances were set to +/-1

and +/-0.6 for the precursor and fragment ions, respectively. The MRM-MS data were

processed and analyzed using the open source software Skyline.29

For global phosphoproteomic analysis of SKBR3 cells performed on the LTQ-XL

instrument, an Agilent 1100 micro-LC separation system was used for performing the

separations. Samples (8-16 µL) were analyzed using an in-house prepared nano-

separation column (100 μm i.d. × 360 μm o.d x 10-12 cm length). The nano-LC flow rate

was generated by splitting the LC pump flow from 10 µL/min to 180-200 nL/min,30 with

a 4 hour concentration gradient of H2O/CH3CN/TFA from 96:4:0.01 v/v (solvent A) to

10:90:0.01 v/v (solvent B). The samples were analyzed via a data-dependent neutral loss

method by performing MS2 scans on the top 5 most intense peaks from each MS scan (5

microscans averaged), and triggering MS3 scans when a neutral loss (24.5 Da, 32.7 Da,

49.0 Da and 98.0 Da, neutral loss mass width +/-1.5 Da) was detected within the top 3

most intense MS2 ion fragments. The ESI voltage was 2 kV, and the data were acquired

over a mass range of 500-2000 m/z. The isolation width, normalized collision energy,

activation Q, and activation time were set at 3 m/z, 35 % (LTQ)/21 % (LTQ-XL), 0.25,

106

and 30 ms, respectively, for inducing collision induced dissociation (CID). The raw data

files were searched against a minimally redundant Homo sapiens protein database from

UniProt using the Discoverer 1.4 software package (Thermo Electron). Minimum to

maximum precursor ion mass range was set to 500-5000 Da, and maximum mass

tolerance for precursor and product ions was 2 Da and 1 Da, respectively.

Minimum/maximum peptide length was 6/144 amino acids, and up to 2 missed cleavages

for fully tryptic fragments were allowed in the search. Phosphorylation was the only

posttranslational modification (PTM) allowed, and a dynamic modification of 79.97 Da

was enabled for Ser, Thr, and Tyr, with maximum 3 PTMs per peptide. The FDRs were 1

% and 3 % for stringent and relaxed searches, respectively.

Microfluidic chip fabrication. The microfluidic chip layouts were designed using

AutoCAD software (Autodesk, San Rafael, CA) and the photomasks were prepared by

HTA Photomask (San Jose, CA). Chips were fabricated from 1.6 mm thick white crown

glass coated with chrome and photoresist (Nanofilm, Shelton, CA) using standard

photolithography and wet chemical etching techniques. In short, the chips were exposed

through the photomask to UV light using a UV exposure system from OAI (San Jose,

CA), developed, and etched with chrome etchant and BOE for the removal of exposed

chrome and glass, respectively. The channel depths were measured with a Dektak

profilometer (Veeco, Plainview, NY). Access holes of ~1 mm i.d. were drilled in the chip

substrates using a rotary tool (Dremel, Racine, WI). The photoresist and chrome layers

from the unexposed areas were further removed with acetone/methanol and chrome

etchant, respectively. The substrate plates were cleaned with sulfuric acid (5 %), DI

107

water, detergent, acetone, and methanol, then boiled in a mixture of water/NH4OH 28

%/H2O2 35 %, and bonded by gradual heating from room temperature to 550 °C. Fused-

silica capillary (20 μm i.d. × 90 µm o.d. × ~1 cm long) electrospray ionization (ESI)

emitters were inserted into one end of the microchip and secured in place with removable

glue (E6000, Eclectic Products, Pineville, LA). The microfluidic phosphopeptide

enrichment microreactor/chamber (50 μm deep × 500 µm wide × 1.5-5 mm long) was

packed with bulk TiO2 (10 μm) particles. To enable selective elution for MS analysis, the

peptides recovered from the TiO2 reactor were captured on a microreactor/chamber of

similar dimensions packed with C18 (5 μm) particles, placed between the TiO2 reactor

and the ESI spraying emitter. Alternatively, for simplicity, some designs featured only a

chamber packed with C18 particles, with the TiO2 being loaded directly in a microfluidic

channel (50 μm deep × 120 µm wide × 10 mm long bed) connected to the C18 chamber.

The particles were loaded in their corresponding chambers by using an isopropanol slurry

for C18, and a mixture of isopropanol/lactic acid (80:20 v/v) for the larger and heavier

TiO2 particles. Lactic acid is often used as an additive to improve the selectivity of TiO2-

based phosphopeptide enrichment. Being a viscous solution, it also proved to be a good

additive for preparing the TiO2 slurry. All particle loading processes were performed

manually using a 250 μL gastight syringe (Hamilton, Reno, MA), and after rinsing the

chip with isopropanol and methanol, the packed chambers were sealed with a two-

component epoxy glue that was cured at 95 °C for 30 min (Epo-Tek, Epoxy Technology,

Inc., Billerica, MA). Connecting ports on the chip were prepared from PEEK unions

(Valco Instruments, Houston, TX) cut evenly into two pieces. Fluid flows were

visualized on the chip with a Nikon epi-fluorescence microscope (Melville, NY).

108

6.3 Results and discussion

Selection of MRM transitions. The SKBR3 cells overexpress the cell surface ERBB2

protein, a receptor tyrosine kinase, which upon ligand binding leads to the activation of a

variety of MAPK signaling pathways that promote proliferation and evasion of apoptosis.

Overexpression of-and mutations in-this protein result in dysregulated downstream

phosphosignaling processes and aberrant proliferation, such as in the case of breast

cancer cells. We have previously reported an on-line, microcapillary reactor for

performing fast, streamlined protein digestion, phosphopeptide enrichment, and ESI-

MS/MS analysis.31 For simple, standard protein mixtures, the methodology provided

similar results to conventional bench-top protocols. In this work, we implemented the

approach on microfluidic devices, and evaluated the ability to enrich and detect select

phosphopeptides from small amounts of complex cellular extracts. For demonstration,

targeted MRM-MS was used for the detection of peptides that become phosphorylated

early in the ERBB2-initiated signaling cascade. To identify such phosphopeptides, an

EGF-stimulated SKBR3 time-point experiment was conducted, in which cell cycle

arrested SKBR3 cells were stimulated with EGF for 1 min, 5 min, 15 min, 30 min, 60

min and 24 h. The experiment led to the identification of 713 proteins matched by 1016

unique phosphopeptide sequences (1415 distinct phosphopeptides), of which 24 proteins

matched by 33 phosphopeptide sequences (45 distinct phosphopeptides) were mapped to

ERBB2/MAPK pathways (Table 6.1). Based on relevance to ERBB2 biological

signaling, quality of tandem mass spectra, and ease of identification as indicated by the

number of total spectral counts that matched the 45 phosphopeptides, a subset of three

double charged peptides were selected for MRM-MS detection from microfluidic TiO2-

109

enriched SKBR3 whole-cell extracts. These peptides represented the receptor tyrosine-

protein kinase ERBB2 (P04626), SHC-transforming protein 1 (P29353) and adapter

molecule CRK (P46108), carrying phosphorylation at Ser998, Tyr427 and Ser41,

respectively. For quantitative MRM-MS analysis, 13C/15N stable isotope-labeled

standards of these phosphopeptides were synthesized with the label (7+ Da) being

incorporated at Leu residues. The amino acid sequences, the protein names and IDs, the

phosphorylated (“*”) and isotopically labeled sites (“^”), the GRAVY values of the non-

phosphorylated counterparts, and the precursor-fragment transitions are provided in

Table 6.2. The transitions that were chosen for MRM-MS analysis were selected

manually from the top 5-10 most intense peaks in the tandem mass spectra of the three

peptides, produced through data dependent neutral loss LC-MS/MS experiments, and the

list of transitions generated automatically in Skyline. Only the transitions identified by

both methods were selected for inclusion in the data acquisition method. This strategy

resulted in the selection of as many as 5-7 transitions per parent ion, minimizing the

probability of false identifications due to contaminants with potentially the same

precursor and fragment m/z values. As a result of the relatively low LTQ-MS instrument

resolution, a wide mass selection window (m/z= ± 1.5 Da) for the product ions was used

at first. However, at this window width, contaminant ions interfered with the detection of

MRM transitions, and the product mass window had to be narrowed to m/z= ±0.6 Da.

This resulted in improved signal-to-noise ratios and elimination of many interfering

transitions.

110

Protein

ID

Protein Descriptions

Sequence#

phosphoRS Site

Probabilities

O75369 Filamin-B LVs#PGSANETSSILVESVTR S(3): 99.4

O75369 Filamin-B LVSPGs#ANETSSILVESVTR S(6): 1.1

O75369 Filamin-B YADEEIPRs#PFK S(9): 99.9

O75676 Ribosomal protein S6 kinase alpha-4 LEPVYs#PPGs#PPPGDPR S(6): 99.8; S(10): 100.0

O96013 Serine/threonine-protein kinase PAK 4 s#LVGTPYWMAPELISR S(1): 100.0

P00533

Epidermal growth factor receptor Fs#NNPALCNVEs#IQWR S(2): 100.0; S(12):

100.0

P00533 Epidermal growth factor receptor GSHQIs#LDNPDYQQDFFPK S(6): 88.8

P01100 Proto-oncogene c-Fos GSSSNEPSSDSLSs#PTLLAL S(14): 87.0

P01100 Proto-oncogene c-Fos KGs#SSNEPSSDSLSSPTLLAL S(3): 68.0

P01100 Proto-oncogene c-Fos KGSs#SNEPSSDSLSSPTLLAL S(4): 14.8

P01100 Proto-oncogene c-Fos KGSSs#NEPs#SDSLSSPTLLAL S(5): 0.1; S(9): 0.0

P04626 Receptor tyrosine-protein kinase erbB-2 FVVIQNEDLGPAs#PLDSTFYR S(13): 100.0

P04626 Receptor tyrosine-protein kinase erbB-2 Gt#PTAENPEYLGLDVPV T(2): 49.9

P04626 Receptor tyrosine-protein kinase erbB-2 GTPt#AENPEYLGLDVPV T(4): 49.9

P04626 Receptor tyrosine-protein kinase erbB-2 GTPTAENPEy#LGLDVPV Y(10): 100.0

P04626 Receptor tyrosine-protein kinase erbB-2 s#GGGDLTLGLEPSEEEAPR S(1): 100.0

P04626 Receptor tyrosine-protein kinase erbB-2 SPLAPs#EGAGs#DVFDGDLGMGAAK S(6): 99.9; S(11): 100.0

P04626 Receptor tyrosine-protein kinase erbB-2 SPLAPs#EGAGSDVFDGDLGMGAAK S(6): 50.0

P04626 Receptor tyrosine-protein kinase erbB-2 SPLAPSEGAGs#DVFDGDLGMGAAK S(11): 100.0

