Red-shifted Azobenzenes and Photocontrol of Protein ... · This thesis elucidates the mechanism by...

Red-shifted Azobenzenes and Photocontrol of Protein Affinity Reagents

by

Amirhossein Babalhavaeji

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Department of Chemistry University of Toronto

© Copyright by Amirhossein Babalhavaeji 2018

ii

Red-shifted Azobenzenes and Photocontrol of Protein Affinity

Reagents

Amirhossein Babalhavaeji

Doctor of Philosophy

Department of Chemistry

University of Toronto

2018

Abstract

Azobenzene derivatives are a class of molecular switches that have found broad use in

photocontrol of various biological systems. A key feature of azobenzenes designed to photoswitch

in vivo is the wavelength of light that is required for their photoisomerization. To pass through

tissue, wavelengths in the red, far-red, or ideally near-infrared region of the spectrum are required.

This thesis elucidates the mechanism by which a tetra-ortho-methoxy aminoazobenzene derivative

exhibits robust red light switching by forming azonium ions at physiological pH. With the design

and characterization of other azobenzenes, it is shown that the photophysical properties of azonium

ions can be tuned to achieve water-stable near-infrared photoswitches. Finally, a generalizable

approach is proposed for obtaining light controllable bioactive agents by modifying the scaffold

of a protein affinity reagent using an azobenzene cross-linker. It is shown that the azobenzene-

cross-linked affinity reagent is a modular system where the photoswitch, and the affinity reagent’s

target can be independently altered to optimize the system for a particular application.

iii

ی نفزودیززج حیرتم از حیات چ دآورد هب اضطرارم اول هب وجو

بودن و رفتن مقصودزین آدمن و دورفتیم هب ارکاه و ندانیم هچ ب هجری خورشیدی( 724عمر خیام نیشابوری )زاده رد اردیبهشت

Coercing me into existence,

life did nothing but confound me.

When compelled to leave, I will still wonder:

Was there a purpose to this inception, pause and departure?

From the quatrains of Omar Khayyam of Nishapur (1048-1131 CE)

iv

Acknowledgements

I owe my completion of the PhD program, as with all my previous achievements, first and foremost

to Farahnaz, my mother, and Fereydoon, my father. Words cannot express my gratitude to them

for all their love and sacrifice.

I am honoured to have been a graduate student of Professor Andrew Woolley. He is a patient

mentor, an influential teacher, and a thoughtful researcher. I cannot thank him enough for guiding

me through the program and for being incredibly supportive and understanding when I had to cope

with the loss of loved ones amidst a heavy work-load. I also thank him for leading a fantastic

research group.

Next, I thank my other family members without whose support I could not have succeeded in

graduate school. Of special mention are my aunt, Faranak, and her husband Hamid who welcomed

me in Toronto and continue to support me in every aspect of my life; my late uncle, Farshid, who

always passionately encouraged me in pursuing higher education; and my aunt, Mojgan, who has

always offered her wisdom when I sought counsel for making important decisions.

Other members of the Woolley laboratory have all contributed to the collegial atmosphere that

helped make my experience of graduate school rewarding. I am lucky to have had such great

colleagues. I would like to give special thanks to Dr. Ahmed Ali who among other members of the

lab taught me some of the skills that I needed to begin my research in biological chemistry.

I thank Dr. Jakeb Reis for our numerous scientific and non-scientific conversations and for

introducing me to some of the best places in Toronto to get coffee. I thank Lulu Lu and Kate

Brechun, for being good friends, beyond good colleagues, and Anna Jaikaran for all the thought

provoking lunch time discussions.

I am privileged to have worked with several undergraduate students and international visiting

researchers in the lab, and to have taught many students in the undergraduate instructional labs.

My interactions with them benefited me in both my knowledge of chemistry and the art of teaching.

I would like to express my gratitude to all of them.

I would also like to thank my research collaborators from whom I learnt a lot, and the community

in the Department of Chemistry.

v

I would also like to acknowledge all my past teachers; in particular, I thank Ms. Mehrnoosh Naderi,

my exceptional Analytical Chemistry instructor at Sharif University in Tehran.

And last, but not least, I give thanks to all other lovely friends, in particular Dr. Ali Eghtesadi and

Ahmad Golaraei, whose support and kindness continue to give me strength in the face of every

challenge.

vi

Table of Contents

Acknowledgements ........................................................................................................................ iv

Table of Contents ........................................................................................................................... vi

List of Tables ...................................................................................................................................x

List of Figures ................................................................................................................................ xi

List of Appendices ....................................................................................................................... xiii

Introduction .................................................................................................................................1

1.1 Remote control of biology with light ...................................................................................1

1.2 Naturally photoswitchable proteins .....................................................................................2

1.3 Azobenzene, a photoswitchable small molecule .................................................................3

1.4 Applications of azobenzene in biology ................................................................................4

1.5 Azobenzenes in vivo ............................................................................................................5

1.6 Challenges of long-wavelength photoswitch design ............................................................6

1.7 Push-pull azo systems ..........................................................................................................7

1.8 Azo dyes that absorb at long wavelengths: C2 bridged azobenzenes (diazocines) .............9

1.9 Heterocycles, BF2 adducts and two-photon chromophores ...............................................10

1.10 Ortho substitution...............................................................................................................12

1.11 Tetra-ortho-methoxy substitution ......................................................................................13

1.12 Other chromophores exhibiting long wavelength switching .............................................14

1.13 Azobenzene-based protein cross-linkers ............................................................................17

1.14 The objectives of this thesis ...............................................................................................18

1.15 References ..........................................................................................................................20

A red light switching azonium ion ............................................................................................27

2.1 Introduction ........................................................................................................................27

2.2 Methods..............................................................................................................................29

2.2.1 Computational Methods .........................................................................................29

vii

2.2.2 Synthesis ................................................................................................................30

2.2.3 Cross-linking of peptide AB15 to obtain 2: ...........................................................30

2.2.4 Electronic absorption spectroscopy: ......................................................................31

2.2.5 Determining the pKa of trans-2: .............................................................................31

2.2.6 Determining the pKa of trans-3: .............................................................................31

2.2.7 Determining the pKa of cis-2: ................................................................................32

2.2.8 Glutathione stability ...............................................................................................33

2.2.9 Photoswitching in whole blood ..............................................................................33

2.3 Results and Discussion ......................................................................................................33

2.4 References ..........................................................................................................................47

Red, far-red, and near infrared photoswitches based on azonium ions .....................................50

3.1 Introduction ........................................................................................................................50

3.2 Materials and methods .......................................................................................................52

3.2.1 Computational methods .........................................................................................52

3.2.2 Sample preparation for UV-Visible spectroscopy .................................................53

3.2.3 Calculation of molar absorption coefficients for 1-4 .............................................54

3.2.4 pH-dependence of absorption spectra ....................................................................55

3.2.5 Thermal relaxation kinetics for 2 ...........................................................................55

3.2.6 Laser flash photolysis experiments ........................................................................56

3.3 Results and discussion .......................................................................................................57

3.3.1 Azonium ions based on the ortho-meta substitution pattern ..................................57

3.3.2 Designing a slow relaxing near-IR azobenzene switch .........................................67

3.4 Conclusions ........................................................................................................................74

3.5 References ..........................................................................................................................75

Modular Design of Optically Controlled Protein Affinity Reagents ........................................79

4.1 Significance of this study ...................................................................................................79

viii

4.2 Introduction ........................................................................................................................80

4.3 Materials and methods .......................................................................................................83

4.3.1 Protein production ..................................................................................................83

4.3.2 Production of 15N-labelled proteins .......................................................................84

4.3.3 Cross-linking of Fynomer with azobenzene cross-linkers .....................................85

4.3.4 Electronic absorption spectroscopy .......................................................................86

4.3.5 Estimates of the percentage of cis/trans in photostationary states (PSS) ...............87

4.3.6 Thermal relaxation rates ........................................................................................87

4.3.7 NMR spectroscopy.................................................................................................88

4.3.8 Activity assay .........................................................................................................89

4.4 Results and discussion .......................................................................................................92

4.5 Conclusions ......................................................................................................................102

4.6 References ........................................................................................................................103

4.7 Addendum: Exploring stilbenes cross-linkers as models for pure cis and trans

azobenzenes .....................................................................................................................107

4.7.1 Introduction ..........................................................................................................107

4.7.2 Materials and Methods .........................................................................................109

4.7.3 Results and Discussion ........................................................................................114

4.7.4 Conclusion ...........................................................................................................118

4.8 References ........................................................................................................................118

Summary and future directions ...............................................................................................120

5.1 Red-shifted azobenzene photoswitches ...........................................................................120

5.2 Photocontrol of protein function using azobenzene cross-linkers ...................................123

5.3 References ........................................................................................................................124

Appendices ..............................................................................................................................126

6.1 Relaxation kinetics for compounds discussed in Chapter 3 .............................................126

ix

6.2 Mass spectra for cross-linked Fynomers..........................................................................133

6.3 Select azonium structures and their calculated max values .............................................135

x

List of Tables

Table 2-1. QTAIM topological parameters for bond critical points shown in Figure 2-5a...........37

Table 2-2. Calculated gas phase free energies of protonations of trans-2’ and cis-2’ .................. 42

Table 2-3. Calculated free energies of the isodesmic reaction ..................................................... 43

Table 3-1. Predicted and observed properties of compounds 1-4. ................................................ 62

Table 3-2. TD-DFT results for compounds 2, 4, 5, 6 and 7 .......................................................... 71

Table 3-3. Selected TD-DFT data for cis and trans isomers of 7 (a model for 8) ........................ 72

Table 4-1. Sample preparation scheme for quantitative 1H-NMR experiments. ........................ 111

Table 4-2. Molar absorption coefficients obtained for trans and cis stilbenes ........................... 112

xi

List of Figures

Figure 1-1. Examples of azobenzene derivatives that act on biomolecules. .................................. 5

Figure 1-2. Some long wavelength azo dyes. ................................................................................. 8

Figure 1-3. Bridged azobenzenes. ................................................................................................. 10

Figure 1-4. Examples of azoheterocyclic, BF2 adduct, and 2-photon absorbing photoswitches. . 11

Figure 1-5. Some ortho-amino-substituted azobenzenes photoswitches. ..................................... 12

Figure 1-6. Tetra-ortho-methoxy (TOM) substituted switch and UV-Vis spectra ....................... 14

Figure 1-7. Non-azo red and near-IR switches based on indigo, DASA, and DHP ..................... 16

Figure 2-1. Neutral and protonated forms of azobenzenes ........................................................... 28

Figure 2-2. UV-Vis spectra of 2 and 3 cross-linked to a peptide at various pHs. ........................ 34

Figure 2-3. UV-Vis spectra of 3 cross-linked to a peptide at various pHs ................................... 34

Figure 2-4. Photoswitching (UV-Vis spectra) of 2 cross-linked to the peptide AB15 ................. 35

Figure 2-5. Calculated azo/azonium structures and resonance representations ............................ 37

Figure 2-6. Molecular graphs showing QTAIM analysis on the trans/cis azonium ions ............. 37

Figure 2-7. Predicted vs. experimental UV-Vis for the neutral and protonated species (2’) ....... 39

Figure 2-8. Rate of thermal relaxation as function of pH ............................................................. 40

Figure 2-9. UV-vis spectra of photoswitching in whole blood ..................................................... 46

Figure 2-10. Stability of 2-cross-linked AB15 peptide in presence of 10 mM glutathione .......... 47

Figure 3-1. UV-Vis spectra of trans 1-4 in aqueous buffer as a function of pH. .......................... 60

Figure 3-2. Computational models of 1-4. .................................................................................... 64

xii

Figure 3-3. Far-red photoisomerization of 2 and thermal cis-to-trans relaxation of 1-4 as a

function of pH in aqueous buffer .................................................................................................. 66

Figure 3-4. Computational model of 5. ......................................................................................... 68

Figure 3-5. Computational model of 6 .......................................................................................... 69

Figure 3-6. Computational model of a p-pyrrolidine derivative of 7 ........................................... 70

Figure 3-7. Simulated spectra of 7 (a model for 8) based on TD-DFT calculations. ................... 71

Figure 3-8. Near-IR photoswitching and UV-Vis spectra of DOM-azo. ...................................... 74

Figure 4-1. Models showing the overall design of photoswitchable Fynomer ............................. 82

Figure 4-2. Photoswitching and UV-Vis spectra of BSBCA-Fynomer ........................................ 94

Figure 4-3. A light vs. dark difference NH-HSQC spectrum of BSBCA-Fynomer ..................... 95

Figure 4-4. BSBCA-Fynomer inhibition of chymase activity ...................................................... 96

Figure 4-5. IC50 curves for photoswitchable vs. unmodified Fynomer ....................................... 97

Figure 4-6. UV irradiation does not affect inhibitory activity of uncross-linked Fynomer ......... 97

Figure 4-7. Thermal relaxation of BSBCA-Fynomer. .................................................................. 98

Figure 4-8. Testing reversibility of photoswtiching of activity on Fynomer-BSBCA. ................ 99

Figure 4-9. UV-Vis spectra of TOM-Fynomer before and after irradiation ............................... 100

Figure 4-10. Difference HSQC spectrum, and activity assay for TOM-Fynomer ...................... 101

Figure 4-11. Overlayed HSQCs of folded states of BSBCA-Fynomer and TOM-Fynomer. ..... 102

Figure 4-12. Computational models for stilbene cross-linkers ................................................... 108

Figure 4-13 Chymase inhibition assay with stil-Fynomer ......................................................... 116

Figure 4-14. CD spectra obtained from trans vs. cis stil-FK11 peptide ..................................... 117

xiii

List of Appendices

6.1 Relaxation kinetics for compounds discussed in Chapter 3 .............................................126

6.2 Mass spectra for cross-linked Fynomers discussed in Chapter 4 ....................................133

6.3 Select azonium structures and their calculated max values .............................................135

1

Introduction

Note: Some sections of this chapter have been reproduced, by permission from The American

Chemical Society, from the following published review article: Dong, M., Babalhavaeji, A., Dong,

M. X., Samanta, S., Beharry, A., & Woolley, G. A. (2015). Red shifting azobenzene photoswitches

for in vivo use. Accounts of Chemical Research, 2015, 48(10), 2662-2670. The text has been

modified for consistency.

The original article can be obtained via the following link:

https://pubs.acs.org/doi/10.1021/acs.accounts.5b00270

1.1 Remote control of biology with light

Using light to direct the activity of molecules is a promising approach that can enable scientists to

better understand and control complex processes that operate in a biological system. It is possible

to tune a light source to a desired wavelength, shine light in a precisely time-controlled manner,

set its intensity to a desired level, and focus it onto a finite area. These features make light a useful

reagent to manipulate a biochemical system.1 If the activity of a biomolecule of interest is

adjustable by its illumination with a specific colour of light, one can decide where, when, and how

much the biomolecule should be functional via controlling irradiation in space and time. This

strategy is useful for elucidating the role of individual components of a biological system or in

therapeutics where drug action could be remotely targeted with light’s high temporal resolution.2-

3 However, while all biomolecules respond to non-destructive irradiation by absorbing certain

wavelengths, they rarely change their function as a result. To enable the use of light in biology,

therefore, methods must be devised to confer light control to a molecule that is otherwise

functionally indifferent to light. This thesis presents a detailed study of several photoswitchable

compounds, based on the azobenzene chromophore, that help advance the goal of optical control

2

of biological systems. Further, it describes a novel application of such compounds in remote

control of protein activity.

1.2 Naturally photoswitchable proteins

Through evolution, living organisms have become able to respond to light in different ways to

adapt to varying environmental stimuli.4 Response to light is often regulated by photosensitive

proteins.5 The most familiar example of a light sensing biological system is the human eye where

proteins called rhodopsin absorb photons in the visible range of the electromagnetic spectrum and

through changing their conformation, initiate a signalling cascade that ultimately leads to vision.6

Several other light-responsive systems derived from bacteria, fungi, and plants have been studied

and efforts have been made to utilize these naturally photo-active systems to engineer

photocontrolled biomolecules.5 Microbial channelrhodopsins7, light-oxygen-voltage (LOV)

domains,8 photo-active yellow protein (PYP)9 and phytochromes10 are examples of naturally light-

switchable proteins that have been repurposed to control the activity of other proteins with light.11-

14 In all these proteins, response to light is mediated through a covalently bound chromophore that

isomerizes upon irradiation at a certain wavelength, and its geometrical change is translated into a

larger conformational change in the protein tertiary structure.15 The conformational change of the

protein then leads to a change of function, such as a different binding affinity to a partner molecule,

or a change in enzymatic activity.5

A simple approach to confer photoswitching to a protein of interest is to design a construct by

fusion of the target to the light-switchable protein. In such a system, it is possible, in principle, to

alter the availability of the target moiety by photoswitching the light-responsive moiety.5 The key

advantage of utilizing natural photo responsive proteins for photocontrol in biology is that they

3

can be genetically encoded in an organism. This is advantageous when the goal of optical control

is to help answer questions about a biological system and minimal external interference is ideal.16

1.3 Azobenzene, a photoswitchable small molecule

There are various small molecules that respond to light and change their conformation. The

photophysical properties of rhodopsin-like chromophores, stilbenes, dithienylethenes, chromenes,

spiropyrans, and fulgides have been studied for decades.17 Some have been used to develop

photoreceptors.18 However, none, other than azobenzenes, have been used extensively as a

photoswitch for optical control.19 The popularity of the azobenzene chromophore results from a

number of factors including its robust photocycle, large conformational change upon

photoisomerization, tunability by derivatization, relative ease of synthesis, and well-studied

photochemistry.19

Scheme 1-1. Isomerization of azobenzene

Azobenzene occurs in two geometrical isomers: The thermodynamically more stable trans isomer

can be photoisomerized to the less stable cis isomer with UV irradiation. The cis isomer can

thermally relax to trans in the dark or can be photochemically isomerized to trans by blue

irradiation. (Scheme 1-1) This process occurs efficiently, without major side reactions, and

continues to occur when azobenzene is derivatized in numerous ways.20

4

1.4 Applications of azobenzene in biology

In 1937, Hartley discovered that shining UV light on trans azobenzene produced the cis isomer.21

In the late 1960s, Erlanger applied azobenzene to the photo-control of enzymes and ion channels.22-

23 This was done by modifying ligands or inhibitors or even allosteric activators with azobenzene

units.24 With advances in molecular and structural biology, it has now become possible to tether

azobenzene derivatives site-specifically to essentially any protein target in living cells, and in some

cases in vivo (e.g. in transparent zebrafish).25 This can be accomplished via reaction with a Cys

residue introduced via site-directed mutagenesis, or by using biorthogonal ligation methods

together with nonsense codon suppression strategies.26-29 Targeted protein modification with

azobenzenes has now led to light switchable Glu receptors,30-31 K+ channels,26 acetylcholine

receptors,32 and GABA receptors33 as well as kinases,29 kinesins,34 and transcription factors35

(Figure 1-1).

In parallel with these developments has been the realization that drugs could be made light

sensitive via incorporation of an azobenzene unit.3, 36-37 “Azologized” drugs now include the

GABA receptor potentiator propofol,38 the opioid receptor agonist fentanyl,39 antibacterial

quinolones,40 and ATP-sensitive potassium channel active sulfonylureas (Figure 1-1).41 While it

remains to be seen how such drugs will behave in whole organisms, the possibility of remote in

vivo optical control of a drug is tantalizing. It directly addresses a primary problem of drug design

by making the drug active only where and when light is applied. “Photopharmacology”, a term

that appeared in a science fiction context in the 1970s,42 has now become a real possibility.3, 43

5

Figure 1-1. Examples of azobenzene derivatives that act on biomolecules. For references, see 1,29

2,44 3,45-46 4,40 and 5.41

1.5 Azobenzenes in vivo

For an azobenzene photoswitch to operate in vivo, a number of criteria have to be met. Besides

being non-toxic, the switch must be resistant to hydrolysis and to reduction of the azo group, which

would render it inoperable. The metabolism of many azobenzenes has been analyzed by scientists

interested in the effects of widely-used azo-based food dyes. It varies greatly with the structure of

the dye and the route of administration (e.g. whether or not it is exposed to gut bacteria).47 This

body of knowledge is likely to be very useful in guiding the development of photopharmaceuticals.

Photoswitching has not been considered in any of these studies, only the metabolism and toxicity

of the stable (trans) forms of the dyes. To address in vivo stability of photoswitching, Beharry and

colleagues monitored photoswitching of a bis-p-amido substituted azobenzene using a fluorescent

6

reporter assay and found photoswitching behaviour was retained for days in developing

zebrafish.48 Bis-p-amido substituted azobenzene is the core unit of several photoswitches that have

been developed to target biomolecules (e.g. 1 and 3 in Figure 1-1). Newly developed switches are

now tested for their stability to reduction by glutathione, the predominant intracellular reductant.

Some switches are stable to reduction and some are not, depending on the specific structure of the

switch in question.49-50 The widely held concern that all azobenzenes would be reduced in vivo51

is too simplistic a view.

Aside from stability in vivo, a key feature that will determine the success of azobenzene-based

tools for biological research and photopharmacology is the wavelength required to cause

photoisomerization. With the exception of the eye (where azobenzene-based drugs to treat certain

types of blindness are making real progress44), the body is opaque throughout most of the visible

range. The UV light needed to cause trans-to-cis isomerization of unmodified azo compounds is

strongly absorbed and scattered by skin. In principle, one can introduce light via fibre optics, as is

done in optogenetics, but this is cumbersome. A straightforward solution is to use light in a

wavelength range where penetration through body tissue is orders of magnitude better – i.e. in the

red to near infrared (near-IR) region.52

1.6 Challenges of long-wavelength photoswitch design

In general, shifting azobenzene absorbance to longer wavelengths leads to faster thermal relaxation

rates.53 Sometimes fast thermal relaxation is useful. For instance, if one is triggering an ion channel

involved in vision, rapid turn-off is critical for real-time responses.44 In other cases though, one

might wish to use a pulse of light to produce a long-lived cis isomer or simply to produce a large

fraction of the cis isomer. Rapid thermal relaxation means that the steady state fraction of the cis

7

isomer is small unless bright light sources with photon fluxes comparable to the thermal back

reaction rate are used.

In addition to lowering the thermal barrier to isomerization, modifications to azobenzene designed

to enable long wavelength switching can alter the relative stabilities of the ground state cis and

trans isomers. An extreme case would be if cis and trans isomers became isoenergetic and were

separated by a small thermal energy barrier. Such a photoswitch would be unable to drive

conformational switching of a target biomolecule. Instead, conformational preferences of the target

molecule would drive thermal equilibration of the switch. Thus, in designing red/near-IR

photoswitches, it is important to consider the effect of modifications on the relative energies of the

cis and trans isomers as well as the barrier between them.

1.7 Push-pull azo systems

Many thousands of azo dyes have been synthesized since the discovery of azobenzene in 1834. A

number of dyes have been found that absorb in the red region and even the near-IR.54-56 A few of

these are shown in Figure 1-2. Although the structures of these dyes vary considerably, long

wavelength absorption has typically been achieved by creating a push-pull system with electron

donating groups on one side of the azo unit and electron withdrawing groups on the other. Because

of their polarized structure, the absorption properties of these dyes are highly solvent dependent.

Historically, most scientists interested in azo dyes did not study possible isomerization events.

Photochromism would be an unwanted feature in a dye where colour fastness is usually prized. It

is likely that these long wavelength dyes have very short lived cis isomeric forms. As a result,

photoisomerization would not be observable without a specialized laser flash photolysis apparatus.

8

The Velasco group has studied the thermal relaxation behaviour of push-pull azobenzene systems

in detail53 and has reported dyes that show thermal relaxation times as fast as 40 ns.57

Figure 1-2. Some long wavelength azo dyes. For references see 6,54 7,55 and 8.56

The structures shown in Figure 1-2 are not easily adapted as components of photoswitches such as

MAG (compound 1, Figure 1-1) or as components of intramolecular cross-linkers (e.g. compound

3, Figure 1-1) where the simple bis-p-amido azobenzene core has been used. Photoswitches usually

require few rotatable bonds in order to preserve the end-to-end distance change that occurs upon

isomerization.58-59 In addition, the large size of these dyes may be a liability in pharmacological

applications where the overall size of a drug is an important feature. Finally, the push-pull design

of these long wavelength azobenzenes gives the molecule an inherent asymmetry. For

photopharmacology, asymmetry is not a concern, but for an intramolecular cross-linker

(compound 3, Figure 1-1) having an asymmetric structure would also require two distinct bio-

orthogonal means of attachment to a target protein to avoid the formation of distinct regioisomers.

9

1.8 Azo dyes that absorb at long wavelengths: C2 bridged

azobenzenes (diazocines)

In 2009 Siewersten et al60 reported the remarkable photochemical properties of a C2 bridged

azobenzene (9). The cis and trans n-π* transitions of this compound are separated by some ~100

nm making photochemical switching unusually complete in either direction. Samanta and

colleagues made a bis-p-amido substituted derivative (10) to allow conjugation to biomolecules61

and in doing this they also made the bis-p-amino substituted version (11). Herges group has

reported similar molecules with amino groups in the 3,3’ positions.62 This molecule (11) showed

significant absorbance in the red region of the spectrum (Figure 1-3) in its trans isomeric form.

Irradiation with red light (600 nm) produced full conversion to the cis isomer (Figure 1-3). The

thermal relaxation rate was on the order of minutes at room temperature in aqueous solution. The

drawback for in vivo use is that, because of the bridge, the cis state (the state produced by red light

in this case) is the thermodynamically more stable form. Unless a photoswitch is truly bistable, i.e.

thermal relaxation is negligible on biological timescales, it is most convenient if the more stable

state is the biologically inactive isomer. Otherwise in the body, in the dark, there is a steady thermal

production of the bioactive isomer. With compound 11, blue light is needed to produce the less

stable trans form. In addition to this limitation, the synthesis of these C2 bridged molecules proved

quite challenging. However, substantial improvements in synthetic methods have now been

reported.63-64 Perhaps if their thermal relaxation rates could be slowed further, these compounds

should be revisited for photopharmacological applications.

