1

Engineering Cell-Permeable Proteins through Insertion of Cell-Penetrating Motifs into Surface Loops

Kuangyu Chen, and Dehua Pei*

Department of Chemistry and Biochemistry, The Ohio State University, 484 West 12th Avenue, Columbus, OH 43210, USA

KEYWORDS Cell-penetrating peptide, enzyme replacement therapy, intracellular biologics, protein delivery, protein engineering

ABSTRACT: Effective delivery of proteins into the cytosol and nucleus of mammalian cells would open the door to a wide range of applications including treatment of many currently intractable diseases. However, despite great efforts from nu-merous investigators and the development of a variety of innovative methods, effective protein delivery in a clinical setting is yet to be accomplished. Herein we report a potentially general approach to engineering cell-permeable proteins by ge-netically grafting a short cell-penetrating peptide to an exposed loop region of a protein of interest. The grafted peptide is conformationally constrained by the protein structure, sharing the structural features of cyclic cell-penetrating peptides and exhibiting enhanced proteolytic stability and cellular entry efficiency. Insertion of an amphipathic motif, Arg-Arg-Arg-Arg-Trp-Trp-Trp, into the loop regions of enhanced green fluorescent protein (EGFP), protein-tyrosine phosphatase 1B (PTP1B), and purine nucleoside phosphorylase (PNP) rendered all three proteins cell-permeable and biologically active in cellular assays. When added into growth medium, the engineered PTP1B dose-dependently reduced the phosphotyrosine levels of intracellular proteins, while the modified PNP protected PNP-deficient mouse T lymphocytes (NSU-1) against toxic levels of deoxyguanosine, providing a potential enzyme replacement therapy for a rare genetic disease.

INTRODUCTION

Cell-permeable proteins would provide powerful research tools as well as novel therapeutics against many of the cur-rently undruggable targets, such as intracellular protein-pro-tein interactions and missing/defective proteins caused by ge-netic mutations. Unfortunately, because of their large sizes and hydrophilic nature, proteins are generally impermeable to the cell membrane. Interestingly, a small number of naturally occurring proteins including the trans-activator of transcrip-tion (Tat) protein of HIV,1,2 the homeobox protein Anten-napedia of Drosophila,3 bacterial toxins,4 and human protein

-synuclein5 have demonstrated the capacity of entering the cytosol of mammalian cells. It was later found that short pep-tides derived from some of these proteins, e.g., Tat47-57 from the HIV Tat protein6 and penetratin from Antennapedia7, are sufficient to mediate the internalization of these proteins. Since these initial discoveries in the 1990’s, hundreds of other short peptides (typically 5-30 residues) of both natural and synthetic origins have been found to have similar cellular en-try capabilities.8-10 These peptides are collectively termed as “cell-penetrating peptides (CPPs)”. The CPPs have been ap-plied to deliver a wide variety of cargoes including small mol-ecule drugs, peptides, proteins, nucleic acids, and nanoparti-cles into mammalian cells.8-10 For example, several CPPs have been genetically fused to the N- or C-terminus of an otherwise impermeable protein to render the latter cell-permeable.11-14 An attractive feature of CPP-mediated protein delivery is that

CPPs are relatively tolerant to the physicochemical properties of the cargo, which are highly diverse among different pro-teins. However, linear CPPs generally have low cytosolic de-livery efficiencies (<5%).15,16 It is now generally accepted that CPP-protein conjugates enter mammalian cells predomi-nantly by endocytic mechanisms and become entrapped in-side the endosome and lysosome, with only a very small frac-tion reaching the cytosol, which is the intended destination for most applications. Another major limitation of CPP-mediated protein delivery is the poor proteolytic stability of the CPP-protein conjugates. CPPs genetically fused to the N- or C-terminus of a cargo protein consist of proteinogenic amino acids and are generally unstructured. Consequently, the fusion protein is highly susceptible to proteolytic degrada-tion in vivo (e.g., during circulation), resulting in rapid loss of the CPP sequence. In fact, some of the fusion proteins are so unstable that they cannot be isolated from bacterial expres-sion systems in their intact forms.14

Several other approaches are also being pursued to deliver recombinant proteins into mammalian cells, including physi-cal methods,17 fusion with bacterial toxins,18 surface modifica-tion,19-21 complexation with cationic peptides and polymers,22-

25 and encapsulation into nanoparticles and liposomes.26,27 Each of these approaches has its advantages but also faces unique challenges. Protein delivery by physical methods is generally incompatible with clinical applications, except for

(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2020. ; https://doi.org/10.1101/2020.01.03.894543doi: bioRxiv preprint

2

topical administration. Fusion of protein cargoes with a bac-terial toxin allows for endocytic uptake specifically into target cells/tissues and efficient endosomal escape, but is highly im-munogenic and has limited capacity (in terms of the number of cargo molecules delivered into each cell). Supercharging proteins by replacing their non-essential surface residues with positively charged amino acids results in highly efficient en-docytic uptake, but the endocytosed proteins remain en-trapped inside the endosomal/lysosomal compartments, with little cytosolic exposure.16 Complexation with cationic poly-mers is possible for negatively charged proteins but less effec-tive for neutral and positively charged ones. Finally, protein delivery by encapsulation into nanoparticles or liposomes may prove challenging for tissues outside the liver, spleen, and kid-ney.28 Clearly, there remains an urgent need for alternative ap-proaches that more effectively deliver proteins into the mam-malian cytosol.

Others29-31 and we32,33 recently discovered that cyclization of CPPs enhances their cellular entry efficiencies. An added ben-efit of peptide cyclization is the greatly improved proteolytic stability. In particular, we reported a family of small amphi-pathic cyclic peptides as exceptionally active CPPs.34 The cy-clic CPPs have been applied to efficiently deliver a variety of drug modalities, including proteins33,35,36 and nucleic acids,37,38 into the mammalian cytosol in vitro and in vivo. However, a drawback of the cyclic CPPs, which contain non-proteino-genic amino acids, is that they must be conjugated to a protein cargo post-translationally. Since site-specific modification of proteins remains a significant challenge,39 especially on an in-dustrial scale, it is highly desirable to integrate the cyclic CPP sequence into the structure of a cargo protein, so that the CPP-protein conjugate can be produced recombinantly. In this study, we show that CPP motifs can indeed be genetically inserted into the loop regions of a cargo protein to mimic the conformations of cyclic CPPs, resulting in proteolytically sta-ble proteins and their effective delivery into the cytosol of mammalian cells.