P05412 Transcription factor AP-1 NSDLLt#SPDVGLLK T(6): 0.7

P05412 Transcription factor AP-1 NSDLLTs#PDVGLLK S(7): 99.2

P11487 Fibroblast growth factor 3 Ly#CATKy#HLQLHPSGR Y(2): 83.5; Y(7): 4.2

P15056 Serine/threonine-protein kinase B-raf RDs#SDDWEIPDGQITVGQR S(3): 50.0

P15056 Serine/threonine-protein kinase B-raf RDSs#DDWEIPDGQITVGQR S(4): 88.6

P16949 Stathmin ASGQAFELILs#PR S(11): 100.0

P16949 Stathmin RAs#GQAFELILs#PR S(3): 100.0; S(12): 100

P16949 Stathmin RAs#GQAFELILSPR S(3): 100.0

P16949 Stathmin SKESVPEFPLs#PPK S(11): 100.0

P17535 Transcription factor jun-D ADGAPSAAPPDGLLAs#PDLGLLK S(16): 99.9

P21359 Neurofibromin ENVELSPTt#GHCNSGRt#R T(9): 94.4; T(17): 18.5

P27361 Mitogen-activated protein kinase 3 IADPEHDHTGFLTEy#VATR Y(15): 98.6

P28482 Mitogen-activated protein kinase 1 VADPDHDHTGFLTEy#VATR Y(15): 86.2

P29353 SHC-transforming protein 1 ELFDDPSy#VNVQNLDK Y(8): 99.4

P31749 RAC-alpha ser/threonine-protein kinase SGs#PSDNSGAEEMEVSLAKPK S(3): 45.3

P46108 Adapter molecule crk Ds#STSPGDYVLSVSENSR S(2): 49.6

P46108 Adapter molecule crk DSs#TSPGDYVLSVSENSR S(3): 46.5

P46108 Adapter molecule crk DSSt#SPGDYVLSVSENSR T(4): 6.5

Q13177 Serine/threonine-protein kinase PAK 2 YLs#FTPPEK S(3): 99.9

Q13541 Eukaryotic transl. init. factor 4E-binding t#PPRDLPTIPGVTSPSSDEPPMEASQSHLR T(1): 94.8

Q14315 Filamin-C s#LTATGGNHVTARVLNPSGAK S(1): 67.1

Q7L7X3 Serine/threonine-protein kinase TAO1 AGs#LKDPEIAELFFK S(3): 100.0

Q92934 Bcl2-associated agonist of cell death HSSy#PAGTEDDEGMGEEPSPFR Y(4): 11.9

Q92934 Bcl2-associated agonist of cell death RMs#DEFVDSFK S(3): 100.0

Q9UER7 Death domain-associated protein 6 DGDKs#PMSSLQISNEK S(5): 98.2

Q9Y6R4 Mitogen-act. prot. kinase kinase kinase 4 LEs#EDDSLGWGAPDWSTEAGFSR S(3): 99.6

Table 6.1 Proteins mapped to ERBB2/MAPK pathways in EGF stimulated SKBR3 cells. Highlighted

rows represent peptides selected for the analysis.

111

Protein/Peptides GRAVY Transitions

(precursor ion → product ions)

P04626/Receptor tyrosine-protein kinase ErbB-2

FVVIQNEDLGPA(*S)P(^L)DSTFYR -0.06

Light 1224.6++ →

(b9+ 1058.6, b10+ 1115.6, y5+ 673.3, y8+ 998.5,

y11+ 1333.6, y12+ 1390.6, y13+ 1503.7)

Heavy 1228.1++ →

(b9+ 1058.6, b10+ 1115.6, y5+ 673.3, y8+

1005.51, y11+ 1340.6, y12+ 1397.6, y13+ 1510.7)

P29353/SHC-transforming protein 1

ELFDDPS(*Y)VNVQN(^L)DK

-0.83

Light 988.4++ →

(y5+ 617.3, y7+ 830.4, y11+ 1356.6, y11++ 678.8,

y14++ 867.4)

Heavy 991.9++ →

(y5+ 624.3, y7+ 837.5, y11+ 1363.6, y11++

682.3, y14++ 870.9)

P46108/Adapter molecule crk

DS(*S)TSPGDYV(^L)SVSENSR -0.84

Light 990.4++ →

(b11+ 1202.5, b13+ 1388.6, y5+ 592.3, y7+

778.4, y13+ 1422.7, y14+ 1509.7)

Heavy 993.9++ →

(b11+ 1209.5, b13+ 1395.6, y5+ 592.3, y7+

778.4, y13+ 1429.7, y14+ 1516.7)

Table 6.2 Phosphopeptides selected for MRM-MS analysis of EGF-stimulated SKBR3 cell extracts.

Precursor-product transitions are listed for both endogenous and stable-isotope labeled synthetic peptide

standards. The GRAVY values are calculated using the non-phosphorylated forms. * refers to

phosphorylated residue, ^ refers to 13C/15N-labeled amino acid.

112

Microfluidic chip design for the detection of phosphopeptides. The detection of

phosphopeptides in the presence of their non-phosphorylated counterparts is challenging

not only due to the heterogeneity and low abundance of phosphorylation, but also due to

the diminished ability to detect peptides in (+) ion mode when the ions carry a (-) charged

group at given experimental conditions. Phosphopeptides can be enriched, however, from

cellular extracts using a variety of procedures. Methodologies relying on the use of TiO2

particles, alone,17-19 or in combination with other fractionation methods such as anion or

cation exchange chromatography12,13,16 have gained particular momentum. The approach

relies on the ability of TiO2 particles to retain the phosphopeptides at low pH (pH~2) and

release them at high pH (pH>10). Unlike other acidic moieties in a peptide, the phosphate

group is partially ionized at pH=2 (pKa1=2.12) and can interact with-and be retained

selectively by-the Lewis acidic sites of TiO2 particles. At high pH, in the presence of

competing anionic species, the phosphopeptides are released from these sites. To reduce

non-specific interactions, the procedure is performed in the presence of organic solvents

and various additives such as o-hydroxybenzoic acids (e.g., 2,5-dihydroxybenzoic

acid)19,20 or more MS-friendly aliphatic hydroxy acids (e.g., lactic acid).17

Phosphopeptides eluted in basic buffer systems are further acidified and desalted on SDB

(styrene-divinylbenzene) and GC (graphite carbon) cartridges or spin tips for recovering

hydrophobic and hydrophilic phosphopeptides, respectively, and analyzed by reversed-

phase LC-MS. Alternative approaches include the use of basic buffer solutions to elute

the peptides from TiO2 directly on C18 traps or LC columns,12,13 of pH gradients for

fractional elution of phosphopeptides from TiO2,32 or even of high pH reversed-phase

113

fractionation, prior to TiO2 enrichment, for separating singly charged phosphorylated

peptides.33

Figure 6.1 Schematic diagrams of the phosphoprotein analysis microfluidic chips. (A) Two-chamber

design. The microfluidic chip has two chambers of identical dimensions (50 μm deep × 500 µm wide × 1.5

mm long) for loading TiO2 and C18 particles, respectively. (B) Single-chamber design. The microfluidic

chip has one chamber for C18 particles (same dimensions as in A), while TiO2 is loaded directly in the

sample delivery channel (50 μm deep × 120 µm wide × 10 mm long bed).

114

Time Operation

Solution composition

Concentration Flow rate

On-chip sample preparation

10 min

Equilibration with lactic acid

H2O/CH3CN/ lactic acid/TFA 18:72:10:0.4 v/v 1 μL/min

10 min Sample loading

H2O/CH3CN/ lactic acid/TFA

18:72:10:0.4 v/v 1 μg/μL cell extract; 1 – 10 nM standards

1 μL/min

10 min

Rinsing with lactic acid

H2O/CH3CN/ lactic acid/TFA

18:72:10:0.4 v/v 1 μL/min

10 min

Lactic acid removal

H2O/CH3CN/ TFA

20:80:0.4 v/v 1 μL/min

15 min

pH change, organic removal

H2O/CH3CN 98:2 v/v 1 μL/min

MRM analysis

One-step

45

min

Simultaneous elution with the detection of all phosphopeptides

Elution solution: H2O/CH3CN/ 28% NH4OH

80:10:10 v/v 300 nL/min

Two-step

45 min

Elution step 1: detection of hydrophilic phosphopeptides

Elution sol. 1: H2O/CH3CN/ 28% NH4OH

90:5:5 v/v 300 nL/min

45 min

Elution step 2: detection of hydrophobic phosphopeptides

Elution sol. 2: H2O/CH3CN/ 28% NH4OH

80:10:10 v/v 300 nL/min

Table 6.3 Steps for the phosphopeptide enrichment and analysis using the microfluidic chip.

CRK SHC1 ERBB2

Elution step 1 1 2

Retention

time (min)

Light 14.6 15.3 20.8

Heavy 14.2 15 21.3

dotp Light 0.91 0.9 0.89

Heavy 0.93 0.87 0.91

rdotp 0.99 0.98 0.98

Total ratio (Light/Heavy) 0.1 0.29 0.38

Table 6.4 Quantitative measurement of heavy and light phosphopeptides using the chip-MRM/MS

approach. The sample was eluted in two steps. Elution step 1: H2O/ACN/28%NH4OH 90:5:5, at a flow

rate of 300 nL/min for 45 min; elution step 2: H2O/ACN/28%NH4OH 80:10:10, at a flow rate of 300

nL/min for 45 min.

115

Based on the above strategies that were advanced for phosphopeptide enrichment with

conventional benchtop instrumentation, and the platforms that were developed for

interfacing microfluidic devices to ESI-MS detection for analyzing cell extracts,34-36 the

microfluidic chips that were designed for phosphoprotein enrichment in this work

encompassed functional elements that enabled (a) direct loading of cellular protein

digests on TiO2 particles for enrichment in phosphopeptides, (b) elution with an

appropriate solvent system, and (c) capture on C18 particles for subsequent elution and

MS analysis using various experimental conditions. The chips comprised either two

microfluidic chambers packed with TiO2 and C18 particles (Figure 6.1A), or in a

simplified version, only a C18 chamber, having the TiO2 particles packed in a

microfluidic channel (Figure 6.1B). The role of the C18 chamber was to capture the

phosphopeptides after elution from TiO2, and enable, in case of need, experimental

conditions that favor phosphopeptide detection (e.g., peptide fractionation or buffer

exchange from basic to acidic conditions). Two processing lines were included in a 2” x

1” chip. The goal was to perform fast analysis with a minimal number of processing

steps, on chips of rather simple design. There were no additional functional elements for

desalting or enrichment. The composition and flow rate of the eluents needed to process

the protein extract digests on the chip is provided in Table 6.3. The TiO2 chip

microreactors were equilibrated first with a mixture of H2O/CH3CN/lactic acid/TFA

(18:72:10:0.4 v/v), the sample was loaded on TiO2 and rinsed on the chip with the same

solution to enable the capture of phosphopeptides and removal of non-phosphorylated

ones, the lactic acid was removed with H2O/CH3CN/TFA (20:80:0.4 v/v), and the pH was

raised to ~7 with a solution of H2O/CH3CN (98:2 v/v). Last, the phosphopeptides were

116

eluted from TiO2 in one or two steps in basic solvent solutions, and captured on the C18

reactor or analyzed on-the-fly by MS. While the entire sample handling procedure

amounted to 90-135 min, increasing the flow rate during rinsing and reducing the

phosphopeptide elution window to the strictly necessary, could have reduced the

processing time to half.

