Selecting for fast Protein-Protein Association as demonstrated on a

2 days ago - Scientist gets 10 years for theft of gene-modified rice. Chinese scientist Weiqiang Zhang was sentenced on April 4 to more than 10 years ...
0 downloads 4 Views 2MB Size
Subscriber access provided by UNIV OF NEW ENGLAND ARMIDALE

Selecting for fast Protein-Protein Association as demonstrated on a Random TEM1 Yeast Library Binding BLIP Ruth Cohen Khait, and Gideon Schreiber Biochemistry, Just Accepted Manuscript • DOI: 10.1021/acs.biochem.8b00172 • Publication Date (Web): 19 Apr 2018 Downloaded from http://pubs.acs.org on April 19, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Selecting for fast Protein-Protein Association as demonstrated on a Random TEM1 Yeast Library Binding BLIP Ruth Cohen-Khait and Gideon Schreiber*

Department of Biomolecular Sciences Weizmann Institute of Science, 76100, Rehovot, Israel *Corresponding author, [email protected]

ACS Paragon Plus Environment

1

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 13

ABSTRACT. Protein-protein interactions mediate the vast majority of cellular processes. Though protein interactions obey basic chemical principles also within the cell, the in vivo physiological environment may not allow for equilibrium to be reached. Thus in vitro measured thermodynamic affinity may not provide a complete picture of protein interactions in the biological context. Binding kinetics composed of the association and dissociation rate constants are relevant and important in the cell. Therefore, changes in protein-protein interaction kinetics have a significant impact on the in vivo activity of the proteins. The common protocol for selection of tighter binders from a mutant library selects for protein complexes with slower dissociation rate constants. Here we describe a method to specifically select for variants with faster association rate constants by using pre-equilibrium selection, starting from a large random library. Towards this end we refine the selection conditions of a TEM1-β-lactamase library against its natural namomolar affinity binder β-lactamase inhibitor protein (BLIP). The optimal selection conditions depend on the ligand concentration and on the incubation time. In addition, we show that a second sort of the library helps to separate signal from noise, resulting in a higher percent of faster binders in the selected library. Fast associating protein variants are of particular interest for drug development and other biotechnological applications.

ACS Paragon Plus Environment

2

Page 3 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

INTRODUCTION. Protein-protein interactions (PPIs) are ubiquitous in the cell. By their nature, PPIs are fast and specific (1). They are formed through shape and chemical complementarity of the interface (2, 3). PPIs are encoded by the genomic DNA, making them prone to evolution and natural selection (4). A key parameter characterizing molecular interactions is the affinity (KD). The affinity constant can be defined either thermodynamically from the concentration ratio of the unbound and bound reaction components when the reaction reaches chemical equilibrium, or kinetically as the ratio between the interaction dissociation (koff) and association (kon) rate constants (5–7). The kinetic definition may be of particular importance in the biological context, where time and concentration play a crucial role (8). For example, in reactions involving multiple sequential components the flux, not the equilibrium (which is not reached) may dictate the outcome (9) (10). Moreover, in the crowded in vivo milieu many processes - such as drug delivery, can act without reaching equilibrium, resulting in particular significance of the reaction kinetics and the rate limiting process (11, 12). Whether kon or koff separately have a significant role on the activity depends on the specific interacting protein pair, on their relative available concentrations, and on the biological context. In some cases, the dissociation rate constant is of particular importance since the residence time of the complex determines activity (13). In other cases, fast “on rates” are desired as they allow for reduced concentrations and faster initiation of binding. For example, in the case of histone deacetylase inhibitors only fast binders result in pro-granulin enhanced levels in human neurons which may prevent frontotemporal dementia (14). For electron transfer complexes the reaction is optimized when on and off rates are fast (15, 16). For RNAse and DNAse inhibitors a fast on rate and slow off rate is desired to maximise the inhibition of these efficient, toxic enzymes (17). For Intrinsically Disordered Proteins, where folding and binding are coupled, changes in on and off rates may affect the mechanism of binding from one where binding drives folding to binding upon folding (9, 10). For type I cytokine receptors a 5000-fold difference in kon was apparently anticorrelated with differences in ligand concentrations of the cytokines erythropoietin, interleukin-4, human growth hormone, and prolactin, leading to similar observed association-rates (18). Selection methods, such as yeast surface display or phage display were traditionally applied to enhance binding affinity, mostly through off rate optimization. For example, the binding affinity between Human Growth Hormone and its receptor, anti-ErbB2 chimeric antibody binding ErbB2 or a FAB based on the Humira® antibody binding the tumor necrosis factor were increased by hundreds of fold with only marginal changes in the on rate (19–21). These selections were performed under equilibrium conditions, and show that equilibrium selection methods select for slower off rates rather than faster on rates. Enhancing kon is an area of protein engineering that registered remarkable success (22, 23), partially due to a detailed understanding of the mechanisms dictating protein complex formation (1, 24) . For example, kon of BLIP binding TEM1 was increased ~100-fold by optimizing the electrostatic complementarity between the two proteins (23). Conversely, very little work has been done to increase “on rates” not through electrostatic optimization. This would require to follow in details the final docking stages of binding, which, using current computational methods is difficult (1) . Here we present a novel strategy for retrieving proteins with enhanced on-rates which does not necessarily depend on any particular property of the interaction. Starting from a large random library one can specifically select for enhanced kon rates by applying pre-equilibrium selection (25).