10

Figure 1-3. (a) Amino-substituted bridged azobenzene photoswitches. (b) UV-Vis spectra of 11

obtained in aqueous solution under dark-adapted conditions and under blue (450 nm) and red

(600 nm) irradiation.

1.9 Heterocycles, BF2 adducts and two-photon chromophores

Although numerous heterocyclic azo compounds have been studied as dyes, these compounds are

relatively underexplored as photoswitches. A recent study by the Fuchter group highlights the

possibility that novel switching behaviour can be obtained with such systems.65 These authors

studied arylazopyrazoles (compound 12 in Figure 1-4) and found thermal cis half-lives of days

together with a large separation of the absorption maxima of the cis and trans isomers.

The Aprahamian group discovered that BF2-adducts of azobenzenes (13) exhibited exceptional

photoswitching characteristics. Complexation of the azo group with BF2 leads to π-π* transitions

(a)

(b)

11

in the visible region.66 By altering the nature of the R group in 13, isomerization in the far red and

even near-IR was recently achieved.67 Unfortunately, these species are converted to hydrazones in

water, thus limiting their applicability in biological systems. However, modifications to the BF2

groups may change the susceptibility to hydrolysis.

An entirely different approach to driving photoisomerization with long wavelength light is to take

advantage of two-photon absorption processes. The two-photon cross section of some bioactive

azobenzene-based photoswitches has been analyzed68 and non-linear optical responses can be

improved by chemical design. For example, 14 includes a naphthalene derivative as an antenna to

enhance two-photon absorption.69 However, these photoswitches can become large and complex,

and require the use of femtosecond pulsed laser systems.70

Figure 1-4. Examples of an azoheterocyclic switch (12),65 a BF2 adduct (13),66-67 and a two-photon

absorbing photoswitch (14).69

12

1.10 Ortho substitution

To develop small azo compounds that showed long wavelength photoswitching while maintaining

slow (seconds to minutes) thermal relaxation rates in water, Sadovski et al introduced amino

groups at positions ortho to the azo group (Figure 1-5).71

Figure 1-5. Some ortho-amino-substituted azobenzenes photoswitches.71 15 is shown for

comparison.

This approach worked in the sense that thermal relaxation rates for ortho substituted compounds

were slower than for corresponding para substituted azobenzenes (e.g. half-lives of 100s vs. less

than 1 s).71 However, as the system became more electron rich, photo-bleaching was observed.

This was not a problem for blue-absorbing (16) and cyan-absorbing compounds (17) but with the

green-absorbing compound 18, colour loss occurred in a few hours under ambient light.

13

1.11 Tetra-ortho-methoxy substitution

Beharry et al replaced ortho-amino groups with methoxy groups with the expectation that the

photo-bleaching observed with ortho-amino substituted azobenzenes might be avoided. Since

methoxy groups are less electron donating than amino groups, they introduced four rather than two

substituents. They made the tetra-ortho-methoxy substituted azobenzene (19), again with two p-

amido moieties to permit linkage to target biomolecules.72 The UV-Vis spectrum of the dark-

adapted, trans isomer is shown in Figure 1-6. The main π- π* transition was blue shifted and the

n-π* transition was red shifted compared to the parent compound without the methoxy substituents

(Figure 1-6). They proposed that this was due to twisting (non-planarity) of the trans isomer and

interaction of the methoxy groups with the N lone pairs on the azo group.72

The unusually shifted n-π* transition is only a feature of the trans isomer of 19; the geometry of

the cis isomer moves the methoxy groups away from the azo nitrogen lone pairs, so the cis n- π*

transition is less affected by the methoxy groups. As a result, the n-π* transitions of cis and trans

isomers are sufficiently separated to allow cis-to-trans isomerization with blue light, and trans-to-

cis isomerization with green light, avoiding the use of UV entirely.72

A second critical feature of compound 19 is that the thermal back reaction is very slow; the half-

life is ~ 2 days at room temperature in aqueous solution.72 Thus, the authors had discovered a

simple derivative that could be switched with green light (like ortho amino compound 18 above),

did not photo-bleach, was stable in water, and had a very slow thermal back reaction.

14

(a) (b)

Figure 1-6. (a) Tetra-ortho-methoxy substituted bis-p-amido azobenzene 19, and the parent

compound 15. (b) Corresponding UV-Vis spectra (trans isomers) in aqueous solution. (c) UV-

Vis spectra of 19 attached to a peptide in cis and trans conformations.

1.12 Other chromophores exhibiting long wavelength switching

In parallel to the efforts on the design of long wavelength absorbing azobenzene molecules, two

other classes of photoswitchable systems have been explored for photocontrol with long

wavelengths. Huang et al have shown that the abundantly available dye indigo can be derivatized

by alkyl substitution at both nitrogen atoms to obtain molecules that undergo isomerization in

acetonitrile with far red wavelengths (660 nm).73 (Figure 1-7 a)

The Read de Alaniz group presented a new platform for designing tunable photoswitches based on

donor-acceptor Stenhouse adducts (DASAs) in which photoisomerization of an alkene bond is

followed by 4π-electrocyclization.74-75 The cyclic product of this two-step process thermally

reverts back to the nonpolar open molecule. (Fig 1-7b) The electron rich (donor) and electron

deficient (acceptor) moieties in the DASA system can be tuned to generate photoswitches that

operate with a variety of wavelengths.76 Effective photochromism of DASAs with near IR (750

nm) has been achieved in organic solvents. While the currently available DASAs are promising

(c)

(a) (b) (c)

15

photoswitches for applications in materials science, their highly solvent-dependent photophysical

character renders them impractical for in vivo applications.76-77

Another study by the Hecht group recently showed that donor-acceptor dihydropyrenes (DHPs)

can also be engineered to photoswitch with near-IR wavelengths. In this system, when dissolved

in acetonitrile the aromatic DHP (Figure 1-7c) is converted upon irradiation at near-IR (780 nm)

to an open isomer where the donor and acceptor moieties are no longer coupled. The less stable

isomer can revert back to the aromatic closed isomer thermally or via UV irradiation. The

photochemical reaction is near quantitative (>95% open).78 Unfortunately, the currently studied

DHP derivatives are not water soluble. Photochromism of DHPs is also highly solvent dependent

which complicates their potential application in vivo.78

16

Figure 1-7. Red and near-IR switches are designed based on N,N’ dialkylated indigo, DASA,

and DHP chromophores. (a) Indigos can be tuned to isomerize with red light. When R=Boc,

effective switching at 660 nm (PSS=80%, 1/2=193 min) is observed in acetonitrile.73 (b) DASA;

by varying the R group red shifted photoswitches are made.74, 76 (c) A DHP derivative that can

effectively switch with near-IR irradiation.78

17

1.13 Azobenzene-based protein cross-linkers

To exploit the photoswitching of a small molecule in direct control of protein structure/function,

various design concepts have been examined. For example a photoswitch can be incorporated in

the structure of an enzyme near its active site to affect its activity79-80 or it can be introduced by

non-specific reaction to side chains of a protein.81 As these methods usually involve small

modifications at the surface of the target protein, they tend to result in small changes in the overall

structure. To achieve a more pronounced change, a molecular switch can be incorporated in the

backbone of the protein but testing this approach on a variety of targets is faced with significant

synthetic challenges.82 Another strategy is to introduce the switching compound, still at the surface

of the target protein, but as an intramolecular cross-link. This can be achieved by equipping the

target protein with Cys residues at pre-defined cross-linking sites and creating a photoswitching

compound with thiol reactive handles at both ends.82 This method has been extensively studied by

the Woolley lab on test alpha-helix peptides and azobenzene based cross-linkers.83 The molecule

of azobenzene undergoes a dramatic end-to-end distance change; a feature that has not been

outperformed by any other class of photoswitches.83 With the photoswitching cross-link concept,

any protein can in principle be photocontrolled if the protein can be cross-linked with the

azobenzene in a way that one state (isomer) favours the bioactive conformation of the target

protein.

It has been shown that a number of key factors need to be carefully considered if one intends to

use photoswitchable cross-linkers for remote control of proteins. The choice of cross-linking site

on the structure is important to ensure maximal conformational change upon switching as well as

the preservation of bioactivity (i.e. often the properly folded state of the protein) in the respective

18

isomer state.59, 83 The selection of cross-linking sites in proteins that contain helices can be guided

by the Ali and colleagues’ detailed analysis.59

The cross-linker molecule should also be as rigid as possible. This is because the presence of

flexible spacer groups between each Cys side chain and the core of the photoswitch can result in a

partial decoupling between the geometry of the switch and the conformation of the molecule.82

Azobenzene cross-linked alpha-helices have been used to achieve photocontrolled DNA

binding.35, 84 Azobenzenes have also been used to optically control the activity of a peptide-based

inhibitor of protein-protein interactions inside cells.85 These examples prove the promise of this

strategy for small bioactive helical structures. However, more studies are needed to establish a

framework for using azo cross-linkers for direct control of the activity of more complex proteins.

1.14 The objectives of this thesis

Photoisomerization of azobenzenes can be applied to control specific biological targets in vivo.

Azo compounds can be used as research tools or, in principle, could act as optically controlled

drugs. Such photopharmaceuticals offer the prospect of targeted drug action and an unprecedented

degree of temporal control.

A key feature of azo compounds designed to photoswitch in vivo is the wavelength of light required

to cause the photoisomerization. To pass through tissue, wavelengths in the red, far-red, or ideally

near infrared region are required. Introducing electron-donating or push/pull substituents at the

para positions delocalizes the azobenzene chromophore and leads to long wavelength absorption

but usually also lowers the thermal barrier to interconversion of the isomers. Fast thermal

relaxation means that it is difficult to produce a large steady state fraction of the cis isomer. Thus,

19

specifically activating or inhibiting a biological process with the cis isomer would require an

impractically bright light source.

It has been found that introducing methoxy substituents at all four ortho positions leads to azo

compounds with a number of unusual properties that are useful for in vivo photoswitching. When

the para substituents are amide groups, the tetra-ortho-methoxy azo compound (19 in Figure 1.6)

shows unusually slow thermal relaxation rates compared to analogues without ortho substituents

and rapidly isomerizes from trans to cis with green light irradiation or slowly with red light

irradiation. Substitution of the amide groups with amine moieties leads to a strong red shift and

effective photoswitching. This study aims to elucidate the underlying reasons for this behaviour

and to use it to design and test other azobenzene photoswitches that have more optimal

photophysical properties for in vivo applications.

In Chapter 2 the mechanism by which the para-piperazino analogue of 19 can operate effectively

with deep red light under physiological conditions is examined. Through computational and

spectroscopic analyses, the remarkable behaviour of this azo switch is explained in terms of its

ability to form red light absorbing azonium ions at neutral pH.

As azonium ions absorb strongly in the red region of the spectrum, the design of even more red

shifted species is pursued by varying the substitution pattern. Chapter 3 details how such

compounds can be designed and characterized by focusing on the azonium ion rather than the

neutral parent molecule.

Developing effective strategies for applying azobenzene cross-linkers for optical control of protein

function is another goal of this work. In Chapter 4, the design, production and characterization of

a photocontrollable protein-based affinity reagent is discussed. It is shown that the azobenzene

20

cross-linked affinity reagent is a modular system where the azo chromophore and the affinity

reagent’s target can be independently altered to optimize the system for a particular application.

1.15 References

1. Deiters, A., Principles and applications of the photochemical control of cellular processes.

ChemBioChem: a European Journal of Chemical Biology 2010, 11 (1), 47-53.

2. Kim, B.; Lin, M. Z., Optobiology: optical control of biological processes via protein

engineering. Biochem Society Transactions 2013, 41 (5), 1183-8.

3. Velema, W. A.; Szymanski, W.; Feringa, B. L., Photopharmacology: beyond proof of

principle. Journal of the American Chemical Society 2014, 136 (6), 2178-91.

4. Riggsbee, C. W.; Deiters, A., Recent advances in the photochemical control of protein

function. Trends in Biotechnology 2010, 28 (9), 468-75.

5. Krauss, U.; Drepper, T.; Jaeger, K. E., Enlightened enzymes: strategies to create novel

photoresponsive proteins. Chemistry 2011, 17 (9), 2552-60.

6. Bruce, V. G., P. R.; Georgeson, M. A., Visual Perception: Physiology, Psychology, and

Ecology. Psychology Press: 2003.

7. Rajasethupathy, P.; Ferenczi, E.; Deisseroth, K., Targeting neural circuits. Cell 2016, 165

(3), 524-34.

8. Herrou, J.; Crosson, S., Function, structure and mechanism of bacterial photosensory LOV

proteins. Nature Reviews Microbiology 2011, 9 (10), 713-23.

9. Imamoto, Y.; Kataoka, M., Structure and photoreaction of photoactive yellow protein, a

structural prototype of the PAS domain superfamily. Photochemistry and Photobiology 2007, 83

(1), 40-9.

10. Ni, M. T. J. M.; Quail, P.H., Binding of phytochrome B to its nuclear signalling partner

PIF3 is reversibly induced by light. Nature 1999, 400, 781-5.

11. Kramer, R. H.; Mourot, A.; Adesnik, H., Optogenetic pharmacology for control of native

neuronal signaling proteins. Nature Neuroscience 2013, 16 (7), 816-23.

12. Pathak, G. P.; Vrana, J. D.; Tucker, C. L., Optogenetic control of cell function using

engineered photoreceptors. Biology of the Cell 2013, 105 (2), 59-72.

13. Morgan, S. A.; Al-Abdul-Wahid, S.; Woolley, G. A., Structure-based design of a

photocontrolled DNA binding protein. Journal of Molecular Biology 2010, 399 (1), 94-112.

21

14. Kaberniuk, A. A.; Shemetov, A. A.; Verkhusha, V. V., A bacterial phytochrome-based

optogenetic system controllable with near-infrared light. Nature Methods 2016, 13 (7), 591-7.

15. Dugave, C. D., L., Cis-Trans isomerization of organic molecules and biomolecules:

implications and applications. Chemical reviews 2003, 103, 2475-2532.

16. Brechun, K. E.; Arndt, K. M.; Woolley, G. A., Strategies for the photo-control of

endogenous protein activity. Current Opinion in Structural Biology 2017, 45, 53-58.

17. Durr, H. B.-L., H., Photochromism: Molecules and Systems. Elsevier Science: 1990.

18. Natali, M.; Giordani, S., Molecular switches as photocontrollable "smart" receptors.

Chemical Society Reviews 2012, 41 (10), 4010-29.

19. Beharry, A. A.; Woolley, G. A., Azobenzene photoswitches for biomolecules. Chemical

Society reviews 2011, 40 (8), 4422-37.

20. Rau, H., Azo Compounds. In Photochromism. Molecules and Systems, Durr, H.; Bouas-

Laurent, H., Eds. Elsevier: Amsterdam, 2003; 165-192.

21. Hartley, G. S., The cis-form of azobenzene. Nature 1937, 140, 281-281.

22. Kaufman, H.; Vratsanos, S. M.; Erlanger, B. F., Photoregulation of an enzymic process by

means of a light-sensitive ligand. Science 1968, 162 (861), 1487-9.

23. Bartels, E.; Wassermann, N. H.; Erlanger, B. F., Photochromic activators of the

acetylcholine receptor. Proceedings of the National Academy of Sciences of the USA 1971, 68 (8),

1820-3.

24. Erlanger, B. F.; Wassermann, N. H.; Cooper, A. G.; Monk, R. J., Allosteric activation of

the hydrolysis of specific substrates by chymotrypsin. European Journal of Biochemistry 1976, 61

(1), 287-95.

25. Wyart, C.; Del Bene, F.; Warp, E.; Scott, E. K.; Trauner, D.; Baier, H.; Isacoff, E. Y.,

Optogenetic dissection of a behavioural module in the vertebrate spinal cord. Nature 2009, 461

(7262), 407-10.

26. Banghart, M.; Borges, K.; Isacoff, E.; Trauner, D.; Kramer, R. H., Light-activated ion

channels for remote control of neuronal firing. Nature Neuroscience 2004, 7, 1381-6.

27. Bose, M.; Groff, D.; Xie, J.; Brustad, E.; Schultz, P. G., The incorporation of a

photoisomerizable amino acid into proteins in E. coli. Journal of the American Chemical Society

2006, 128 (2), 388-9.

28. Szymanski, W.; Beierle, J. M.; Kistemaker, H. A.; Velema, W. A.; Feringa, B. L.,

Reversible photocontrol of biological systems by the incorporation of molecular photoswitches.

Chemical Reviews 2013. 113(8):6114-78

22

29. Tsai, Y. H.; Essig, S.; James, J. R.; Lang, K.; Chin, J. W., Selective, rapid and optically

switchable regulation of protein function in live mammalian cells. Nature Chemistry 2015, 7 (7),

554-61.

30. Volgraf, M.; Gorostiza, P.; Numano, R.; Kramer, R. H.; Isacoff, E. Y.; Trauner, D.,

Allosteric control of an ionotropic glutamate receptor with an optical switch. Nature Chemical

Biology 2006, 2 (1), 47-52.

31. Levitz, J.; Pantoja, C.; Gaub, B.; Janovjak, H.; Reiner, A.; Hoagland, A.; Schoppik, D.;

Kane, B.; Stawski, P.; Schier, A. F.; Trauner, D.; Isacoff, E. Y., Optical control of metabotropic

glutamate receptors. Nature Neuroscience 2013, 16 (4), 507-16.

32. Tochitsky, I.; Banghart, M. R.; Mourot, A.; Yao, J. Z.; Gaub, B.; Kramer, R. H.; Trauner,

D., Optochemical control of genetically engineered neuronal nicotinic acetylcholine receptors.

Nature Chemistry 2012, 4 (2), 105-11.

33. Lin, W. C.; Davenport, C. M.; Mourot, A.; Vytla, D.; Smith, C. M.; Medeiros, K. A.;

Chambers, J. J.; Kramer, R. H., Engineering a light-regulated GABAA receptor for optical control

of neural inhibition. ACS Chemical Biology 2014, 9 (7), 1414-9.

34. Kumar, K. R.; Kamei, T.; Fukaminato, T.; Tamaoki, N., Complete ON/OFF

photoswitching of the motility of a nanobiomolecular machine. ACS Nano 2014, 8 (5), 4157-65.

35. Ali, A. M.; Forbes, M. W.; Woolley, G. A., Optimizing the photocontrol of bZIP coiled

coils with azobenzene crosslinkers: Role of the crosslinking site. ChemBioChem: a European

Journal of Chemical Biology 2015.

36. Pittolo, S.; Gomez-Santacana, X.; Eckelt, K.; Rovira, X.; Dalton, J.; Goudet, C.; Pin, J. P.;

Llobet, A.; Giraldo, J.; Llebaria, A.; Gorostiza, P., An allosteric modulator to control endogenous

G protein-coupled receptors with light. Nature Chemical Biology 2014, 10 (10), 813-5.

37. Broichhagen, J.; Frank, J. A.; Trauner, D., A roadmap to success in photopharmacology.

Accounts of Chemical Research 2015, 48 (7), 1947-60.

38. Stein, M.; Middendorp, S. J.; Carta, V.; Pejo, E.; Raines, D. E.; Forman, S. A.; Sigel, E.;

Trauner, D., Azo-propofols: photochromic potentiators of GABA(A) receptors. Angewandte

Chemie International Edition 2012, 51 (42), 10500-4.

39. Schonberger, M.; Trauner, D., A photochromic agonist for mu-opioid receptors.

Angewandte Chemie International Edition 2014, 53 (12), 3264-7.

40. Velema, W. A.; van der Berg, J. P.; Hansen, M. J.; Szymanski, W.; Driessen, A. J.; Feringa,

B. L., Optical control of antibacterial activity. Nature Chemistry 2013, 5 (11), 924-8.

41. Broichhagen, J.; Frank, J. A.; Johnston, N. R.; Mitchell, R. K.; Smid, K.; Marchetti, P.;

Bugliani, M.; Rutter, G. A.; Trauner, D.; Hodson, D. J., A red-shifted photochromic sulfonylurea

for the remote control of pancreatic beta cell function. Chemical Communications 2015, 51 (27),

6018-21.

23

42. Jones, D. E. H., Ariadne. New Scientist 1976, 70 (1003), 560.

43. Borowiak, M.; Nahaboo, W.; Reynders, M.; Nekolla, K.; Jalinot, P.; Hasserodt, J.;

Rehberg, M.; Delattre, M.; Zahler, S.; Vollmar, A.; Trauner, D.; Thorn-Seshold, O.,

Photoswitchable inhibitors of microtubule dynamics optically control mitosis and cell death. Cell

2015, 162 (2), 403-11.

44. Tochitsky, I.; Polosukhina, A.; Degtyar, V. E.; Gallerani, N.; Smith, C. M.; Friedman, A.;

Van Gelder, R. N.; Trauner, D.; Kaufer, D.; Kramer, R. H., Restoring visual function to blind mice

with a photoswitch that exploits electrophysiological remodeling of retinal ganglion cells. Neuron

2014, 81 (4), 800-13.

45. Bellotto, S.; Chen, S.; Rentero Rebollo, I.; Wegner, H. A.; Heinis, C., Phage selection of

photoswitchable peptide ligands. Journal of the American Chemical Society 2014, 136 (16), 5880-

3.

46. Jafari, M. R.; Deng, L.; Kitov, P. I.; Ng, S.; Matochko, W. L.; Tjhung, K. F.; Zeberoff, A.;

Elias, A.; Klassen, J. S.; Derda, R., Discovery of light-responsive ligands through screening of a

light-responsive genetically encoded library. ACS chemical biology 2014, 9 (2), 443-50.

47. Brown, M. A.; Devito, S. C., Predicting azo-dye toxicity. Critical Reviews in

Environmental Science and Technology 1993, 23 (3), 249-324.

48. Beharry, A. A.; Wong, L.; Tropepe, V.; Woolley, G. A., Fluorescence imaging of

azobenzene photoswitching in vivo. Angewandte Chemie International Edition 2011, 50 (6), 1325-

7.

49. Samanta, S.; Beharry, A. A.; Sadovski, O.; McCormick, T. M.; Babalhavaeji, A.; Tropepe,

V.; Woolley, G. A., Photoswitching azo compounds in vivo with red light. Journal of the American

Chemical Society 2013, 135 (26), 9777-84.

50. Samanta, S.; McCormick, T. M.; Schmidt, S. K.; Seferos, D. S.; Woolley, G. A., Robust

visible light photoswitching with ortho-thiol substituted azobenzenes. Chemical Communications

2013, 49 (87), 10314-6.

51. Heckel, A.; Mayer, G., Light-responsive nucleic acids for the spatiotemporal control of

biological processes. Chemical Biology of Nucleic Acids, Mayer, G., Ed. Wiley: Chichester 2010,

279-306.

52. Cheong, W. F.; Prahl, S. A.; Welch, A. J., A review of the optical properties of biological

tissues. IEEE Journal of Quantum Electronics 1990, 26 (12), 2166-2185.

53. Garcia-Amoros, J.; Velasco, D., Recent advances towards azobenzene-based light-driven

real-time information-transmitting materials. Beilstein Journal of Organic Chemistry 2012, 8,

1003-1017.

54. Bello, K. A.; Griffiths, J., Azo dyes with absorption bands in the near-infrared. Journal of

the Chemical Society, Chemical Communications 1986, (22), 1639-1640.

24

55. Li, Y.; Patrick, B. O.; Dolphin, D., Near-infrared absorbing azo dyes: Synthesis and X-ray

crystallographic and spectral characterization of monoazopyrroles, bisazopyrroles, and a boron-

azopyrrole complex. Journal of Organic Chemistry 2009, 74 (15), 5237-5243.

56. Salvador, M. A.; Almeida, P.; Reis, L. V.; Santos, P. F., Near-infrared absorbing

delocalized cationic azo dyes. Dyes and Pigments 2009, 82 (2), 118-123.

57. Garcia-Amoros, J.; Diaz-Lobo, M.; Nonell, S.; Velasco, D., Fastest Thermal Isomerization

of an Azobenzene for Nanosecond Photoswitching Applications under Physiological Conditions.


58. Burns, D. C.; Flint, D. G.; Kumita, J. R.; Feldman, H. J.; Serrano, L.; Zhang, Z.; Smart, O.

S.; Woolley, G. A., Origins of helix-coil switching in a light-sensitive peptide. Biochemistry 2004,

43 (49), 15329-38.

59. Beharry, A. A.; Chen, T.; Al-Abdul-Wahid, M. S.; Samanta, S.; Davidov, K.; Sadovski,

O.; Ali, A. M.; Chen, S. B.; Prosser, R. S.; Chan, H. S.; Woolley, G. A., Quantitative analysis of

the effects of photoswitchable distance constraints on the structure of a globular protein.

Biochemistry 2012, 51 (32), 6421-31.

60. Siewertsen, R.; Neumann, H.; Buchheim-Stehn, B.; Herges, R.; Nather, C.; Renth, F.;

Temps, F., Highly efficient reversible Z-E photoisomerization of a bridged azobenzene with

visible light through resolved S(1)(n pi*) absorption bands. Journal of the American Chemical

Society 2009, 131 (43), 15594-5.

61. Samanta, S.; Qin, C. G.; Lough, A. J.; Woolley, G. A., Bidirectional photocontrol of

peptide conformation with a bridged azobenzene derivative. Angewandte Chemie International

Edition 2012, 51 (26), 6452-6455.

62. Sell, H.; Nather, C.; Herges, R., Amino-substituted diazocines as pincer-type photochromic

switches. Beilstein Journal of Organic Chemistry 2013, 9, 1-7.