Effect of high pH on the detection of phosphopeptides. The LC-MS time-point

experiments that enabled the selection of MRM transitions were conducted from TiO2-

enriched, SDB/GC desalted, and acidified phosphopeptide samples. MS detection

occurred from acidic LC eluents. All phosphopeptide purifications steps were conducted

independently, with the sample being isolated between each step and prepared in an

adequate solvent system for the next step. On the chip however, the isolation of samples

(or “parking”) after each step is not always feasible, and upstream effluents may need to

be discarded through downstream functional elements, risking contamination and altering

of performance. On the chips developed for this study, the removal of non-

phosphorylated peptides, contaminants and additives was performed on the TiO2

particles, with all effluents being discarded through the C18 chamber. Phosphopeptide

elution from TiO2, followed by subsequent capture on C18 particles, was performed with

NH4OH of various concentrations. MS detection was then performed either directly from

the basic eluent system, or after lowering the pH once the peptides were captured on the

C18 bed. Therefore, due to different benchtop/chip experimental conditions, the ability to

detect the phosphopeptides of interest was verified first from both acidic and basic

solutions of 500 nM peptide concentrations, by using capillary infusion experiments

117

(Figure 6.2). Under both conditions, the peptides of interest were detectable as double

charged species, confirming the validity of previously chosen LC-MRM transitions for

chip analysis. This confirmation was critical, as the MRM transitions would be

completely different if the charge state of the precursor ions is changed.

Figure 6.2 Full mass scans acquired during the infusion of a mixture of three stable isotope-labeled

synthetic phosphopeptides. (A) Acidic infusion solution: H2O/CH3CN/TFA 50:50:0.01 v/v; (B) Basic

infusion solution: H2O/CH3CN/NH4OH (28 %) 45:50:5 v/v. Conditions: infusion performed at 300 nL/min,

sample concentration 500 nM.

118

Figure 6.3 Microfluidic chip MRM-MS analysis of an SKBR3 cell extract digest spiked with three

heavy standards. The protein cell extract was digested overnight with trypsin and subjected to C18

cleanup. The cell extract was dissolved in H2O/CH3CN/lactic acid/TFA 18:72:10:0.4 v/v (1 μg/μL) and

spiked with the 3 heavy standards (1 nM). The sample (10 μL) was pumped through a two-chamber chip

loaded with a bed of 1.5 mm TiO2 and 1.5 mm C18 particles. A basic solution [H2O/CH3CN/NH4OH (28

%) 88:2:10 v/v] was used to elute the phosphopeptides from TiO2 onto C18, and an acidic solution

[H2O/CH3CN/TFA 50:50:0.01 v/v] was used for elution from C18 particles for MS analysis. Detailed

experimental conditions for sample processing on the chip, with a one-step bulk elution of

phosphopeptides, are provided in Table 6.2. Only the ERBB2 phosphopeptide was detected.

119

A two-chamber chip packed with TiO2 and C18 particles was next evaluated with a digest

of SKBR3 proteins desalted on C18 cartridges (1 μg/μL), and spiked with the three heavy

isotope labeled standards (1 nM). The sample was loaded in H2O/CH3CN/lactic acid/TFA

(18:72:10:0.4 v/v) on the TiO2 bed, the retained phosphopeptides were eluted from TiO2

onto C18 in one step with a solution of H2O/CH3CN/28% NH4OH (88:2:10 v/v), and an

acidic solution of H2O/CH3CN/TFA (50:50:0.01 v/v) was used for final elution from C18

and MRM-MS analysis (see all the detailed steps in Table 6.3). Using these conditions,

only the ERBB2 phosphopeptide was detected in both heavy and light forms (Figure

6.3), while the other SHC and CRK phosphopeptides could not be detected at all,

indicating that the two missing phosphopeptides were either lost during the sample

processing steps, or not eluted from the TiO2 particles. Based on the pKa values of

phosphorylated amino acids (pThr/pSer: pKa = 7.12, 2.12) and (pTyr: pKa = 7.0, 1.0),37

and the MRM-MS results obtained from acidic and basic solution infusion experiments,

MS detectability was not deemed to represent a problem at low and high pH values for

these peptides. The effect of pH on the interactions between phosphopeptides and the

reversed-phase C18 and TiO2 particles was, therefore, next investigated. Two sets of

experiments were performed: (a) one that tested the ability of C18 particles to retain

phosphopeptides when loaded in acidic or basic solvent systems, and (b) one that tested

the ability of TiO2 particles to release the phosphopeptides in basic eluent mixtures of

various pH values. For evaluating the performance of C18 particles, the phosphopeptide

standards were loaded onto C18 columns using acidic (pH ~4) or basic (pH ~10) media,

rinsed with 15 μL acidic solution (H2O/CH3CN/TFA 98:2:0.01), and eluted at 300

nL/min in H2O/CH3CN/TFA 50:50:0.01 v/v for ESI-MRM/MS analysis. The results

120

presented in Figure 6.4 reveal that all phosphopeptides were detectable under low pH

loading conditions, but only one under high pH loading, even if MS detection occurred

from (+) ESI-favorable acidic solutions. At pH values of 4 and 10, the phosphate groups

carry either one or two (-) charges that can contribute to a substantial increase in

hydrophilicity, so retention on the reversed-phase C18 column appeared to be mostly

controlled by the hydrophobic properties of the peptide backbone. According to the

GRAVY (grand average of hydropathy)38 values presented in Table 6.2, the two

undetected SHC1 and CRK peptides are hydrophilic, with GRAVY values of (-0.83)-(-

0.84), while the ERBB2 peptide is more hydrophobic (GRAVY index of -0.06). Protein

GRAVY values range from -2 to +2, the more positive values being indicative of more

hydrophobic proteins. At a loading pH of 10, only the more hydrophobic ERBB2 peptide

was retained on the C18 column. While high pH loading conditions on C18 columns

were detrimental to the detection of phosphopeptides, such high pH values were

necessary for facilitating elution from the TiO2 columns. To find the minimum pH

needed for eluting phosphopeptides from TiO2 particles, the same sample was loaded

onto TiO2 and eluted with solutions of pH values ranging from 9 to 11. The results are

shown in Figure 6.5. While at pH 9 the signal for phosphopeptides was very small, and

at pH 10 the phosphopeptides eluted within an extended elution window, at pH 11 all

three peptides could be detected within a window of a few minutes. The eluting power

and ion intensities increased progressively with the increase of the eluent pH. However,

the lowest pH value (~10) needed for the elution of phosphopeptide from TiO2 was

clearly too high for subsequent retention of hydrophilic phosphopeptides on C18

particles.

121

Figure 6.4 Retention of stable isotope-labeled synthetic phosphopeptides on C18 particles using

loading solutions of different pH. (A) Acidic loading solution: H2O/CH3CN/TFA 98:2:0.01 v/v, pH ~4;

(B) Basic loading solution: H2O/CH3CN/NH4OH (28 %) 97:2:1 v/v, pH ~10. Conditions: 100 fmol sample

was loaded on packed C18 columns (100 μm i.d. fused silica capillaries packed with a 5 mm bed of C18

particles); peptide elution/detection was performed at 300 nL/min in a solution of CH3CN/H2O/TFA

50:50:0.01 v/v.

122

Figure 6.5 Elution of stable isotope-labeled synthetic phosphopeptides from TiO2 particles using

eluents of different pH. (A) H2O/CH3CN/NH4OH (28 %) 70:30:0.1 v/v, pH ~9; (B) H2O/CH3CN/NH4OH

(28 %) 69:30:1 v/v, pH ~10; and (C) H2O/CH3CN/NH4OH (28 %) 60:30:10 v/v, pH ~11. Conditions:

phosphopeptide standards (10 nM) were prepared in a solution of H2O/CH3CN/lactic acid/TFA

18:72:10:0.4 v/v, and 10 µL sample (100 fmol) was loaded on TiO2 columns (100 μm i.d. fused silica

capillaries packed with a 10 mm bed of TiO2 particles); peptide elution/detection was performed at 300

nL/min in solutions A, B or C.

123

Figure 6.6 Two-step sequential elution of on-column enriched stable isotope-labeled synthetic

phosphopeptides. (A) 1st elution step: H2O/CH3CN/ NH4OH (28 %) 90:5:5 % v/v; (B) 2nd elution step:

H2O/CH3CN/NH4OH (28 %) 80:10:10 v/v. Conditions: phosphopeptide standards (10 nM) were prepared

in a solution of H2O/CH3CN/lactic acid/TFA 18:72:10:0.4 v/v, and 10 µL sample (100 fmol) was loaded on

TiO2/C18 columns (100 μm i.d. fused silica capillaries packed with a bed of 5 mm TiO2 and 5 mm C18

particles); peptide elution/detection was performed at 300 nL/min.

124

Two-step sequential elution of hydrophilic and hydrophobic phosphopeptides. To identify

experimental conditions that enable effective elution from TiO2 and capture on C18

particles, a few other buffer compositions were tested. For example, C18 sample loading

in solutions of NH4HCO3 (250 mM)/CH3CN (2 %), pH 8-9, enabled retention on C18

columns, but the buffer system could not elute the phosphopeptides from TiO2; NH4OH

(10 %)/CH3CN (2 %) solutions enabled elution from TiO2 of all three peptides, but could

not elute completely the ERBB2 peptide from C18; while NH4OH (5-10 %)/CH3CN (10-

30 %) enabled elution from both columns and detection of all three peptides. High

CH3CN content (30 %) in the TiO2 elution solvent mixture prevented the retention of any

peptide on C18 particles, however, concentrations above 5 % enabled a mild separation

of phosphopeptides, even on the short C18 chambers that were used in this study. Based

on these findings, a two-step sequential elution approach of phosphopeptides from TiO2

particles was developed (Table 6.3). The approach prevented the undesired loss of

hydrophilic phosphopeptides, while increasing detection sensitivity. In a 1st step, a

solution consisting of 5 % NH4OH (28 %), 5 % CH3CN, and 90 % H2O was used to elute

phosphopeptides from TiO2 and load them on the C18 column. The hydrophilic peptides

that could not be captured by the C18 particles under these conditions were analyzed

directly, on-the-fly, by MRM-MS, while the more hydrophobic peptides were retained on

the C18 particles. In the 2nd step, a stronger elution solvent system comprised of 10 %

NH4OH (28 %), 10 % CH3CN and 80 % H2O, was used to recover the phosphopeptides

with high affinity for both TiO2 and C18. Depending on the peptide backbone structure

and hydrophobicity, with 5 % and 10 % CH3CN concentration in the elution solvent

system, the fractionation of phosphopeptides on the C18 column could be achieved, as

125

well. However, it must be recognized that a fine interplay between the concentration of

CH3CN, NH4OH and dimensions of TiO2/C18 microreactors will play a role in fine-

tuning the ability to recover and separate the phosphopeptides. Therefore the optimization

of the system for specific samples will be required. The performance of this approach for

the analysis of the three heavy phosphoppetide standards is shown in Figure 6.6. Not

surprisingly, the two hydrophilic CRK/SHC phosphopeptides were detected in the first

step, and the hydrophobic ERBB2 peptide was detected only during the second step.