ACS Paragon Plus Environment

3

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 13

Figure 1. Kinetic selections of the TEM1 random library against Y-BLIP (A) Schematic description of the selection procedure. (B) FACS plot of TEM1 random library sorted against Y-BLIP at designated concentrations and times of incubation. Sort I and II are the FACS plots after 1 and 2 sorts applying 50 nM Y-BLIP for 20 s. In the upper panel the black dots represent the no Y-BLIP control and the red 5 s incubation with Y-BLIP. In all other plots the black dots represent the signal after incubation with YBLIP at the designated times and concentrations. The numbers in each plot are the percent yeast that bind Y-BLIP in the given condition. (C) Fraction Y-BLIP binding the TEM1 library upon incubation with 50, 200 and 1000 nM Y-BLIP for 5-1800 s. (D) Fraction binders of the TEM1 libraries created after FACS sorting using 50 nM or 1 µM Y-BLIP for the designated times. The highest percent binders were in the library generated after two sorts using 50 nM Y-BLIP for 20 sec (designated 20+20 in the figure). Error bars (SE) were calculated from at least 6 repeats of each measurement. The selected libraries were compared in all cases using 50 nM Y-BLIP for 20 seconds.

ACS Paragon Plus Environment

4

Page 5 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Table 1: Selected clones after FACS II from the fastest selected library. Sequence Transcomp kon T140A T188S I208L G218R 2.2 I280V 2.2 A126T E239K 11.7 E110G E239G 41.3 N100Y 4.6 E104Q K111R 3.3 R93H E110V 9.6 K51R E110G 32.6 I47F K111N S124R 1.6 L102R A237T K111R V30Q H112R T128S E171K E197K A185S A217S E239G E110G E197G R204Q S82T K111R T128L Mutations on 15 clones from fastest selected library (FACS II). For clones where TransComp calculations were possible the relative to WT (4.6x104 M-1s-1) kon values are shown.

DEVELOPMENT OF THE METHOD. We describe the method in details and show how to best optimize it, while discussing the parameters that are of particular importance for enrichment of a library of fast binding protein variants.