63. Tellkamp, T.; Shen, J.; Okamoto, Y.; Herges, R., Diazocines on molecular platforms.

European Journal of Organic Chemistry 2014, (25), 5456-5461.

64. Joshi, D. K.; Mitchell, M. J.; Bruce, D.; Lough, A. J.; Yan, H. B., Synthesis of cyclic

azobenzene analogues. Tetrahedron 2012, 68 (41), 8670-8676.

65. Weston, C. E.; Richardson, R. D.; Haycock, P. R.; White, A. J.; Fuchter, M. J.,

Arylazopyrazoles: azoheteroarene photoswitches offering quantitative isomerization and long

thermal half-lives. Journal of the American Chemical Society 2014, 136 (34), 11878-81.

66. Yang, Y.; Hughes, R. P.; Aprahamian, I., Visible light switching of a BF2-coordinated azo

compound. Journal of the American Chemical Society 2012, 134 (37), 15221-15224.

67. Yang, Y.; Hughes, R. P.; Aprahamian, I., Near-infrared light activated azo-BF2 switches.

Journal of the American Chemical Society 2014, 136 (38), 13190-3.

25

68. Carroll, E. C.; Berlin, S.; Levitz, J.; Kienzler, M. A.; Yuan, Z.; Madsen, D.; Larsen, D. S.;

Isacoff, E. Y., Two-photon brightness of azobenzene photoswitches designed for glutamate

receptor optogenetics. Proceedings of the National Academy of Sciences of the United States of

America 2015, 112 (7), E776-85.

69. Izquierdo-Serra, M.; Gascon-Moya, M.; Hirtz, J. J.; Pittolo, S.; Poskanzer, K. E.; Ferrer,

E.; Alibes, R.; Busque, F.; Yuste, R.; Hernando, J.; Gorostiza, P., Two-photon neuronal and

astrocytic stimulation with azobenzene-based photoswitches. Journal of the American Chemical

Society 2014, 136 (24), 8693-701.

70. Sakadzic, S.; Demirbas, U.; Mempel, T. R.; Moore, A.; Ruvinskaya, S.; Boas, D. A.;

Sennaroglu, A.; Kaertner, F. X.; Fujimoto, J. G., Multi-photon microscopy with a low-cost and

highly efficient Cr:LiCAF laser. Optics express 2008, 16 (25), 20848-63.

71. Sadovski, O.; Beharry, A. A.; Zhang, F.; Woolley, G. A., Spectral tuning of azobenzene

photoswitches for biological applications. Angewandte Chemie International Edition 2009, 48 (8),

1484-6.

72. Beharry, A. A.; Sadovski, O.; Woolley, G. A., Azobenzene photoswitching without

ultraviolet light. Journal of the American Chemical Society 2011, 133 (49), 19684-19687.

73. Huang, C.-Y.; Bonasera, A.; Hristov, L.; Garmshausen, Y.; Schmidt, B. M.; Jacquemin,

D.; Hecht, S., N,N′-disubstituted indigos as readily available red-light photoswitches with tunable

thermal half-lives. Journal of the American Chemical Society 2017, 139 (42), 15205-15211.

74. Helmy, S.; Leibfarth, F. A.; Oh, S.; Poelma, J. E.; Hawker, C. J.; Read de Alaniz, J.,

Photoswitching using visible light: A new class of organic photochromic molecules. Journal of

the American Chemical Society 2014, 136 (23), 8169-8172.

75. Lerch, M. M.; Wezenberg, S. J.; Szymanski, W.; Feringa, B. L., Unraveling the

photoswitching mechanism in donor-acceptor stenhouse adducts. Journal of the American


76. Hemmer, J. R.; Poelma, S. O.; Treat, N.; Page, Z. A.; Dolinski, N.; Diaz, Y. J.; Tomlinson,

W.; Clark, K. D.; Hooper, J. P.; Hawker, C. J.; Read de Alaniz, J., Tunable visible and near infrared

photoswitches. Journal of the American Chemical Society 2016.

77. Dong, M.; Babalhavaeji, A.; Collins, C. V.; Jarrah, K.; Sadovski, O.; Dai, Q.; Woolley, G.

A., Near-infrared photoswitching of azobenzenes under physiological conditions. Journal of the

American Chemical Society 2017, 139 (38), 13483-13486.

78. Klaue, K.; Garmshausen, Y.; Hecht, S., Taking photochromism beyond visible: direct one‐photon NIR photoswitches operating in the biological window. Angewandte Chemie International

Edition 2018, 57 (5), 1414-1417.

79. Liu, D.; Karanicolas, J.; Yu, C.; Zhang, Z.; Woolley, G. A., Site-specific incorporation of

photoisomerizable azobenzene groups into ribonuclease S. Bioorganic & Medicinal Chemistry

Letters 1997, 7 (20), 2677-2680.

26

80. Nakayama, K.; Endo, M.; Majima, T., Photochemical regulation of the activity of an

endonuclease BamHI using an azobenzene moiety incorporated site-selectively into the dimer

interface. Chemical Communications 2004, (21), 2386-2387.

81. Willner, I.; Rubin, S., Control of the structure and functions of biomaterials by light.


82. Woolley, G. A., Photocontrolling peptide α helices. Accounts of Chemical Research 2005,

38 (6), 486-493.

83. Woolley, G. A.; Beharry, A. A., Azobenzene photoswitches for biomolecules. Chemical

Society Reviews 2011, 40 (8), 4422-4437.

84. Guerrero, L.; Smart, O. S.; Weston, C. J.; Burns, D. C.; Woolley, G. A.; Allemann, R. K.,

Photochemical regulation of DNA‐binding specificity of MyoD. Angewandte Chemie

International Edition 2005, 44 (47), 7778-7782.

85. Nevola, L.; Martín‐Quirós, A.; Eckelt, K.; Camarero, N.; Tosi, S.; Llobet, A.; Giralt, E.;

Gorostiza, P., Light‐regulated stapled peptides to inhibit protein–protein interactions involved in

clathrin‐mediated endocytosis. Angewandte Chemie International Edition 2013, 52 (30), 7704-

7708.

27

A red light switching azonium ion

Authors’ contributions: This chapter is reproduced, by permission from Wiley, from the

following published article: Samanta, S., Babalhavaeji, A., Dong, M. X., & Woolley, G. A.

(2013). Photoswitching of ortho‐Substituted Azonium Ions by Red Light in Whole Blood.

Angewandte Chemie International Edition, 52(52), 14127-14130. The text has been modified for

consistency. The synthetic work, performed by SS and MD, is not included in this thesis. AB

performed the computational and spectroscopic analyses. GAW supervised the project.

The original article can be accessed via the following links:

https://onlinelibrary.wiley.com/doi/abs/10.1002/anie.201306352

2.1 Introduction

Photo-control using red light is highly desirable for biological applications since red wavelengths

are the only part of the visible spectrum that can effectively penetrate tissue.1-2 Efforts to develop

optogenetic and optochemical genetic tools that are red-shifted range from exploring natural

biodiversity in the search for red-shifted opsins3 to the conjugation of chemical photoswitches to

upconverting nanoparticles.4 Samanta and colleagues recently reported that azobenzenes with

bulky polar substituents in all four positions ortho to the azo group could undergo red-light driven

photo-isomerization. However, the compounds require intense red light or long irradiation times

to reach the photostationary state because the absorption coefficients for wavelengths above 600

nm are very small.5

Azonium ions formed by amino-substituted azobenzenes (Figure 2-1) are well known species that

have strong absorbance in the red region of the spectrum.6-9 Two features, however, make typical

azonium ions difficult to use as photoswitches in a biological context. First, most azonium ions,

such as protonated methyl orange (1), have pKas in the range of 1.5-3.56-8, 10 so that the azonium

species is hardly present at the neutral pH normally encountered in vivo. Second, the cis-to-trans

thermal isomerization rate of azonium ions is fast, with cis half lives in the μs range, so that

28

production of a significant concentration of the cis isomer is difficult without very bright light

sources.11 The rapid thermal isomerization of the cis azonium species is attributed to the decreased

double bond character of the N-N bond due to the contribution of resonance structure 1(d) (Figure

2-1).12

Figure 2-1. Neutral and protonated forms of methyl orange (1) and the ortho substituted (2) and

unsubstituted (3) aminoazobenzene cross-linked peptides studied here. Doubly-protonated

species can also form.

We report here that introduction of methoxy substituents to all four positions ortho to the azo group

in an aminoazobenzene derivative has a remarkable effect on the photochemistry in that it enables

photoswitching of the related azonium ion at neutral pH with red light.

29

2.2 Methods

2.2.1 Computational Methods

DFT and TD-DFT calculations were performed using the Gaussian 09 suite of programs.13 All

optimizations were followed by harmonic oscillator frequency calculations at the same level of

theory to verify the absence of imaginary frequencies.

Piperidino analogues of the azobenzene moieties of 2 and 3 were used for simplicity (Figure 2-5

and Scheme 2-1) and will be referred to as 2’ and 3’, respectively. Initial geometries of the trans

and cis isomers of 2’ were constructed using GaussView 5. Optimizations were performed using

both B3LYP14 and M05-2X15 functionals. The 6-311++G(d,p) basis set was used for all

calculations. The initial geometries of azonium species in both cis and trans forms were generated

using GaussView 5.0 by manually adding one hydrogen to the azo group of the optimized neutral

structures and setting the N-H bond lengths equal to 1.02 Å. The NNH angle and CNNH dihedral

angle were set to 120⁰ and 180⁰ respectively. These structures were then re-optimized at the same

levels of theory and using the same basis set.

To predict the UV-Vis spectra in vacuo optimized geometries were re-optimized using B3LYP

functional and 6-311++G(d,p) basis set, using the SMD polarizable continuum model.16 TD-DFT

calculations at the same level of theory followed. GaussView was used to visualize the predicted

transitions by applying Gaussian curves to each one of them.

The results of optimizations in vacuo at B3LYP/6-311++G(d,p) were analyzed for bond critical

points (BCPs) using the AIM2000 program.17

30

2.2.2 Synthesis

The tetra-ortho-methoxy substituted aminoazobenzene derivative 4 (Chart 2-1) that permits two-

point attachment to a thiol containing target biomolecule was synthesized by my colleagues. (For

details, see the original paper cited at the beginning of this chapter.)

Chart 2-1 tetra-ortho-methoxy aminoazobenzene cross-linker.

2.2.3 Cross-linking of peptide AB15 to obtain 2:

The peptide AB15 (sequence: Ac-WGCAEAAAREAAAREAACRQ-NH2) was prepared using

standard Fmoc-based solid-phase synthetic methods. Intramolecular cross-linking of cysteine

residues on AB15 with freshly prepared azobenzene cross-linker 4 (Chart 2-1) was performed in

50% DMSO as follows: a solution of 0.5 mM purified peptide and 2 mM cross-linker in 50 mM

Tris buffer at pH 8 was stirred at 40C under a nitrogen atmosphere for 16-20 hours. The

completion of the reaction was judged by MALDI mass spectrometry. The reaction was dried

under high vacuum, and the cross-linked peptide was purified by reversed-phase HPLC on a semi-

preparative RX-C8 column (Zorbax, 9.4 mm ID × 255 mm) with a linear gradient of 10-65%

acetonitrile/water (containing 0.1% trifluoroacetic acid) over a period of 25 min. Cross-linked

AB15 (2) was eluted at 47% acetonitrile. ESI-MS: m/z calculated for C114H173N38O34S2: 2683.9,

observed: 2682.4.

31

2.2.4 Electronic absorption spectroscopy:

Photostationary state spectra were recorded using an Agilent 8453 diode array spectrophotometer

equipped with a temperature controlled cuvette holder (Agilent 89090A). A high intensity red LED

was used to drive isomerization (LedEngin LZ4-40R200-0000, 700 mA, 635 nm, 80 mW/cm2).

The temperature was maintained at 22.0C during the experiment.

2.2.5 Determining the pKa of trans-2:

UV-Vis spectra of an aqueous solution of 2 were scanned at various pHs using a Perkin Elmer

Lambda 35 spectrophotometer coupled to a temperature controlled cuvette holder (Quantum

Northwest, Inc.). Temperature was maintained at 22.0C during all experiments. At each step small

quantities of hydrochloric acid or sodium hydroxide were added to the universal buffer solution

(i.e. CAPSO, tris, MES, sodium acetate, 25 mM with respect to each) and the pH was measured

using a microelectrode (MI-710, Microelectrodes Inc.) directly in the cuvette. Baseline-corrected

absorbance values at λ=560 nm were then used to obtain the pKa of the trans azonium species by

plotting pH vs. log((1-L)/L) where L was the ratio of absorbance at 560 nm to the maximum

absorbance at the same wavelength in the set of spectra. The intercept gives the pKa according to

the Henderson-Hasselbalch equation.

2.2.6 Determining the pKa of trans-3:

UV-Vis spectra of an aqueous solution of 3 were used to determine the pKa of trans-3 using the

same procedure as described for trans-2. Baseline-corrected absorbance values at λ=604 nm were

used. Fitting the data into Henderson-Hasselbalch equation gave an apparent pKa of 1.5.

Temperature was maintained at 22.0C during all experiments.

32

2.2.7 Determining the pKa of cis-2:

Thermal relaxation rates were measured by monitoring the absorbance of an aqueous solution of

2 at λ=540 nm as a function of time after removal of red light or amber light irradiation. A green

LED (LedEngin LZ4-41G100, 700 mA, 537 nm, 64 mW/cm2) was used as a light source for

measurement. Relative absorbance values were recorded using a photomultiplier tube (Oriel,

Newport Corporation) connected to a digital oscilloscope (Handyscope HS3, TiePie Engineering).

A band-pass filter (XMV532BP40, QuantaMax) transmitting at 532nm ± 20nm was placed in front

of the detector to eliminate other wavelengths, including scattered light coming from the

isomerization light source. The sample was irradiated by a red (LedEngin LZ4-40R200-0000, 700

mA, 635 nm, 80 mW/cm2) or an amber LED (LedEngin LZ4-40A100-0000, 700 mA, 591 nm,

30mW/cm2) with a 90 degree angle to the measuring beam, to produce a photostationary state. The

output of the detector was recorded immediately after triggering the amber light off. Fitting the

PMT signal vs. elapsed time to a monoexponential decay process gave the observed kinetic

constants for thermal relaxation at various pHs. The temperature of the sample was maintained at

22.0C during all measurements. Plotting the observed kinetic constants as a function of pH and

fitting the data with equation 1 gave the pKa of cis-2.

𝑘𝑜𝑏𝑠 = 𝑘𝑛 ∗ (10−𝑝𝐾𝑐𝑖𝑠

(10−𝑝𝐻)+(10−𝑝𝐾𝑐𝑖𝑠)) + 𝑘𝑎𝑧 ∗ (

10−𝑝𝐻

(10−𝑝𝐻)+ (10−𝑝𝐾𝑐𝑖𝑠)) Eq. 1

To confirm that the rate of cis to trans conversion was dominated by thermal relaxation, thermal

relaxation rates were measured at two different intensities of the incident beam. Also, other

wavelengths of the measuring beam were examined. Data obtained in these tests confirmed that

the measuring beam used in this case does not significantly affect the observed rates.

33

2.2.8 Glutathione stability

A solution of 2 (10-15 µM) in 100 mM sodium phosphate buffer, pH 7 was mixed with 0.25 M

glutathione (prepared by dissolving solid reduced glutathione in 0.5 M sodium phosphate buffer

adjusted the pH to 7 by adding NaOH solution) to give a final concentration of 10 mM reduced

glutathione (GSH) and the resulting solution was incubation at 25C under nitrogen atmosphere

for 14 h. During this incubation period, the absorbance at 580 nm was measured as a function of

time in a Perkin-Elmer Lambda 25 spectrophotometer.

2.2.9 Photoswitching in whole blood

A solution of 2 (1 mM) in Tris-buffered saline (TBS), pH 7.35 was mixed directly 1:1 with whole

sheep blood (sterile, defibrinated, Cedarlane Labs Inc.). To measure absorbance directly of this

solution, a 0.001 cm path length cell was employed. To obtain spectra undistorted by scattering

effects from the red blood cells a OLIS/Cary 14 spectrophotometer with CLARITY integrating

cavity system for scatter-free measurement of whole cell spectra (OLIS Inc. Bogart, GA, USA)

was employed. Just prior to measurement, samples were diluted 1000-fold in TBS, pH 7.35 to fill

the instrument cavity and spectra and kinetics were obtained at room temperature (~23C).

2.3 Results and Discussion

Attachment to a peptide ensures water solubility and prevents self-association of the dye as well

as mimicking the target environment for the photoswitch. Azobenzene derivatives attached as

cross-linkers to peptides and proteins have been used to drive conformational and functional

changes in a variety of targets.1, 18

34

Figure 2-2 shows UV-Vis spectra obtained over the range of pH 5-9 of 2 and 3 (the non-methoxy

substituted counterpart) in their trans states. For the tetra-ortho-methoxy substituted compound 2,

the azonium ion (λmax = 560 nm) is produced with an apparent pKa of 7.2. The corresponding non-

ortho-substituted species 3 does not become protonated in this pH range (Figure 2-2b). The

azonium ion of 3 is only seen below pH 3 (Figure 2-3).

Figure 2-2. UV-Vis spectra of (a) 2 and (b) 3 in aqueous buffer at the indicated pHs (22C).

Figure 2-3. UV-Vis spectra of 3 in aqueous buffer at the indicated pHs (22C).

35

Irradiation of the azonium peak of 2 at pH 7.5 with red light (635 nm, 80 mW/cm2) produces a

marked photochromism (Figure 2-4a). The strong absorbance of the azonium species (~20,000 M-

1cm-1 at 600 nm, pH 7.0) results in rapid (~1 s) production of the photostationary state (Figure 2-

4b). After removing the red light irradiation, the dark state spectrum recovers thermally in a

monoexponential manner with a half-life of ~10 s at pH 7.5 (Figure 2-4b). This process can be

repeated over many cycles with no apparent photobleaching (Figure 2-4c).

Figure 2-4. (a) UV-Vis spectra of 2: dark-adapted (solid line), and with red light irradiation

(dashed line). (pH 7.5, 22C) (b) Photoisomerization under red light (indicated by the red bar)

followed by thermal recovery in the dark (black bar). A 540 nm measuring beam is used. (c)

Multiple photoswitching cycles show no evidence of photobleaching (pH 7.5, 22C, measuring

relative absorbance at 540 nm).

36

The thermal isomerization of methyl orange (1) has been studied in detail using laser flash

photolysis techniques by Barra, Sanchez and colleagues.11 These authors observed half-lives for

thermal recovery via the azonium ion on the order of 2-3 μs. Thus, under comparable conditions,

the thermal recovery rate of 2 is ~106 times slower. The consequence of this dramatically slowed

thermal isomerization rate is that substantial populations of the thermally less stable cis species

can be produced by red LED illumination under physiological conditions.

In an effort to understand the remarkable behaviour of 2, we carried out DFT calculations to find

minimum energy structures. We used piperidino analogues of the azobenzene moiety of 2 and 3,

to which we refer as 2’ and 3’, to simplify and reduce the run-time of the DFT calculations. The

neutral species of 2’ (2a in Figure 2-1) is non-planar (Figure 2-5a) with substantial twisting of the

rings as was observed in X-ray crystal structures of other neutral tetra-ortho-methoxy species

reported earlier.5 Calculations predict the azonium species of 2’ (2c and 2d in Figure 2-1),

however, is planar (Figure 2-5b). Planarity facilitates resonance stabilization and H-bonding

interactions as diagrammed in Figure 2-5c.

37

Figure 2-5. Calculated minimum energy structures of trans neutral (a) and protonated (b) forms

of a model compound representative of 2. (c) Resonance representations showing effects that

stabilize the azonium ion.

To confirm the possibility of H-bonding based on the obtained electronic structure as diagrammed

in Figure 2-5b, we performed atoms-in-molecules (AIM) analysis.19-20 In the trans azonium

species, two bond critical points (BCPs) were found along the path interconnecting the hydrogen

atom attached to the azo group and the two nearby oxygen atoms belonging to the ortho-methoxy

substituents. Such BCPs were not detected in the models of the cis azonium ion (Figure 2-6). The

topological properties of these BCPs are provided in Table 2-1. Consistent with the criteria

proposed by Koch and Popelier,19 the values of electron density and its Laplacian at both of these

BCPs lie within the range of ordinary H-bonds (ρ(r)=0.002–0.04 a.u., ∇2ρ(r)= 0.015–0.150 a.u.).

The electronic kinetic energy density is higher than the electronic potential energy density at both

BCPs. It is proposed that such hydrogen bonds should be classified as weak to medium in

strength.20-21 BCP1, which completes a 6-membered ring, exhibits a higher local electron density

and so is predicted to be a stronger H-bond than BCP2; only BCP1 is shown in Figure 2-5.

Table 2-1. QTAIM topological parameters (all in a.u.) obtained at B3LYP/6-311++G(d,p) level

of theory. ρ(r), ∇2ρ(r), G(r), H(r), and V(r) represent electron probability density, its Laplacian,

Figure 2-6. Molecular graphs generated by AIM2000 software, showing QTAIM analysis

results for the azonium ion models in trans (a) and cis (b) forms. Bond critical points (BCPs)

show the presence of H-bonds in the trans isomer.

(a) (b)

38

kinetic electron energy density, total electron energy density and potential electron energy

density values at the bond critical points respectively. BCP1 and BCP2 are the two bond critical

points shown in molecular graph in Figure 2-5a.

ρ(r) 𝛁2ρ(r) G(r) H(r) V(r)

BCP 1 0.0316 0.129 0.0301 0.0023 -0.0278

BCP 2 0.0210 0.099 0.0208 0.0039 -0.0168

Stabilization of azonium species by a single ortho substituent capable of H-bonding has been

reported previously.22-23 In 2, however, these effects apparently raise the pKa by ~ 4-5 pH units

(compared to 3), so that the azonium species is the dominant species at physiological pHs.

39

The red light induced photochromism observed (Figure 2-3) implies the trans azonium species can

absorb red light and isomerize to produce a cis species that has a lower absorbance coefficient. To

simulate the UV-Vis spectra of the four species (i.e. cis, trans/protonated, neutral), re-optimization

of the structures was performed using the SMD implicit solvation model to better capture the effect

of the solvation.16 On the resulting structures, time-dependent DFT data were used to generate the

predicted UV-Vis spectra shown in Figure 2-7 by applying a Gaussian function with 0.4 eV peak

half-width at half-height to each transition.

In Figure 2-7 the spectra of neutral and protonated trans-2 were obtained by performing singular

value decomposition (SVD) on spectral data at various pHs by using the Henderson-Hasselbalch

equation describing the equilibrium between the two species.

Figure 2-7. Predicted vs. experimental UV-Visible spectra of the neutral and protonated species

(2” and 2). Spectra denoted as experimental are derived by performing singular value

decomposition. Calculated absorption coefficients were scaled by factors of 0.4 and 0.5 for the

neutral and protonated species, respectively. TDDFT results therefore suggest that both neutral

and protonated cis species have lower absorbance coefficients in the red region than the trans

azonium ion.

40

Figure 2-8 shows calculated minimum energy structures for the protonated and neutral cis species

(Figure 2-8a,b). Unlike the trans azonium species, the proton on the azo moiety of the cis azonium

ion is too far from the ortho methoxy substituents to form an effective H-bond, suggesting the pKa

of the cis azonium species is lower than the trans. Thus, photoisomerization to the cis azonium

species may be accompanied by conversion to the neutral cis state.

Figure 2-8. Cis protonated (a) and neutral (b) forms of a model compound representative of 2.

(c) Observed rate constant for thermal relaxation as function of pH. Limiting half-lives are ~1

min for the neutral species and ~100 ms for the cis azonium ion. (d) Kinetic scheme showing

isomerization and protonation of 2 (R) represents the linkage to the peptide. The data in (c) are

fit to the following equation: 𝒌𝒐𝒃𝒔 = 𝒌𝒏 𝟏𝟎−𝒑𝑲𝒄𝒊𝒔

𝟏𝟎−𝒑𝑯+𝟏𝟎−𝒑𝑲𝒄𝒊𝒔 + 𝒌𝒂𝒛

𝟏𝟎−𝒑𝑯

𝟏𝟎−𝒑𝑯+𝟏𝟎−𝒑𝑲𝒄𝒊𝒔

(d)

41

A neutral cis isomer would be expected to undergo thermal relaxation to the trans state much more

slowly than the cis azonium species.11-12 We measured the rate of thermal recovery as a function

of pH (Figure 2-8c) and found a sigmoidal dependence to the observed rate constant. This

behaviour is consistent with the simple kinetic scheme shown in Figure 2-8(d) in which

protonation/deprotonation rates are fast compared to rates of thermal relaxation and the neutral cis

species undergoes relaxation (kn) ~1000 times more slowly than the cis azonium ion (kaz). From

these data the cis azonium ion is estimated to have a pKa of 5.7, lower by 1.5 pH units than the

trans azonium ion.

We turned to DFT calculations again to evaluate the proton affinity of the cis isomer compared to

that of the trans isomer. The gas phase proton affinities of 2’ (i.e. the chemical reaction shown in

Scheme 2-1a) in trans and cis forms were estimated based on calculations using B3LYP and M05-

2X functionals and the 6-311++G(d,p) basis set. Using DFT calculations to obtain the free energy

of a single proton is not possible as it does not possess any electrons. Its free energy can be

obtained, however, by standard thermodynamic equations. For a proton, the molar enthalpy,

according to its definition Hm=Em+PV can be calculated as the sum of molar translational energy,

3/2RT, and PV=RT. Molar entropy can also be derived from Sackur-Tetrode equation for

monoatomic ideal gas.24 Such a treatment gives the value of -6.28 kcal/mol at 298.15 K and 1 atm

which is consistent with experimental data.25 This approach for calculating the free energy of a

proton has been used by several authors.16, 25-27 This value has been used here to calculate proton

affinities (Table 2-2). Both functionals predict the gas phase proton affinity of the trans isomer of

2’ to be ~ 3 kcal/mol greater than that of the cis isomer.