Signals from the hydrophilic phosphopeptides were also observable at the beginning (RT

~ 1 min) of the second elution step, but overall, their intensities were relatively low, no

more than 25 % of the intensities in the first step.

MRM analysis of phosphopeptides from SKBR3 breast cancer cell extracts. The efficacy

of the proposed approach was tested for the quantitative analysis of phosphorylated

peptides from real biological samples. The three heavy standards were spiked (10 nM)

into a tryptic digest of SKBR3 cell extract (1 μg/μL). The spiked sample was processed

on the chip following the protocol described above, and the results are shown in Figure

6.7. The chip of simpler design, comprising a C18 chamber (5 mm) and a channel (10

mm) packed with TiO2, was used in the analysis (Figure 6.1B). All three

phosphopeptides, including both heavy and light forms, were confidently detected and

identified. As expected, the two hydrophilic peptides were eluted in the first elution step,

while the more hydrophobic ERBB2 phosphopeptide was detected in the second step.

The results of such LC-MS/MRM experiments can be validated in two dimensions, based

on the relative intensities of ion transitions and the retention time. The former can be

126

evaluated using dotp and rdotp parameters (Table 6.4). The dotp measures the correlation

between the detected product ion intensities and the fragment ions intensities in a

standard spectral library, while the rdotp reflects the similarity of relative intensities of

the transitions between heavy and light phosphopeptides. The calculated dotp values

ranged from 0.87 to 0.93, and the rdotp values were no less than 0.98, indicating a high

confidence of detection of both heavy and light forms for all target peptides. The heavy

and light forms of the same phosphopeptide have the same physicochemical properties,

and therefore they should elute at the same retention time. The slight discrepancies

between the heavy and light peptides (Figure 6.7 and Table 6.4) were attributed to the

relatively large elution window of these peptides, and less than ideal spray stability at 5-

10 % organic solvent in the eluent, a factor that resulted in spiked sample peaks. For

every experiment, the heavy standard was loaded at the same known concentration of 10

nM (100 fmol for each standard), so the concentration of the light phosphopeptides could

be inferred by using the ratios of total signal intensities between the heavy and light

forms. The measured light-to-heavy ratios were 0.1, 0.29, and 0.38 for the 3

phosphopeptides, leading to an estimated concentration of the endogenous

phosphopeptides of <10 nM. As the data were generated with a relatively low resolution

LTQ instrument, the sensitivity and selectivity of the approach is expected to improve

substantially when the chip is coupled to a high-end, high resolution and sensitivity mass

spectrometer.

127

Figure 6.7 Two-step sequential elution of phosphopeptides from an on-chip enriched SKBR3 cell

extract spiked with stable isotope-labeled synthetic peptides. Upper panel: Light, endogenous peptides

from SKBR3; Lower panel: Heavy, standard peptide spikes. Conditions: the protein cell extract was

digested overnight with trypsin and subjected to C18 cleanup; the cell extract was dissolved in

H2O/CH3CN/lactic acid/TFA 18:72:10:0.4 v/v (1 μg/μL) and spiked with the 3 heavy standards (10 nM);

10 μL sample was loaded onto a one-chamber chip (5 mm C18 chamber, 10 mm TiO2 bed packed in a

microfluidic channel); 1st elution step: H2O/CH3CN/ NH4OH (28 %) 90:5:5 % v/v (panels A, B, D and E);

2nd elution step: H2O/CH3CN/NH4OH (28 %) 80:10:10 v/v (panels C and F); peptide elution was performed

at 300 nL/min.

128

6.4 Conclusions

In this work, we developed an integrated microfluidic platform for the targeted detection

and analysis of phosphopeptides involved in signaling networks. The approach relies on

balancing the hydrophilic properties imparted to peptides by the presence of

phosphorylated groups, with the hydrophobic properties of the peptide backbone, to

sequentially capture, enrich and release phosphopeptides from complex cellular extracts.

The performance of the approach was demonstrated using tryptic digests of SKBR3

proteins spiked with stable isotope-labeled standards. Low nanomolar concentrations of

endogenous phosphopeptides involved in MAPK signaling could be detected via MRM-

MS. The deregulation of the three selected phosphoproteins are frequently observed in

many cancers, for example, ERBB2, an essential component of a neuregulin-receptor

complex, is overexpressed in up to 30% of breast cancers, correlating with poor patient

survival. More importantly, the peptides that were selected from the three

phosphoproteins carry phosphosites that have known effects in a wide range of biological

processes in different cells, for example, the phosphorylated Y-427 of SHC1 is reportedly

related to the altered cell growth, motility, and apoptosis, and the phosphorylated S-41 of

CRK can induce carcinogenesis and cell motility. Although only SKBR3 cell line was

used in this study, the selected targets can also be used for the analysis of other cells,

including both cancer and normal cells. Overall, the advantages of this approach include:

(a) simplicity, as the microfluidic design relies only on the use of microfluidic reactors of

simple construction; (b) ease of operation, as no off-line cell extract fractionation is

required, and sample processing is enabled solely by sequential loading/elution on and

from the microfluidic reactors; (c) speed, as the total time for sample processing is less

129

than 1-2 hours; and (d) versatility, as integration into more complex workflows that

include analytical separations and enable fast MS detection,39,40 to further increase

analytical power, is feasible.

Acknowledgments. This work was supported by the National Science Foundation, award

DBI-1255991 to IML.

6.5 References

1. He, X.; Chen, Q.; Zhang, Y.; Lin, J.M. Recent advances in microchip-mass

spectrometry for biological analysis. TrAC, Trends Anal. Chem. 2014, 53, 84-97.

2. Feng, X.; Liu, B.F.; Li, J.; Liu, X. Advances in coupling microfluidic chips to mass

spectrometry. Mass Spectrom. Rev. 2015, 34, 535-557.

3. Wang, X.; Yi, L.; Mukhitov, N.; Schrell, A.M.; Dhumpa, R.; Roper, M.G.

Microfluidics-to-mass spectrometry: a review of coupling methods and

applications. J. Chromatogr., A. 2015, 1382, 98-116.

4. Chao, T.C.; Hansmeier, N. Microfluidic devices for high‐ throughput proteome

analyses. Proteomics. 2013, 13, 467-479.

5. Hung, P. J.; Lee, P. J.; Sabounchi, P.; Lin, R.; Lee, L. P. Continuous perfusion

microfluidic cell culture array for high-throughput cell-based assays. Biotechnol.

Bioeng. 2004, 89, 1–8.

6. Toh, Y.-C.; Lim, T. C.; Tai, D.; Xiao, G.; Noort, D. V.; Yu, H. A microfluidic 3D

hepatocyte chip for drug toxicity testing. Lab Chip 2009, 9, 2026.

130

7. Zhang, C.; Xu, J.; Ma, W.; Zheng, W. PCR microfluidic devices for DNA

amplification. Biotechnol. Adv. 2006, 24, 243–284.

8. Dunn, J.D.; Reid, G.E.; Bruening, M.L. Techniques for phosphopeptide enrichment

prior to analysis by mass spectrometry. Mass Spectrom. Rev. 2010, 29, 29-54.

9. Fíla, J.; Honys, D. Enrichment techniques employed in phosphoproteomics. Amino

Acids 2011, 43, 1025–1047.

10. Min, Q.; Chen, X.; Zhang, X.; Zhu, J.J. Tailoring of a TiO2 nanotube array-integrated

portable microdevice for efficient on-chip enrichment and isotope labeling of serum

phosphopeptides. Lab Chip. 2013, 13, 3853-3861.

11. Mohammed, S.; Kraiczek, K.; Pinkse, M.W.; Lemeer, S.; Benschop, J.J.; Heck, A.J.

Chip-Based Enrichment and NanoLC− MS/MS Analysis of Phosphopeptides from

Whole Lysates. J. Proteome Res. 2008, 7, 1565-1571.

12. Polat, A.N.; Kraiczek, K.; Heck, A.J.; Raijmakers, R.; Mohammed, S. Fully

automated isotopic dimethyl labeling and phosphopeptide enrichment using a

microfluidic HPLC phosphochip. Anal. Bioanal. Chem. 2012, 404, 2507-2512.

13. Raijmakers, R.; Kraiczek, K.; de Jong, A.P.; Mohammed, S.; Heck, A.J. Exploring

the Human Leukocyte Phosphoproteome Using a Microfluidic Reversed-Phase−

TiO2− Reversed-Phase High-Performance Liquid Chromatography Phosphochip

Coupled to a Quadrupole Time-of-Flight Mass Spectrometer. Anal. Chem. 2010, 82,

824-832.

14. Rudin, T.; Tsougeni, K.; Gogolides, E.; Pratsinis, S.E. Flame aerosol deposition of

TiO2 nanoparticle films on polymers and polymeric microfluidic devices for on-chip

phosphopeptide enrichment. Microelectron. Eng. 2012, 97, 341-344.

131

15. Ellinas, K.; Tsougeni, K.; Boulousis, G.; Speliotis, T.; Tserepi, A.; Gogolides, E.

Phosphopeptide enrichment and separation in an affinity microcolumn on a silicon

microchip: Comparison of sputtered and wet-deposited TiO 2 stationary-phase. Sens.

Actuators, B. 2013, 188, 1073-1079.

16. Ikeguchi, Y.; Nakamura, H. Determination of Organic Phosphates by Column-

Switching High Performance Anion-Exchange Chromatography Using On-Line

Preconcentration on Titania. Anal. Sci. 1997, 13, 479–483.

17. Sugiyama, N.; Masuda, T.; Shinoda, K.; Nakamura, A.; Tomita, M.; Ishihama, Y.

Phosphopeptide Enrichment by Aliphatic Hydroxy Acid-modified Metal Oxide

Chromatography for Nano-LC-MS/MS in Proteomics Applications. Mol. Cell.

Proteomics 2007, 6, 1103–1109.