As starting point we used a large (10 ) yeast library displaying TEM1-βlactamase with an average of 4 mutations per clone that were generated through random mutagenesis using mutazyme (GeneMorph II random mutagenesis kit catalog # 200550). The yeast (Eby100) library was created following the procedure described by Benatuil (26). The DNA amplification was done with Taq DNA-polymerase as described by Chao (27). The mutant library was created on a stabilized variant of the TEM-1 βlactamase previously described (28), allowing for more mutations to be accumulated on TEM1 without losing its structural fold. The library size was estimated to be ~108 by plating serial dilutions on selection plates lacking Tryptophan. The library was further examined for the correct gene insertion and for the average amount of mutations introduced per gene by sequencing 20 single clones. This TEM1 library was selected against the b-lactamase inhibiting protein (BLIP), which binds TEM1 with nM affinity, kon of ~105 M-1s-1 and koff of ~10-4 s-1 (29, 30). BLIP (bait) was conjugated to the fluorescent protein YPET enabling direct FACS sorting without the need for a secondary ligand (Figure 1A). The unique features of our selection are the short incubation times combined with relatively low ligand concentrations, resulting in selection of fast binding variants. In addition to addressing the importance of these two key parameters determining the association rate constant (time and concentration), we also address an important technical issue of separating signal from noise, which arises in this kind of selections due to the use of low ligand concentration. We demonstrate that a second selection round on the same library leads to significantly better enrichment, with the resulting library containing mostly fast associating clones. The experimental results were further backed by simulations. 8

FACS sorts were initiated after incubation with different concentrations of Y-BLIP, at incubation times varying from seconds to 30 min. All incubations (except for 30 min, which was done on ice) were done at room temperature. After removing excess ligand through washing, all libraries were kept on ice as a pellet until resuspension for FACS analysis. The overall selection scheme is presented in Figure 1A. Figure 1B and 2A show the FACS data after different incubation times. The black dots in the top panel show the background, without the addition of Y-BLIP, while the red dots represent incubation with 50 nM Y-BLIP for 5 sec. In all other panels the black dots show the FACS data upon addition of Y-BLIP for the designated times. The numbers in each panel ACS Paragon Plus Environment

5

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 13

show the fraction of bound Y-BLIP at each condition (from 0.05-0.66), with the fraction of yeast binding Y-BLIP increasing at longer incubation times. A fraction of 0.66 (after incubation with 1 µM Y-BLIP for 30 min) represents the fraction of yeast that is able to bind Y-BLIP, while the remaining 0.34 are background (as shown in the top panel). Figure 1C shows the fraction of yeast bound after 5-1800 sec incubation at three Y-BLIP concentrations. In addition to the fraction bound, the strength of the signal is also increasing (x-axis Figure 1B and 2A). The strength of the signal is proportional to the number of Y-BLIP proteins binding a single yeast cell. Each vial contains ~108 yeast (from 1 ml grown to OD600~2 concentrated to 50 µL). Assuming that each yeast has on average ~100000 surface TEM1 proteins (31), the concentration of TEM1 proteins displayed on the yeast is ~0.5 µM. Therefore, 50 nM Y-BLIP is sufficient to bind to only 10% of the surface TEM1 proteins at equilibrium, while 1 µM Y-BLIP is saturating the yeast surface. This

Figure 2. Histograms of TEM1 random libraries incubated with Y-BLIP for the designated times. The first Gaussian represents the 35% of yeast clones that do not bind Y-BLIP (see Figure 1B). (A) Y-BLIP signal after different incubation times. (B) Y-BLIP signal after one (FACS II) or two (FACS III) sorts against 50 nM BLIP for 20 sec. (C and D) Simulating the expected FACS signal. The simulations were done using the ProK II software (Applied Photophysics) using a standard reversible binding model assuming that 1% of the library is made of 5x faster binding clones with 100,000 thousand copies of TEM1 displayed per yeast and 108 yeast in 50 µl used for selection (taken from the experimental conditions used here). (C) Signal deconvolution (50 nM Y-BLIP for 20 s) into background (generated from a Gaussian distribution around the experimentally obtained median value), the slow (99%) and fast (1%) binding clones. (D) The simulated selection process, with FACS II and III are as in (B).