RESULTS AND DISCUSSION

General Design Considerations. We chose to fuse CPP motifs to the loop regions (instead of the N- or C-terminus) of cargo proteins for several reasons. First, insertion of a short CPP peptide into a surface loop or replacement of the original loop sequence with a CPP should constrain the CPP sequence into a “cyclic” conformation, greatly enhancing its proteolytic stability. Second, the “cyclic” conformation of a loop-embed-ded CPP may mimic that of a cyclic CPP and enhance its cel-lular entry efficiency. Third, previous studies have shown that insertion of proper peptide sequences into surface loops of a protein often causes only minor destabilization of the protein structure.40

Another important consideration is the CPP sequence. CPPs are thought to escape the endosome by binding to the intraluminal membrane and inducing CPP-enriched lipid do-mains to bud off the endosomal membrane as tiny vesicles, which then disintegrate into amorphous lipid/CPP aggregates inside the cytoplasm.34 Amphipathic CPPs likely facilitate en-dosomal escape by stabilizing the budding neck structure, which features simultaneous positive and negative membrane curvatures in orthogonal directions (or negative Gaussian cur-vature).41 It is hypothesized that the hydrophobic group(s) may insert into the lipid bilayer to generate positive curvature,

while the arginine residues bind to and bring phospholipid head groups together, inducing negative membrane curva-ture. In addition, the most active cyclic CPPs [e.g., cyclo(Phe-phe-Nal-Arg-arg-Arg-arg-Gln),34 where phe is D-phenylala-nine, Nal is L-naphthylalanine (Nal), and arg is D-arginine] contain D- as well as L-amino acids at roughly alternating po-sitions. Recombinant production of proteins necessitates the use of genetically encoded amino acids. Although some of the non-proteinogenic amino acids in cyclic CPPs (e.g., Nal) may be incorporated into proteins by expanding the genetic code,42 we decided to use only the 20 proteinogenic amino acids as building blocks in this study for technical simplicity. We chemically synthesized a small panel of amphipathic peptides containing 3 or 4 arginine and 2 or 3 phenylalanine and/or tryptophan residues and cyclized them via a disulfide bond between two cysteine residues flanking the CPP sequence. The peptides were labeled with a pH-sensitive dye, naphthofluo-rescein (NF), and HeLa cells treated with the peptides were analyzed by flow cytometry. Because NF is nonfluorescent in-side the acidic environments of the endosome and lysosome, the cellular fluorescence as determined by flow cytometry re-flects the cytosolic entry efficiency of the peptide.43 Among the peptides tested, cyclo(CWWWRRRC) showed the highest cytosolic entry efficiency and was selected for protein delivery experiments.

Intracellular Delivery of EGFP by Inserting CPPs into a Surface Loop. We first tested the loop insertion strategy on enhanced green fluorescent protein (EGFP), whose intrinsic fluorescence facilitates the identification of properly folded mutants as well as the assessment of cellular entry efficiency. Loop 9 of EGFP (amino acids 171-176) was pre-viously shown to be highly tolerant to peptide insertion.44 We therefore inserted the CPP motif WWWRRR between Asp173 and Gly174 of EGFP in both orientations (Figure 1a). For the RRRWWW insertion, we deleted the two acidic residues in the loop, Glu172 and Asp173, which may otherwise partially neu-tralize the positive charges of the CPP and reduce its cell-pen-etrating activity. Fortuitously, in addition to the desired con-structs, insertion mutagenesis also generated a construct con-taining an extra arginine residue, RRRRWWW, likely as the result of frame shift mutation during homologous recombina-tion of the PCR products in bacterial cells. The EGFP insertion mutants generated in this study and their properties are sum-marized in Table 1. Both wild-type and mutant forms of EGFP were expressed in E. coli and purified to near homogeneity in high yields. Although the mutant proteins showed slightly re-duced fluorescence intensity (10-50%) relative to wild type EGFP, their excitation and emission maxima remained essen-tially unchanged (Figure S1).

To determine the cellular entry efficiency of EGFP and the insertion mutants, HeLa cells were treated with 5 µM protein for 2 h in the presence of 10% fetal bovine serum (FBS), washed, and analyzed by flow cytometry. While EGFPW3R3 showed no significant improvement in cellular uptake com-pared to WT EGFP, EGFPR3W3 and EGFPR4W3 entered the cells with 19- and 26-fold higher efficiency than EGFPWT (Table 1). To confirm the flow cytometry results, we also treated HeLa cells with the EGFP variants (5 µM) for 2 h and imaged the cells by live-cell confocal microscopy. The strongest fluores-cence was observed in cells treated with EGFPR4W3, followed by EGFPR3W3 and EGFPW3R3, whereas cells treated with WT EGFP showed no detectable intracellular fluorescence (Figure

(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2020. ; https://doi.org/10.1101/2020.01.03.894543doi: bioRxiv preprint

3

1b). The punctate fluorescence patterns observed for the three mutant proteins suggest that they were taken up by the cell through endocytosis and a substantial fraction of the internal-ized proteins remained entrapped inside the endosomes and lysosomes. Unfortunately, neither flow cytometry nor confo-cal microscopy revealed whether any of the internalized pro-teins reached the cytosol. The poorer cellular uptake of EGFPW3R3 than EGFPR3W3 is likely caused by the presence of two negatively charged residues adjacent to the CPP motif in the former (Table 1). It is also possible that the WWWRRR motif (when inserted into loop 9 of EGFP) binds the plasma membrane less effectively than RRRWWW.