42

Table 2-2. Calculated gas phase free energies of protonations of trans-2’ and cis-2’

ΔG⁰ (kcal/mol)

Functional/Basis set trans isomer cis isomer

B3LYP/6-311++G(d,p) -264 -261

M05-2X/6-311++G(d,p) -258 -255

To estimate the stabilization caused by H-bonds and planarity of phenyl rings in the trans isomer

of 2, the isodesmic reaction given in Scheme 2-1b was used. Calculations of the non-methoxy

analogues in both trans and cis forms and their respective azonium species were performed using

the same method. The aqueous free energies of these reactions were estimated using the

thermodynamic cycle given in Scheme 2-2, where A and B represent the species of interest and

their non-methoxy analogues respectively. ΔGg and ΔGaq represent the free energies of the reaction

in gas phase and aqueous solution respectively. ΔGs denotes the solvation free energy of each

(a)

(b)

Scheme 2-1. (a) Protonation of 2’ (b) isodesmic proton transfer between 2’ and its non

methoxy analogue.

43

species. According to this cycle, the free energy of the isodesmic reaction for each species can be

expressed as follows:

ΔGaq = ΔGg + [ΔGs (AH+) + ΔGs (B) – ΔGs (A) – ΔGs (BH+)]

Scheme 2-2. Thermodynamic cycle used for obtaining the free energy of the isodesmic reaction

in aqueous phase.

The free energy of solvation for each species is calculated as the difference of self-consistent

reaction field energy, including non-electrostatic effects, obtained from an optimization using the

SMD model and the electronic energies obtained from an in vacuo optimization as recommended

by Ho et al.28 No correction for the change of standard state is needed as the number of moles of

reactants and products are equal. The ΔGaq values obtained for the cis and trans species is given in

Table 2-3.

Table 2-3. Calculated free energies of the isodesmic reaction

ΔG aq (kcal/mol)

trans isomer -8.2

cis isomer -2.8

stabilization free energy -5.4

44

Note that the solution-phase free energy of the isodesmic reaction for the trans isomer shows good

agreement with the experimental results. Using the experimentally determined apparent pKa values

of trans-2 and trans-3 this value is calculated to be -7.9 kcal/mol. (vs. -8.2 calculated, Table 2-3)

A similar thermodynamic cycle can be used for the reaction given in Scheme 2-3 to calculate the

acid dissociation constant of cis-2, using trans-2 as a reference and considering the acid

dissociation constant of trans-2 to be equal to that of trans-2’. In this case, the pKa of cis-2 is

predicted to be 5.3 which is only 0.4 logarithmic unit different than what was obtained for cis-2

experimentally.

∆𝐺𝑎𝑞 = 2.303𝑅𝑇(𝑝𝐾𝑡𝑟𝑎𝑛𝑠−𝟑 − 𝑝𝐾𝑡𝑟𝑎𝑛𝑠−𝟐)

= 2.303 ∗ 1.987 ∗ 10−3𝑘𝑐𝑎𝑙𝐾−1𝑚𝑜𝑙−1 ∗ 298.15𝐾 ∗ (1.47 − 7.24) = −7.87

𝑝𝐾𝑐𝑖𝑠−𝟐 =∆𝐺𝑎𝑞

2.303𝑅𝑇+ 𝑝𝐾𝑡𝑟𝑎𝑛𝑠−𝟐 =

−2.62 𝑘𝑐𝑎𝑙𝑚𝑜𝑙−1

2.303 ∗ 1.987 ∗ 10−3𝑘𝑐𝑎𝑙𝐾−1𝑚𝑜𝑙−1 ∗ 298.15𝐾+ 7.24

= 5.32

Scheme 2-3. Proton transfer between trans-2’ and cis-2’. Having trans-2’ as a reference and

using trans-2 experimental pKa, a pKa of 5.3 was obtained for cis-2’ based on DFT results.

Although B3LYP gas-phase free energies and aqueous solvation free energies of charged species

obtained from treating the solvent only by dielectric continuum models are not generally reliable

for predicting absolute acid dissociation constants,25 here, the calculation of relative dissociation

45

constants appears to produce an effective cancellation of errors that leads to excellent agreement

with experiments in the present case. Based on these results, the stabilization energy presented in

Table 2-3 can be taken as a realistic estimate of a combination of H-bonds and geometrical effects

present in the trans azonium species which give rise to its higher pKa compared to the cis isomer,

which in turn allows for deprotonation after trans to cis photoisomerization at physiological pH.

In summary, the million fold decrease in the rate of thermal relaxation of 2 compared to typical

azonium ions therefore appears due to two factors, (i) an intrinsic effect of the tetra-ortho

substitution pattern on the thermal barrier, and (ii) deprotonation accompanying conversion to the

cis isomer at physiological pHs.

Red light has significantly higher penetration through biological tissue primarily because

wavelengths above ~600 nm can avoid absorbance by hemoglobin.29 To test directly the possibility

of using 2 as a photoswitch for controlling conformational changes and ultimately bioactivity in

vivo, we mixed 2 directly with whole blood. Using a Clarity UV-Vis instrument designed for

measuring absorbance in highly scattering samples we were able to observe photoswitching of 2

directly in blood (Figure 2-9).

46

Figure 2-9. (a) UV-vis spectra of whole blood (dotted line), 2 (solid line) and 2 (1 mM)

(dash/dot line) in whole blood. Each sample was diluted 1000 fold in Tris-buffered saline, pH 7.4

before scanning. (b) Thermal relaxation after red light irradiation (80 mW/cm2, 635 nm) of blood

only (dotted line) and blood containing 2 (solid line) (half-life 3.4 s).

The tetra-ortho-methoxy substitution pattern creates an opportunity for developing a class of azo

compounds whose isomerization can be triggered by long wavelength light. Azonium species are

known with absorbance maxima higher than 660 nm7, 9 and with further tuning of the pKa

difference between cis and trans isomers, azo compounds might be developed for manipulating

biomolecules in vivo with near infrared light. As might be anticipated based on previous work,5

there is some sensitivity to reduction by high concentrations of glutathione however as the

compound is reduced by 10 mM GSH with a half-life of ~1.4 h (Figure 2-10). This compound (4)

may be best suited for use in extracellular and/or oxidizing environments in vivo. For instance

blood-borne peptide hormones and growth factors could be coupled to an azonium based

photoswitch and manipulated non-invasively with a high degree of spatiotemporal control.

47

Figure 2-10. Stability of 2 in presence of 10 mM glutathione. The half-life of bleaching is ~1.4

h. Black line denotes monoexponential fit of the bleaching.

2.4 References



Chemical Reviews 2013, 113 (8), 6114-78.

2. Shcherbo, D.; Shemiakina, II; Ryabova, A. V.; Luker, K. E.; Schmidt, B. T.; Souslova, E.

A.; Gorodnicheva, T. V.; Strukova, L.; Shidlovskiy, K. M.; Britanova, O. V.; Zaraisky, A. G.;

Lukyanov, K. A.; Loschenov, V. B.; Luker, G. D.; Chudakov, D. M., Near-infrared fluorescent

proteins. Nature Methods 2010, 7 (10), 827-9.

3. Chow, B. Y.; Han, X.; Dobry, A. S.; Qian, X.; Chuong, A. S.; Li, M.; Henninger, M. A.;

Belfort, G. M.; Lin, Y.; Monahan, P. E.; Boyden, E. S., High-performance genetically targetable

optical neural silencing by light-driven proton pumps. Nature 2010, 463 (7277), 98-102.

4. Liu, J.; Bu, W.; Pan, L.; Shi, J., NIR-triggered anticancer drug delivery by upconverting

nanoparticles with integrated azobenzene-modified mesoporous silica. Angewandte Chemie

International Edition 2013.




6. Stoyanov, S.; Antonov, L.; Stoyanova, T.; Petrova, V., Ammonium azonium tautomerism

in some N,N-dialkylaminoazodyes—II Compounds containing more than two protonation sites.

Dyes and Pigments 1996, 32 (3), 171-185.

48

7. Sawicki, E., Physical properties of the aminoazobenzene dyes .VIII. Absorption spectra in

acid solution. Journal of Organic Chemistry 1957, 22 (9), 1084-1088.

8. Sawicki, E., Physical properties of aminoazobenzene dyes .V. The C/A ratio. Journal of

Organic Chemistry 1957, 22 (6), 621-625.

9. Yagupolskii, L.; Gandelsman, L., Effect of substituents on the color of dialkylamino azo

dyes and their salts. Journal of General Chemistry of the USSR 1965, 35, 1259-1266.

10. Zenhausern, A.; Zollinger, H., Assoziation Bei Azoverbindungen II - Einfluss der

assoziation auf das tautomere gleichgewicht von 4-dimethylamino-azobenzol-derivaten. Helvetica

Chimica Acta 1962, 45 (6), 1890-98.

11. Sanchez, A. M.; Barra, M.; de Rossi, R. H., On the mechanism of the acid/base-catalyzed

thermal cis-trans isomerization of methyl orange. Journal of Organic Chemistry 1999, 64 (5),

1604-1609.

12. Azuki, M.; Morihashi, K.; Watanabe, T.; Takahashi, O.; Kikuchi, O., Ab initio GB study

of the acid-catalyzed cis-trans isomerization of methyl yellow and methyl orange in aqueous

solution. Journal of Molecular Structure THEOCHEM 2001, 542, 255-262.

13. Frisch, M. J., G. W. T., H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G.

Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P.

Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota,

R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A.

Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V.

N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S.

Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken,

C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C.

Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador,

J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski,

and D. J. Fox Gaussian 09, revision B.01, Gaussian, Inc.: Wallingford CT, 2009.

14. Becke, A. D., Density functional thermochemistry 3. The role of exact exchange. Journal

of Chemical Physics 1997, 98, 5648-5652.

15. Zhao, Y.; Schultz, N. E.; Truhlar, D. G., Design of density functionals by combining the

method of constraint satisfaction with parametrization for thermochemistry, thermochemical

kinetics, and noncovalent interactions. Journal of Chemical Theory and Computation 2006, 2 (2),

364-382.

16. Marenich, A. V.; Cramer, C. J.; Truhlar, D. G., Universal solvation model dased on solute

electron density and on a continuum model of the solvent defined by the bulk dielectric constant

and atomic surface tensions. Journal of Physical Chemistry B 2009, 113 (18), 6378-6396.

17. Biegler-Konig, F.; Schonbohm, J., Update of the AIM2000-program for atoms in

molecules. Journal of Computational Chemistry 2002, 23 (15), 1489-1494.

49


Society Reviews 2011, 40 (8), 4422-37.

19. Koch, U.; Popelier, P. L. A., Characterization of C-H-O hydrogen-bonds on the basis of

the charge-density. Journal of Physical Chemistry. 1995, 99 (24), 9747-9754.

20. Fuster, F.; Grabowski, S. J., Intramolecular hydrogen bonds: the QTAIM and ELF

characteristics. Journal of Physical Chemistry A 2011, 115 (35), 10078-10086.

21. Rozas, I.; Alkorta, I.; Elguero, J., Behavior of ylides containing N, O, and C atoms as

hydrogen bond accepters. Journal of the American Chemical Society 2000, 122 (45), 11154-11161.

22. Gregory, P.; Thorp, D., Electronic absorption spectra of 2-substituted-4-Nn-diethylamino-

4'-nitroazobenzene dyes and their monoacid cations - applicability of dewar rules to these and

related dyestuffs. Journal of the Chemical Society Perkin Transactions 1 1979, (8), 1990-2000.

23. Ross, W. C. J.; Warwick, G. P., Aryl-2-halogenoalkylamines .17. The ultraviolet

absorption spectra of some 4-di-(2-chlorethyl)aminoazobenzenes in neutral and acidic ethanol

solutions. Journal of the Chem Society 1956, (Jun), 1719-1724.

24. Zemansky, M. W., Heat and Thermodynamics. 7th ed.; McGraw Hill: New York, 1997.

25. Alongi, K. S.; Shields, G. C., Theoretical calculations of acid dissociation constants: A

review article. Annual Reports in Computational Chemistry 2010, 6, 113-118.

26. Liptak, M. D.; Shields, G. C., Accurate pK(a) calculations for carboxylic acids using

complete basis set and gaussian models combined with CPCM continuum solvation methods.


27. Marenich, A. V.; Ding, W. D.; Cramer, C. J.; Truhlar, D. G., Resolution of a challenge for

solvation modeling: Calculation of dicarboxylic acid dissociation constants using mixed discrete-

continuum solvation models. Journal of Physical Chemistry Letters 2012, 3 (11), 1437-1442.

28. Ho, J. M.; Klamt, A.; Coote, M. L., Comment on the correct use of continuum solvent

models. Journal of Physical Chemistry A 2010, 114 (51), 13442-13444.

29. Cheong, W. F.; Prahl, S. A.; Welch, A. J., A review of the optical properties of biological

tissues. IEEE Journal of Quantum Electronics 1990, 26 (12), 2166-2185.

50

Red, far-red, and near infrared photoswitches based

on azonium ions

Authors’ contributions: This chapter is reproduced, by permission from The Royal Society of

Chemistry, from the following published article: Dong, M.; Babalhavaeji, A.; Hansen, M. J.;

Kálmán, L.; & Woolley, G. A. (2015). Red, far-red, and near infrared photoswitches based on

azonium ions, Chemical Communications, 51(65):12981-4; and by permission from The American

Chemical Society, from the following published article: Dong, M.; Babalhavaeji, A.; Collins,

C.V.; Jarrah, K.; Sadovski, O.; Qiuyun D.; & Woolley, G. A. (2018). Near-Infrared

Photoswitching of Azobenzenes under Physiological Conditions, Journal of the American

Chemical Society, 139(38):13483-6. The text has been modified for consistency. In the first paper,

MD and MJH carried out the synthetic work which is not included in this thesis. AB carried out

the computational and spectroscopy experiments. LK and AB performed laser-flash photolysis.

GAW supervised the project. In the second article, MD, CVC, KJ, and OS synthesized the

compounds. Note that the synthetic aspects are not covered in this thesis. AB, CVD, KJ performed

characterization experiments. QD performed crystallography, which will not be discussed. AB

performed the computational analysis. GAW supervised the project. The original articles can be

accessed via the following links:

http://pubs.rsc.org/en/Content/ArticleLanding/2015/CC/C5CC02804C

https://pubs.acs.org/doi/10.1021/jacs.7b06471

3.1 Introduction

Molecules that undergo reversible light driven changes in structure are of increasing interest for

applications in fields ranging from molecular electronics to neurobiology.1-3 For biological

applications in particular, considerable efforts have been made to produce molecules that absorb

in the red region of the visible spectrum since these wavelengths enable much greater tissue

penetration and less phototoxicity than shorter wavelengths.4-9 Ideally, however, molecular

photoswitches designed for in vivo use should operate in the near infrared (IR) optical window.6,

10 The edges of the near IR window depend on the tissue in question, but are generally set by

haemoglobin absorption at shorter wavelengths and water absorption at longer wavelengths. The

best penetration is often observed near 730 nm.11 Although some near IR absorbing azo

compounds are known, these are complex, flexible, molecules that would be difficult to apply as

51

photoswitches.12-13 Aprahamian and colleagues described near IR absorbing photoswitches based

on BF2-modified azo compounds.10 These compounds showed excellent photochemical

characteristics, however, their tendency to hydrolyse in aqueous solution presents a serious

drawback to their application in a biological context.

In Chapter 2 the azonium ion 1 (Chart 3-1) that absorbs red light and undergoes trans-cis

photoisomerization was described. It relaxes to the trans isomer in the dark on the timescale of

seconds so that pulses of red light could be used to drive multiple isomerization cycles. Resonance

stabilization of the azonium cation together with intramolecular H-bonding between the azonium

proton and methoxy groups ortho to the azo unit mean that the trans azonium species is present at

pH 7 in aqueous solution, whereas typically azonium ions are only formed from aminoazobenzenes

at pH < 3.5.14-16 The geometry of the cis isomer prevents formation of this intramolecular H-bond

so that the cis azonium pKa is lower than that of the trans isomer.17 As a result, depending on the

pH, trans to cis isomerization is accompanied by proton dissociation to produce the neutral cis

species, thereby slowing thermal reversion to the trans form. This feature is useful for the

photochemical production of a large fraction of this cis isomer to control molecular targets. Since

compound 1 acts as an effective red light switch that can operate in biological milieu, we used it

as a starting point to try to design azonium ion based switches that would operate in the near-IR.

Chart 3-1. Azobenzene with methoxy groups in all ortho positions and para amino groups has a

high pKa and forms red-light absorbing azonium ions at physiological pH.

52

There is a rich literature describing substituent effects on the absorption wavelengths of

azobenzenes and, to a lesser extent, the corresponding azonium ions.18 We wished to keep at least

one methoxy substituent in an ortho position to preserve the possibility of intramolecular H-

bonding of the trans azonium ion. We therefore considered changes at para and meta positions.

Varying the nature of the amine in the para position alters the degree to which the para-amino

nitrogen lone pairs delocalize into the ring system.19 A second phenomenon, called the distribution

rule of auxochromes was recognized early by Kaufmann20 and described in detail by Wizinger.21

This rule predicts that a second auxochrome (e.g. a methoxy substituent) in a meta position should

shift absorbance to longer wavelengths. The phenomenon is not explained using heuristic rules of

resonance and historically was the subject of much discussion,21-22 but nowadays is predicted by

standard quantum chemical calculations. In an effort to produce photoswitches operating at longer

wavelengths than (1), we decided to examine the role of the methoxy substitution pattern as well

as the nature of the para amino group in the photochemical behaviour of these azonium-based

photoswitches.

3.2 Materials and methods

The details of the synthesis of azobenzenes and their attachment to glutathione can be found in the

publications cited at the top of this chapter.

3.2.1 Computational methods

DFT and time-dependent DFT calculations were carried out on Sharcnet (Compute Canada)

supercomputers using Gaussian 09 package.23 Gauss View 5 was used to sketch the initial

geometries. For saving on calculation time, pyrrolidino and piperidino moieties were used instead

of the full amine moieties.

53

For 2, 3, and 4 (Chart 3-2), input coordinates were generated by manually modifying the optimized

structures of 1 in the neutral and protonated forms from our previous study.17

For 5, 6, and 7 (Chart 3-4), optimized structures of 2 were used to draw initial coordinates. For

neutral trans geometries, the optimized structures of the corresponding azonium ions were used to

produce initial coordinates. For each structure, the azonium proton was removed and the charge

and multiplicity were corrected.

For cis structures, two conformations were manually sketched by changing dihedral angles that

involved the azonium moiety in the optimized trans geometry. Dihedral angles from the models

of 6 were used. 17 Neutral input structures were produced by simply deleting the azonium proton

and changing the charge and multiplicity.

All structures were optimized in vacuo using B3LYP 24 hybrid functionals and the 6-31+G(d,p)

basis set. Frequency calculations were performed at the same level of theory to confirm that the

optimization results represented local minima on the energy landscape. An exhaustive

conformational search was not carried out.

The optimized geometries of all structures were subjected to TD-DFT calculations using the same

functionals and basis set, assuming the first 15 singlet excitations and by applying the SMD

solvation model to implicitly approximate the effect of water. TD-DFT data were used to generate

the simulated UV-Vis spectra by applying a Gaussian function with 0.3 eV peak half-width at half-

height to each transition.

3.2.2 Sample preparation for UV-Visible spectroscopy

Solutions were prepared in a mixture of 4 buffers (CAPSO, TRIS, MES, sodium acetate, 25 mM

each, termed a universal buffer) to ensure the pH could be easily adjusted between 2 and 11 by

54

addition of small quantities of concentrated hydrochloric acid and sodium hydroxide. The pH was

measured by a glass combination micro-electrode (MI-710, Microelectrodes Inc.)

3.2.3 Calculation of molar absorption coefficients for 1-4

Absorbance spectra of the GSH adduct of each compound (1-4) in a 1 cm quartz cell were obtained

in the universal buffer solution described above using a Perkin Elmer Lambda-35 instrument

coupled to a Peltier temperature controller (Quantum Northwest). The concentration of the samples

was then determined by quantitative amino acid analysis using a Water Pico-tag system as follows

(performed at the SPARC BioCentre at the Hospital for Sick Children, Toronto): Samples were

dried in pyrolyzed borosilicate tubes in a vacuum centrifugal concentrator and subjected to vapour

phase hydrolysis by 6N HCl with 1% phenol at 110°C for 24 hours under a pre-purified nitrogen

atmosphere. After hydrolysis, excess HCl was removed by vacuum, hydrolyzates were washed

with redrying solution (Waters, Inc) and derivatized with phenylisothiocyanate (PITC) to produce

phenylthiocarbamyl (PTC) amino acids. Derivatized amino acids are redissolved in phosphate

buffer and transferred to injection vials which are loaded into an autosampler for automatic

injection. PTC derivatives were separated by reverse phase HPLC. PTC derivatives of Gly and

Glu (derived from hydrolysis of GSH) were quantified by comparison with standards. Molar

absorption coefficients at a given wavelength were calculated by dividing absorbance values

determined by UV-Vis measurements by concentrations determined by quantitative amino acid

analysis. Note that estimation of azonium ion absorption coefficients is complicated by the

possible coexistence of doubly protonated species.

55

3.2.4 pH-dependence of absorption spectra

UV-Visible spectra were acquired using a Perkin Elmer Lambda-1050 instrument coupled to a

Peltier temperature controller (PTP-6, Perkin Elmer). The temperature was maintained at 20C.

All spectra were baseline-corrected, assuming zero absorption at 800 nm.

3.2.5 Thermal relaxation kinetics for 2

A solution of 2 was prepared in the buffer mixture described above. For each measurement, fresh

samples with a known pH were injected into a Z-shaped flow cell (FIAlab) with a light path length

of 2.5 mm. A 530 nm LED, 700 mA, 5.6 mW/cm2 was used as a source for the measuring beam

while isomerization was achieved by irradiation with 455 nm (Thorlabs) or 660 nm (Mightex

Systems Inc.) high power LEDs.

Thermal relaxation rates were measured by monitoring absorbance after removal of the blue light

source. Relative absorbance values were recorded using a photomultiplier tube (Oriel, Newport

Corporation) connected to a digital oscilloscope (Handyscope HS3, TiePie Engineering). Two

linear variable band-pass filters (LVF-HL, Ocean Optics) transmitting at 540 nm ± 20nm were

placed before and in front of the sample to eliminate other wavelengths, including scattered light

coming from the isomerization light source. A third band-pass filter transmitting at 530nm ± 20nm

(QuantaMax) was placed in front of the detector to further block the scattered blue light. The output

of the detector was recorded immediately after triggering the blue light off. Fitting the

photomultiplier tube signal vs. elapsed time to a monoexponential decay process gave the observed

kinetic constants for thermal relaxation at various pHs. These data are shown in Appendix 6.1.

Kinetic constants used were the average of at least three decay curves. The temperature of the

sample was maintained near 22C during all measurements. To confirm that the rate of cis to trans

conversion was dominated by thermal relaxation, relaxation rates were measured at two different

56

intensities of the incident beam. Also, other wavelengths of the measuring beam were examined.

Data obtained in these tests confirmed that the measuring beam used in this case does not

significantly affect the observed rates.

3.2.6 Laser flash photolysis experiments

Solutions of 3 and 4 were prepared as described above and placed in a 1 cm quartz cuvette at room

temperature (~22C).

The isomerization rates of these two compounds were measured with a miniaturized laser-flash

photolysis system (LFP-112; Luzchem Research Inc. Ottawa,ON, Canada) as described earlier.4

Samples were excited with pulses of 25 mJ/pulse output energy and 3-5 ns pulse width at 532 nm

from a frequency doubled Nd:YAG laser (Minilite II; Continuum, Santa Clara, CA, U.S.A.) The

laser was operated with a 1 Hz repetition rate to allow plenty of time to the transitional signals to

recover.

Various measuring beam wavelengths were tried between 380 nm to 700 nm to obtain an

acceptable signal to noise ratio as the differential absorption spectra of trans and cis species were

unknown. Eventually, transient measurements were carried out at 550 nm for 3 and 480 nm for 4.

25 measurements were averaged for 3 at each pH point. The number of measurements that were

averaged for 4 varied between 10 to 50 depending on the relaxation rate and the signal-to-noise at

a given pH. The reproducibility of the kinetic traces indicated that numerous cycles of

photoisomerization could be repeated without significant photobleaching.

Relaxation data were fit to single exponential functions to obtain decay rate constants at each pH.

These data are shown in appendix 6.1. As with compound 2, relaxation data at most pHs are

dominated by a single exponential decay process. We assume the time constant derived from these

57

fits corresponds to the cis-to-trans thermal isomerization rate constant because this is expected to

be much larger than the trans-to-cis rate constant. This is because calculations indicate the trans

azonium ions (and neutral species) are ~10-15 kcal/mol more stable than the cis species.

3.3 Results and discussion

3.3.1 Azonium ions based on the ortho-meta substitution pattern

Compounds 2, 3, and 4 (Chart 3-2) were designed to preserve the possibility of H-bonding of the

azonium proton to an ortho-methoxy group in the trans isomer (as in 1) but to have longer

wavelength absorption due to the presence of a better para donating substituents (2 vs. 1, 4 vs. 3)

or an ortho-meta methoxy substitution pattern (3 and 4 vs. 1 and 2). In addition to affecting the

wavelength of absorption, these changes are expected to alter the pKa of the azonium ion formed

in each case.