ACS Paragon Plus Environment

6

Page 7 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Table 2: Biophysical characterization of selected mutants Trans Comp konmut/ HyPare koff d KDe x10- Tmf
 Enzymatic 10 a b c -4 -1 Mutant konwt kon kon x10 s M ᵒC Activityg E104Q K111R 8.2 0.8 3.3 1.6 0.9 54 0.011 A126T E239K 4.1 0.8 12 1.1 1.3 56 0.011 E110G E197G R204Q 3.7 1.5 1.2 1.5 54 0.012 R93H E110V 3.5 1.4 10 1.1 1.5 51 0.011 E110G E239G 2.7 1.4 41 1.0 1.8 55 0.010 V30Q H112R T128S E171K E197K 2.7 1.1 1.5 2.7 51 0.005 K51R E110G 2.6 1.5 33 1.4 2.6 50 0.010 WT 1.0 1.2 5.7 55 0.008 a Stopped flow measurements as described in the main text. The values are relative to WT (3x105 M-1s-1). The standard error of individual relative values is 20%. 
 b Relative kon values as calculated using HyPare (34). c TransComp calculations relative to WT kon (see Table 1). d From SPR measurements done at six different analyte concentrations. Standard error is 25%. e KD = koff / kon, with koff values taken from SPR and kon from stopped flow. Standard error is 32%. f Protein thermal stability, standard error is 0.6 0C g Enzymatic activity was determined as the initial slope of centa catalysis using 5 nM TEM1 and 375 µM substrate. Standard error is 3.7%

rough calculation fits the 10-fold stronger fluorescence signal (x-axis) observed per cell when using 1 µM Y-BLIP over 50 nM (Figure 2A). The 1% yeast cells having the strongest fluorescence signal (x-axis) were selected from each binding condition using FACS sort. The selected library had a size of up to ~104 different clones (the size may be lower due to clone repetitions). A reference library was also sorted using as bait 1 µM Y-BLIP for 30 minutes to assure that it reaches equilibrium. This library was sorted for the best 50% binders. This library served as the reference library for the performance of all the other libraries. To find the optimal FACS sort conditions, the libraries sorted after incubation with 50 nM and 1 µM YBLIP for 5-1800 sec were incubated with 50 nM Y-BLIP for 20 sec and the fraction of TEM1 clones binding Y-BLIP were measured (Figure 1D). The highest fraction of binders was found for the library sorted against 50 nM Y-BLIP for 20 seconds. Repeating the selection for a second time under the same conditions (50 nM Y-BLIP for 20 seconds) increased further the percent of binding clones (Figure 1B and D and 2B). Selections against 1 µM Y-BLIP gave worse results, with a second round of selection against 50 nM Y-BLIP for 20 seconds resulting in lesser improvement. The results emphasize that two rounds of selection are preferable over one, and that the optimal incubation time and concentration of the bait are important parameters to achieve optimal results. Figures 2C and D show the results of simulating (using ProK II, Applied Photophysics) the expected FACS sort results using as input a mix of 99% of TEM1 clones that bind at WT kon and 1% clones that bind 5-times faster. Indeed, the simulated FACS curves mimic the experimental curves (Figures 2A and B versus C and D), including the enrichment of faster binders. The simulations suggest that both a low bait concentration and a short incubation time contribute to the

ACS Paragon Plus Environment

7

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 13

selection of faster binders from the general library, as is also shown experimentally. VALIDATION OF THE METHOD. To explore the sequence and binding constants of individual clones from the best performing library (sorted two times against 50 nM Y-BLIP after 20 seconds incubation), 15 clones were sequenced and analysed by TransComp – a software designed for kon prediction (22) (Table 1). TransComp predicted an increase in kon for the selected clones. Seven of those clones were expressed as SUMO fusion. In short, the genes were expressed in E. coli strain BL21, at 37 oC until OD660~1 and then incubated with 200 µM IPTG at 16 oC overnight. The bacteria were pelleted and sonicated. The protein in the supernatant was bound to Ni beads and suspended in SUMO protease buffer (50 mM HEPES pH=7.5, 0.3 M NaCl, 10 mM imidazol, 2 mM MgCl2, 10% glycerol and 1 mM DTT). 1:1000 protease to protein was added for 16 hours at 4_C, which cleaves at the N-ter of the protein (32). The cleaved proteins were then dialyzed into PBS (10 mM phosphate buffer, pH 7.4, 150 mM NaCl) and frozen in small aliquots. kon, koff, KD (calculated as koff/kon), thermal-stability and enzymatic activity values were determined (Table 2). kon rate constants were determined under second order conditions with the concentrations of both proteins being 0.25 µM using a stopped flow (Applied Photophysics) as described in (31). Stopped-flow results were shown to provide a more reliable measure for kon values then Surface Plasmon Resonance (SPR) (23). The measurements were done using either PBS or 20 mM HEPES buffer pH 7.4 without NaCl (Table 3 and Figure 3). koff rate constants were determined using the ProteON XPR36 (Bio-Rad) in PBS-Tween (0.005% Tween 20). The ligands were bound to a Bio-Rad GLC SPR chips using the high affinity biotin-avidin interaction (Figure 3). Surface regeneration was done by two short cycles of 8M urea as established previously for this complex (30) . The analyte was injected at 6 different concentrations, and the affinity and kinetic binding constants were determined by the inbuilt software using the Langmuir reaction model (25). b-