Generation of Cell-Permeable Protein Tyrosine Phosphatase 1B. To demonstrate cytosolic entry as well as the generality of our approach, we next chose to deliver the catalytic domain (amino acids 1-321) of protein-tyrosine phos-phatase 1B (PTP1B) into mammalian cells. Tyrosine phosphor-ylation is generally restricted to cytosolic and nuclear proteins or the cytosolic domains of transmembrane proteins. Any per-turbation of the phosphotyrosine (pY) levels of cellular pro-teins would therefore provide definitive evidence for func-tional delivery of PTP1B into the cytosolic space. Moreover, any change in the pY level can be conveniently detected by immunoblotting with an anti-pY antibody. Inspection of the PTP1B(1-321) structure45 revealed five solvent exposed loop re-gions as potential sites for CPP grafting (Table 2). These loops are distal from the catalytic or allosteric site of PTP1B. Se-quence alignment with other members of the PTP family re-vealed a high degree of sequence variation in these loop re-gions,46 suggesting that modification of these loops is likely to be tolerated with regard to the folding and catalytic function of PTP1B. For each loop, the CPP sequence was inserted in both orientations, WWWRRRR and RRRRWWW, resulting in a total of 10 loop insertion mutants (Table 2). The mutant pro-teins were designated as “1-5W” and “1-5R”, based on the site of insertion (i.e., “1-5” for loops 1-5, respectively) and the CPP orientation (“W” for WWWRRRR and “R” for RRRRWWW). To ensure an overall positive charge at the modified loops, some of the acidic residues in the original loop regions were deleted. In some cases, glycine residues were added to both sides of the CPP sequence to maintain a minimal level of loop flexibility. The 3D structures of the 10 PTP1B mutants were predicted by using the online protein fold recognition server Phyre 2.47 All 10 mutants were predicted to have wild-type pro-tein fold with the CPP sequences displayed at the protein sur-face (Figure S2). For loop 1, 3, and 5 insertion mutants, the CPP motifs adopted “cyclic-like” topology with the side chains fac-ing the solvent, whereas in Loop 2 and 4 mutants, the CPPs showed a less constrained structure.

The PTP1B mutants were generated by the one-step poly-merase chain reaction method.48 To quickly assess their solu-bility and catalytic activity, each of the mutants was expressed in 5 mL of Escherichia coli cell culture and the crude cell ly-sates were analyzed by SDS-PAGE. When expressed at re-duced temperature, all 10 insertion mutants produced pre-dominantly soluble proteins indicating that insertion of CPP into the loops did not disrupt the global folding of PTP1B (Fig-ure S3). The phosphatase activities in the cell lysates were quantitated by using p-nitrophenyl phosphate (pNPP; 0.5 mM) as the substrate. Four out of the 10 mutants (PTP1B1W, PTP1B1R, PTP1B2R, and PTP1B4R) showed catalytic activities that are 25-60% of wild type PTP1B, while the rest were much less active (Figure 2). Note that the total PTP activity in a cell ly-sate is governed by both the expression level and the specific activity of a given mutant.

The four most active PTP1B mutants (PTP1B1W, PTP1B1R, PTP1B2R, and PTP1B4R) were expressed in E. coli cells on a large scale and purified to near homogeneity by metal affinity chro-matography (all mutants contained an C-terminal histidine tag). The four mutants showed different yields of soluble pro-tein, likely caused by different folding efficiency and proteo-lytic stabilities. The specific activities of the mutants were de-termined with the purified proteins and compared to that of wild type PTP1B. With the exception of PTP1B1R, which had

Table 1. Structures and Properties of EGFP Variants a

Name Loop 9 Sequencea Protein

Fluorescence Intensity (%)

Cellular Uptake

Efficiency (%)

EGFP IEDGSV 100 100 ± 63

EGFPW3R3 IEDWWWRRRGS

V 87 38 ± 57

EGFPR3W3 IRRRWWWGSV 43 1910 ± 90 EGFPR4W3 IRRRRWWWGSV 52 2560 ± 330

aInserted CPP sequences are shown in boldfaced letters. Cel-lular uptake efficiency values are relative to that of WT EGFP (100%) and have been corrected for the lower quantum yields of the mutants.

Figure 1. (a) Structures of WT and mutant EGFP showing the location of loop 9 and the inserted CPP motif. (b) Live-cell confocal microscopy images of HeLa cells after treat-

ment with WT or mutant EGFP (5 M) for 2 h in the pres-ence of 1% FBS.

Loop 9

EGFP

R

R

RR

WW

W

EGFPR4W3

EGFPR4W3EGFPR3W3

EGFPW3R3EGFPWT

a

b

20 µm

(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2020. ; https://doi.org/10.1101/2020.01.03.894543doi: bioRxiv preprint

4

both low yield and poor specific activity, the other three mu-tants were isolated in good yields and showed similar or higher catalytic activities to/than wild type PTP1B (Table 2).

To assess the cell permeability of the PTP1B mutants, NIH 3T3 fibroblast cells were treated with wild-type or mutant PTP1B for 2 h and lysed. The crude cell lysates were separated by SDS-PAGE and the gel was immunoblotted with anti-pY antibody 4G10. While the untreated cells and cells treated with wild-type PTP1B showed very similar pY protein patterns, the cells after treatment with PTP1B2R and PTP1B4R exhibited greatly reduced global pY levels (Figure 3A). Reduced pY levels were also apparent for cells treated with PTP1B1W and PTP1B1R, although to a less extent compared to those of PTP1B2R and PTP1B4R. Further, 3T3 cells treated with different concentra-tions of PTP1B2R exhibited dose-dependent reduction of the pY level for most proteins, with some of the proteins (e.g., those at ~30, 37, 41, 47, and 55 kD) being more sensitive to PTP1B than others (Figure 3B). Similar effect on pY levels was also observed in HeLa cervical cancer and A549 lung cancer cells (Figure S4). These data indicate that the PTP1B mutants effectively entered the cytosol of mammalian cells and were biologically active in dephosphorylating tyrosine residues of

Table 2. Structures and Properties of WT and Mutant PTP1B

Protein Insertion Site Original Sequencea Sequence after CPP Grafting a Isolated Yield (mg/L of culture)

Specific Activity (%)b

PTP1B 10.4 100

PTP1B1W Loop 1 60-HQEDND-65 60-HQWWWRRRRND-70 4.9 310 ± 23

PTP1B1R 60-HQRRRRWWWND-70 0.28 8.4 ± 0.4

PTP1B2W Loop 2

128-KEEKE-132 128-KWWWRRRRKE-137 ND ND

PTP1B2R 128-KRRRRWWWKE-137 3.2 135 ± 10

PTP1B3W Loop 3 163-LTTQE-167 163-LTGWWWRRRRGTQE-176 ND ND

PTP1B3R 163-LTGRRRRWWWGTQE-176 ND ND

PTP1B4W Loop 4

206-PEHGP-210 206-PWWWRRRRHGP-216 ND ND

PTP1B4R 206-PRRRRWWWHGP-216 4.5 218 ± 19

PTP1B5W Loop 5 75-EEAQ-78 75-GWWWRRRRAQ-84 ND ND

PTP1B5R 75-GRRRRWWWAQ-84 ND ND

aAcidic residues deleted during CPP insertion are underlined. Inserted CPP sequences are shown in boldfaced letters. bAll activities were tested with pNPP as the substrate and are relative to that of WT PTP1B (100%). ND, not determined.