18. Pinkse, M. W. H.; Uitto, P. M.; Hilhorst, M. J.; Ooms, B.; Heck, A. J. R. Selective

Isolation at the Femtomole Level of Phosphopeptides from Proteolytic Digests Using

2D-NanoLC-ESI-MS/MS and Titanium Oxide Precolumns. Anal. Chem. 2004, 76,

3935–3943.

19. Larsen, M. R.; Thingholm, T. E.; Jensen, O. N.; Roepstorff, P.; Jørgensen, T. J. D.

Highly Selective Enrichment of Phosphorylated Peptides from Peptide Mixtures

Using Titanium Dioxide Microcolumns. Mol. Cell. Proteomics 2005, 4, 873–886.

20. Jensen, S. C. B. S.; Larsen, M. R. Evaluation of the impact of some experimental

procedures on different phosphopeptide enrichment techniques. Rapid Commun.

Mass Spectrom. 2007, 21, 3635–3645.

21. Humphrey, S. J.; Azimifar, S. B.; Mann, M. High-throughput phosphoproteomics

reveals in vivo insulin signaling dynamics. Nat. Biotechnol. 2015, 33, 990–995.

132

22. Deutsch, E.W.; Lam, H.; Aebersold, R. PeptideAtlas: a resource for target selection

for emerging targeted proteomics workflows. EMBO Rep. 2008, 9, 429-434.

23. Craig, R.; Cortens, J.P.; Beavis, R.C. Open source system for analyzing, validating,

and storing protein identification data. J. Proteome Res. 2004, 3, 1234-1242.

24. Jones, P.; Côté, R.G.; Cho, S.Y.; Klie, S.; Martens, L.; Quinn, A.F.; Thorneycroft, D.;

Hermjakob, H. PRIDE: new developments and new datasets. Nucleic Acids Res.

2007, 36, 878-883.

25. Picotti, P.; Lam, H.; Campbell, D.; Deutsch, E.W.; Mirzaei, H.; Ranish, J.; Domon,

B.; Aebersold, R. A database of mass spectrometric assays for the yeast

proteome. Nature methods. 2008, 5, 913-914.

26. Kirkpatrick, D.S.; Gerber, S.A.; Gygi, S.P. The absolute quantification strategy: a

general procedure for the quantification of proteins and post-translational

modifications. Methods. 2005, 35, 265-273.

27. Kennedy, J.J.; Yan, P.; Zhao, L.; Ivey, R.G.; Voytovich, U.J.; Moore, H.D.; Lin, C.;

Pogosova-Agadjanyan, E.L.; Stirewalt, D.L.; Reding, K.W.; Whiteaker, J.R.

Immobilized metal affinity chromatography coupled to multiple reaction monitoring

enables reproducible quantification of phospho-signaling. Mol. Cell. Proteomics.

2016, 15, 726-739.

28. Zawadzka, A.M.; Schilling, B.; Held, J.M.; Sahu, A.K.; Cusack, M.P.; Drake, P.M.;

Fisher, S.J.; Gibson, B.W. Variation and quantification among a target set of

phosphopeptides in human plasma by multiple reaction monitoring and SWATH‐

MS2 data‐independent acquisition. Electrophoresis. 2014, 35, 3487-3497.

133

29. MacLean, B.; Tomazela, D.M.; Shulman, N.; Chambers, M.; Finney, G.L.; Frewen,

B.; Kern, R.; Tabb, D.L.; Liebler, D.C.; MacCoss, M.J. Skyline: an open source

document editor for creating and analyzing targeted proteomics

experiments. Bioinformatics. 2010, 26, 966-968.

30. Sarvaiya, H.A.; Yoon, J.H.; Lazar, I.M. Proteome profile of the MCF7 cancer cell

line: a mass spectrometric evaluation. Rapid Commun. Mass Spectrom. 2006, 20,

3039-3055.

31. Deng, J.; Lazar, I.M. Proteolytic digestion and TiO2 phosphopeptide enrichment

microreactor for fast MS identification of proteins. J. Am. Soc. Mass Spectrom.

2016, 27, 686-698.

32. Park, S.-S.; Maudsley, S. Discontinuous pH gradient-mediated separation of TiO2-

enriched phosphopeptides. Anal. Biochem. 2011, 409, 81–88.

33. Batth, T.S.; Francavilla, C.; Olsen, J.V. Off-line high-pH reversed-phase fractionation

for in-depth phosphoproteomics. J. Proteome Res. 2014, 13, 6176-6186.

34. Armenta, J.M.; Dawoud, A.A.; Lazar, I.M. Microfluidic chips for protein differential

expression profiling. Electrophoresis 2009, 30, 1145-1156.

35. Dawoud, A.A.; Sarvaiya, H.A.; Lazar, I.M. Microfluidic platform with mass

spectrometry detection for the analysis of phosphoproteins. Electrophoresis 2007, 28,

4645-4660.

36. Lazar, I.M. Recent advances in capillary and microfluidic platforms with MS

detection for the analysis of phosphoproteins. Electrophoresis 2009, 30, 262-275.

134

37. Kumar, Y.; Khachane, A.; Belwal, M.; Das, S.; Somsundaram, K.; Tatu, U.

ProteoMod: A new tool to quantitate protein post-translational

modifications. Proteomics 2004, 4, 1672–1683.

38. Kyte, J.; Doolittle, R.F. A simple method for displaying the hydropathic character of

a protein. J. Mol. Biol. 1982, 157, 105–132.

39. Lazar, I.M.; Rockwood, A.L.; Lee, E.D.; Sin, J.C.H.; Lee, M.L. High-speed TOFMS

detection for capillary electrophoresis. Anal. Chem. 1999, 71, 2578-2581.

40. Shen, Y.; Smith, R.D; Unger, K.K.; Kumar, D.; Lubda, D. Ultrahigh-Throughput

Proteomics Using Fast RPLC Separations with ESI-MS/MS. Anal. Chem. 2005, 77,

6692–6701.

135

CHAPTER 7: Microfluidic Reactors for Cell Stimulation, Lysis and Sample

Collection for MS Analysis

Jingren Deng, Shreya Ahuja and Iulia M. Lazar*

Department of Biological Sciences, Virginia Tech

1981 Kraft Drive, Blacksburg, VA 24061, USA

*Corresponding author: malazar@vt.edu

Author contributions

Designed the experiments: JD IML

Performed the experiments: JD SA

Performed COMSOL simulations: JD

Analyzed the data: JD IML

Wrote the paper: JD IML

Conceived and coordinated the study: IML

Key words: microfluidics, mass spectrometry, cell lysis, fast cell processing

136

Abstract

Analysis of intracellular proteins following external stimulation can further the

understanding of cell regulatory mechanisms. Mass spectrometry has emerged as one of

the most powerful analytical tools to perform this task. However, traditional bench-top

sample preparation methods require large devices and complicated operations, while the

existing microfluidic approaches focus mainly on single-cell processing, generating

insufficient sample to perform mass spectrometry . Here, we report novel integrated

microfluidic devices that enable cell stimulation and fast cell lysis. These microfluidic

devices, to the best of our knowledge, have the largest loading capacity so far, around 105

cells per chip. Cells were transferred into the chip using a syringe or pump, stimulated by

infusing EGF, and lysed either on-chip with a strong electrical field or off-chip using

sonication. The generated proteins were digested and analyzed with HPLC/MS, and the

results showed that the identified proteins were from essentially all cellular locations,

indicating that complete lysis was achieved. More than 70 KEGG pathways with an

enrichment P-value lower than 0.05 were represented. Overall, the devices are capable of

rapid cell stimulation and reagent-free cell lysis, and have promising potential to be

integrated into other lab-on-chip devices for more broad and complex applications.

7.1 Introduction

Cells are the elementary unit of life and they are constantly exposed to various stimuli

from the external microenvironment. Signals transduced into cells and can induce

changes in the composition of the cellular contents, such as the protein expression levels

and post-translational modifications1. Analysis of intracellular proteins is playing an

137

increasingly important role in understanding cell regulatory mechanisms, biomarker

characterization, and drug target discovery 2,3. Over the past decades, mass spectrometry

(MS) has emerged as one of the most powerful and also widely used technologies due to

its extremely high sensitivity and specificity, and feasibility of coupling with other

analytical technologies, such as HPLC, to perform a comprehensive analysis of complex

biological samples. Unfortunately, despite MS analysis can be highly automated and

rapid, the conventional sample preparation methods are often labor-intensive and time-

consuming. In addition, various lytic agents, such as SDS, Triton X-100 or proteinase K4,

are often needed to disrupt the cell membrane, resulting in the contamination of the

sample and extra clean-up steps. Therefore, studying the cell responses to external stimuli

in a high-throughput manner continues to be a challenge.

In recent years, microfluidics has gained dramatic interest and provided novel approaches

for a broad range of studies in molecular biology, clinical medicine and biomedical

sciences5-7. The main advantages of microfluidics include integration, miniaturization,

automation, and low sample and reagent consumption. So far, a number of microfluidic-

based cell lysis methods have been proposed. These methods can be classified in five

major groups: chemical lysis8-10, mechanical lysis11,12, electrical lysis13,14, optical lysis15,

and thermal lysis16,17. Among them, electrical/electroporation based cell lysis

microfluidic device have been the most widely studied, due to the ability to carry out

efficient and rapid cell lysis in a highly controlled manner without the risk of introducing

chemical contaminants. For example, Jokilaakso et al.13 developed a microfluidic chip

with a silicon nanowire and nanoribbon biological field effect transistors to position and

138

lyse single HT-29 human colon carcinoma cells. The cells were lysed within as little as 2

ms by applying a voltage of 600-900 mV at 10 MHz across the shortened source-drain

and the back gate of the transistor. Islam et al.14 successfully designed and used a

microfluidic device coupled with a commercially available nanoporous membrane to

perform electrophoretic concentration and electrical lysis of bacteria. The lysis efficiency

was determined to be ~90% with a potential of 300 V applied for 3 min. Many

microfluidic devices were designed for the purpose of DNA extraction18,19. Geng et al.18

reported a chip that enabled cell capture, electrical lysis, and DNA purification. As a

result, they were able to obtain a DNA extraction efficiency of 36% and 45% for

eukaryotic and bacterial cells, respectively, which was comparable to the widely used

chemical lysis, using high-intensity pulses (~2 kV/cm, a total duration of 1 s). Although

most of the proposed methods can achieve rapid reagent-free cell lysis, none of them,

unfortunately, can generate enough sample for performing comprehensive MS analysis.

Current microfluidic methods focus mainly on the manipulation and analysis of very

limited amount of cells, often at the single-cell level. Therefore, to enable reliable MS

analysis, an optimal method would integrate cell stimulation and rapid electrical cell lysis

for a relatively large number of cell (105 – 106).