Figure 3. SPR traces and stopped flow measurements of TEM1 selected mutant clones. (A) SPR traces measured under standard conditions in a XPR36 ProteON (BioRad) (25) for seven selected fast binding TEM1 clones and WT. The presented traces are from applying 500 nM of each of the TEM1 variants on a BLIP immobilized surface. To obtain affinity and kinetic values shown in Table 2, 5 additional TEM1 concentrations were applied. Due to the slow dissociation rate the dissociation phase lasted for 1 hr, while the association took 2 min therefore the figure is mostly relevant for dissociation phase comparison. (B) Stopped Flow measurements of 3 faster binding clones as determined in 20 mM HEPES pH 7.4. The data were fitted using a second order equation with the concentration of both proteins being 0.25 µM. The fitting results are shown in Tables 2 and 3.

ACS Paragon Plus Environment

8

Page 9 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

lactamase activity was estimated with the commercially available yellow chromogenic substrate CENTA (Calbiochem # 219475), which absorbance at 405 nm rises upon catalysis by b-lactamase. As a measure of relative activity, the initial slope at the linear phase of the reaction (steady state) was determined using 5 nM enzyme and 375 µM substrate. Protein Thermal stability was determined by the method described in (33). The assay is based on the tendency of Sypro Orange (sigma S5692) to bind hydrophobic areas, which are exposed upon protein unfolding due to a gradual increase in temperature (0.050 C per second from 250 C to 990 C) controlled by the Applied Biosystems ViiA 7 RT PCR instrument. The assay was optimized for b-lactamase, with the measurements done in 20 µl volume using 10 µM protein in PBS buffer with the dye diluted 500fold (25). Reported values for individual measurements were calculated from 2-6 repeats of the different experiments. The seven purified and analysed TEM1 selected clones bind BLIP faster than the wild-type in general agreement with the computational predictions. The dissociation rate constant, protein thermal-stability and enzymatic activity were not altered by the selection (Figure 3A and Table 2). The increase in the association rate constants were even more pronounced when the ionic strength was reduced from the physiological value of 150 mM to 20 mM, (Table 3, Figure 3B). Note that kon of the wild-type complex did not increase in low salt conditions while it did so for the selected mutations, emphasising the limited role of electrostatic steering in the original interaction (Table 3) (23). This suggests that electrostatic forces contribute to the increased association rate constants of the selected mutations. Still, the electrostatic contribution towards faster binding explains only part of the increased rate constants, as can be seen from the limited correlation between the calculated kon values using HyPare (which calculates only the electrostatic contribution to association (34)) or TransComp (which takes into account also the non-electrostatic contributions (22)) versus the experimental results (Table 2). It should be noted that these two predictors performed much better when predicting rate enhancement of engineered rather than selected mutations in protein interfaces (22, 23, 34). These results emphasise that the selection procedure developed by us for faster binders will select for clones irrespectively of the reason they bind faster. Thus, a combination of different chemical forces not necessarily related to electrostatic steering will also contribute to the observed results. Table 3: kon values for selected mutants at high and low salt concentration. 150 mM No Salt NaCl Sequence 1.1 WT 1.0 5.1 E110G E197G R204Q 3.7 8.3 A126T E239K 4.1 12.1 E104Q K111R 8.2 Relative kon (to WT at 150 mM NaCl) as measured using a stopped flow in 25 mM HEPES buffer pH 7.4. The measurements were repeated 2 times for each clone with an average standard error of 20%. The values are relative to WT (3x105 M-1s-1).