Figure 2. Phosphatase activity in the crude lysates of E. coli BL21(DE3) cells expressing WT or mutant PTP1B. Data shown represent the mean and SEM of three independent experiments and are normalized to that of WT PTP1B (100%). BL21, untransformed cells.

BL

21

WT

1R

1W 2

R2W 3

R3W 4

R4W 5

R5W

0

2 5

5 0

7 5

1 0 0

1 2 5

1 5 0

Ph

os

ph

ata

se

Ac

tiv

ity

%

Figure 3. Effect of WT and mutant PTP1B on the global pY levels in NIH 3T3 cells. (a) Anti-pY Western blot analysis of NIH 3T3 cells after treatment for 2 h with wild-type or

mutant PTP1B (2.1 M for PTP1B1R due to limited availabil-ity and 3.0 µM for all other proteins) in the presence of 1% serum. (b) Dose-dependent reduction of global pY levels

as a function of PTP1B2R concentration (0.5-5 M). The membrane was re-blotted with anti-GAPDH antibody to ensure equal sample loading. M, molecular weight mark-ers; C, control cells without PTP1B treatment.

(kD)

0.5 2 55C (µM)

2RWT

250

100

75

150

37

25

50

M

anti-pY

C WT 1W1R 2R 4RM

GAPDH

(kD)

100

75

150

37

25

50

250

anti-pY

a

b

GAPDH

(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2020. ; https://doi.org/10.1101/2020.01.03.894543doi: bioRxiv preprint

5

intracellular proteins. These PTP1B variants may provide a useful tool for cell signaling research by globally reducing the pY levels of intracellular proteins.

Intracellular Delivery of Purine Nucleoside Phos-phorylase as Potential Enzyme Replacement Ther-apy. Purine nucleoside phosphorylase (PNP) is an enzyme in-volved in purine metabolism, by converting inosine into hy-poxanthine and guanosine into guanine, plus ribose phos-phate.49 Mutations that result in PNP deficiency cause defec-tive T-cell (cell-mediated) immunity but can also affect B-cell immunity and antibody responses.50 A potential treatment of this rare genetic disease is to deliver enzymatically active PNP into the cytosol of patient cells. Examination of the homotri-meric structure of PNP51 revealed three solvent exposed loops that are also distal from the active site, namely His20-Pro25, Asn74-Gly75, and Gly182-Leu187. We inserted the CPP motif RRRRWWW into each of these loop regions to produce three PNP variants (Table 3). For the third insertion mutant (182-187), an acidic residue (Glu183) was removed to maximize overall positive charges at the loop sequence. Pilot expression experiments under different induction conditions revealed that CPP insertion at site 1 or 2 resulted in insoluble proteins, whereas insertion at site 3 produced a partially soluble pro-tein, PNP3R, which was purified to near homogeneity following the same procedure as for wild-type PNP. PNP3R has similar catalytic activity to the wild-type enzyme (Table 3).

Cellular entry of PNP3R was first examined by treating HeLa cells for 5 h with 5 µM fluorescein-labeled PNP3R or wild-type

PNP (PNPWT) and imaging the cells by live-cell confocal mi-croscopy. Cells treated with PNP3R showed readily visible green fluorescence signals inside the cells, whereas cells treated with PNPWT showed no detectable fluorescence under the same experimental condition (Figure 4a). Note that the proteins were intentionally labeled at a low stoichiometry (0.1-0.2 dye/protein) to minimize any protein precipitation or de-naturation. To further assess the cellular entry efficiency of PNP3R, PNP-deficient mouse T lymphocytes (NSU-1) were treated with 1 µM PNPWT or PNP3R for 2 h and washed exhaust-ively to remove extracellular proteins. The cells were lysed and the PNP activities in cytosolic fractions were quantified by us-ing a commercial PNP enzymatic assay kit. While the un-treated NSU-1 cells had no significant PNP activity, treatment of NSU-1 cells with PNP3R resulted in 1.35-fold higher PNP ac-tivity than that of normal S49 cells (100%; Figure 4b). Under the same condition, NSU-1 cells treated with PNPWT showed an activity that was 16% relative to that of S49 cells. The latter activity is likely due to incomplete removal of the extracellular PNP activity by the washing procedure, as NSU-1 cells are non-adherent cells and it was difficult to remove the extracellular fluids completely during washing.

Finally, we tested the capacity of PNP3R to correct the met-abolic defects of NSU-1 cells caused by PNP deficiency. PNP-deficient cells (e.g., NSU-1) are sensitive to deoxyguanosine (dG) toxicity. As shown in Figure 4c, NSU-1 cells failed to grow

in the presence of 25 M dG, while in the absence of dG the cell density increased from 1 x 105 to 2.3 x 106 cells/mL in 72 h.

Figure 4. Cellular entry and biological activity of PNP3R. (a) Live-cell confocal images of HeLa cells after treatment with 5 µM fluorescein-labeled PNPWT (top) or PNP3R (bottom) for 5 h in the presence of 1% FBS. Left panels, FITC fluorescence; right panels, overlap of FITC signals with the DIC images of the same cells. (b) PNP activities in cell lysates derived from S49 (wild-type PNP) or NSU-1 cells with and without treatment with PNPWT or PNP3R (1 µM). Representative data (mean ± SD) from three independent experiments are shown. (c) Protection of NSU-1 cells against dG toxicity. NSU-1 cells were treated with PBS (no protein), 3 µM PNPWT, or 3 µM PNP3R for 6 h at 37 ˚C, washed exhaustively, and incubated with trypsin-EDTA for 3 min.