Chart 3-2. Azobenzenes with varying the ortho/meta methoxy, and para amine substitution

patterns

Calculations were performed using density functional methods (B3LYP (6-31+G**)) to optimize

geometry and TD-DFT to calculate absorption wavelength maxima. These computational methods

have been used successfully with related compounds.17, 25 For sketching the neutral structures of 3

58

and 4, the ortho methoxy groups were placed on the opposite sides of the phenyl rings. For the

protonated forms of 3 and 4, all possible rotamers shown in Chart 3-3 were calculated to determine

the thermodynamically most stable geometry based on the sum of electronic and thermal free

energies.

Chart 3-3. Conformers with different intramolecular H-bonding patterns.

The conformers indicated by ii in Chart 3-3 allowed for two hydrogen bonds and had the lowest

calculated free energy after optimization. For both compounds, the free energy of conformers i,

iii, and iv in the gas phase were calculated to be ~12, ~3.5, and ~7.5 kcal/mol higher than ii,

respectively. Conformation ii was therefore chosen for the prediction of maximum absorption

wavelengths for 3 and 4.

TD-DFT calculations predict that these compounds should absorb at longer wavelengths than

(1)(Table 3-1). Calculating effects of the substitution patterns on pKas is complex26 and was not

attempted here. (Refer to section 2.3)

59

We used a bicyclic pyrrolidine in compound 4 to preserve the symmetry of the compound and to

maximize the end-to-end distance change upon trans to cis isomerization; both these features are

important for applying these compounds as cross-linkers for peptides and proteins.27 The bicyclic

pyrrolidine version of 2 proved to be insoluble so that it was replaced by the monocyclic amino-

pyrrolidine in that case 2. These different pyrrolidine moieties are expected to behave similarly in

terms of electron donation to the azo system. In each case, the compound was linked at both ends

to the short tripeptide glutathione (GSH). This confers water solubility on the photoswitch and also

provides a test of the operational sensitivity of the compound to reduction by thiols. While azonium

ions can, in general, be reduced by thiols17, 28 the rate of reduction depends on the pH and thiol

concentration which vary widely depending on the particular biological environment in question.29

In practice, we find that if a GSH adduct can be readily synthesized, the rate of reduction is slow

enough that the compound can be used under typical physiological conditions.

The absorption properties of compounds 2, 3, and 4 were examined in aqueous buffer as a function

of pH. These compounds were found to be stable for weeks at room temperature in aqueous

solutions near neutral pH, an important requirement for biological applications. Figure 3-1 shows

UV-Vis absorption spectra for each newly synthesized compound together with those obtained

previously for compound 1. In the high pH limit, the spectrum in each case is assumed to

correspond to that of the neutral (unprotonated) species. As the pH is lowered, a species with long

wavelength absorption appears and this is assumed to be the singly protonated azonium ion.

Contributions to the spectra from doubly protonated species (azonium plus ammonium) are

indicated by dashed grey lines in Figure 3-1 where these dominate the spectra.

60

Figure 3-1. (Left) UV-Vis spectra of the trans forms of compounds 1-4 (linked to GSH) in

aqueous buffer as a function of pH. In each case, the spectrum at the high pH limit is shown as a

black dotted line. Lowering the pH produces the azonium ion (shown as a series of grey solid

61

lines) until a maximum amount of the azonium ion is obtained (solid black line). Further

decreases to the pH produce doubly protonated species (azonium plus ammonium). The spectra

of these species are indicated by dashed grey lines. (Right) pH dependencies of the azonium ions

formed in each compound. Solid lines represent the fits to Eq. 1 (compounds (1, 2, and 3) and

Eq. 2 for compound 4. The dotted arrows mark the pKas of the azonium ions).

To estimate the apparent pKa’s for the trans form of each compound, the absorbance values at the

maximum absorption wavelengths corresponding to the azonium ions (Table 3-1) were plotted

against the pH. For compounds 1, 2, and 3 these data were fitted to Eq. 1, which describes a single

protonation event.

𝐴𝑏𝑠𝜆 = [𝐴]𝑡𝑜𝑡 (𝜀𝐴 1 − (1 (1 + 10(𝑝𝐻−𝑝𝐾𝑎)⁄ )) + 𝜀𝐻𝐴(1 (1 + 10(𝑝𝐻−𝑝𝐾𝑎))⁄ )) Eq. 1

Where [A]tot is the total concentration of the compound (in any ionization state), εA is the molar

absorption coefficient of the neutral species, εHA is the molar absorption coefficient of the singly

protonated azonium ion. For compound 4, data were fitted to Eq. 2, which describes two

protonation events:

𝐴𝑏𝑠𝜆 = [𝐴]𝑡𝑜𝑡(𝜀𝐴(𝑓𝑁) + 𝜀𝐻𝐴(𝑓𝑃1) + 𝜀𝐻𝐻𝐴(𝑓𝑃2)) Eq. 2

Where εHHA is the molar absorption coefficient of the doubly protonated azonium/ammonium ion

and 𝑓𝑁, 𝑓𝑃1, 𝑓𝑃2 are the fractions of neutral, singly protonated and doubly protonated species

respectively:

𝑓𝑁 = 10(𝑝𝐻−𝑝𝐾2). 10(𝑝𝐻−𝑝𝐾1)/(10(𝑝𝐻−𝑝𝐾2) + 1 + (10(𝑝𝐻−𝑝𝐾2). 10(𝑝𝐻−𝑝𝐾1)))

62

𝑓𝑃1 = 10(𝑝𝐻−𝑝𝐾2)/(10(𝑝𝐻−𝑝𝐾2) + 1 + (10(𝑝𝐻−𝑝𝐾2). 10(𝑝𝐻−𝑝𝐾1)))

𝑓𝑃2 = 1/(10(𝑝𝐻−𝑝𝐾2) + 1 + (10(𝑝𝐻−𝑝𝐾2). 10(𝑝𝐻−𝑝𝐾1)))

The constants pK1 and pK2 refer to the dissociation of the single protonated and doubly pronated

species respectively.

The molar absorption coefficients of the trans azonium forms of 1-4 are collected in Table 3-1.

These are similar to that found for methyl orange (55,000 M-1cm-1).30 TD-DFT calculations also

predict large oscillator strengths for these compounds.

Table 3-1. Predicted (B3LYP (6-31++G**) and observed properties of compounds 1-4. For

compounds 3 and 4, only the calculation results for rotamer ii are presented. (See methods)

Compound

Predicted

λmax

(trans

neutral

form)

Observed

λmax

(trans

neutral

form)

Predicted

λmax

(trans

azonium

form)

Observed

λmax

(trans

azonium

form)

Oscillator

Strength

(neutral)

Oscillator

Strength

(azonium)

ε (λmax)

(trans

azonium

form)

Apparent

pKa of

azonium

ion

1 522 nm

384 nm

460 nm

365 nm 545 nm 563 nm

0.666,

0.502 1.54

47,000

M-1cm-1 7.5

2 518 nm,

384 nm

491 nm,

410 nm 544 nm 610 nm

0.638,

0.664 1.50

51,000

M-1cm-1 8.9

3 497 nm 490 nm 631 nm 630 nm 0.704 1.34 40,000

M-1cm-1 2.6

4 485 nm 517 nm 634 nm 670 nm 1.37 1.39 37,000

M-1cm-1 5.4

Increasing the donating ability of the para amino substituent moves absorption from the red region

into the far-red (compound 2 vs. 1). Introducing the ortho-meta substitution pattern produces

azonium ions absorbing in the far-red (3) and near IR (4). Compound 4 shows significant

absorbance in the near IR window at 730 nm.

Whether the azonium ion is present at physiological pHs near ~7, depends on the pKa. The apparent

pKas of these compounds are collected in Table 3-1. Increasing the donating ability of the para

63

amino substituent increased the pKa of the trans azonium ion of 2 to 8.9 (compared to 7.5 for 1).

A consequence of this change is that the azonium ion dominates at physiological pHs near ~7 so

that red and far-red wavelengths are absorbed strongly. Replacing two ortho methoxy groups with

an ortho-meta substitution pattern (e.g. 3 vs. 1 or 4 vs. 2) causes a significant drop in the pKa of

the corresponding trans azonium ions, particularly in the case of 3. The removal of an ortho

methoxy group removes a possible resonance contributor for stabilization of the positive charge.

In addition, it removes a potential H-bond acceptor; however, calculations suggest that H-bonded

species still predominate suggesting the latter is not a major reason for the drop in the azonium

pKa. Instead, the pKa drop may result from the presence of meta methoxy group which forces the

para piperazino substituent of 3 to rotate so that the para piperazino nitrogen atom is significantly

less able to act as an electron donor. Calculated structures of 3 (Figure 3-2) confirm that the

piperazino units are highly twisted. Consistent with this proposal, the less sterically demanding

pyrrolidino substituent (compound 4) shows a significantly higher pKa (Table 3-1).

64

Figure 3-2. Models showing optimized structures of compounds 1-4 in their trans neutral (left

column) and azonium (right column) forms. Strong H-bonds are indicated by yellow dotted lines.

The steric clash leading to twisting of the 6-membered ring in 3 is highlighted by arrows. The

neutral forms of 1 and 2 are highly twisted. All other compounds are relatively planar throughout

the aromatic system.

We then examined the photoswitching behaviour of this series of compounds using laser flash

photolysis techniques. Hundreds of rounds of trans-cis photoisomerization occurred in each case

and photocyclization of the type observed with simple azonium ions by Lewis31 and discussed by

Mallory32 does not seem to occur with these compounds. The rate of cis to trans thermal relaxation

increases as the pH is lowered because cis azonium species relax significantly more quickly than

their neutral counterparts.33 Figure 3-3 shows the observed first order rate constants for thermal

65

cis-to-trans relaxation as a function of pH for compounds 1, 2, 3 and 4 in aqueous solution. These

data are fit to Equation 3 where 𝑘𝑜𝑏𝑠 is the measured (fitted) rate constant at a particular pH, 𝑘𝑛 is

the rate constant for thermal isomerization of the neutral cis species, 𝑘𝑎𝑧 is the rate constant for

thermal isomerization of the cis azonium species and 𝑝𝐾𝑐𝑖𝑠 refers to the cis azonium ion. Where

the fit is poorly constrained due to lack of data the fit is shown as a dotted line.

𝑘𝑜𝑏𝑠 = 𝑘𝑛 ∗ (10−𝑝𝐾𝑐𝑖𝑠

(10−𝑝𝐻)+(10−𝑝𝐾𝑐𝑖𝑠)) + 𝑘𝑎𝑧 ∗ (

10−𝑝𝐻

(10−𝑝𝐻)+ (10−𝑝𝐾𝑐𝑖𝑠)) Eq. 3

In general, the rate constants are bounded at the low pH end by the intrinsic relaxation rate of the

cis azonium ion and at the high pH end by the intrinsic relaxation rate of the neutral species in

each case. Not all pH/rate combinations were experimentally accessible. At a given pH the

observed rate reflects both the intrinsic thermal relaxation rate as well as the pKa of the cis azonium

ion.17 At physiological pH (~7) the tetra-ortho methoxy substituted species 1 and 2 show cis

lifetimes on the order of 1 s and ~100 ms respectively.

66

Figure 3-3. (a) Far-red photoisomerization of 2. Absorbance changes observed with pulses of

660 nm light (aqueous buffer, pH 8.8, 20C) (b) Observed rate constants for thermal cis-to-trans

relaxation of compounds 1-4 as a function of pH in aqueous buffer. The kinetic traces of thermal

relaxation were fit to single exponential decay functions to obtain rate constants. Circles,

diamonds, squares, and triangles show the observed rate constants for compounds, 1, 2, 3, and 4,

respectively. The lines represent the fit to Eq. 3 yielding pKa values of the cis azonium ions of

5.7, 8.6, 4.1, and 7.6 respectively.

Compound 2 thus functions as a far-red photoswitch that operates at physiological pHs.

Compounds, 3 and 4 have lifetimes of ~ 1 ms and 10 µs respectively at pH 7, however only

compound 4 has significant near IR absorbance at this pH. Compound 4 could thus function as a

near IR switch with rapid (10 µs) thermal relaxation.

67

3.3.2 Designing a slow relaxing near-IR azobenzene switch

Whereas tetra-ortho methoxy substitution is required for slow thermal reversion, a meta methoxy

substitution pattern enhances the long wavelength absorbance of azonium ions. A meta methoxy

substitution places steric constraints on the nature of the para substituent so that a pyrrolidino

group is best able to act as an electron donor in this position. Varying these substituents allow the

construction of azonium-based photoswitches with different degrees of far-red absorbance,

thermal relaxation rate and pKa tuned for desired applications.

While rapid thermal relaxation is ideal for certain applications34-35 slower relaxation permits

substantial fractions of the cis isomer to be produced at low irradiation powers and is useful when

a large degree of photo-control of single target biomolecules is desired.1 For photopharmacological

applications in vivo, the rate of the thermal back reaction to the stable isomer should be slow

enough to permit accumulation of the cis isomer under constant near-IR illumination, but fast

enough that spatial targeting is not lost. Based on MRI and PET measurements of mean transit

times for blood in the cerebral vasculature, for example, the appropriate time constant for thermal

relaxation should be in the range of 0.1-10 s.36-37

The previous analysis of the derivatives bearing four methoxy substituents showed that when all

four substituents were in ortho positions relative to the azo unit, and there was an amino group in

the para position (e.g. 2), the thermal half-life of the cis isomer was appropriate (~20 s) but the

wavelength maximum was too short so that absorbance at wavelengths >700 nm was minimal. An

ortho-meta arrangement of the methoxy groups (4), in contrast, produced a substantial red-shift,

but led to a very short (μs) cis isomer thermal half-life and a pKa for the azonium ion too low for

use at pH 7.0.

68

Chart 3-4. Design of slow-relaxing azonium forming photoswitches with ortho,meta- methoxy

subsitution pattern.

The neutral trans form of tetra-ortho substituted azo compounds such as 2, exhibits a highly

twisted geometry, a feature that is absent in the ortho-meta derivative. The steric hindrance that

produces this twisting may also increase the thermal energy barrier for cis-to-trans isomerization.

Thus, to obtain slow thermal relaxation together with absorbance in the bio-optical window, both

tetra-ortho substitution and meta-methoxy substitution would appear necessary, for example as in

5, which bears methoxy substituents at all ortho and all meta positions. Computational modeling

showed that steric clash of the methyl groups in 5 (Chart 3-4) would lead to loss of conjugation

between methoxy oxygen lone pairs and the ring systems. (Figure 3-4)

Figure 3-4. Energy minimized structure of 5.

Locking the conformation of the methoxy groups by creating dioxane rings (compound 6) solves

this problem. (Figure 3-5) Computational analysis indicated that para amino substituted

derivatives of 6 would have the desired red shift. However, as a result of the extremely poor

solubility of precursors to 6 in a variety of solvents, the structure had to be modified.

69

Figure 3-5. Energy minimized (DFT) structure of 6.

The substitution pattern shown in compound 7 (Chart 3-4), in which there is one meta oxygen

substituent per ring, was then considered. Computational models predicted that all four trans

azonium conformations of 7 would be planar with respect to the phenyl rings, similar to the

reported models for 2.17 The distance between the azonium proton and the nearby oxygens is

similar to compound 2, which suggests the existence of a similar H-bonding pattern.

Due to the asymmetry of the compound, distinct conformations exist with the dioxane rings on the

same side or on different sides of the molecule. (Chart 3-5) For the trans azonium ion, the free

energy difference between the least and the most stable conformations was calculated to be 0.9

kcal/mol (Table 3-2). The lowest energy conformation is shown in Figure 3-6(b) with a stabilizing

H-bond indicated.

70

Chart 3-5. Different conformations for the trans azonium ion formed by 7

Figure 3-6. DFT-calculated low energy conformations of a p-pyrrolidine derivative of 7 in the

(a) neutral trans form (b) trans azonium form, (c) neutral cis form, (d) cis azonium form.

Predicted max values (TDDFT) were very close for all conformations. The most stable structure

(conformation 3) was predicted to absorb at 597 nm. This wavelength is significantly longer than

what was calculated for 2 (544 nm) but lower than 4 (634 nm) (Table 3-2). Based on these results,

substantial absorbance >700 nm was expected for 7. Simulated spectra based on TD-DFT results

of 7 are shown in Figure 3-7.

71

Figure 3-7. Simulated spectra of 7 (a model for 8) based on TD-DFT calculations.

Table 3-2. TD-DFT results for compounds 2, 4, 5, 6 and 7

Compound max, trans azonium Energy difference from most

stable conformer (kcal/mol)

Compound 2 544 nm -

Compound 4 634 nm -

Compound 5 608 nm -

Compound 6 629 nm -

Compound 7, conformation 1 596 nm 0.89


Compound 7, conformation 3 597 nm 0


72

Both neutral and azonium species in their cis forms absorb less strongly as predicted by lower

oscillator strengths (Table 3-3). This is what has been experimentally observed for similar

compounds.

Table 3-3. Selected TD-DFT data for cis and trans isomers of 7 (a model for 8)

max

trans

azoni

um

Oscillato

r strength

max

trans

neutra

l

Oscillato

r strength

max cis

azoniu

m

Oscillato

r strength

max

cis

neutra

l

Oscillato

r strength

ΔG

(trans_to_cis

) kcal/mol

azonium

ΔG

(trans_to_cis

) kcal/mol

neutral

Compound 7

(model for 8) 597 1.27 530 0.70 623 0.68 531 0.32 12.5 8.7

Since the nature of the amine in the para position can tune the wavelength maximum and also the

azonium pKa, a variety of derivatives of 7 were analyzed. (For details see original publication cited

at the beginning of this chapter) Compound 8 (Chart 3-6) in which the para substituent is a

pyrrolidine group bearing a solubilizing sulfone moiety exhibited the longest wavelength

absorbance tail, together with a pKa near neutral. The observed pKa of 6.7 (measured by my

colleagues) means that ~25% of the molecules are protonated at a physiological pH of 7.2. The

molar absorptivity of the azonium ion is high (138,000 M-1cm-1 at 600 nm) so that strong

absorbance is seen at wavelengths >700 nm at physiological pH.

73

Chart 3-6. The near-IR absorbing DOM-azo compound (8). Sulfone groups improve solubility.

We then tested the ability of DOM-azo to photoswitch upon exposure to near-IR light. A high

intensity LED (17 mW/cm2) with a maximum emission at 720 nm was used to irradiate a solution

of DOM-azo at pH 7.2, and a 540 nm LED source was used as a measuring beam (the spectra of

these LED sources is shown in Fig. 3a for reference). A photomultiplier tube was used to monitor

changes in absorbance over time. Figure 3-8(c) shows photoswitching time courses in response to

720 nm irradiation interleaved with periods of thermal relaxation in the dark. Clear, rapid

photoswitching is seen (τon = 0.9 ± 0.1 s) together with thermal relaxation (τoff 0.7 ± 0.1 s). These

time courses are rapid enough to permit spatial localization of the cis isomer in vivo. We also tested

the stability of DOM-azo in water by measuring absorbance scans over extended periods of time

and found no significant changes over >6 h.

74

Figure 3-8. UV-Vis spectra of DOM-azo (8) as a function of pH. The solid black line is the

limiting spectrum of the azonium ion at low pH (<pH 5.0); the dotted black line is the spectrum

of the neutral trans isomer obtained at pH 10.0. Spectra at intermediate pHs are shown in gray.

(b) Absorbance (680 nm) plotted vs. pH from the data in part (a); the fitted pKa = 6.7. (c)

Photoswitching with near-IR light under physiological conditions (aqueous buffer solution (20%

DMSO), pH 7.2). Absorbance at 540 nm was monitored vs. time while the sample was exposed

to 720 nm near-IR light (indicated by red bars above the trace) or left in darkness (black bars).

On and off time constants were extracted from exponential fits (two such fits are indicated). The

spectral outputs of the near-IR LED used for switching (red line) and the green LED used to

monitor absorbance (green line) are shown (red line) in part (a).

3.4 Conclusions

By tailoring the structure of an azobenzene with appropriate substituents, trans-to-cis

photoswitching was achieved with near-IR light (720 nm) from a continuous LED source under

physiological conditions. The core photoswitch structure developed here may form a basis for

75

future photopharmaceuticals. It could directly enable near-IR responses in current photoswitchable

tethered ligands for ion channels and receptors,38 in photoswitchable cross-linkers39 or in

azobenzene -based drug delivery systems.40

3.5 References


Society Reviews 2011, 40 (8), 4422-37.

2. Feringa, B.; Browne, W. R., Molecular Switches. 2nd Edition; Wiley-VCH 2011, 476.

3. Velema, W. A.; Szymanski, W.; Feringa, B. L., Photopharmacology: beyond proof of

principle. Journal of the American Chemical Society 2014, 136 (6), 2178-91.






6. Izquierdo-Serra, M.; Gascon-Moya, M.; Hirtz, J. J.; Pittolo, S.; Poskanzer, K. E.; Ferrer,

E.; Alibes, R.; Busque, F.; Yuste, R.; Hernando, J.; Gorostiza, P., Two-photon neuronal and

astrocytic stimulation with azobenzene-based photoswitches. Journal of the American Chemical

Society 2014, 136 (24), 8693-701.

7. Kienzler, M. A.; Reiner, A.; Trautman, E.; Yoo, S.; Trauner, D.; Isacoff, E. Y., A red-

shifted, fast-relaxing azobenzene photoswitch for visible light control of an ionotropic glutamate

receptor. Journal of the American Chemical Society 2013, 135 (47), 17683-6.

8. Broichhagen, J.; Frank, J. A.; Johnston, N. R.; Mitchell, R. K.; Smid, K.; Marchetti, P.;

Bugliani, M.; Rutter, G. A.; Trauner, D.; Hodson, D. J., A red-shifted photochromic sulfonylurea

for the remote control of pancreatic beta cell function. Chemical Communications 2015, 51 (27),

6018-21.

9. Nishioka, H.; Liang, X. G.; Kato, T.; Asanuma, H., A photon-fueled DNA nanodevice that

contains two different photoswitches. Angewandte Chemie International Edition 2012, 51 (5),

1165-1168.

10. Yang, Y.; Hughes, R. P.; Aprahamian, I., Near-infrared light activated azo-BF2 switches.


11. Wang, L. V.; Wu, H.-i., Biomedical Optics: Principles and Imaging. Wiley: 2007; p 376.

76

12. Li, Y.; Patrick, B. O.; Dolphin, D., Near-infrared absorbing azo dyes: Synthesis and X-ray

crystallographic and spectral characterization of monoazopyrroles, bisazopyrroles, and a boron-

azopyrrole complex. Journal of Organic Chemistry 2009, 74 (15), 5237-5243.

13. Bello, K. A.; Griffiths, J., Azo dyes with absorption bands in the near-infrared. Journal of

the Chemical Society, Chemical Communications 1986, (22), 1639-1640.

14. Guha, S. K.; Jha, O. P., Studies on some azonium derivatives. Journal of the Indian

Chemical Society 1968, 45 (5), 365-&.

15. Stoyanov, S.; Antonov, L.; Stoyanova, T.; Petrova, V., Ammonium azonium tautomerism

in some N,N-dialkylaminoazodyes—II. Compounds containing more than two protonation sites.

Dyes and Pigments 1996, 32 (3), 171-185.

16. Stoyanova, T.; Stoyanov, S.; Antonov, L.; Petrova, V., Ammonium-azonium tautomerism

in some N,N-dialkylaminoazo dyes—I. General considerations. Dyes and Pigments 1996, 31 (1),

1-12.

17. Samanta, S.; Babalhavaeji, A.; Dong, M. X.; Woolley, G. A., Photoswitching of ortho-

substituted azonium ions by red light in whole blood. Angewandte Chemie International Edition

2013, 52 (52), 14127-30.

18. Sawicki, E., Physical Properties of the Aminoazobenzene Dyes .8. Absorption spectra in

acid solution. Journal of Organic Chemistry 1957, 22 (9), 1084-1088.

19. Hallas, G.; Marsden, R.; Hepworth, J. D.; Mason, D., The effects of cyclic terminal groups

in 4-aminoazobenzene and related azo dyes—I. Electronic absorption-spectra of some monoazo

dyes derived from N-phenylpyrrolidine and N-phenylpiperidine. Journal of the Chemical Society

Perkin Transactions 2 1984, (1), 149-153.

20. Kauffmann, H.; Kugel, W., A proposition of distribution of auxochromes and azo

compounds. Berichte Der Deutschen Chemischen Gesellschaft 1911, 44, 2386-2389.

21. Kokkinos, K.; Wizinger, R., Law of Distribution of Auxochromes in Azo Bodies—I. Di-

and tetramethoxy-azobenzenes. Helvetica Chimica Acta 1971, 54 (1), 330-334.

22. Wizinger, R., Der verteilungssatz der auxochrome (Distribution rule of auxochromes).

Chimia 1965, 19 (5), 339-50.

23. Frisch, M. J., G. W. T., H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G.

Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P.

Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota,

R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A.

Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V.

N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S.

Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken,

C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C.

Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador,

77

J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski,

and D. J. Fox Gaussian 09, revision B.01, Gaussian, Inc.: Wallingford CT, 2009.

24. Becke, A. D., Density functional thermochemistry 3. The role of exact exchange. Journal

of Chemical Physics 1997, 98, 5648-5652.

25. Bleger, D.; Schwarz, J.; Brouwer, A. M.; Hecht, S., o-Fluoroazobenzenes as readily

synthesized photoswitches offering nearly quantitative two-way isomerization with visible light.


26. Alongi, K. S.; Shields, G. C., Theoretical calculations of acid dissociation constants: A

review article. Annual Reports in Computational Chemistry 2010, 6, 113-118.




Biochemistry 2012, 51 (32), 6421-31.

28. Samanta, S.; McCormick, T. M.; Schmidt, S. K.; Seferos, D. S.; Woolley, G. A., Robust

visible light photoswitching with ortho-thiol substituted azobenzenes. Chemical Communications

2013, 49 (87), 10314-6.