Herein, we report the development of an integrated microfluidic platform that enables

cell stimulation, cell lysis, and protein extraction for MS analysis. This device, to the best

of our knowledge, has the largest capacity for cell loading, stimulation, and lysis. We also

optimized the geometric design of the channel layouts to allow rapid and even

distribution of cell stimulation solutions (i.e., with growth factors) over the entire cell

139

loading chamber. Both COMSOL simulation and fluorescent dyes were used to further

evaluate the flow behavior inside the chip. Cell lysis was performed immediately after

growth factor incubation either on-chip by applying a strong electrical field of ~2 kV/cm

or off-chip by using sonication. The efficiency of cell lysis was assessed by comparing

the cell morphology under a microscope before and after lysis. The cell lysate was

flushed out, collected, digested, and analyzed using HPLC/MS, and the results showed

that the identified proteins represent essentially all cellular locations, such as membrane,

mitochondria, and nucleus, indicating complete lysis of cells. Overall, this device is

capable of rapid cell stimulation and reagent-free cell lysis, and has promising potential

to be integrated into other lab-on-chip devices for more broad and complex applications.

7.2 Material and methods

Materials. SKBR3 human breast cancer cells, HBEC-5i human brain endothelial cells,

HMC3 microglia cells, phosphate buffered saline (PBS), and trypsin/EDTA were

purchased from the American Tissue Culture Collection (ATCC, Manassas, VA). Cell

culture media including McCoy’s 5A, EMEM, and DMEM/F12 (1:1) were obtained from

Life Technologies (Carlsbad, CA), fetal bovine serum (FBS) from Gemini Bio-Products

(West Sacramento, CA), Normocin from InvivoGen (San Diego, CA), human epidermal

growth factor (hEGF) from PeproTech (Rocky Hill, NJ). Sequencing grade modified

trypsin was purchased from Promega Corporation (Madison, WI). Endothelial cell

growth supplement (ECGS), urea, dithiothreitol (DTT), acetone, acetic acid,

trifluoroacetic acid (TFA), ammonium bicarbonate, protease inhibitor solution,

ammonium hydroxide were purchased from Sigma-Aldrich (St. Louis, MO). SPEC-PT-

140

C18 and SPEC-PT-SCX solid extraction pipette tips, and 5 µm/C18 Zorbax particles

were purchased from Agilent Technologies (Santa Clara, CA), and fused silica capillary

columns from Polymicro Technologies (Phoenix, AZ). HPLC-grade acetonitrile and

methanol were from Fisher Scientific (Fair Lawn, NJ), and DI water was prepared with a

MilliQ Ultrapure water system (Millipore, Bedford, MA). Buffered oxide etch (BOE) and

chromium etchant from Transene Company (Danvers, MA), MF-319 developer from

Rohm and Haas (Philadelphia, PA), and sulfuric acid from Mallinckrodt (St. Louis, MO).

Cell culture and processing. SKBR3 human breast cancer cells were cultured using

McCoy’s 5A medium supplemented with 10% FBS. HBEC5i cells were cultured in

DMEM/F12 supplemented with 10 % FBS and ECGS (40 µg/mL). Both cells were

incubated at 37 °C with 5% CO2 atmosphere. Normocin was added to the cell culture

media to a final concentration of 0.1 mg/ml to prevent contamination of bacteria, fungi,

and mycoplasma. The cells were harvested at full confluence by trypsinization, washed

with cold PBS, and used immediately for on-chip testing or stored at -80 °C until use. For

on-chip cell manipulation, fresh cells were resuspended in PBS or culture medium (~1:1

packed cell volume:PBS, v/v), transferred into a syringe and loaded into the chip. The

cells were stimulated by infusing EGF through the chip for ~3 min and lysed either on-

chip using a strong electric field strength (~2 kV/cm) or off-chip using sonication. For

on-chip lysis, the cell lysates were flushed out and collected as the protein samples. For

off-chip lysis, the intact cells were flushed out and lysed by intermittent sonication of 2

min in ice-cooled water bath. After centrifuging for 5 min at 16,000 rpm, the supernatant

was collected as the protein sample.

141

Tryptic digestion. The protein extracts were denatured/reduced with 8 M urea at ~57 °C

for 1 h, diluted 10-fold with NH4HCO3 (50 mM), digested with trypsin at 37 °C for

overnight, and quenched with glacial CH3COOH (100:1 digestion solution:glacial

CH3COOH, v/v). Alkylation was not performed as it may generate side-products. The

digested samples were cleaned-up using SPEC-PT-C18 and SPEC-PT-SCX solid

extraction pipette tips, brought to close to dryness in a vacuum centrifuge, and

resuspended in a solution of H2O/CH3CN/TFA 98:2:0.01 v/v to a final concentration 2

μg/μL, and further analyzed by nano-HPLC-MS.

LC-MS/MS analysis and data processing. The samples were analyzed using an LTQ mass

spectrometer purchased from Thermo Electron (San Jose, CA). The MS was operated in

the positive-ion mode with a capillary voltage of ~2kV for electrospray ionization (ESI).

Nano-LC was performed with an Agilent 1100 or 1260 micro-LC separation system and

in-house prepared nano-separation columns (100 μm i.d. × 360 μm o.d. × 10-12 cm

length, packed with 5 µm/C18 Zorbax particles). The nano-LC flow rate was generated

by splitting the LC pump flow from 10 μL/min to ~180-200 nL/min, with a 4-hour

concentration gradient of H2O/CH3CN/TFA from 96:4:0.01 v/v (solvent A) to

10:90:0.01 v/v (solvent B). Data-dependent acquisition was used to perform zoom/MS2

scan on the top 5 most intense peaks from each MS scan (5 microscans averaged), and

the data were acquired over a mass range of 500-2000 m/z. The collision-induced

dissociation (CID) parameters were 3 m/z ion isolation width, 35 % normalized

collision energy, 0.25 activation Q, 30 ms activation time, and the threshold for

142

triggering MS2 scans 100 counts. Conditions for data dependent analysis included: 5

m/z zoom scan width, 1.5 m/z exclusion mass width, dynamic exclusion at repeat

count 1, repeat duration of 30 s, exclusion list size 200, and exclusion duration 60 s.

Raw data files were analyzed with the Thermo Proteome Discoverer 1.4.0.288, using the

Sequest HT search engine (Thermo Electron) for performing searches against a Homo

sapiens protein database from UniProt (August 2015 download) comprising 20,197

reviewed/non-redundant sequences. Parameters for the database searches included: 500-

5000 mass range, precursor ion tolerance 2 Da, fragment ion tolerance 1 Da, b/y/a ion

fragments only, fully tryptic fragments with up to 2 missed cleavages allowed, and no

PTM allowed. Three technical replicates were performed for each sample. The peptide

false discovery rate (FDR) settings were 3 % and 1 % for relaxed and stringent database

searches, respectively. FDRs were calculated with the Target Decoy PSM Validator node

based on Xcorr vs. charge state values.

Chip fabrication. Chip layouts were designed using AutoCAD software (Autodesk, San

Rafael, CA) and flow simulation was performed using COMSOL Multiphysics software

(COMSOL AB, Stockholm, Sweden) to optimize the design. The photomasks were

prepared by HTA Photomask (San Jose, CA), the substrates were 1.6 mm thick white

crown glass coated with chrome and photoresist from Nanofilm (Shelton, CA), and the

UV light was purchased from OAI (San Jose, CA). Chips were fabricated using standard

photolithography and wet chemical etching techniques: the substrates were covered by

the photomask and exposed to UV, developed by developer, and etched using chromium

etchant and BOE to remove the exposed chrome and glass, respectively. The channel

143

depths were measured using a Dektak profilometer (Veeco, Plainview, NY), and etching

and measuring steps were repeated until the desired depths achieved. Connecting holes of

~1 mm i.d. were drilled in the substrates with a rotary tool (Dremel, Racine, WI). The

remaining photoresist and chrome layers were removed with acetone/methanol and

chromium etchant, respectively. The substrates were cleaned with sulfuric acid (5%),

distilled DI water, detergent, acetone, methanol, and boiled in a mixer of distilled DI

water/28% NH4OH/35% H2O2 (2:1:1, v/v), and bonded by gradual heating from room

temperature to 550 °C. A number of different chips featured 1–4 cell loading/lysis

chambers (50 μm deep × 600 μm – 1 mm wide × 1 cm long) were designed and

fabricated. The loading/lysis chambers were surrounded by filter channels (3 μm deep × 5

μm wide × 50 μm long) to allow the passage of buffer and the retention of cells.

Connecting ports on the chip were prepared from PEEK unions (Valco Instruments,

Houston, TX) cut evenly into two pieces. Cells were manually loaded into the chip using

1 mL disposable syringes (Becton Dickinson, Franklin Lakes, NJ) through the ports.

Platinum wires purchased from VWR were used as electrodes by connecting to an in-

house built 6-channel high voltage power supply controlled by LabVIEW 7 Express

software (National Instruments, Austin, TX) to enable fast cell lysis. Fluid flows and cell

behaviors on the chip were monitored with a Nikon epi-fluorescence microscope

(Melville, NY).

7.3 Results and Discussion

Design of the devices. The vast majority of the previous work on microfluidic cell lysis

focused on single-cell analysis. However, many proteins that play important roles in cell

144

functioning may be present at very low copy numbers (less than 1000 copies per cell).

With single-cell analysis, their abundances are far below the sensitivity limits of the

commonly used analytical methods, including mass spectrometry. To solve this problem,

we developed devices especially suitable for handling a relatively large amount of cells.

In this paper, two chip designs are presented and discussed. The schematic of the first

design is shown in Figure 7.1. The lysis chamber has the following dimensions: length=1

cm, width= 600 μm, and depth=50 μm (Figure 7.1A). Four identical chambers are placed

apart in equal distance and connected by symmetrical branch channels to the inlet port. A

total number of five identical collecting channels are placed on both sides of the four

chambers, and they are connected by fine filters that are perpendicular to the chambers on

the cover layer. As can be seen in Figure 7.1B, different dimensions and layouts of the

fine filters were designed and tested to achieve efficient and uniform cell loading and

stimulation. Regardless of the layouts, all the fine filters were etched to the same depth of

~3 μm. The backpressure can increase significantly if the filter is too shallow, and the

cells can be easily squeezed in if the filter is too deep. For chip operation, the port 1 in

Figure 7.1A was used as the inlet for cell loading and EGF stimulation, port 2 was the

outlet, and both of them were connected to electrodes for cell lysis. The cell lysate was

recovered by backflushing the device from port 2 to port 1. The schematic of the other

design is shown in Figure 7.2A. The lysis chamber has the following dimensions:

length=1 cm, width= 1 mm, and depth=50 μm. Unlike the first design, this device has

separate inlets and outlets for cell loading and EGF stimulation, and their directions are

perpendicular to each other. Besides, only one cell loading chamber is placed on each

chip to reduce the distance between the EGF stimulation inlet and outlet. For chip

145

operation, port 1 was the inlet for cell loading, 2 and 3 acted as the outlets during cell

loading, 4 and 5 were the inlet and outlet for EGF stimulation, respectively. In case of on-

chip electrical cell lysis, 1 and 6 were connected to electrodes. To collect lysate or intact

stimulated cells, when off-chip cell lysis was performed, the deviced was backflushed

from 2 and 3 to 1. Given an average cell size of ~15 μm in diameter, the volumn-based

theoretical loading capacity is ~105-106 cells per chip, which can provide enough sample

for the downstream proteomic analysis.