The cells were seeded at a density of 1 x 105 cells/mL in DMEM containing 25 M dG and cell growth (cell counts) was moni-tored for 72 h. Cells without protein or dG treatment were used as positive control.

b ca

PNP3R

PNPWT

20 µm0 2 4 4 8 7 2

0

5 0

1 0 0

1 5 0

2 0 0

2 5 0

G r o w t h T i m e ( h )

Ce

ll D

en

sit

y (

10

4/m

L)

U n t r e a t e d + P B S

U n t r e a t e d + d G

P N PW T

+ d G

P N P3 R

+ d G

N o n e N o n e P N PW T

P N P3 R

0

5 0

1 0 0

1 5 0

2 0 0

PN

P A

ctiv

ity

%

S 4 9 N S U - 1

Table 3. Structures and Properties of PNP Insertion Mutants

Protein Insertion Site

Original Sequencea Sequence after CPP Insertiona Soluble Protein? Enzyme Activity (pmol/µg/min)

WT - - - Soluble 465 ± 4

PNP1R 1 20-HTKHRP-25 20-HTKRRRRWWWHRP-32 Insoluble -

PNP2R 2 74-NG-75 74-NRRRRWWWG-82 Insoluble -

PNP3R 3 182-GEQREL-187 182-GRRRRWWWQREL-193 Soluble 441 ± 9

aAcidic residues deleted during CPP insertion are underlined. Inserted CPP sequences are shown in boldfaced letters.

(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2020. ; https://doi.org/10.1101/2020.01.03.894543doi: bioRxiv preprint

6

When NSU-1 cells were pretreated with 3 M PNP3R for 6 h, washed exhaustively to remove any extracellular PNP3R, and

then challenged with 25 M dG, they exhibited a growth curve similar to that of the untreated cells (no dG, no protein). Un-der the same conditions, NSU-1 cells treated with PNPWT showed only a small amount of growth (13%) relative to the untreated control, likely due to incomplete removal of PNPWT from the growth medium. Thus, PNP3R, but not PNPWT, can effectively rescue PNP-deficient cells against dG toxicity. PNP3R may be further developed into a novel, intracellular en-zyme replacement therapy. All previous enzyme replacement therapies involved extracellular or lysosomal enzymes.52

Serum Stability of Loop Insertion Mutants. Insertion of amphipathic CPP sequences (e.g., RRRRWWW) into sur-face loops may decrease the thermodynamic stability of a pro-tein as well as generates potential new cleavage sites for pro-teases (e.g., trypsin and chymotrypsin). Both factors can po-tentially reduce the metabolic stability of the mutant proteins. We thus tested the proteolytic stabilities of wild-type EGFP, PTP1B, and PNP as well as the biologically active mutants de-rived from this study by incubating them in human serum for varying periods of time (0-16 h) and quantitating the amounts of remaining intact protein by SDS-PAGE analysis (Figure S5). The wild-type proteins were all highly stable in serum, exhib-iting t1/2 values of >16 h (Figure 5). Among the seven mutant proteins tested, six showed comparable or slightly reduced stability relative to their wild-type counterparts; only one mu-tant, PTP1B1W, showed substantially more rapid degradation than wild-type PTP1B. We also examined the serum stability of PNPWT and PNP3R by monitoring their enzymatic activities as a function of the incubation time and observed no signifi-cant loss of PNP activity for either protein after 16 h of incu-bation (Figure S6). Since linear CPPs (e.g., Phe-Nal-Arg-Arg-Arg-Arg) are highly susceptible to proteolytic degradation (se-rum t1/2 ≤30 min),33,53 these data demonstrate that insertion of short CPP sequences into protein loops greatly increases their proteolytic stability, likely as a result of restrained confor-mations.

Comparison to Other Protein Delivery Methods and Mechanistic Implications. The loop-insertion approach of-fers a number of advantages over other reported protein engi-neering/delivery methods, not the least of which is its simplic-ity. In our case, once a recombinant protein is purified (e.g., from an E. coli cell lysate), it may be used directly as a biolog-ical probe or therapeutic agent. Additionally, whereas post-translational conjugation of a protein with a CPP (or other chemical entities) often results in a mixture of different spe-cies, the loop-insertion method produces a single species of well-defined structure. Compared to other protein surface re-modeling methods (e.g., supercharging19,20 and esterifica-tion21), our method involves relatively minor changes to the protein structure and should be applicable to a broader range of proteins. The resulting mutant proteins are more likely to retain the original protein fold/activity and less likely to elicit immune responses. Finally, the CPP motifs grafted to protein loops are structurally constrained and relatively stable against proteolytic degradation, whereas fusion of CPP sequences to the N- or C-terminus of a protein, as practiced by previous in-vestigators, generates recombinant proteins that frequently lack sufficient metabolic stability for clinical applications.

Compared to posttranslational conjugation of proteins with some of the cyclic CPP(s),33,35 the loop-insertion mutants from

this work appear to have lower cytosolic entry efficiencies. When examined by confocal microscopy, the cell-permeable proteins derived from this study produced punctate intracel-lular fluorescence patterns (Figures 1b and 4a), indicating in-complete endosomal escape of these proteins. We envision that their endocytic uptake and/or endosomal escape effi-ciency may be improved by exploring other proteinogenic amino acids (in addition to Arg and Trp) for the CPP motif as well as different insertion sites on the protein surfaces. After all, bacteria and viruses have evolved highly efficient systems to deliver their proteins and nucleic acids into the cytosol of mammalian cells. Non-proteinogenic amino acids (e.g., Nal) may also be leveraged to improve the efficiency of the CPP motifs.42 We have previously shown that substitution of Nal for Phe (or Trp) greatly improves the cytosolic entry efficiency of cyclic CPPs [e.g., cyclo(Phe-Nal-Arg-Arg-Arg-Arg-Gln) and cyclo(Phe-phe-Nal-Arg-arg-Arg-arg-Gln)].32,34

Our study also has important mechanistic implications. As discussed above, proteins derived from diverse organisms (from bacteria to man) have demonstrated the capacity to en-ter the cytosol of mammalian cells, but their mechanism(s) of

Figure 5. Serum stability of wild-type and mutant forms of EGFP (a), PTP1B (b), and PNP (c). Values reported rep-resent the amounts of intact protein remaining at indi-cated time points and are relative to that at time zero (100%).