29. Jiang, X.; Yu, Y.; Chen, J.; Zhao, M.; Chen, H.; Song, X.; Matzuk, A. J.; Carroll, S. L.;

Tan, X.; Sizovs, A.; Cheng, N.; Wang, M. C.; Wang, J., Quantitative imaging of glutathione in

live cells using a reversible reaction-based ratiometric fluorescent probe. ACS Chemical Biology

2015, 10 (3), 864-74.

30. Boily, J. F.; Seward, T. M., On the dissociation of methyl orange: Spectrophotometric

investigation in aqueous solutions from 10 to 90 degrees C and theoretical evidence for

intramolecular dihydrogen bonding. Journal of Solution Chemistry 2005, 34 (12), 1387-1406.

31. Lewis, G. E., Photochemical reactions of azo compounds .1. spectroscopic studies of the

conjugate acids of cis-azobenzene and trans-azobenzene. Journal of Organic Chemistry 1960, 25

(12), 2193-2195.

32. Mallory, F. B.; Mallory, C. W., Photocyclization of stilbenes and related molecules.

Organic Reactions 2005, 30 (1), 1–456.

33. Sanchez, A. M.; Barra, M.; de Rossi, R. H., On the mechanism of the acid/base-catalyzed

thermal cis-trans isomerization of methyl orange. Journal of Organic Chemistry 1999, 64 (5),

1604-1609.

34. Garcia-Amoros, J.; Diaz-Lobo, M.; Nonell, S.; Velasco, D., Fastest thermal isomerization

of an azobenzene for nanosecond photoswitching applications under physiological conditions.


35. Vapaavuori, J.; Goulet-Hanssens, A.; Heikkinen, I. T. S.; Barrett, C. J.; Priimagi, A., Are

two azo groups better than one? Investigating the photoresponse of polymer-bisazobenzene

complexes. Chemistry of Materials 2014, 26, 5089–5096.

78

36. Ibaraki, M.; Ito, H.; Shimosegawa, E.; Toyoshima, H.; Ishigame, K.; Takahashi, K.; Kanno,

I.; Miura, S., Cerebral vascular mean transit time in healthy humans: a comparative study with

PET and dynamic susceptibility contrast-enhanced MRI. Journal of Cerebral Blood Flow and

Metabolism 2007, 27 (2), 404-13.

37. Carrera, E.; Jones, P. S.; Iglesias, S.; Guadagno, J. V.; Warburton, E. A.; Fryer, T. D.;

Aigbirhio, F. I.; Baron, J. C., The vascular mean transit time: a surrogate for the penumbra flow

threshold? Journal of Cerebral Blood Flow and Metabolism 2011, 31 (4), 1027-35.

38. Rullo, A.; Reiner, A.; Reiter, A.; Trauner, D.; Isacoff, E. Y.; Woolley, G. A., Long

wavelength optical control of glutamate receptor ion channels using a tetra-ortho-substituted

azobenzene derivative. Chemical Communications 2014, 50 (93), 14613-5.

39. Martin-Quiros, A.; Nevola, L.; Eckelt, K.; Madurga, S.; Gorostiza, P.; Giralt, E., Absence

of a stable secondary structure is not a limitation for photoswitchable inhibitors of beta-

arrestin/beta-Adaptin 2 protein-protein interaction. Chemical Biology 2015, 22 (1), 31-7.

40. Wang, D.; Wu, S., Red-light-responsive supramolecular valves for photocontrolled drug

release from mesoporous nanoparticles. Langmuir 2016, 32 (2), 632-6.

79

Modular Design of Optically Controlled Protein Affinity Reagents

Authors’ contributions: This chapter, excluding the addendum, is reproduced, by permission

from The Royal Society of Chemistry, from a published article: Babalhavaeji, A. & Woolley, G.

A. (2018). Modular Design of Optically Controlled Protein Affinity Reagents, Chemical

Communications, 54(13):1591-1594. The text has been has been modified for consistency. AB

carried out the experiments and analysis and GAW supervised the project.

The original article can be accessed via the following link:

http://pubs.rsc.org/en/content/articlelanding/2018/cc/c7cc07391g

4.1 Significance of this study

In the previous chapters of this thesis, I described how relatively low-cost computational methods

could aid in the rational design of azobenzene photoswitches for optical control of biology. I

showed that DFT and TD-DFT can be used to predict some key photophysical and thermodynamic

properties of azo photoswitches. This allowed us to choose promising substitution patterns on the

azo chromophore. By spectroscopic analysis, I demonstrated that the design strategy allowed us to

obtain red-shifted azonium ions with sufficiently high pKa to exist in high proportions under

physiological conditions. Cis to trans relaxation kinetics ranged from µs to ~1 s. These studies

complemented the toolkit of azo photoswitches with novel structures that can operate with long

wavelengths of light and are therefore suitable for in vivo use. In parallel work, I sought to apply

azobenzene-based cross-linkers to photocontrol protein-protein interactions. This chapter

describes how two azobenzenes with different photophysical character could be used to optically

control the conformation and function of a versatile family of protein-based affinity reagents:

Fynomers.

80

4.2 Introduction

Photopharmacology, the control of drugs with light, offers the prospect of avoiding off-target

effects of bioactive compounds by limiting their spatial and temporal activation.1

Photopharmaceuticals have been developed, thus far to preclinical stages only, that modulate ion

channel and GPCR function,2-4 that regulate tyrosine kinase activity,5 that control glucose

homeostasis,6 and that can restore visual responses in damaged retinas.7 Despite remarkable

successes in individual cases, a general approach for converting a pharmaceutical to its

photoswitchable version is not available. One may attempt “azologization”,8-9 or structure-based

design,10-12 yet in practice, each new target for control has often been found to require screening

and optimization.10, 13-14 It may be expected that bioavailability and metabolic stability will also

have to be optimized in each case as for typical small molecule drugs.6, 15

In contrast, antibody-based drugs, which now constitute a significant fraction of new therapeutics,

use the same basic three dimensional structure for all targets.16 Although fine-tuning of the

structure of antibody-based drugs has improved their efficacy, their development does not require

an extensive structure - activity analysis and is thus modular.17-18 In addition to IgGs, which are

large and bivalent and contain domains not involved in target recognition,19 a number of smaller,

protein-based affinity reagents have been developed including affibodies,20 avimers,21 anticalins,22

DARPins,23 and Fynomers24 among others.25-26 These domains can be recombinantly produced in

high yields. The residues involved in their binding interfaces can be randomized through phage

display and other selection techniques to obtain proteins that specifically bind to virtually any

desired target. As with IgGs, these structures can be therapeutic or diagnostic tools.25-27 Since the

tertiary structure of these proteins remains essentially unchanged through the selection process for

81

different targets, if there were a means for making the scaffold itself photoswitchable, this may

provide a general, modular, solution to making light-switchable therapeutics.

We chose Fynomers as a test scaffold on which to attempt to confer optical control. Fynomers are

derived from the SH3 domain of the human protein Fyn by variation of 10 residues in the RT and

n-Src loops.28-29 They are small, stable, cysteine-free and non-immunogenic domains of about 55

amino acids.24 We showed previously that the folding of the native Fyn SH3 domain can be photo-

controlled by introducing the biocompatible30 azobenzene-based intramolecular cross-linker

BSBCA (Figure 4-1).31 BSBCA is symmetrical such that only one species is formed upon cross-

linking; also, a minimum number of single bonds connect the azo double bond and the protein

backbone enhancing conformational coupling of the photoswitch to the protein. The cross-linking

site can be chosen such that the thermodynamically stable trans isomer of the cross-linker promotes

protein unfolding. Irradiation drives trans-to-cis photoisomerization of the cross-linker leading to

a decrease in its end-to-end distance thereby promoting protein folding. Importantly, the protein

remains soluble in its unfolded form under physiological conditions so that photo-control of

protein folding is completely reversible.31

82

Figure 4-1. Models showing the overall design. The Fynomer is held in an unfolded state (off-

state) by the trans form of the azobenzene-based cross-linker (shown in grey sticks) (top left).

Irradiation produces the cis state (pink sticks) and allows the Fynomer to fold into its active form

and bind chymase (bottom). (Fynomer-chymase complex PDB: 4AG1)(Inset: Chemical

structures of BSBCA and TOM photoswitches)

83

4.3 Materials and methods

4.3.1 Protein production

pET24b+ plasmids containing the sequence that codes for the 4C-E4 Fynomer,29 as well as its

L3CL29C double mutant were produced by Biobasic Inc. In each plasmid the coding sequence

was followed by a 6-His tag for purification purposes. The modified loops are underlined.

Fynomer sequence:

MRGSGVTLFVALYDYNATRWTDLSFHKGEKFQILEFGPGDWWEARSLTTGETGYIPSN

YVAPVDSIQGEQKLISEEDLHHHHHH*

L3CL29C-Fynomer sequence:

MRGSGVTCFVALYDYNATRWTDLSFHKGEKFQICEFGPGDWWEARSLTTGETGYIPSN

YVAPVDSIQGEQKLISEEDLHHHHHH*

For producing each Fynomer, chemically competent E. coli (BL21*DE3) cells were transformed

with the respective plasmid using the heat shock method and plated on agar/lysogeny broth (LB)

plates containing 50 µg/mL kanamycin. A colony was used for inoculating 50 mL of LB culture

media on the following day. The culture was grown overnight at 37°C in the presence of 50 µg/mL

kanamycin. The overnight culture was transferred to 1 L LB containing the same concentration of

kanamycin and incubated at 37°C on a shaker (250 rpm) until an optical density (OD) of 0.6 was

reached. Protein expression was induced by the addition of isopropyl-β-D-thiogalactopyranoside

(IPTG) to a final concentration of 750 µM. The induced culture was then incubated/shaken for 4

hours at the same temperature. The culture was centrifuged at 4000 rpm for 1 hour and the

harvested cell pellets were resuspended in 30 mL of 100 mM phosphate buffered 6 M GuCl

84

solution (pH=8.0, containing 10 mM imidazole) and was stored at -20°C. On a different day, the

cell lysis mixture was thawed on an orbital shaker for two hours at room temperature. Cell debris

was removed by centrifugation at 12000 rpm for 2 hours. The crude supernatant was filtered

through a 0.45 µm syringe filter and then applied to a Ni-NTA column which was pre-equilibrated

with the same buffer. Once loaded with the His-tagged protein, the Ni-NTA resin was washed with

5 mL of the basic GuCl buffer three times. The protein was eluted by lowering the pH to 4.5 using

a solution of 6 M GuCl, 200 mM sodium acetate. The expression yield was found to be 5 mg

purified protein per litre of culture on average. The eluate was then used for either the cross-linking

reaction or further purification for the enzymatic assays by reversed-phase HPLC (vide infra).

4.3.2 Production of 15N-labelled proteins

A colony of transformed BL21*DE3 cells was used for inoculating 25 mL of LB media. The

culture was grown overnight at 37°C in the presence of 50 µg/mL kanamycin, then 6 mL of the

overnight culture was centrifuged. The pellet was resuspended in 50 mL of M9 minimal media

containing 50 µg/mL kanamycin and was incubated/shaken until the optical density reached

OD=0.6. The culture was centrifuged and the pellet was resuspended in 1 L of M9 media,

supplemented with 1 g of 15NH4Cl (Cambridge Isotope Laboratories, Inc.), 1 mL of 15N Bioexpress

growth mixture (Cambridge Isotope Laboratories, Inc.), 10 mg of thiamine, 10 mg of biotin, 0.3%

D-glucose, 1 mM MgSO4, 1 mM CaCl2 and 50 µg of kanamycin. The culture was induced after an

OD of 0.6 was reached. Subsequent steps were identical to those described above for unlabelled

proteins.

85

4.3.3 Cross-linking of Fynomer with azobenzene cross-linkers

Affinity-purified Fynomer was buffer exchanged into pH 8.0, 8M GuCl (10 mM phosphate, 10

mM imidazole) and concentrated using centrifuge concentrator tubes (Millipore Sigma, 3000

MWCO). The absorbance of the resulting protein stock solution was measured at 280 nm to

determine protein concentration. (Molar absorption coefficients for wild-type and mutant

Fynomers were obtained from the ProtParam online tool: 22460 M-1cm-1)

The protein was then treated with 10 equivalents of Tris(2-carboxyethyl)phosphine (TCEP) and

incubated at room temperature for at least 30 minutes to ensure the cysteine thiols were fully

reduced. Stock solutions of BSBCA and TOM (synthesized as described previously32-33) were

prepared in anhydrous DMSO (Sigma). To cross-link the Fynomer with BSBCA, 3 equivalents of

BSBCA were added to the reduced protein and the reaction mixture was shaken at 50°C for 12

hours. To produce TOM-Fynomer, reduced protein was first diluted with DMSO such that the final

concentration of DMSO would be at least 30% (v/v). Then 2 equivalents of TOM were added and

the mixture was shaken at 50°C for 6 hours. Another 5 equivalents of TCEP followed by 1

equivalent of TOM were added to the mixture and shaking was continued for 12 more hours. The

step-wise addition of cross-linker in this case prevented the formation of species with two

azobenzenes attached.

The reaction mixture was purified by reversed-phase HPLC using a C18 semi-preparatory column

(Zorbax, RxC18) using 0.1% (v/v) TFA in water and 0.1% (v/v) TFA in acetonitrile as solvents

with a linear gradient of 5% to 70% acetonitrile in the course of 25 minutes. The eluate was

monitored using a dual wavelength Waters detector at 280 nm (chromophore: protein and

azobenzene) and 360 nm (only azobenzene). The cross-linked Fynomers eluted at ~52%

acetonitrile. The eluate was then lyophilized and reconstituted using 50 mM sodium phosphate pH

86

8.0. The purified product was analyzed with ESI or MALDI mass spectrometry. The expected and

observed mass values are as follows:

BSBCA-Fynomer: Expected 10089.0, Observed (QTOF ESI-MS) 10088.9, BSBCA-15N-

Fynomer: Expected 10206.7, Observed (QTOF ESI-MS) 10205.4, TOM-Fynomer: Expected

10049.0, Observed (QTOF ESI-MS) 10049.0, TOM-15N-Fynomer: Expected 10166.7, Observed

(MALDI-TOF-MS) 10167.7. Refer to Appendix 6.2 for mass spectra.

4.3.4 Electronic absorption spectroscopy

UV-Visible spectra were recorded on Perkin Elmer Lambda 35, Shimadzu UV-2401PC, or a diode

array spectrophotometer (Ocean Optics Inc., USB4000). Temperature was maintained at 22°C for

all measurements (Quantum Northwest), using 10 mm or 1.5 mm quartz cuvettes (Hellma

Analytics).

Photoisomerization: For BSBCA-Fynomer, protein samples in 50 mM sodium phosphate pH 8

were dark adapted by incubation at 37°C overnight in light protected containers. UV irradiation

was performed by placing a 365 nm LED above the sample tube for 1 minute (897-LZ440U610

LedEngin LED, San Jose, CA, USA, operating at 68 mW/cm2). For blue light irradiation, a 445

nm LED was used for 1 minute (Luxeon III Star LED Royal Blue Lambertian operating at 40

mW/cm2 at 700 mA). For TOM-Fynomer, a fan-cooled bright red LED was used for 30 minutes

(LedEngin LZ4-40R200-0000, 700 mA, 635 nm, 90 mW/cm2).

87

4.3.5 Estimates of the percentage of cis/trans in photostationary states

(PSS)

Spectra for purely cis BSBCA and cis TOM (both conjugated to peptide FK11) from the

literature32, 34 were used to approximate the cis isomers’ contribution to absorbance at max in the

dark adapted spectra (367 nm and 400 nm for BSBCA-Fynomer and TOM-Fynomer, respectively).

It is also assumed that the fraction of cis isomer in the fully dark-adapted samples is negligible.

This allows for the calculation for %cis using the following equation and the absorbance values at

these wavelengths:

𝑃𝑆𝑆 𝑐𝑖𝑠% = 𝐴𝑃𝑆𝑆 − 𝐴𝑝𝑢𝑟𝑒 𝑐𝑖𝑠

𝐴𝑑𝑎𝑟𝑘−𝑎𝑑𝑎𝑝𝑡𝑒𝑑 − 𝐴𝑝𝑢𝑟𝑒 𝑐𝑖𝑠

4.3.6 Thermal relaxation rates

For studying the effect of chymase binding on the relaxation rate of the azobenzene, a sample of

BSBCA-Fynomer was mixed with chymase to reach final concentrations of 4 µM and 4.5~9.0 µM,

with respect to BSBCA-Fynomer and chymase, respectively. The final solution contained 12.5%

glycerol, 20 mM sodium phosphate, 20 mM Tris, and 0.4 M NaCl, and its pH was measured to be

7.8. The sample was placed in a 1 cm cuvette. The cuvette was irradiated with UV (365 nm) for

15 seconds, gently mixed by pipetting, then immediately capped with Parafilm to prevent

evaporation, and placed inside the spectrophotometer. The sample was periodically scanned (250

to 600 nm for each scan, integration time: 2s; scan speed: 480 nm/min) with 20 minute intervals

for 6 hours. Similar kinetic measurements were performed on a sample of 4 µM BSBCA-Fynomer

in the same buffer with the same pH.

88

4.3.7 NMR spectroscopy

Lyophilized isotopically labelled compounds were dissolved in 10 mM HEPES (pH 7.2), 10%

D2O. They were then centrifuged at 15000 rcf for 10 minutes to ensure that no insoluble precipitate

was present in the final samples. All HSQC experiments were performed at CSICOMP

(Department of Chemistry, University of Toronto) using an Agilent DD2 700 MHz spectrometer

equipped with an HFCN cold probe. An NH HSQC Watergate pulse sequence from the Varian

Biopack library was used. All spectra were acquired at 25°C with 128 transients and 64 increments

in the 15N dimension. Sample concentrations of ~200 µM were used since higher concentrations

promoted self-association of the analyte and decreased the efficiency of switching due to higher

optical density. Spectra were processed using the NMRPipe processing suite.35 FID signals were

zero filled to double the original data size and apodized using a sine window function prior to

Fourier transformation. In the indirect dimension, linear prediction was applied to double the

original data size. To achieve the difference spectra, subtraction of FIDs was done by the addNMR

program of NMRPipe. Analysis was aided by NMRViewJ (One Moon Scientific).

Dark-adapted BSBCA-Fynomer was prepared by incubating a foil-wrapped NMR tube containing

the sample at 37°C overnight. The foil-wrap was removed immediately prior to insertion in the

magnet to minimize exposure of the sample to ambient light. To irradiate a sample inside the

magnet, a 1 mm optical fibre (Thorlabs) was inserted through the top of the NMR tube. The

aforementioned LEDs were placed directly in front of the other end of the tube (with a 2 mm gap

to avoid touching). The following irradiation methods were employed to ensure that the

photostationary was reached for each sample.

BSBCA-Fynomer, UV: 30 minutes, inside the magnet

89

TOM-Fynomer, Red: 1 hour inside the magnet following offline irradiation of the tube for

four hours

TOM-Fynomer, Blue: 1 hour inside the magnet

In all three cases, light was kept on during acquisition.

The spectra for both the dark-adapted and UV-irradiated states exhibited some unresolved signal

in the centre of the spectrum consistent with the formation of soluble aggregates at higher

concentrations needed for 2D-NMR. Subtraction of the processed spectrum files resulted in a

spectrum with two sets of cross-peaks with opposite phases which correspond to the changes of

peak intensities after irradiation. The disappearance of the unresolved peaks after subtraction

suggested that the soluble aggregates did not undergo significant changes upon irradiation.

4.3.8 Activity assay

The activity assay was based on measurement of the production of para-nitroaniline (Scheme 4-

1).

Scheme 4-1. The reaction used in chymase kinetic assay

Chymase and the substrate peptide were purchased from Sigma (CS1140). A stock solution of the

substrate peptide was prepared in DMSO. Substrate working solutions with various concentrations

were prepared by diluting the stock solutions with a phosphate buffer (50 mM sodium phosphate,

90

pH 8.0). DMSO was added to maintain the same final DMSO percentage in all dilutions for

consistency.

All kinetic measurements were performed in triplicate. For each kinetic measurement, 100 µL of

the enzyme, or enzyme/Fynomer solution was added to 50 µL of substrate solution in a 1 cm UV-

rated plastic micro-cuvettes (Sigma, BR759235) and immediately placed in the Lambda 35

instrument equipped with a moving cuvette rack. Absorbance at 400 nm (=12300 M-1cm-1)36 was

measured every 30 seconds for 10 to 15 minutes following the initiation of the reaction. The

spectrophotometer slit width and integration time were adjusted to 0.5 nm and 0.5 s to minimize

the effect of the measuring light on the azobenzene.

Uninhibited reaction:

A 333-fold dilution of ~9 M chymase was prepared with the addition of phosphate buffer and the

assay buffer provided in the chymase assay kit (Sigma, CS1140). The final percentage of the assay

buffer was 20%.

Inhibition assays:

Enzyme/Fynomer solutions were first made by diluting the chymase stock (~9 µM) 333-fold with

the phosphate and assay buffers, and adding the inhibitor solution (unmodified, dark-adapted

BSBCA-Fynomer, or TOM-Fynomer). Once mixed with the substrate solution, the final

percentage of the assay kit buffer was 17% v/v, and the nominal enzyme concentration was ~18

nM. For the estimation of apparent IC50 values for the BSBCA-Fynomer, Triton-X100 (Sigma)

was added to achieve a final concentration of 0.02% v/v to help prevent the formation of

aggregates. The unmodified Fynomer was also treated with Triton to the same final concentration

for consistency.

91

All samples were made in a dark room under a dim red light.

For the lit-state of BSBCA-Fynomer, the enzyme/inhibitor working solution was divided into two

aliquots. The first aliquot was irradiated with UV light for 30 seconds while inside a temperature

controller jacket. The LED was fan-cooled during the experiment. The light was then switched off

and the solution was mixed by inversion to ensure that the percentage of cis-azo is not higher near

the top of the tube (which is closer to the light source). The solution was then quickly mixed with

the substrate and the measurement initiated.

IC50 data (Figure 4-4a, Figure 4-5) were fitted to a standard 4-parameter equation29:

Vo = D + ((A-D)/(1+10^((log([BSBCA-Fynomer])-log(IC50))*B)))

Rate data as a function of substrate concentration (Figure 4-4b) were fitted to the Michaelis-

Menten (MM) equation and the MM equation for competitive inhibition using the global fitting

method in IgorPro (Wavemetrics). The total concentration of inhibitor was adjusted to give an

observed rate similar to that seen in IC50 data, (~80 nM). Irradiated state data were included in the

global fitting by fitting to the following modified MM (competitive inhibition) equation:

Vo = Vmax*[S]/((Km*(1 + (0.2*[I]/Ki_trans) + (0.8*[I]/Ki_cis))) + [S]);

Which assumes the BSBCA-Fynomer is 80% cis when irradiated based on UV-Vis data (see

above). Mean initial rates obtained from triplicate measurements were subjected to t-test statistical

analysis. The analysis confirmed that the difference between the rates corresponding to dark-

adapted and UV-irradiated inhibitors was significant with 99.9% confidence.

Vmax , kcat, and Km values for chymase in the absence of inhibitor were: Vmax = 110 ± 15 nM/s; kcat

= 7 ± 1 s-1 (if [chymase] is 15 nM), Km = 225 ± 20 µM. These values are similar to those reported

92

previously for the same substrate.37 We note these fits for Ki values are approximate since the

concentration of chymase in these assays is ~ 15-20 nM (similar to the fitted Kis).

For the irradiated states of TOM-Fynomer, the enzyme/inhibitor working solution was first

irradiated with red light for 30 minutes. After the measurement of the red-irradiated samples, the

same working solution was irradiated with blue light for 5 minutes. Both LEDs were fan-cooled

during irradiation.

Control experiment with unmodified Fynomer:

To ensure that UV irradiation has no effect on the enzyme or any other component of this system,

the inhibition assay was performed using the unmodified Fynomer with and without UV

irradiation.

Testing the reversibility of activity change:

A solution of enzyme and BSBCA-Fynomer was prepared as described above and was divided

into two aliquots. The first aliquot was irradiated with blue light (445 nm) for 1 minute. A 100 µL

aliquot of the enzyme/inhibitor solution was mixed with 50 µL of substrate solution for each of

the triplicate cuvettes and absorbance was monitored as a function of time. The second aliquot was

irradiated with UV light (365 nm) for 30 seconds and immediately irradiated with blue light (445

nm) for 1 minute and then mixed with substrate in the same fashion.

4.4 Results and discussion

We wished to test if this intramolecular cross-linking approach, which appears to depend primarily

on the distance between cross-linker attachment sites (Cys residues) in the structure, as well as

local mobility,38 was modular, i.e. could it be used to switch activity of a Fynomer active against

93

a potential therapeutic target? We chose a Fynomer that was developed by Schlatter et al to inhibit

human chymase by binding near its active site.29 Chymase is a serine protease that is secreted by

mast cells and is involved in cardiovascular diseases and pathological inflammatory conditions.39-

40 Based on previous work, we chose a Fynomer cross-linking site defined by introduction of a

pair of Cys residues (L3C, L29C) to enable intramolecular reaction with a thiol-reactive cross-

linker. The Cys-to-Cys distance of this site is 8~16 Å (S-to-S atom) in the folded protein.38

When cross-linked using BSBCA (Figure 4-1) the Cys residues are expected to be too close

together to accommodate the trans isomer, so that protein unfolding should occur. Irradiation to

produce the cis-isomer should produce an increase in the folded, and therefore active, fraction.

This direction of photoswitching (trans-off/cis-on) is preferred since it leads to the greatest change

in the concentration of the active isomer after irradiation.41 The cross-linking site is on the opposite

side of the Fynomer protein from the RT and n-Src loops that were randomized to produce chymase

selectivity (Figure 4-1), so that the cross-linker is not expected to directly interfere with binding

(although, vide infra).