CONCLUSION. In this paper we present a library-based method to select for faster binders originating from a large random library. Enrichment in fast binding clones is achieved by incubating the library against low ligand concentrations for very short incubation periods and selection of the 1% best binders. Still, there is a limit in stringency, as it may lead to high signal to noise and low separation of faster binding clones (as seen in Figure 1B and D). In our case the best results were achieved at incubation conditions were 20% of the yeast clones bound to BLIP, from

ACS Paragon Plus Environment

9

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 13

Figure 4. Illustration of the protocol developed here for selection of fast binders. Ovals represent yeast cells expressing random variants of TEM1 on their surface (squares). Red squares – represent TEM1 bound to Y-BLIP, while white squares represent unbound TEM1.

which the highest 1% were selected. The optimal selection conditions were achieved by varying both the ligand concentration and incubation time. Three ligand (Y-BLIP) concentrations were examined: a high concentration of 1 µM, a medium concentration of 200 nM and a low concentration of 50 nM. Optimal selection conditions were impossible to achieve with the high YBLIP concentration, since the library reached saturation even at very short incubation times such as 5 seconds (Figure 1B). In our case the lowest, 50 nM Y-BLIP concentration gave the best enrichment for fast binders. A second round of selection further improved the enrichment. Simulating the process (Figures 2C and D) shows clearly that the reason for this is that the second round of selection purges the slower binding clones from the selection, further increasing the percent of faster binders in the library. Overall, the experiments show that after two rounds of selections all the tested clones bound faster. A summary of the method to select for fast binders is given in Figure 4 and comprised the following steps: 1. Identify the ligand concentration where ~10% of the bait protein is bound at equilibrium and the incubation time needed to achieve binding to 20% of the binding capacity at the optimal concentration (which is here 50 nM, 20 sec for TEM1-BLIP).

2. Select the library at the ligand concentration and incubation times (pre-equilibrium) determined in step 1. One should consider short incubation periods of under one minute. Verify the results by selection and optimisation of fraction binders after short incubation times (and compare with the equilibrium selected library). 3. Sort and select for the best 1% of binders. 4. Grow the selected fast library, and repeat the same selection procedure for further enrichment of the library with fast binders. 5. Isolate individual clones from the selected library for analysis.

ACS Paragon Plus Environment

10

Page 11 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

It should be noted that the method will perform only for interacting protein pairs that have a sufficiently slow off-rate for the FACS selection to be done under pre-equilibrium conditions. We estimate that koff of 0.02 s-1 or slower would suffice.

ACKNOWLEGMENTS This work was supported by a grant of the Israel Science Foundation (1549/14)

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.