0 5 10 15 200

20

40

60

80

100

120

Incubation Time (h)

Rem

ain

ing

Pro

tein

(%

)

EGFPWT

EGFPW3R3

EGFPR3W3

EGFPR4W3

0 5 10 15 200

20

40

60

80

100

Incubation Time (h)

Rem

ain

ing

Pro

tein

(%

)

PNPWT

PNP3R

0 5 10 15 200

20

40

60

80

100

120

Incubation Time (h)

Rem

ain

ing

Pro

tein

(%

)

PTP1BWT

PTP1B2R

PTP1B1W

PTP1B4R

a

b

c

(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2020. ; https://doi.org/10.1101/2020.01.03.894543doi: bioRxiv preprint

7

cellular entry has not yet been resolved. In the case of bacterial toxins, it is well established that the proteins are brought into the early endosome by receptor-mediated endocytosis and some of the toxins are capable of directly crossing the endoso-mal membrane into the cytosol.54,55 The mechanism of endo-somal escape, however, remains controversial despite decades of intensive studies. A popular hypothesis states that, as the endosomal pH decreases, the transmembrane domain of the toxin undergoes a conformational change and inserts into the

endosomal membrane to form an -helical or -barrel pore, through which the effector domain translocates from the en-dosome to the cytosol in its denatured state.54,55 Although our current findings do not disprove the pore-formation mecha-nism (for bacterial toxins) per se, they demonstrate that pro-teins are capable of escaping the endosome by a mechanism(s) other than pore formation. None of the proteins examined in this work (EGFP, PTP1B, and PNP) are known to undergo acid denaturation and membrane insertion to form a pore. In fact, PTP1B is very stable at the endosomal pH (4.5-6.5) and cata-lytically most active at pH 5.5.56 It is also highly unlikely that insertion of RRRRWWW into surface loops would endow the above proteins the capability of pore formation. Rather, we believe that the inserted CPP motif functions as a “cyclic CPP” – by binding first to the plasma membrane to facilitate endo-cytic uptake of the proteins and then to the endosomal mem-brane to induce vesicle budding and collapse.41 We further hy-pothesize that, like the cyclic CPPs, bacterial toxins may exit the endosome by inducing vesicle budding and collapse from the endosomal membrane.57

CONCLUSION

In this study, we have demonstrated that proteins may be rendered cell-permeable by inserting a short CPP motif into one of their surface loop regions. For the three proteins inves-tigated in this study (EGFP, PTP1B, and PNP), a total of 16 mu-tants were generated (Table 4). Among these mutants, eight (50%) produced soluble proteins in good yields, seven of which (44%) exhibited wild type-like biochemical activity and six of which (38%) were cell-permeable. Importantly, for each of the three proteins tested in this work, at least one soluble, cell-permeable, and biologically active mutant was obtained. We conclude that insertion of CPP motifs into the surface loops of proteins represents a general approach to engineering cell-permeable proteins.

ASSOCIATED CONTENT

Supporting Information

Experimental details and additional data are available free of charge online at http://pubs.acs.org.

AUTHOR INFORMATION

Corresponding Author

*To whom correspondence should be addressed. Phone: (614) 688-4068; e-mail: pei.3@osu.edu.

Author Contributions

D.P. conceived the project, D.P. and K.C. designed the exper-iments, K.C. carried out the experiments, and D.P. and K.C. wrote the manuscript.

Funding Sources

Financial support from the National Institutes of Health (GM122459 and CA234124) is gratefully acknowledged.

Notes

The authors declare no competing financial interests.

ACKNOWLEDGMENT

We thank Dr. Z. Qian and A. Sahni for their technical assis-tance during confocal microscopy experiments.

ABBREVIATIONS

CPP, cell-penetrating peptide; dG, deoxyguanosine; EGFP, en-hanced green fluorescent protein; PNP, purine nucleoside phosphorylase; PTP, protein tyrosine phosphatase.

REFERENCES

(1) Green, M.; Loewenstein, P. Autonomous functional domains of chemically synthesized human immunodeficiency virus Tat trans-activator protein. Cell 1988, 55, 1179-1188.

(2) Frankel, A.; Pabo, C. Cellular uptake of the Tat protein from human immunodeficiency virus. Cell 1988, 55, 1189-1193.

(3) Joliot, A.; Pernelle, C.; Deagostini-Bazin, H.; Prochiantz, A. Antennapedia homeobox peptide regulates neural morpho-genesis. Proc. Natl. Acad. Sci. U. S. A. 1991, 88, 1864-1868.

(4) Murphy, J. Mechanism of diphtheria toxin catalytic domain delivery to the eukaryotic cell cytosol and the cellular factors that directly participate in the process. Toxins 2011, 3, 294-308.

(5) Masuda-Suzukake, M.; Nonaka, T.; Hosokawa, M.; Oikawa, T.; Arai, T.; Akiyama, H.; Mann, D.; Hasegawa, M. Prion-Like

spreading of pathological -synuclein in brain. Brain 2013, 136, 1128-1138.

(6) Vivès, E.; Brodin, P.; Lebleu, B. A truncated HIV-1 Tat protein basic domain rapidly translocates through the plasma mem-brane and accumulates in the cell nucleus. J. Biol. Chem. 1997, 272, 16010-16017.

(7) Derossi, D.; Joliot, A. H.; Chassaing, G.; Prochiantz, A. The third helix of the Antennapedia homeodomain translocates through biological membranes. J. Biol. Chem. 1994, 269, 10444-10450.

(8) Gupta, B.; Levchenko, T. S.; Torchilin, V. P. Intracellular de-livery of large molecules and small particles by cell-penetrat-ing proteins and peptides. Adv. Drug Deliv. Rev. 2005, 57, 637-51.

Table 4. Summary of Loop Insertion Mutants Generated in This Study

Protein No. of Mutant Constructs Generated

No. of Soluble Proteins No. of Proteins with WT-Like Activity

No. of Cell-Permeable Proteins

EGFP 3 3 3 2

PTP1B 10 4 3 3

PNP 3 1 1 1

Total 16 8 7 6

(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2020. ; https://doi.org/10.1101/2020.01.03.894543doi: bioRxiv preprint

8

(9) Futaki, S. Membrane-permeable arginine-rich peptides and the translocation mechanisms. Adv. Drug Deliv. Rev. 2005, 57, 547-558.

(10) Wadia, J. S.; Dowdy, S. F. Transmembrane delivery of protein and peptide drugs by TAT-mediated transduction in the treatment of cancer. Adv. Drug Deliv. Rev. 2005, 57, 579-596.