146

Figure 7.1 Schematic diagram of the first chip design (A) Schematic of the microfluidic chip with

multiple cell loading chambers. Each cell loading chamber has the following dimensions: L=1 cm, W=600

μm, D=50 μm (B) Schematic of the cover layer of the chip. The insert shows the four different designs of

the cover layer.

147

Figure 7.2 Schematic diagram and simulation of the second chip design. (A)Schematic of the

microfluidic chip featuring single cell loading chamber and perpendicular EGF infusing channels. The

chamber has the following dimensions: L=1 cm, D=60 um, W=1 mm. (B)(C)(D) On-chip time-dependent

EGF concentration profiles simulated using COMSOL. EGF solution is infused from the left inlet to the

right outlet at a flow rate of 1 µL/min. Red indicates high concentration, and blue indicates low

concentration. Three time points are presented: 0 s, 30 s, 50 s.

148

Figure 7.3 COMSOL simulation of the on-chip EGF infusing process. EGF solution is infused from the

left inlet to the right outlet at a flow rate of 1 µL/min. Red indicates high concentration, and blue indicates

low concentration. The concentration profile is presented in three time points: (A)Time=10 s; (B)Time=30

s; (C)Time=70 s.

149

COMSOL simulation of EGF infusion. The time-dependent EGF concentration profile

during stimulation was simulated using COMSOL for both designs. For simplicity, the

models were first constructed in 2-D and then converted into 3-D by adding a uniform

depth of 50 μm for all channels, though the actual depth of the fine filters were only ~3

μm. For this reason, the width of filters was reduced accordingly in the model to make

the volume of the filter channels unchanged. The model was built with the following

settings: incompressible laminar flow with no slip on the boundary, water as the material,

average mesh density, and all coefficients with default values. The concentration of the

EGF stock solution was set at 1 unit concentration and the flow rate was 1 µL/min. For

the first design, the concentration profiles at three time points, 10 s, 30 s, and 70 s, are

shown in Figure 7.3A-C. Blue indicates low EGF concentration, and red refers to high

concentration. The results show clearly that the four parallel chambers have the same

flow profiles at all time points, indicating that using multiple parallel chambers can be an

efficient method to increase capacity. However, one drawback of this design is also

obvious that the infusion is relatively slow inside the chambers, which may lead to

uneven EGF exposure times for cells at different locations, especially the end of the

chamber that seems to be a dead corner. As also observed, the parabolic flow profile in

the middle of chamber (Figure 7.3B) gradually became flattened as observed at chamber

end (Figure 7.3C). To accelerate the infusion and also enable more evenly distribution of

EGF, the second design was developed and the simulation results are shown in Figure

7.2B-D. The key feature of this device is that it distributes EGF solution into hundreds of

filter channels before the EGF reaches the chamber, which greatly facilitates an even

distribution of EGF across the chamber. Besides, since the stimulation channels are

150

perpendicular to the cell loading chamber, the traveling distance for EGF solution to

cover the entire chamber is significantly reduced from 1 cm to 1 mm. As a result, it only

took about 30 s for EGF to cover the entire chamber, counting from the time when EGF

just entered the chamber (Figure 7.2C) to when the EGF reached the other side of the

chamber (Figure 7.2D). It also needs to be noted that the time-scale of the infusion is

highly dependent on the flow rate. The input used in the model, 1 µL/min, was a

conservative choice for both designs, and hence shorter time-scales for the infusion

process can be achieved if a higher flow rate is used.

Flow visualization. To better understand the concentration profiles during stimulation,

rhodamine was used to visualize the flow. The rhodamine was prepared at a

concentration of 26 µM using 50 µM NH4HCO3 solution, as the fluorescence of

rhodamine is highly pH dependent and a basic environment is needed. A chip of the first

design (Figure 7.1) was first flushed with 50 µM NH4HCO3 solution to remove any air

bubble in the channels, and rhodamine solution was infused by a syringe pump at a flow

rate of 2 µL/min and monitored under a microscope. The flow profiles at two areas, the

head and the tail of the chamber, were recorded at four time points, as shown in Figure

7.4. The view fields at time zero were completely dark (Figure 7.4A), indicating the

absence of rhodamine. At time 3 min (Figure 7.4B), weak fluorescence can be observed

at the head of the chamber, but not at the tail, indicating the entering of rhodamine. This

delay of 3 min was caused by the dead volume of the tubing that connects the chip to the

syringe. After that, both chambers of the chip were quickly and evenly filled with

rhodamine, as we can see in Figure 7.4C that the head area was lightened up at time 4

151

min, and the entire chip also became bright within another 1 min (Figure 7.4D). Overall,

it took no more than 2 min, a total time of 5 min minus 3 min delay caused by the dead

volume, for rhodamine to infuse throughout the chamber, and this was in line with the

COMSOL simulation (Figure 7.3). As this visualization was performed using an empty

chip, it would not be a surprise to see the flow behave differently if the chip is loaded

with cells. For this reason, we performed another visualization of the flow in a chip

loaded with latex beads. The average size of the beads was around 11.6 µm in diameter,

roughly the same as SKBR3 and HBEC5i cells. The same chip was used under the same

conditions, and the results are shown in Figure 7.5. Based on Figure 7.5A/B, a delay of

3 min was also observed for the entering of rhodamine. But clearly it was more difficult

for the rhodamine to infuse in the fully packed chamber, to reach the tail area. At time 5

min (Figure 7.5C), intense fluorescent light was observed at the head of the chamber,

while the tail was almost dark. Even after another 15 min, the tail area only became

partially bright (Figure 7.5D). It was also obvious that the rhodamine solution leaked out

through the shallow perpendicular filters into the large collection channel, and then

slowly diffused back into the cell chamber. Therefore, for a fully loaded chip, the

distance for stimulation should be short enough to avoid uneven distribution of the

stimuli. One possible way of achieving this goal is to apply stimuli to the cell chamber in

a perpendicular direction, as in the second chip design (Figure 7.2).

152

Figure 7.4 Visualization of flow in an empty chip. A rhodamine solution of 26 uM was pumped through

the chip at a flow rate of 2 µL/min. Four time points are shown here: (A)Time=0 min; (B)Time=3 min;

(C)Time=4 min; (D)Time=5 min. Brighter area indicates a higher concentration of rhodamine.

153

Figure 7.5 Visualization of flow in a chip fully pakced with latex beads of 11.6 µm in diameter. A

rhodamine solution of 26 uM was pumped through the chip at a flow rate of 2 µL/min. Four time points

are shown here: (A)Time=0 min; (B)Time=3 min; (C)Time=5 min; (D)Time=20 min. Brighter area

indicates a higher concentration of rhodamine.

154

Figure 7.6 On-chip processing of HBEC5i cells. (A)Freshly harvested cells were centrifuged,

resuspended in medium as a hightly concentrated slurry, and loaded in a chip of the first deisgn (B)

Electrical cell lysis was performed by applying a electric field of 2 kV/cm for 2-3 min

155

Sample processing-cell loading, stimulation, and lysis. Retaining the cells in the chamber

can sometimes be difficult as mammalian cells have no rigid cell wall and can be

squeezed into the filters under a slight pressure. Even for filters with a depth of merely

1.6 µm, SKBR3 and HBEC5i cells were observed to enter during cell loading, which

resulted in the increase of backpressure and eventually the blockage of the entire chip.

Therefore, the key to enable a high-density cell loading was to use a highly concentrated

cell slurry, which can be loaded like a block, in one shot. To prepare the slurry, harvested

cells were centrifuged and resuspended in medium with a ratio of 1 : 1 (packed cell

volume : medium). The loading procedure and performance were the same for both

designs. As shown in Figure 7.6A, HBEC5i cells were fully loaded in a chip of the first

design without clogging the filters. Given the difficulty of infusing EGF along the long

axis of the chamber, cell stimulation was not performed with this device. Immediately

after cell loading, two electrodes with a potential of 3000-4000 V were connected to the

device to perform electrical lysis. Based on the average cell size, we determined the

threshold for cell lysis was ~1500 V/cm, resulting in a required potential of at least ~3000

V as the distance between the two electrodes was ~2 cm. Figure 7.6B shows the result of

a successful lysis that happened within 2-3 min. Clearly, no intact cells were present in

the field of view, indicating that an efficient and fast lysis process can be achieved.

However, we also noticed relatively high variances from run to run, and from chip to

chip. When applying such a strong electrical field on a microfluidic system, air bubble

formation and Joule heating can be serious problems. The most general formula to

describe the Joule’s Law of Heating is:

156

where P is the energy converted from electrical energy to thermal energy per unit time,

(VA-VB) refers to the potential drop, and I is the current. As the cell medium contains

various salts, for example, DMEM has at least 6 major inorganic salts with

concentrations ranging from 0.8 mM to 110 mM, this results in high conductivity and

hence a high level of current traveling in the system. Combined with a high potential

drop, the temperature can be increased quickly and dramatically in the confined

chambers. The electrolysis of water into oxygen and hydrogen was also observed during

lysis, and the generated air bubbles can be dragged into the chamber by electroosmotic

flow to result in poor or even complete loss of electrical continuity. Because of the

challenges associated with on-chip electrical lysis, we considered and tested an

alternative fast off-chip cell lysis method using sonication. A chip of the second design

was used to perform the test of cell loading, EGF stimulation, and off-chip cell lysis by

sonication. The results are shown in Figure 7.7. High-density SKBR3 cell loading was

accomplished without any trouble (Figure 7.7A), and EGF solution with protease

inhibitors and phosphatase inhibitors was infused through the chamber in a perpendicular

direction at a flow rate of ~1 µL/min. As we can see in Figure 7.7B, all the cells were

pushed downwards uniformly by the EGF, demonstrating evenly distributed EGF

solution through the chamber. After stimulation for 3 min, the cells were immediately

flushed out using a hypotonic buffer (~100-150 µL) containing 10 mM

NH4HCO3/CH3OH (5 %) then collected into a 1.5 mL Eppendorf vial and subjected to

sonication with an ice-cold water bath. Figure 7.7C shows that the chamber was totally

empty after flushing, which means the sample can be effectively recovered. The total

processing time for flusing and collecting the cells, and the sonication lysis was <5 min.

157

Due to the presence of protease inhibitors and phosphatase inhibitors, the phospho-

signaling events resulted from the EGF stimulation can be fixed and captured at the

moment the cell lysed.