Schreiber, G., Haran, G., and Zhou, H. X. (2009), Chem Rev 109, 839-860. Perkins, J. R., Diboun, I., Dessailly, B. H., Lees, J. G., and Orengo, C. (2010), Structure 18, 12331243. Conte, L. L., Chothia, C., and Janin, J. (1999), J. Mol. Biol. 285, 2177-2198. Szklarczyk, D., Franceschini, A., Wyder, S., Forslund, K., Heller, D., Huerta-Cepas, J., Simonovic, M., Roth, A., Santos, A., Tsafou, K. P., Kuhn, M., Bork, P., Jensen, L. J., and von Mering, C. (2015), Nucleic Acids Res 43, D447-52. Schreiber, G. (2002), Curr Opin Struct Biol 12, 41-47. Janin, J. (1995), Biochimie 77, 497-505. Stanley, N., Esteban-Martín, S., and De Fabritiis, G. (2014), Nat Commun 5, 5272. van der Merwe, P. A., Bodian, D. L., Daenke, S., Linsley, P., and Davis, S. J. (1997), J Exp Med 185, 393-403. Shammas, S. L., Crabtree, M. D., Dahal, L., Wicky, B. I., and Clarke, J. (2016), J Biol Chem 291, 6689-6695. Zhou, H. X. (2012), Trends Biochem Sci 37, 43-48. Davidi, D., Noor, E., Liebermeister, W., Bar-Even, A., Flamholz, A., Tummler, K., Barenholz, U., Goldenfeld, M., Shlomi, T., and Milo, R. (2016), Proc Natl Acad Sci U S A 113, 3401-3406. Zotter, A., Baeuerle, F., Dey, D., Kiss, V., and Schreiber, G. (2017), J. Biol. Chem. 292, 1583815848. Shammas, S. L., Travis, A. J., and Clarke, J. (2014), Proc Natl Acad Sci U S A 111, 12055-12060. She, A., Kurtser, I., Reis, S. A., Hennig, K., Lai, J., Lang, A., Zhao, W. N., Mazitschek, R., Dickerson, B. C., Herz, J., and Haggarty, S. J. (2017), Cell Chem Biol 24, 892-906.e5. Prudencio, M., and Ubbink, M. (2004), J Mol Recognit 17, 524-539. Spencer, A. L., Bagai, I., Becker, D. F., Zuiderweg, E. R., and Ragsdale, S. W. (2014), J Biol Chem 289, 29836-29858. Schreiber, G., Buckle, A. M., and Fersht, A. R. (1994), Structure 2, 945-951. Pang, X., Qin, S., and Zhou, H. X. (2011), Biophys J 101, 1175-1183. Bernat, B., Sun, M., Dwyer, M., Feldkamp, M., and Kossiakoff, A. A. (2004), Biochemistry 43, 6076-6084. Hu, D., Hu, S., Wan, W., Xu, M., Du, R., Zhao, W., Gao, X., Liu, J., Liu, H., and Hong, J. (2015), PLoS One 10, e0129125. Votsmeier, C., Plittersdorf, H., Hesse, O., Scheidig, A., Strerath, M., Gritzan, U., Pellengahr, K., Scholz, P., Eicker, A., Myszka, D., Coco, W. M., and Haupts, U. (2012), MAbs 4, 341-348. Qin, S., Pang, X., and Zhou, H. X. (2011), Structure 19, 1744-1751. Selzer, T., Albeck, S., and Schreiber, G. (2000), Nat Struct Biol 7, 537-541. Plattner, N., Doerr, S., De Fabritiis, G., and Noé, F. (2017), Nat Chem 9, 1005-1011. Cohen-Khait, R., and Schreiber, G. (2016), Proc. Natl. Acad. Sci. U. S. A. 113, 14982-14987. Benatuil, L., Perez, J. M., Belk, J., and Hsieh, C.-M. (2010), Protein Eng. Des. Sel. 23, 155-159.

ACS Paragon Plus Environment

11

Biochemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

27. 28. 29. 30. 31. 32. 33. 34.

Page 12 of 13

Chao, G., Lau, W. L., Hackel, B. J., Sazinsky, S. L., Lippow, S. M., and Wittrup, K. D. (2006), Nat Protoc 1, 755-768. Dellus-Gur, E., Toth-Petroczy, A., Elias, M., and Tawfik, D. S. (2013), J Mol Biol 425, 2609-2621. Strynadka, N. C., Jensen, S. E., Johns, K., Blanchard, H., Page, M., Matagne, A., Frere, J. M., and James, M. N. (1994), Nature 368, 657-660. Albeck, S., and Schreiber, G. (1999), Biochemistry 38, 11-21. Boder, E. T., and Wittrup, K. D. (2000), Methods Enzymol 328, 430-444. Frey, S., and Goerlich, D. (2014), J. Chromatogr. A 1337, 95-105. Lavinder, J. J., Hari, S. B., Sullivan, B. J., and Magliery, T. J. (2009), J Am Chem Soc 131, 37943795. Shaul, Y., and Schreiber, G. (2005), Proteins 60, 341-352.

ACS Paragon Plus Environment

12

Page 13 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

For Table of Contents Use Only

ACS Paragon Plus Environment

13