(11) Fawell, S.; Seery, J.; Daikh, Y.; Moore, C.; Chen, L.; Pepinsky, B.; Barsoum, J. Tat-mediated delivery of heterologous pro-teins into cells. Proc. Natl. Acad. Sci. 1994, 91, 664-668.

(12) Schwarze, S.; Ho, A.; Vocero-Akbani, A.; Dowdy, S. F. In vivo protein transduction: Delivery of a biologically active protein into the mouse. Science 1999, 285, 1569-1572.

(13) Säälik, P.; Elmquist, A.; Hansen, M.; Padari, K.; Saar, K.; Viht, K.; Langel, Ü.; Pooga, M. Protein cargo delivery properties of cell-penetrating peptides. A comparative study. Bioconju-gate Chem. 2004, 15, 1246– 1253.

(14) Patel, S. G.; Sayers, E. J.; He, L.; Narayan, R.; Williams, T. L.; Mills, E. M.; Allemann, R. K.; Luk, L. Y. P.; Jones, A. T.; Tsai, Y. H. Cell-penetrating peptide sequence and modi-fication dependent uptake and subcellular distribution of green florescent protein in different cell lines. Sci. Rep. 2019, 9, 6298.

(15) LaRochelle, J. R.; Cobb, G. B.; Steinauer, A.; Rhoades, E.; Schepartz, A. Fluorescence correlation spectroscopy reveals highly efficient cytosolic delivery of certain penta-Arg pro-teins and stapled peptides. J. Am. Chem. Soc. 2015, 137, 2536-2541.

(16) Verdurmen, W. P. R.; Mazlami, M.; Plückthun, A. A quanti-tative comparison of cytosolic delivery via different protein uptake systems. Sci. Rep. 2017, 7, 13194.

(17) Stewart, M.; Langer, R.; Jensen, K. Intracellular delivery by membrane disruption: mechanisms, strategies, and con-cepts. Chem. Rev. 2018, 118, 7409-7531.

(18) Rabideau, A.; Pentelute, B. Delivery of non-native cargo into mammalian cells using anthrax lethal toxin. ACS Chem. Biol. 2016, 11, 1490-1501.

(19) Cronican, J. J.; Thompson, D. B.; Beier, K. T.; McNaughton, B. R.; Cepko, C. L.; Liu, D. R. Potent delivery of functional proteins into mammalian cells in vitro and in vivo using a supercharged protein. ACS Chem. Biol. 2010, 5, 747-752.

(20) Fuchs, S. M.; Raines, R. T. Arginine grafting to endow cell-permeability. ACS Chem. Biol. 2007, 2, 167-170.

(21) Mix, K. A.; Lomax, J. E.; Raines, R. T. Cytosolic delivery of proteins by bioreversible esterification. J. Am. Chem. Soc. 2017, 139, 14396–14398.

(22) Lv, J.; Fan, Q.; Wang, H.; Cheng, Y. Polymers for cytosolic protein delivery. Biomaterials 2019, 218, 119358.

(23) Morris, M. C.; Depollier, J.; Mery, J.; Heitz, F.; Divita, G. A peptide carrier for the delivery of biologically active proteins into mammalian cells. Nat Biotechnol. 2001, 19, 1173-1176.

(24) Akishiba, M.; Takeuchi, T.; Kawaguchi, Y.; Sakamoto, K.; Yu, H.-H.; Nakase, I.; Takatani-Nakase, T.; Madani, F.; Gräslund, A.; Futaki, S. Cytosolic antibody delivery by lipid-sensitive endosomolytic peptide. Nat. Chem. 2017, 9, 751–761.

(25) Erazo-Oliveras, A.; Najjar, K.; Dayani, L.; Wang, T.-Y.; John-son, G. A.; Pellois, J.-P. Protein delivery into live cells by in-cubation with an endosomolytic agent. Nat Methods 2014, 11, 861–867.

(26) Lee, Y.; Luther, D. C.; Kretzmann, J. A.; Burden, A.; Jeon, T.; Zhai, S.; Rotello, V. M. Protein delivery into the cell cytosol using non-viral nanocarriers. Theranostics 2019, 9, 3280-3292.

(27) Chatin, B.; Mevel, M.; Devalliere, J.; Dallet, L.; Haudebourg, T.; Peuziat, P.; Colombani, T.; Berchel, M.; Lambert, O.; Edelman, A.; Pitard, B. Liposome-based formulation for in-tracellular delivery of functional proteins. Mol. Ther.-Nucleic Acids 2015, 4, e244.

(28) Shi, B.; Keough, E.; Matter, A.; Leander, K.; Young, S.; Car-lini, E.; Sachs, A. B.; Tao, W.; Abrams, M.; Howell, B.; Sepp-Lorenzino, L. Biodistribution of small interfering RNA at the organ and cellular levels after lipid nanoparticle-mediated delivery. J. Histochem. Cytochem. 2011, 59, 727–740.

(29) Lättig-Tünnemann, G.; Prinz, M.; Hoffmann, D.; Behlke, J.; Palm-Apergi, C.; Morano, I.; Herce, H. D.; Cardoso, M. C. Backbone rigidity and static presentation of guanidinium groups increases cellular uptake of arginine-rich cell-pene-trating peptides. Nat. Commun. 2011, 2, 453–459.

(30) Mandal, D.; Nasrolahi Shirazi, A.; Parang, K. Cell-penetrat-ing homochiral cyclic peptides as nuclear-targeting molecu-lar transporters. Angew. Chem. Int. Ed. 2011, 50, 9633–9637.

(31) Cascales, L.; Henriques, S. T.; Kerr, M. C.; Huang, Y. H.; Sweet, M. J.; Daly, N. L.; Craik, D. J. Identification and characterization of a new family of cell-penetrating peptides: cyclic cell-penetrating peptides. J. Biol. Chem. 2011, 286, 36932–36943.

(32) Qian, Z.; Liu, T.; Liu, Y.-Y.; Briesewitz, R.; Barrios, A. M.; Jhiang, S. M.; Pei, D. Efficient delivery of cyclic peptides into mammalian cells with short sequence motifs. ACS Chem. Biol. 2013, 8, 423–431.