We expressed and purified the Cys-modified Fynomer and introduced an intramolecular cross-link

with BSBCA. The photophysical behaviour of the product was similar to that of the BSBCA-cross-

linked wild-type Fyn SH3 domain. Irradiation at 365 nm efficiently generated a large fraction of

cis azobenzene: >80% at the photostationary state as judged by comparison with spectra for pure

cis and trans isomers (Figure 4-2). In the dark, the cis isomer thermally relaxed with a half-life of

~100 min at 22°C. Irradiation with blue light quickly established a photostationary state comprised

of ~80% trans (Figure 4-2).

94

Figure 4-2. UV-Vis spectra of BSBCA-Fynomer when dark adapted (solid black), 365 nm-

irradiated (dotted purple), and 445 nm irradiated (dashed blue).

We then characterized the effect of azobenzene switching on the conformation of the Fynomer

using 15N-1H HSQC NMR experiments. The spectrum of the dark-adapted BSBCA-Fynomer

showed narrow peak dispersion centred around 8 ppm on the proton axis, as well as a set of less

intense but more dispersed peaks consistent with the coexistence of folded and unfolded states in

slow exchange on the NMR time-scale as was observed with BSBCA-cross-linked wild-type Fyn

SH3.31 Using a fibre-optic cable, the solution was then irradiated inside the NMR spectrometer

with a 365 nm beam and a second HSQC spectrum was recorded using the same acquisition

parameters. The intensity of the more dispersed cross-peaks increased after irradiation, consistent

with a transition to a more folded ensemble (Figure 4-3). Thus, trans-to-cis isomerization of

BSBCA drives folding of the chymase-binding Fynomer in the same manner as it does with wild-

type Fyn SH3. This observation indicates that the binding specificity to the Fyn domain can be

95

changed by exchange of the RT and n-Src loops without affecting the ability of a photoswitchable

intramolecular cross-link to control the tertiary fold, i.e. at least in this case, the system behaves in

a modular fashion.

Figure 4-3. A light vs. dark difference NH-HSQC spectrum of BSBCA-Fynomer. Purple peaks

are those that become more intense after UV-irradiation whereas the black peaks (associated

with the unfolded state) are those that decrease in intensity.

We then tested whether the BSBCA-Fynomer would exhibit photo-controlled inhibition of

chymase activity. We used an assay in which chymase catalyses the hydrolysis of a substrate

peptide to produce para-nitroaniline (which can be quantified spectrophotometrically). Binding of

the unmodified Fynomer to chymase inhibits this reaction as demonstrated by Schlatter et al.29

Figure 4-4 shows the effect of BSBCA-linked Fynomer on the initial rate of this reaction. These

data indicate that the UV-irradiated state of the BSBCA-Fynomer acts as a stronger chymase

inhibitor than the dark-adapted state. The IC50 values observed with high substrate concentration

96

are 97 ± 2 nM for the irradiated state and 126 ± 4 nM for the dark-adapted state (Figure 4-4a).

Global fitting of inhibition data as a function of substrate concentration (Figure 4-4b) gave an

estimated Ki of 23 ± 8 nM for the pure cis isomer of BSBCA-Fynomer vs 85 ± 10 nM for the trans

isomer. The dark-adapted BSBCA-Fynomer retains significant inhibitory ability, consistent with

the persistence of some folded structure (vide supra), while the UV-irradiated BSBCA-Fynomer

is not as active as unmodified Fynomer (Ki = 12 nM; Figure 4-5), so that the degree of switchability

is less than ideal (vide infra). The light-dependent change in BSBCA-Fynomer activity is

nevertheless fully consistent with the NMR data indicating the protein was more folded, and

thereby more active, in the UV-irradiated state.

Figure 4-4. BSBCA-Fynomer inhibition of chymase activity. (a) IC50 curves for dark-adapted

(IC50: 126 ± 4 nM), UV-irradiated (97 ± 2 nM) and (calculated) pure cis states (79 ± 2 nM)

([substrate]=376 µM). (b) Michaelis-Menten plot showing chymase-catalysed peptide hydrolysis

as a function of substrate concentration in the absence of inhibitor and in the presence of

BSBCA-Fynomer (~80 nM). The inhibitory activity of BSBCA-Fynomer increases upon UV

irradiation (Ki-trans = 85 ± 10 nM; Ki-cis = 23 ± 8 nM).

97

Figure 4-5. IC50 curves for dark-adapted (black line), (126 ± 4 nM), UV-irradiated (pink line),

(97 ± 2 nM), (calculated) pure cis state (dotted blue line), (79 ± 2 nM) and unmodified Fynomer

(green line), (12 ± 1 nM).

We verified that UV irradiation did not affect either chymase or the Fynomer in the absence of an

azobenzene cross-linker (Figure 4-6).

Figure 4-6. UV irradiation does not cause a significant change of activity of chymase when the

Fynomer is not cross-linked with BSBCA.

98

One key advantage of photoswitchable systems is their reversibility. The BSBCA-Fynomer

undergoes thermal cis-to-trans relaxation with a half-life of ~100 min (Figure 4-7). The target does

not appear to substantially alter the isomerization properties of the system. Blue light irradiation

to produce ~80% trans state rapidly (<1 min) reversed the effect of UV irradiation on BSBCA-

Fynomer chymase inhibition (Figure 4-8).

Figure 4-7. (a) Thermal relaxation of BSBCA-Fynomer (22°C, phosphate/Tris buffer containing

12.5% glycerol and 0.4 M NaCl, pH 7.8). Absorbance at 360 nm was monitored. Solid curve

shows fitted mono exponential curve (τ½=101 ± 15 min). Addition of chymase (b) does not

significantly alter the relaxation rate (τ½=96 ± 15 min).

99

Figure 4-8. No significant change in the rate of catalysis by chymase is observed following the

two irradiation patterns. This indicates that the isomerization-driven change in the inhibitory

activity of Fynomer is reversible.

Although UV-switchable bioactive molecules have uses in certain settings,42 the photo-control of

bioactivity is ideally effected by switches that operate at long wavelengths where light better

penetrates tissue and is less damaging to cells.43 To examine whether another photoswitch with a

different operating wavelength range could be used with the same system, we cross-linked the

Fynomer with a tetra-ortho-methoxy analogue of BSBCA: TOM (Figure 4-1). This azobenzene

derivative strongly absorbs green light and undergoes similar end-to-end distance changes to

BSBCA when it isomerizes.33 Despite a very low absorptivity in the red region of the spectrum,

the slow thermal back reaction of TOM (τ½ up to 2 days) also allows trans-to-cis photoswitching

with red light (635 nm).34 Relaxation can then be achieved photochemically, with a brief exposure

to blue light, which quickly establishes a photostationary state composed of ~85% trans.33 We

measured the absorption spectra of the TOM-cross-linked Fynomer (TOM-Fynomer) after

100

irradiating the sample with red light and blue light (Figure 4-9) and obtained spectra similar to

those reported for model peptides cross-linked with TOM in aqueous solution.34

Figure 4-9. UV-Vis spectra of TOM-Fynomer when dark adapted (solid black), 635 nm-

irradiated (dotted red), and 445 nm-irradiated (dashed blue).

We then performed NH-HSQC NMR experiments on isotopically labelled TOM-Fynomer. Red

light was continuously shone on the sample prior to and during NMR acquisition via an optical

fibre to generate a state estimated to be 70% cis. Blue irradiation via the optical fibre produced a

photostationary state that was predominately trans (76%). The NMR NH-HSQC difference

spectrum of red and blue-irradiated TOM-Fynomer showed an increase in the intensity of the

101

folded cross-peaks in the cis-enriched state (Figure 4-10a), very similar to that observed for UV-

irradiated BSBCA-Fynomer (Figure 4-11).

Figure 4-10. (a) Difference HSQC spectrum showing peaks whose intensity increases under red

light in red, and whose intensity increases in blue light in blue. (b) Kinetic traces of the chymase

assay in the presence of TOM-Fynomer (1.5 µM) after red and blue irradiation. The red-irradiated

trace has a lower slope, which indicates higher inhibitory activity.

102

Figure 4-11. The folded state (UV-irradiated) of BSBCA-Fynomer (purple cross-peaks)

corresponds closely to the folded state (red-irradiated) of TOM-Fynomer (red cross-peaks).

Finally, we evaluated the inhibitory activity of TOM-Fynomer with the same enzymatic assay.

Comparing the kinetic traces of a chymase reaction in the presence of TOM-Fynomer irradiated

with red or blue light, it was evident that the red-irradiated TOM-Fynomer was a better chymase

inhibitor (Figure 4-10b). The observation was again fully consistent with the NMR data (Figure 4-

10a).

4.5 Conclusions

This research presents an approach for optical control of a versatile family of affinity reagents. By

cross-linking a chymase-binding Fynomer with either a UV-switchable or a red/blue light

switchable azobenzene-based cross-linker, Fynomer folding and function were remotely altered

by light. The observation that different azobenzene-based photoswitches exhibited the same global

impact on the Fynomer structure and activity indicated that the system is also modular in terms of

the photoswitch component. Clearly affinity, degree of affinity switching, irradiation wavelength,

103

and thermal relaxation rate of this first generation system must be substantially improved before

in vivo use might be considered. The affinity of chymase to the BSBCA-Fynomer in the more

active cis state is still ~10-fold less than the unmodified Fynomer (Figure 4-5) suggesting some

unfavourable interaction, perhaps between the cross-linker itself and a chymase loop (Figure 4-1).

The degree of affinity switching appears limited primarily by the incomplete ability of the trans

cross-linker to unfold and inactivate the Fynomer.38 Cross-linkers that undergo larger end-to-end

distance changes could be tested at this or at different cross-linking sites on the Fynomer scaffold

to improve the degree of affinity switching. Finally, for in vivo use, a red or near-IR absorbing and

fast-relaxing photoswitch is more desirable than BSBCA.43 Nevertheless, the modularity we

observe here implies that these parameters could all be optimized independently. Alternatively

other affinity reagent scaffolds, for example multimeric scaffolds,23 may show enhanced

switchability through cooperative effects.44 This study thus sets up a roadmap for the design of

optically-controlled protein affinity reagents.

4.6 References

1. Lerch, M. M.; Hansen, M. J.; van Dam, G. M.; Szymanski, W.; Feringa, B. L., Emerging

targets in photopharmacology. Angewandte Chemie International Edition 2016, 55 (37), 10978-

99.

2. Rullo, A.; Reiner, A.; Reiter, A.; Trauner, D.; Isacoff, E. Y.; Woolley, G. A., Long

wavelength optical control of glutamate receptor ion channels using a tetra-ortho-substituted

azobenzene derivative. Chemical Communications 2014, 50 (93), 14613-5.

3. Pittolo, S.; Gomez-Santacana, X.; Eckelt, K.; Rovira, X.; Dalton, J.; Goudet, C.; Pin, J. P.;

Llobet, A.; Giraldo, J.; Llebaria, A.; Gorostiza, P., An allosteric modulator to control endogenous

G protein-coupled receptors with light. Nature Chemical Biology 2014, 10 (10), 813-5.

4. Agnetta, L.; Kauk, M.; Canizal, M. C. A.; Messerer, R.; Holzgrabe, U.; Hoffmann, C.;

Decker, M., A Photoswitchable dualsteric ligand controlling receptor efficacy. Angewandte

Chemie International Edition 2017, 56 (25), 7282-7287.

5. Ferreira, R.; Nilsson, J. R.; Solano, C.; Andreasson, J.; Grotli, M., Design, synthesis and

inhibitory activity of photoswitchable RET kinase inhibitors. Scientific Reports 2015, 5, 9769.

104

6. Mehta, Z. B.; Johnston, N. R.; Nguyen-Tu, M. S.; Broichhagen, J.; Schultz, P.; Larner, D.

P.; Leclerc, I.; Trauner, D.; Rutter, G. A.; Hodson, D. J., Remote control of glucose homeostasis

in vivo using photopharmacology. Scientific Reports 2017, 7 (1), 291.

7. Polosukhina, A.; Litt, J.; Tochitsky, I.; Nemargut, J.; Sychev, Y.; De Kouchkovsky, I.;

Huang, T.; Borges, K.; Trauner, D.; Van Gelder, R. N.; Kramer, R. H., Photochemical restoration

of visual responses in blind mice. Neuron 2012, 75 (2), 271-82.

8. Schoenberger, M.; Damijonaitis, A.; Zhang, Z.; Nagel, D.; Trauner, D., Development of a

new photochromic ion channel blocker via azologization of fomocaine. ACS Chemical

Neuroscience 2014, 5 (7), 514-8.

9. Broichhagen, J.; Frank, J. A.; Trauner, D., A roadmap to success in photopharmacology.

Accounts of Chemical Research 2015, 48 (7), 1947-60.

10. Szymanski, W.; Ourailidou, M. E.; Velema, W. A.; Dekker, F. J.; Feringa, B. L., Light-

controlled histone deacetylase (HDAC) inhibitors: Towards photopharmacological chemotherapy.

Chemistry 2015, 21 (46), 16517-16524.

11. Wutz, D.; Gluhacevic, D.; Chakrabarti, A.; Schmidtkunz, K.; Robaa, D.; Erdmann, F.;

Romier, C.; Sippl, W.; Jung, M.; Konig, B., Photochromic histone deacetylase inhibitors based on

dithienylethenes and fulgimides. Organic and Biomolecular Chemistry 2017, 15 (22), 4882-4896.

12. Reisinger, B.; Kuzmanovic, N.; Loffler, P.; Merkl, R.; Konig, B.; Sterner, R., Exploiting

protein symmetry to design light-controllable enzyme inhibitors. Angewandte Chemie

International Edition 2014, 53 (2), 595-8.

13. Hansen, M. J.; Velema, W. A.; de Bruin, G.; Overkleeft, H. S.; Szymanski, W.; Feringa,

B. L., Proteasome inhibitors with photocontrolled activity. ChemBioChem: a European Journal of

Chemical Biology 2014, 15 (14), 2053-7.

14. Velema, W. A.; Hansen, M. J.; Lerch, M. M.; Driessen, A. J.; Szymanski, W.; Feringa, B.

L., Ciprofloxacin-photoswitch conjugates: A facile strategy for photopharmacology. Bioconjugate

Chemistry 2015, 26 (12), 2592-7.

15. Meanwell, N. A., Improving drug design: An update on recent applications of efficiency

metrics, strategies for replacing problematic elements, and compounds in nontraditional drug

space. Chemical Research in Toxicology 2016, 29 (4), 564-616.

16. Carter, P. J., Potent antibody therapeutics by design. Nature Reviews Immunology 2006, 6

(5), 343-57.

17. AlDeghaither, D.; Smaglo, B. G.; Weiner, L. M., Beyond peptides and mAbs— current

status and future perspectives for biotherapeutics with novel constructs. Journal of Clinical

Pharmacology 2015, 55 Suppl 3, S4-20.

18. Moore, G. L.; Chen, H.; Karki, S.; Lazar, G. A., Engineered Fc variant antibodies with

enhanced ability to recruit complement and mediate effector functions. MAbs 2010, 2 (2), 181-9.

105

19. Brinkmann, U.; Kontermann, R. E., The making of bispecific antibodies. MAbs 2017, 9

(2), 182-212.

20. Nord, K.; Gunneriusson, E.; Ringdahl, J.; Stahl, S.; Uhlen, M.; Nygren, P. A., Binding

proteins selected from combinatorial libraries of an alpha-helical bacterial receptor domain. Nature

Biotechnology 1997, 15 (8), 772-7.

21. Silverman, J.; Liu, Q.; Bakker, A.; To, W.; Duguay, A.; Alba, B. M.; Smith, R.; Rivas, A.;

Li, P.; Le, H.; Whitehorn, E.; Moore, K. W.; Swimmer, C.; Perlroth, V.; Vogt, M.; Kolkman, J.;

Stemmer, W. P., Multivalent avimer proteins evolved by exon shuffling of a family of human

receptor domains. Nature Biotechnology 2005, 23 (12), 1556-61.

22. Vogt, M.; Skerra, A., Construction of an artificial receptor protein ("anticalin") based on

the human apolipoprotein D. ChemBioChem: a European Journal of Chemical Biology 2004, 5

(2), 191-9.

23. Pluckthun, A., Designed ankyrin repeat proteins (DARPins): binding proteins for research,

diagnostics, and therapy. Annual Review of Pharmacology and Toxicology 2015, 55, 489-511.

24. Grabulovski, D.; Kaspar, M.; Neri, D., A novel, non-immunogenic Fyn SH3-derived

binding protein with tumor vascular targeting properties. J. Biol. Chem. 2007, 282 (5), 3196-204.

25. Jost, C.; Pluckthun, A., Engineered proteins with desired specificity: DARPins, other

alternative scaffolds and bispecific IgGs. Current Opinion in Structural Biology 2014, 27, 102-12.

26. Simeon, R.; Chen, Z., In vitro-engineered non-antibody protein therapeutics. Protein Cell

2017.

27. Azhar, A.; Ahmad, E.; Zia, Q.; Rauf, M. A.; Owais, M.; Ashraf, G. M., Recent advances

in the development of novel protein scaffolds based therapeutics. International Journal of

Biological Macromolecules 2017, 102, 630-641.

28. Brack, S.; Attinger-Toller, I.; Schade, B.; Mourlane, F.; Klupsch, K.; Woods, R.; Hachemi,

H.; von der Bey, U.; Koenig-Friedrich, S.; Bertschinger, J.; Grabulovski, D., A bispecific HER2-

targeting FynomAb with superior antitumor activity and novel mode of action. Molecular Cancer

Therapeutics 2014, 13 (8), 2030-9.

29. Schlatter, D.; Brack, S.; Banner, D. W.; Batey, S.; Benz, J.; Bertschinger, J.; Huber, W.;

Joseph, C.; Rufer, A.; van der Klooster, A.; Weber, M.; Grabulovski, D.; Hennig, M., Generation,

characterization and structural data of chymase binding proteins based on the human Fyn kinase

SH3 domain. MAbs 2012, 4 (4), 497-508.

30. Beharry, A. A.; Wong, L.; Tropepe, V.; Woolley, G. A., Fluorescence imaging of

azobenzene photoswitching in vivo. Angewandte Chemie International Edition 2011, 50 (6), 1325-

7.

31. Zhang, F.; Zarrine-Afsar, A.; Al-Abdul-Wahid, M. S.; Prosser, R. S.; Davidson, A. R.;

Woolley, G. A., Structure-based approach to the photocontrol of protein folding. Journal of the


106

32. Zhang, Z.; Burns, D. C.; Kumita, J. R.; Smart, O. S.; Woolley, G. A., A water-soluble

azobenzene cross-linker for photocontrol of peptide conformation. Bioconjugate Chemistry 2003,

14 (4), 824-829.






35. Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.; Bax, A., NMRPipe: A

multidimensional spectral processing system based on UNIX pipes. Journal of Biomolecular NMR

1995, 6 (3), 277-293.

36. Bru, R.; Walde, P., Product inhibition of α‐chymotrypsin in reverse micelles. European

Journal of Biochemistry 1991, 199 (1), 95-103.

37. Smith, E. T.; Perry, E. T.; Sears, M. B.; Johnson, D. A., Expression of recombinant human

mast cell chymase with Asn-linked glycans in glycoengineered pichia pastoris. Protein Expression

and Purification 2014, 102, 69-75.




Biochemistry 2012, 51 (32), 6421-31.

39. McEuen, A. R.; Sharma, B.; Walls, A. F., Regulation of the activity of human chymase

during storage and release from mast cells: the contributions of inorganic cations, pH, heparin and

histamine. Biochimica et Biophysica Acta 1995, 1267 (2-3), 115-21.

40. Yahiro, E.; Miura, S.; Imaizumi, S.; Uehara, Y.; Saku, K., Chymase inhibitors. Current

Pharmaceutical Design 2013, 19 (17), 3065-71.

41. Woolley, G. A., Photocontrolling peptide α helices. Accounts of Chemical Research 2005,

38 (6), 486-93.



Chemical Reviews 2013. 113(8):6114-78.

43. Dong, M.; Babalhavaeji, A.; Collins, C. V.; Jarrah, K.; Sadovski, O.; Dai, Q.; Woolley, G.

A., Near-infrared photoswitching of azobenzenes under physiological conditions. Journal of the


44. Woolley, G. A., Protein machines: an open and shut cage. Nature Nanotechnology 2013, 8

(12), 892-3.

107

4.7 Addendum: Exploring stilbenes cross-linkers as models for pure cis and trans azobenzenes

4.7.1 Introduction

For studying the effect of azobenzene cross-linkers on protein structure and activity, it is possible

with most switches to achieve a state where practically all azo moieties are in their more stable

isomer state (usually trans). This is achieved by dark adaptation at room or elevated temperatures.1

However, at the photostationary state (PSS), with most azo derivatives, a mixture of isomers is

achieved.2 This is the case, for example, with BSBCA and TOM, the two azobenzenes used in this

chapter.3-5 The limitation is due to thermal as well as photochemical back reactions.6 It would be

advantageous to be able to measure the activity of cross-linked proteins when the cross-linker is

entirely trans or entirely cis. This would allow one to estimate the maximum possible change in

the structure and/or activity of the protein that could be achieved with a given photoswitch

geometry (i.e. the change in end-to-end distance upon isomerization) and at a given cross-linking

position. Such knowledge would be particularly useful since, depending on the type of

structural/functional protein assay, relaxation of the less stable isomer (usually cis) can also

complicate the measurement of the protein’s activity when the switch is highly enriched in the less

stable isomer state.

One approach to producing a cross-linked protein system in either pure trans or pure cis form is to

use a molecule that is close in geometry to the azobenzene photoswitch under study but can exist

in either state with a very slow interconversion rate. This addendum describes our attempt to use

stilbene cross-linkers (with similar structure to the BSBCA and TOM azobenzenes, Figure 4-12)

for this purpose.

108

Figure 4-12. Models of stilbene derivatives in trans (left) and cis (right) obtained from DFT

optimizations. The chloroacetamide groups have been replaced with acetamides for simplification.

The distance between the methyl groups’ carbon atoms is 17.0 Å in the trans stilbene cross-linker

(compared to trans TOM: 16.8 Å) and 13.0 Å in the cis stilbene cross-linker (cis TOM: 12.4 Å).

Stilbenes exhibit a high barrier to rotation along the ethenyl moiety, compared to the azo moiety

in azobenzenes, which slows down thermal relaxation to the extent that the cis isomer can be

isolated and stored without noticeable thermal isomerization.7-8 We made trans and cis

dichloroacetamido stilbenes to use them as non-switchable analogues of BSBCA and TOM in

either isomer form. By cross-linking the chymase binding Fynomer with these stilbenes, we sought

to compare the inhibitory activities of the two isomers. We observed, however, that the Fynomers

bearing these stilbene isomers had identical potency, within error, but had poorer potency

compared to azo-cross-linked samples. We postulate that the highly non-polar character of the

stilbene cross-linkers result in the misfolding or partial unfolding of the protein such that any effect

of the length of the cross-linker is masked by a larger overall decrease in its function.

109

4.7.2 Materials and Methods

4.7.2.1 Synthesis of cis and trans stilbene cross-linkers

Scheme 4-2. Synthesis of cis and trans stilbene cross-linkers

The precursor trans-4,4'-diaminostilbene was purchased from Sigma (D25206). Ten milligrams

(48 µmol) was dissolved in 15 mL methanol (MeOH). While working under amber light (590 nm),

1.5 mL of the stock solution was placed in a tube inside a temperature-controlled jacket (Quantum

Northwest) set to 15C and irradiated at 370 nm for 5 min using an LED (897-LZ440U610

LedEngin, CA, USA, operating at 68 mWcm-2). The LED was positioned ~1.5 cm above the

surface of the liquid. MeOH was removed using a rotary evaporator at room-temperature and the

solid was re-dissolved in dichloromethane (DCM) and was purified using thin-layer

chromatography (TLC) with a mobile phase consisting of DCM and 2% MeOH. Rf of cis: 0.55

and trans: 0.80.

The cis layer was marked by a brief illumination with a dim UV light source and scraped and

suspended in MeOH (twice, 10 mL each time). The suspension was filtered, and MeOH was

removed. Purity was checked by 1H-NMR.

110

About 2 mg of cis-4,4'-diaminostilbene was dissolved in 2 mL THF. Three equivalents of

triethylamine was added and the solution was cooled in ice for 15 min, while stirring. Three eq. of

2-chloroacetyl chloride was added drop-wise. The solution was stirred at room temperature for 3

h. A drop of water was added to neutralize any unreacted acid chloride. Extraction was performed

using ethylacetate/water and all solvents were removed. The solid was then placed under high

vacuum overnight. 1H-NMR confirmed the cis cross-linker was highly pure with negligible signal

from the byproduct chloroacetic acid. The same procedure was used to convert trans-4,4'-

diaminostilbene to its chloroacetamido derivative.

4.7.2.2 Determining molar absorption coefficients

Para-nitrophenol (PNP) was used as an internal reference standard. The intensity of 1H-NMR

peaks of PNP and those of the cross-linkers yield the ratio of the concentrations of the two

molecules. The concentration of PNP in its original stock can be accurately quantified from its

absorbance at 400 nm where its molar absorption coefficient is known to be 18465 M-1cm-1 9 in 1

N NaOH. By measuring the UV-Vis spectrum of a now known concentration of stilbene in DMSO,

its molar absorption coefficients can be obtained.

~ 10 mg of para-nitrophenol (PNP) was dissolved in 1.00 mL D6-DMSO (Sigma, 1.03424 EMD

Millipore). A fraction of this stock solution was diluted 100 times in 0.1 M NaOH (aq) and its UV-

Vis spectrum was recorded in a 0.15 cm quartz cuvette at 20.0C. Using the molar absorption

coefficient of 18,465 M-1cm-1 at 400 nm, the stock concentration was determined to be 74.1 mM.