Figure 7.7 Sample processing on a chip of the second design. (A) SKBR3 cells loaded in the chip; (B)

SKBR3 cells stimulated with EGF solution infused at a flow rate of ~1 µL/min for ~ 3 min; (C) Stimulated

SKBR3 flushed out of the chip with ~200 µL hypotonic buffer for off-chip sonication lysis.

158

HPLC-MS analysis and data interpretation. The cell lysates, from both on-chip and off-

chip lysis, were digested, cleaned-up, and analyzed using HPLC-MS. The sample

originated from each chip was dissolved in a solvent composed of H2O/CH3CN/TFA

from 98:2:0.01 v/v to a final volume of 50 µL before injected into MS. For the sample

collected from on-chip lysis of HBEC5i cells, a total number of 485 proteins were

detected from two HPLC-MS runs, among which, 264 and 243 proteins were from

cytoplasm and nucleus, respectively, indicating complete lysis of the cells. Proteins

involved in critical phospho-signaling pathways, such as ERBB, MAPK, and PI3K-Akt

were also identified, which implies the potential application of this device for the study of

phospho-signaling pathways. For the off-chip lysis, three biological replicates with two

HPLC-MS runs for each were performed. In total, 614 proteins were detected and

identified, which represent 34 KEGG pathways with P-value lower than 0.05. To further

interrogate the phospho-signaling events, phosphopeptide enrichment along with more

sensitive detection method should be applied. For this purpose, this device could be

coupled to another microfluidic device, such as the one that we have previously reported,

which enables sensitive detection and quantification of phosphopeptide using TiO2-based

and multiple reaction monitoring (MRM)-MS detection.

7.4 Conclusions

In this report, we demonstrated microfluidic devices for cell stimulation and fast cell lysis

using either electrical fields on-chip or sonication off-chip. The dimensions of the devices

were large enough to handle 105-106 cell each time, generating sufficient amount of

sample to perform downstream proteomic analyses. The cell stimulation process was

159

simulated using COMSOL, and visualized using a fluorescent dye. By monitoring the

flow and cell movements in the chip, the results were in line with each other. Complete

electrical cell lysis could be achieved, though, the reproducibility was not ideal mainly

due to the lack of ability to apply a sufficiently high voltage to induce uniform lysis. But

this problem can be addressed by a number of methods, such as using embedded

electrodes with short distance to reduce the required potential, or having geometric

variations on the chip channel to amplify the electrical field strength and enable cell lysis

at a certain area. For off-chip lysis by sonication, the process can be accomplished in <5

min without introducing any lytic agent, demonstrating its potential applications in the

study of fast cellular events. These devices can be integrated or coupled to other

microfluidic devices for more broad and complex applications.

Acknowledgments. This work was supported by the National Science Foundation and in

part by the National Institutes of Health, through awards NSF/DBI-1255991 and

NIH/NIGMS-1R01GM121920-01A1 to IML.

7.5 References

1. M. R. Junttila, S.-P. Li, J. Westermarck. Phosphatase-mediated crosstalk between

MAPK signaling pathways in the regulation of cell survival. FASEB J. 2008, 22, 954.

2. C. Choudhary, M. Mann. Decoding signalling networks by mass spectrometry-based

proteomics. Nat. Rev. Mol. Cell Biol. 2010, 11, 427.

3. W. Kolch, A. Pitt. Functional proteomics to dissect tyrosine kinase signalling pathways

in cancer. Nat. Rev. Cancer 2010, 10, 618.

160

4. X. Chen, D. Cui, C. Liu, H. Cai. Microfluidic Biochip for Blood Cell Lysis. Chinese J.

Anal. Chem. 2006, 34, 1656.

5. M. S. Islam, A. Aryasomayajula, P. R. Selvaganapathy. A review on macroscale and

microscale cell lysis methods. Micromachines 2017, 8, DOI 10.3390/mi8030083.

6. G. Du, Q. Fang, J. M. J. den Toonder. Microfluidics for cell-based high throughput

screening platforms-A review. Anal. Chim. Acta 2016, 903, 36.

7. L. Nan, Z. Jiang, X. Wei. Emerging microfluidic devices for cell lysis: a review. Lab

Chip 2014, 14, 1060.

8. D. J. Eastburn, A. Sciambi, A. R. Abate. Ultrahigh-throughput mammalian single-cell

reverse-transcriptase polymerase chain reaction in microfluidic drops. Anal. Chem. 2013,

85, 8016.

9. E. A. Schilling, A. E. Kamholz, P. Yager. Cell lysis and protein extraction in a

microfluidic device with detection by a fluorogenic enzyme assay. Anal. Chem. 2002, 74,

1798.

10. S. C. Kim, I. C. Clark, P. Shahi, A. R. Abate. Single-Cell RT-PCR in Microfluidic

Droplets with Integrated Chemical Lysis. Anal. Chem. 2018, 90, 1273.

11. J. Kim, J. W. Hong, D. P. Kim, J. H. Shin, I. Park. Nanowire-integrated microfluidic

devices for facile and reagent-free mechanical cell lysis. Lab Chip 2012, 12, 2914.

12. S.-S. Yun, S. Y. Yoon, M.-K. Song, S.-H. Im, S. Kim, J.-H. Lee, S. Yang. Handheld

mechanical cell lysis chip with ultra-sharp silicon nano-blade arrays for rapid

intracellular protein extraction. Lab Chip 2010, 10, 1442.

161

13. N. Jokilaakso, E. Salm, A. Chen, L. Millet, C. D. Guevara, B. Dorvel, B. Reddy, A.

E. Karlstrom, Y. Chen, H. Ji, Y. Chen, R. Sooryakumar, R. Bashir. Ultra-localized single

cell electroporation using silicon nanowires. Lab Chip 2013, 13, 336.

14. M. S. Islam, A. Shahid, K. Kuryllo, Y. Li, M. J. Deen, P. R. Selvaganapathy.

Electrophoretic concentration and electrical lysis of bacteria in a microfluidic device

using a nanoporous membrane. Micromachines 2017, 8, DOI 10.3390/mi8020045.

15. P. A. Quinto-Su, H.-H. Lai, H. H. Yoon, C. E. Sims, N. L. Allbritton, V.

Venugopalan. Examination of laser microbeam cell lysis in a PDMS microfluidic channel

using time-resolved imaging. Lab Chip 2008, 8, 408.

16. L. A. Marshall, L. L. Wu, S. Babikian, M. Bachman, J. G. Santiago. Integrated

printed circuit board device for cell lysis and nucleic acid extraction. Anal. Chem. 2012,

84, 9640.

17. S. Baek, J. Min, J.-H. Park. Wireless induction heating in a microfluidic device for

cell lysis. Lab Chip 2010, 10, 909.

18. T. Geng, N. Bao, N. Sriranganathanw, L. Li, C. Lu. Genomic DNA extraction from

cells by electroporation on an integrated microfluidic platform. Anal. Chem. 2012, 84,

9632.

19. S. K. Jha, R. Chand, D. Han, Y.-C. Jang, G.-S. Ra, J. S. Kim, B.-H. Nahm, Y.-S.

Kim. An integrated PCR microfluidic chip incorporating aseptic electrochemical cell

lysis and capillary electrophoresis amperometric DNA detection for rapid and

quantitative genetic analysis. Lab Chip 2012, 12, 4455.

162

CHAPTER 8: Summary and Outlook

By combining two powerful technologies, MS and microfluidics, new opportunities and

insights to proteomics and phosphoproteomics can be obtained. In this thesis, I described

my project in developing a novel MS-Microfluidics approach and also demonstrated its

application in fast sampling and sensitive detection of cellular phosphorylation events.

The study was accomplished in four major steps. First of all, we proposed the use of

accelerated enzymatic reactions for protein analysis, and assessed its performance by

evaluating the impact of the enzymatic digestion time on the quality of generated

peptides (Chapter 4). The results showed that a rapid digestion process could be superior

to established protocols in terms of achieving protein and proteome coverage, with no

sacrifice in the ability to perform efficient tandem mass spectrometric analysis. Second,

we designed a microreactor that enables fast proteolytic digestion and selective

enrichment in phosphopeptides, followed by on-line ESI-MS detection (Chapter 5). The

performance was evaluated using both bovine standards and SKBR3 cell extract, and the

results were comparable to, if not better than, the outcomes resulting from the

conventional method. Third, we developed an integrated microfluidic platform for the

targeted detection and analysis of phosphopeptides involved in signaling networks

(Chapter 6). The performance of the approach was demonstrated using tryptic digests of

SKBR3 proteins spiked with stable isotope-labeled standards, and low nanomolar

concentrations of endogenous phosphopeptides involved in MAPK signaling could be

detected. Lastly, microfluidic devices that enable live cell loading, stimulation, and fast

163

lysis, were developed and demonstrated (Chapter 7). These developed devices (Chapter

5, 6, 7) represent the key elements to implement fast sampling, processing, and analysis

of cellular signaling events. Each of the approaches is the first of its kind and shares a

number of advantages over the conventional strategies, including but not limited to: (a)

less sample and reagent consumption, making them the preferred methods for the

analysis of scarce sample; (b) simplicity, as the microfluidic designs rely only on the use

of microfluidic reactors of simple construction; (b) ease of operation, as all the functional

elements are embedded in chip, and sample processing is enabled solely by sequential

loading/elution on and from the microfluidic reactors; (c) speed, as the on-chip

opreations, such as cell lysis, tryptic digestion, and phosphopeptide enrichment, occur

within a few minutes and the total sample processing time is less than 1-2 hours; and (d)

versatility, as any given microfluidic device has the potential to be integrated into more

complex workflows to further increase the analytical power.

Our studies laid the groundwork for technology development and hence will facilitate

further understanding of the dysregulated phospho-signaling pathways in aberrant cancer

cells. Many cancers are found to have deregulated signaling pathways at a time-frame

from within seconds to tens of hours upon external stimulations. Thus, it is crucial to

monitor, identify, and compare all the changes, such as protein expression level,

modification status, and the dynamic profile, to gain more comprehensive insights into

cancer development. Many proteins involved in phospho-signaling pathways are highly

conservative, and their expression level and modification status have been shown to

correlate with the stage of many diseases. This characteristic suggests that these

164

phosphoproteins can serve as potential biomarkers for cancer diagnosis and prognosis,

and hence our strategy can facilitate the discovery of biomarkers that can be used for the

early diagnosis of cancers and prediction of tumor progression, drug response, and

clinical outcome. The future work will mainly focus on three aspects: (a) To further

optimize and integrate the system, including increase the efficiency, sensitivity,

specificity, and throughput; (b) to perform quantitative analysis of the temporal and

spatial profiles of phosphoproteins that play critical roles in signaling pathway; (c) to

enable parallel monitoring and real-time proteomic analysis of cells that are exposed to

various external stimulations or drug treatments.