(33) Qian, Z.; LaRochelle, J. R.; Jiang, B.; Lian, W.; Hard, R. L.; Sel-ner, N. G.; Luechapanichkul, R.; Barrios, A. M.; Pei, D. Early endosomal escape of a cyclic cell-penetrating peptide allows effective cytosolic cargo delivery. Biochemistry 2014, 53, 4034–4046.

(34) Qian, Z.; Martyna, A.; Hard, R. L.; Wang, J.; Appiah-Kubi, G.; Coss, C.; Phelps, M. A.; Rossman, J. S.; Pei, D. Discovery and mechanism of highly efficient cyclic cell-penetrating pep-tides. Biochemistry 2016, 55, 2601-2612.

(35) Sethuraman, N.; Ruth, J.; Tartaglia, L.; Pei, D.; Qian, Z. Com-positions and methods for treating mitochondrial neurogas-trointestinal encephalopathy. Patent Application WO/2019/165183 (2019).

(36) Herce, H. D.; Schumacher, D.; Schneider, A. F. L.; Ludwig, A. K.; Mann, F. A.; Fillies, M.; Kasper, M. A.; Reinke, S.; Krause, E.; Leonhardt, H.; Cardoso, M. C.; Hackenberger, C. P. R. Cell-permeable nanobodies for targeted immunolabelling and antigen manipulation in living cells. Nat. Chem. 2017, 9, 762-771.

(37) Soudah, T.; Khawaled, S.; Aqeilan, R. I.; Yavin, E. AntimiR-155 cyclic peptide–PNA conjugate: synthesis, cellular uptake, and biological activity. ACS Omega 2019, 4, 13954-13961.

(38) Cai, B.; Kim, D.; Akhand, S.; Sun, Y.; Cassell, R. J.; Alpsoy, A.; Dykhuizen, E. C.; Van Rijn, R. M.; Wendt, M. K.; Krusemark, C. J. Selection of DNA-encoded libraries to protein targets within and on living cells. J. Am. Chem. Soc. 2019, 141, 17057-17061.

(39) Spicer, C.; Davis, B. Selective chemical protein modifica-tion. Nat. Commun. 2014, 5, 4740.

(40) Scalley-Kim, M.; Minard, P.; Baker, D. Low free energy cost of very long loop insertions in proteins. Protein Sci. 2003, 12, 197-206.

(41) Dougherty, P. G.; Sahni, A.; Pei, D. Understanding cell pene-tration of cyclic peptides. Chem. Rev. 2019, 119, 10241–10287.

(42) Noren, C. J.; Anthony-Cahill, S. J.; Griffith, M. C.; Schultz, P. G. A general method for site-specific incorporation of unnat-ural amino acids into proteins. Science 1989, 244, 182-188.

(43) Qian, Z.; Dougherty, P. G.; Pei, D. Monitoring the cytosolic entry of cell-penetrating peptides using a pH-sensitive fluor-ophore. Chem. Commun. 2015, 51, 2162– 2165.

(44) Pavoor, T. V.; Cho, Y. K.; Shusta, E. V. Development of GFP-based biosensors possessing the binding properties of anti-bodies. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 11895–11900.

(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2020. ; https://doi.org/10.1101/2020.01.03.894543doi: bioRxiv preprint

9

(45) Barford, D.; Flint, A. J.; Tonks, N. K. Crystal structure of hu-man protein tyrosine phosphatase 1B. Science 1994, 263, 1397-404.

(46) Yang, J.; Liang, X.; Niu, T.; Meng, W.; Zhao, Z.; Zhou, G. Crys-tal structure of the catalytic domain of protein-tyrosine phosphatase SHP-1. J. Biol. Chem. 1998, 273, 28199-28207.

(47) Kelley, L.; Mezulis, S.; Yates, C.; Wass, M.; Sternberg, M. The Phyre2 web portal for protein modeling, prediction and anal-ysis. Nat. Protoc. 2015, 10, 845-858.

(48) Qi, D.; Scholthof, K.-B. G. A one-step PCR-based method for rapid and efficient site-directed fragment deletion, insertion, and substitution mutagenesis. J. Virol. Methods 2008, 149, 85-90.

(49) Erion, M. D.; Stoeckler, J. D.; Guida, W. C., Walter, R. L., Ea-lick, S. E. Purine nucleoside phosphorylase. 2. Catalytic mechanism. Biochemistry 1997, 36, 11735–11748.

(50) Markert, M. L. Purine nucleoside phosphorylase deficiency. Immunodefic. Rev. 1991, 3, 45–81.

(51) dos Santos, D. M.; Canduri, F.; Pereira, J. H.; Vinicius Ber-tacine Dias M.; Silva, R. G.; Mendes, M. A.; Palma, M. S.; Basso, L. A.; de Azevedo, W. F. Jr; Santos, D. S. Crystal struc-ture of human purine nucleoside phosphorylase complexed with acyclovir. Biochem. Biophys. Res. Commun. 2003, 308, 553-559.

(52) Concolino, D.; Deodato, F.; Parini, R. Enzyme replacement therapy: efficacy and limitations. Ital. J. Pediatr. 2018, 44, 120.

(53) Qian, Z.; Xu, X.; Amacher, J. F.; Madden, D. R.; Cormet-Bo-yaka, E.; Pei, D. Intracellular delivery of peptidyl ligands by reversible cyclization: Discovery of a PDZ domain inhibitor that Rescues CFTR Activity. Angew. Chem. Int. Ed. 2015, 54, 5874-5878.

(54) Murphy, J. R.; Harrison, R. J. Mechanisms of bacterial protein toxin entry into the target cell cytosol. Drug Discovery To-day: Dis. Mech. 2006, 3, 267–272.

(55) Tilley, S. J.; Saibil, H. R. The mechanism of pore formation by bacterial toxins. Curr. Opin. Struct. Biol. 2006, 16, 230–236.

(56) Peters, G. H.; Branner, S.; Møller, K. B.; Andersen, J. N.; Møller, N. P. H. Enzyme kinetic characterization of protein tyrosine phosphatases. Biochimie 2003, 85, 527-534.

(57) Pei, D.; Buyanova, M. Overcoming endosomal entrapment in drug delivery. Bioconjugate Chem. 2019, 30, 273–283.

(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2020. ; https://doi.org/10.1101/2020.01.03.894543doi: bioRxiv preprint

10

Insert Table of Contents artwork here

(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2020. ; https://doi.org/10.1101/2020.01.03.894543doi: bioRxiv preprint