1~2 mg of each cross-linker was dissolved in 300 µL D6-DMSO (Cambridge Isotope Laboratories,

Inc.). Three different volumes of this solution were mixed with known amounts of para-

nitrophenol stock in the same solvent according to Table 4.1.

111

Table 4-1. Sample preparation scheme for quantitative 1H-NMR experiments.

Stilbene stock PNP (74.1 mM) D6-DMSO

i 50 20 130

ii 75 20 105

iii 100 20 80

10 µL of each stilbene stock (cis and trans) was diluted 100 times and its UV-Vis spectrum was

recorded on a Perkin Elmer Lambda 35 instrument at 20.0C. A similar measurement was also

made with the same dilution factor but at a final solvent mixture of 50% (v/v) DMSO in sodium

phosphate buffer (5 mM, pH 8.0).

The solutions were transferred to a 3 mm NMR tube. 1H-NMR was recorded on a 500 MHz

instrument (Oxford Instruments) with the temperature controller set to 25.0C and without

spinning the tube. An array of relaxation delays were recorded and a delay of 7.0 s (d1) was judged

to be sufficient for full relaxation after 90o pulses. Receiver gain was optimized and 64 scans were

recorded. These parameters were used for all samples of both isomers.

Spectra of each sample of each isomer were processed using MNOVA (Mestrelab Research,

Spain) and the chemical shift and intensity data exported as a text file. Integration was calculated

using Igor Pro (Wavemetrics).

Trans: The peak at 7.1 ppm corresponds to the H atoms of the ethane group of stilbenes (x2) and

the multiplet centred at 6.9 ppm corresponds to two hydrogens on PNP phenyl ring, ortho to the

hydroxyl group. The ratio of integrated intensites for these two peaks was used to calculate the

concentration of stilbene in each sample. These two signals result from the same number of protons

112

on each molecule and as such the ratio of integrals is identical to the ratio of the concentrations of

the two molecules.

Analysis of the NMR spectrum of the cis isomer was performed in the same manner except that

the chemical shift of the H atoms on the ethene moiety appear as a singlet at 6.5 ppm.

From the stilbene concentration in each sample, the original stock concentration was calculated

using the dilution factors used for samples i, ii and iii. From the absorbance reading for each isomer

and using the Beer-Lambert law, three molar absorption coefficients were obtained for each isomer

at its lambda max. (Table 4.2)

Table 4-2. Molar absorption coefficients obtained for trans and cis stilbene cross-linkers in DMSO

and in a mixture of DMSO and aqueous buffer

DMSO 50% DMSO, 50% Sodium

Phosphate buffer pH 8

Trans stilbene cross-linker

(M-1cm-1)

51900 ± 400

(at max 348 nm)

37500 ± 300

( at max 341 nm)

Cis stilbene cross-linker

(M-1cm-1)

22400 ± 400

( at max 312 nm)

22400 ± 400

( at max 311 nm)

4.7.2.3 Cross-linking Fynomer with trans and cis stilbenes

Fynomer was expressed and purified in the manner described in section 4.3.1 of this chapter. The

protein was treated with 10 equivalents of tris(2-carboxyethyl)phosphine (TCEP) and incubated at

room temperature for at least 15 minutes. To cross-link the Fynomer with the stilbene derivatives,

2 equivalents of cross-linker were added to the reduced protein sample in a 10 mM sodium

phosphate buffer (pH 8) and DMSO mixture such that the final concentration of DMSO was 70%

(v/v). The reaction mixture was shaken at 40C for 12 hours. Five extra equivalents of TCEP were

added followed by 1 additional equivalent of cross-linker (3 eq in total) and the mixture was

113

allowed to shake at the same temperature for 6 more hours. Purification was carried out using

reverse-phase HPLC with a C18 semi-preparatory column (Zorbax, RxC18) using 0.1% (v/v) TFA

water and 0.1% TFA (v/v) acetonitrile as solvents with a linear gradient of 5% to 70% acetonitrile

over 25 minutes. The eluate was monitored using a dual wavelength Waters detector at 280 nm

(chromophore: protein and stilbene) and 350 nm (chromophore: only stilbene). The cross-linked

Fynomers eluted at ~50% acetonitrile. The eluate was then lyophilized and reconstituted using 50

mM sodium phosphate pH 8.0. The purified product was analyzed with ESI-MS. The expected

and observed masses for both products were 9926.9.

4.7.2.4 Protein activity assay

To test the activity of chymase in the presence of stilbene-cross-linked Fynomer, a similar assay

was performed according to the procedure outlined in chapter 4 for TOM-Fynomer. A substrate

concentration of 376 µM was used. The assay was first performed at stil-Fynomer (trans or cis)

and unmodified Fynomer concentrations of 300 nM. In a separate measurement, trans and cis stil-

Fynomers at 1000 nM were also analyzed.

4.7.2.5 Cross-linking with peptide FK11

Peptide FK11 with the sequence Acetyl-WGEACAREAAAREAACRQ-NH2 was made using

standard Fmoc-based solid phase peptide synthesis using a CEM Liberty automated machine and

was purified using reversed phase HPLC. Two solutions of 3.0 mM peptide were treated each with

10 equivalents of TCEP followed by a 15 minute incubation. Each solution was then mixed with

3 equivalents of trans or cis stilbene cross-linkers and DMSO to the final concentration of 50%

(v/v) and allowed to stir at 45C overnight. Cross-linked peptides were purified using reversed-

phase HPLC on a C-8 semi-preparatory column (Zorbax, 9.4mm) using 0.1% (v/v) TFA water and

0.1% TFA (v/v) acetonitrile as solvents with a linear gradient of 5% to 70% acetonitrile in 25

114

minutes. The eluate was monitored using a dual wavelength Waters detector at 280 nm

(chromophore: tryptophan and other aromatic side chains) and 350 nm (chromophore: stilbene).

For both reaction mixtures, product eluted at ~49% acetonitrile. Cross-linked peptides were then

lyophilized. ESI-MS: expected mass: 2280.5, observed mass: 2280.4.

4.7.2.6 Circular dichroism

Solutions of cis and trans stilbene-FK11 were prepared by dissolving each purified compound in

2.5 mM sodium phosphate buffer (pH 7.0). The concentration of the cis-stilbene-FK11 and trans-

stilbene-FK11 were measured from their UV-Visible spectra to be 200 µM and 300 µM

respectively. Circular dichroism spectra were recorded on an OLIS RSM 1000 CD spectrometer

that was equipped with a Quantum Northwest Peltier temperature controller. 1 mm cuvettes were

used and spectra were recorded at 20.0C. The acquired ellipticity data were converted to MRE

(mean residue ellipticity).

4.7.3 Results and Discussion

The cross-linkers were made by using trans-4,4'-diaminostilbene as starting material. A one step

reaction with chloroacetylchloride yields the chloroacetyl cross-linker in the trans state. Attempts

to photochemically convert the trans cross-linker to its cis isomer failed. Based on UV-Vis

spectrometry and 1H-NMR, UV-irradiation of the trans cross-linker results in significant

constitutional changes. Interestingly, when the compound is dissolved in D2O, the major product

of the photoreaction seems to be one with the deuterated solvent added across the stilbene’s double

bond whereas in DMSO, multiple breakdown products appear. Conversion of the diaminostilbene

precursor to its cis isomer was possible by UV-irradiation in methanol, however, in which the trans

precursor is slightly soluble (~0.5 mg/mL). Irradiation time was optimized by monitoring the

appearance of the product by TLC. It was observed that prolonged irradiation could cause

115

degradation of both the reagent and product. Since all these stilbene derivatives are very poorly

soluble in a variety of solvents, TLC was used for purification. It was also observed that ambient

light could slowly lead to the photochemical relaxation of cis-4,4'-diaminostilbene when this

compound was dissolved in methanol. As a result, all purification was performed under amber

light (590 nm). cis-4,4'-diaminostilbene was then transformed into its cross-linker version via the

same method used for the trans isomer.

Once the two cross-linkers were made, chymase binding Fynomer was cross-linked with both

isomers to obtain two modified proteins: trans stil-Fynomer and cis stil-Fynomer.

To test whether cis stil-Fynomer would show different inhibitory potency compared to its trans

counterpart, the same chymase assay described in chapter 4 could be employed. While in the case

of azobenzene-cross-linked Fynomers, light causes the conversion between the two isomer states,

here we are dealing with two isolated samples of cis and trans stilbene cross-linked-Fynomers.

Hence, the concentration of each sample must be accurately known before their respective

inhibitory activities can be compared. To achieve this, we measured the molar absorption

coefficients of the stilbene cross-linkers via quantitative 1H-NMR and UV-Vis spectroscopy.10 P-

Nitrophenol, which can be quantified colourimetrically, was used as an internal reference in

solutions of each cross-linker in deuterated solvent. From the NMR spectra, the ratios of the areas

under peaks corresponding to the protons of the reference compound, p-nitrophenol, and those that

corresponded to the stilbene allowed for the quantification of the stilbene. From the now quantified

samples of each stilbene, solutions were made in DMSO and in a mixture of DMSO and sodium

phosphate buffer to obtain molar absorption coefficients in either solvent. A sample of cross-linked

Fynomer could be solubilized in the same DMSO/buffer solvent. The effect of the presence of

protein on the molar absorption coefficient is assumed to be negligible.5

116

An inhibition assay was carried out similar to the assay described in chapter 4 for TOM-Fynomer.

Figure 4-13 shows the kinetic traces obtained from this experiment. The effect of trans-stil-

Fynomer and cis-stil-Fynomer on the initial rate of the hydrolysis of the substrate peptide in the

presence of chymase can be compared with the trace obtained from a similar reaction in which an

unmodified Fynomer is used at the same concentration. While both trans and cis stil-Fynomers did

inhibit the activity of chymase, they did not differ in the extent of inhibition beyond statistical

margins of error. The kinetic measurement was also performed at a higher concentration of stil-

Fynomer, with both the cis and trans compounds. Similarly, no meaningful difference was

observed at this concentration. The results also show that the degree of inhibition of chymase is

lower compared to similar concentrations of dark adapted BSBCA-Fynomer (more unfolded, less

potent).

Figure 4-13. Kinetic traces of the chymase assay when uninhibited (dashed black), and in the

presence of trans (blue) and cis (red) stil-Fynomer (300 nM), and unmodified Fynomer (grey) (300

nM).

117

To understand the reason behind this behaviour, we attempted to study the effect of the two isomers

of the stilbene by cross-linking a designed helical peptide, FK11, with each. The peptide, which

contains two cysteine residues at positions i, i+11, has been used as a test model for the analysis

of the effect of azobenzenes such as TOM.5 TOM-cross-linked FK11 has been shown to undergo

unfolding, i.e. a decrease in alpha-helicity, upon trans to cis isomerization of the azo cross-linker.5

Unlike the data obtained for trans TOM-FK11, the CD spectrum of trans stil-FK11 does not appear

with the pattern expected for a typical alpha-helix. Nonetheless, there is a distinct change in the

CD spectrum of cis-stil-FK11 vs its trans variant (Figure 4-14).

Figure 4-14. CD spectra obtained from FK11 peptide when cross-linked with trans stilbene cross-

linker (solid line) and cis stilbene cross-linker (dashed line).

It can be postulated that the stilbenes in either trans or cis form tend to disrupt the secondary

structure of the peptide as a result of their low polarity. The discrepancy between the behaviour of

118

stilbene-Fynomers and their azo counterparts would thus be accounted for by blaming the

hydrophobic nature of the stilbene core which can destabilize the tertiary structure of Fynomer.

Such destabilization could lead the protein to partially unfold and render it impotent regardless of

the geometry of the cross-linker attached to it. This would not be surprising as even with the highly

soluble BSBCA, cross-linked Fynomer was prone to aggregation (See section 4.3.7).

The observations presented here, however, are not sufficient to draw these conclusions with high

confidence. One could argue for instance, that the CD spectra of stilbene-cross-linked FK11 may

have become convoluted due to the interference of the stilbene moiety itself in the CD signal. If

this were true, the structure of the trans-stilbene-FK11 may very well be helical, despite its unsual

CD pattern. More in depth characterization is needed. The structure of stil-Fynomer could be

analyzed using NMR methods, provided that stil-Fynomer is soluble enough to permit such a

study.

4.7.4 Conclusion

We conclude that the stilbene corss-linkers are not suitable mimics for azobenzenes in their pure

cis and pure trans isomer states. For this purpose, the tetra-fluoro derivative of BSBCA could be

considered as it offers near quantitative cis content at the PSS, more than 90% trans content when

photochemically relaxed, and exhibits extremely slow thermal isomerization (1/2 years).11

4.8 References

1. Woolley, G. A.; Beharry, A. A., Azobenzene photoswitches for biomolecules. Chemical

Society Reviews 2011, 40 (8), 4422-4437.

2. Dong, M.; Babalhavaeji, A.; Samanta, S.; Beharry, A. A.; Woolley, G. A., Red-shifting

azobenzene photoswitches for in vivo use. Accounts of Chemical Research 2015, 48 (10), 2662-

70.

119

3. Zhang, Z.; Burns, D. C.; Kumita, J. R.; Smart, O. S.; Woolley, G. A., A water-soluble

azobenzene cross-linker for photocontrol of peptide conformation. Bioconjugate Chemistry 2003,

14 (4), 824-829.







Society Reviews 2011, 40 (8), 4422-37.

7. Likhtenshtein, G. Stilbene photoisomerization. Stilbenes 2009.

8. Momotake, A.; Arai, T., Photochemistry and photophysics of stilbene dendrimers and

related compounds. Journal of Photochemistry and Photobiology C: Photochemistry Reviews

2004, 5 (1), 1-25.

9. Bowers, G. N.; McComb, R. B.; Christensen, R. G.; Schaffer, R., High-purity 4-

nitrophenol: purification, characterization, and specifications for use as a spectrophotometric

reference material. Clinical Chemistry 1980, 26 (6), 724.

10. Saito, T.; Nakaie, S.; Kinoshita, M.; Ihara, T.; Kinugasa, S.; Nomura, A.; Maeda, T.,

Practical guide for accurate quantitative solution state NMR analysis. Metrologia 2004, 41 (3),

213.

11. Bléger, D.; Schwarz, J.; Brouwer, A. M.; Hecht, S., o-Fluoroazobenzenes as readily

synthesized photoswitches offering nearly quantitative two-way isomerization with visible light.


120

Summary and future directions

The molecule of azobenzene photoisomerizes with UV light, which is destructive to biological

systems. However the azobenzene absorbance spectrum is highly sensitive to structural

modifications and this has allowed researchers to develop various azobenzenes that isomerize with

longer wavelengths of light. Azobenzenes can be used effectively in vivo only if the colour of light

that is needed to operate them is in the biological window, i.e. far red to near-IR. An ideal

azobenzene photoswitch can rapidly isomerize when irradiated at a wavelength in the near-IR

range to produce a significant fraction of the less stable state. Once irradiation is stopped, the

molecule must be able to relax quickly enough to enable photocontrol with high spatial precision.1

5.1 Red-shifted azobenzene photoswitches

The present thesis reviewed the steps we took to achieve near-IR photoswitching of a water stable

azobenzene. We first studied the photophysical properties of a tetra-ortho-methoxy substituted

azobenzene (2 in Chapter 2) that exhibited effective switching at 630 nm, and rapid thermal

relaxation. We concluded that the red light absorbing species was not the neutral azobenzene

molecule; it was rather the azonium ion that was abundant at physiological pH due to its unusually

high pKa. With detailed spectroscopic and computational analysis, we learned that a fortunate

combination of high pKa, red-shifted azonium absorbance spectrum, a marked difference in the

pKa of cis vs. trans, and a slow thermal relaxation of the cis isomer due to the tetra-ortho-methoxy

groups, led to effective red-light switching (Chapter 2).

Subsequently, we focused on the azonium species. We varied the substitution pattern on the

azobenzene core to achieve photoswitching at longer wavelengths. DFT and TD-DFT

computational studies facilitated design by allowing us to predict a number of key properties of

121

these structures and choose the promising ones to pursue experimentally. While the predicted

electronic transitions by TD-DFT (in vacuo) calculations in most cases significantly deviated from

the experimental results, the relative red-shift in the azonium species’ spectra was often accurate.

The structural and H-bonding details of the energy-optimized neutral and cis species gave us

qualitative insight into how different substitution patterns could affect the relaxation kinetics of

the photoswitches. For instance, the degree to which the neutral species of a trans ortho-substituted

azobenzene was skewed about the N=N bond would give us a qualitative measure of the energy

barrier for isomerization, hence the thermal relaxation rate.

By introducing ortho-meta-methoxy substituents along with cyclic amino groups at para positions,

we achieved red-shifted azonium ions, including one with an absorbance spectrum that extended

into the near-IR zone (4 in Chapter 3). The pKa of this azobenzene was ~5, and its cis half-life was

~10 µs at neutral pH.

As the utility of the ortho-meta-methoxy azonium switches was limited by their very fast relaxation

rates, we aimed to tune their relaxation kinetics by introducing alkoxy groups at the ortho positions.

We developed an azobenzene derivative that combined the benefits of tetra-ortho methoxy (slow

relaxing cis), meta methoxy (red-shifted azonium), and para amino (high pKa and red-shifted

azonium) substituents. The molecule, nicknamed DOM, photoswitched when irradiated by bright

near-IR LEDs and relaxed on the time scale of ~1 s which is suitable for in vivo applications and

photopharmacology.2

With the DOM molecule, we have achieved several criteria of an ideal azobenzene photoswitch:

water-stability, near-IR switching at physiological pH, and appropriate thermal relaxation rate.

However, DOM’s absorbance change upon photoswitching with near-IR is not dramatic. This

suggests that the percentage of cis at PSS may not be ideal for certain applications. To improve

122

the cis content at PSS, higher absorptivity at the near-IR range is needed. As explained in chapter

3, the DOM azonium species (7 in chapter 3) is less red-shifted compared to the fast-relaxing

ortho-meta analogue (4 in Chapter 3). Computational models (refer to section 3.3.2) showed that

compound 6 (Chapter 3), would have a similar max to 4 (Chapter 3). This compound was the

original goal but could not be synthesized due to poor solubility. It would be interesting to explore

the properties of a less symmetrical derivative of the hexacyclic structure to improve its solubility

while maintaining all four dioxane bridges.

Other functionalities such as amino or alkylamino groups with substitution patterns similar to

DOM or its predecessors would also be interesting to explore. For instance, by substitution of

methoxy groups with dimethylamino groups at two meta positions, the structure shown in Chart

5-1a is predicted to have an even more red-shifted max compared to its ortho-meta-methoxy

version (Chart 5-1b). Based on these preliminary calculations, it would be worthwhile performing

computational analysis on structures such as Chart 5-1c. Such a change in structure may result in

an increase in the azonium absorptivity in the near-IR region while maintaining the pKa and the

relaxation rate.

Chart 5-1. Introduction of amines in the meta positions

A different approach would be to alter the nature of the para amine groups to further redshift the

absorbance of the azonium species. Chart 5-2 shows an example of such modification which is

predicted to take the max of the azonium species all the way to ~800 nm range. Such a dramatic

123

change can be attributed to greater delocalization which would also affect the pKa and relaxation

rates (see Appendix 6.3).

Chart 5-2. Higher conjugation is predicted to further redshift azonium absorbance

5.2 Photocontrol of protein function using azobenzene cross-linkers

Using azobenzenes as protein cross-linkers to obtain photocontrolled proteins was another goal of

this research. In Chapter 4 we described a modular design for photocontrollable protein affinity

reagents based on the Fynomer (Fyn SH3) scaffold.3 We examined a single cross-linking site on a

chymase inhibiting Fynomer with two azobenzene cross-linkers that had similar end-to-end

distances in both of their isomers. What remains to be studied is whether a more suitable location

on the protein exists for cross-linking. The computational methodology that was recently published

by Blacklock et al.4 could be used to select other cross-linking sites to obtain more dramatic

photoswitching of folding and function. Evaluation of other azobenzene structures, including the

red-light switching compound 1 (Chapter 3) should be considered.

The fact that Fynomer’s binding to chymase is mediated through its flexible RT and nSrc loops

may present an inherent limitation on the degree of photocontrol with this system. Partial unfolding

of the protein may not effectively prevent the loop residues from interacting with the target protein.

Therefore, even if a higher degree of unfolding is achieved via changing the attachment sites, the

light-dependent change in function may not improve substantially. Other affinity reagents with

124

more rigid structures (e.g. the 3 helix bundle protein Z scaffold)5 could be a better choice for

azobenzene cross-linking.

Cross-linking the protein of interest with multiple azobenzene photoswitches is another approach

to enhance the photocontrol of activity. This strategy has been successfully used in the case of

light-gated nano-cages where cross-linking ATP-driven group II chaperonin molecules with ~ 8

copies of azobenzene photoswitches resulted in effective opening/closing of the molecular cage.6

Another important area that needs to be studied in more depth is the dependence of the

photophysical properties of azobenzenes when they are used as protein cross-linkers. It is known

that the absorbance spectra and thermal relaxation rates of azobenzenes could change depending

on the local chemical environment.7 If we are to move towards using the new generation of

photoswitchable cross-linkers based on azonium ions, the pKa (and consequently the possibility of

using long wavelengths) could change once they are attached to the target protein. Could the effect

of the local chemical environment on pKa and absorbance spectra be systematically studied? This

should be an important topic for further research. Such information would be extremely useful in

selecting the appropriate photoswitch, would aid in choosing the appropriate position for cross-

linking, and would perhaps help in optimizing a particular system by altering the protein sequence

as well.

5.3 References

1. Dong, M.; Babalhavaeji, A.; Samanta, S.; Beharry, A. A.; Woolley, G. A., Red-Shifting

azobenzene photoswitches for in vivo use. Accounts of Chemical Research 2015, 48 (10), 2662-

70.

2. Ibaraki, M.; Ito, H.; Shimosegawa, E.; Toyoshima, H.; Ishigame, K.; Takahashi, K.; Kanno,

I.; Miura, S., Cerebral vascular mean transit time in healthy humans: a comparative study with

PET and dynamic susceptibility contrast-enhanced MRI. Journal of Cerebral and Blood Flow

Metabolism 2007, 27 (2), 404-13.

125

3. Schlatter, D.; Brack, S.; Banner, D. W.; Batey, S.; Benz, J.; Bertschinger, J.; Huber, W.;

Joseph, C.; Rufer, A.; van der Klooster, A.; Weber, M.; Grabulovski, D.; Hennig, M., Generation,

characterization and structural data of chymase binding proteins based on the human fyn kinase

SH3 domain. MAbs 2012, 4 (4), 497-508.

4. Blacklock, K. M.; Yachnin, B. J.; Woolley, G. A.; Khare, S. D., Computational design of

a photocontrolled cytosine deaminase. Journal of the American Chemical Society 2018, 140 (1),

14-17.

5. Nord, K.; Gunneriusson, E.; Ringdahl, J.; Stahl, S.; Uhlen, M.; Nygren, P. A., Binding

proteins selected from combinatorial libraries of an alpha-helical bacterial receptor domain. Nature

Biotechnology 1997, 15 (8), 772-7.

6. Hoersch, D.; Roh, S.-H.; Chiu, W.; Kortemme, T., Reprogramming an ATP-driven protein

machine into a light-gated nanocage. Nature Nanotechnology 2013, 8 (12), 928-932.




126

Appendices

6.1 Relaxation kinetics for compounds discussed in Chapter 3

The following represent thermal relaxation data for compound 2 in Chapter 3. Each panel shows

data obtained at the pH indicated below the panel. Monoexponential fits are shown together with

calculated residuals. The time constant obtained (k) is listed below each panel. Selected traces

are shown plotted together in the last panel with a logarithmic time axis.

127

The following represent thermal relaxation data for compound 3 (chapter 3). Each panel shows

data obtained at the pH indicated below the panel. Monoexponential fits are shown together with

calculated residuals. The time constant obtained (k) is listed below each panel. Selected traces

are shown plotted together in the final panel with a logarithmic time axis.

129

The following represent thermal relaxation data for compound 4. Each panel shows data obtained

at the pH indicated below the panel. Monoexponential fits are shown together with calculated

residuals. The time constant obtained (k) is listed below each panel. Selected traces are shown

plotted together in the final panel with a logarithmic time axis.

130

.

133

6.2 Mass spectra for cross-linked Fynomers

The following are mass spectra (ESI-QTOF and MADLI-TOF) for crosslinked protein species

discussed in chapter 4.

ESI-MS spectrum of HPLC purified BSBCA-Fynomer

134

ESI-MS spectrum of HPLC purified TOM-Fynomer ESI-MS spectrum of HPLC purified BSBCA-15N-Fynomer

MALDI-MS spectrum of HPLC purified TOM-

15N-Fynomer

135

6.3 Select azonium structures and their calculated max values

The following chart represents a number of azonium structures for which DFT optimization and

TD-DFT calculations (B3LYP/6-31+G(d,p)) have been performed. In cases where different

rotamers could result in different H-bonding patterns, the structure with the lowest free energy

has been chosen. The azonium form of DOM (compound 7 in Chapter 3) is given at the

beginning of the table for reference.

Number Structure Predicted

max

(Azonium)

1

597

2

585

3

603

4

595

136

5

585

6

595

7

588

8

553

9

573

137

10

670

11

618

12

611

13

608

14

545

138

15

494

16

555

17

555

18

580

19

556

139

20

612

21

650

22

592

23

606

24

665

140

25

805

26

894

Red-shifted Azobenzenes and Photocontrol of Protein ... · This thesis elucidates the mechanism by...

Documents

Transcript of Red-shifted Azobenzenes and Photocontrol of Protein ... · This thesis elucidates the mechanism by...