Chapter 7 Epitope-Specific Binder Design by Yeast Surface Display Jasdeep K. Mann and Sheldon Park Abstract Yeast surface display is commonly used to engineer affinity and design novel molecular interaction. By alternating positive and negative selections, yeast display can be used to engineer binders that specifically interact with the target protein at a defined site. Epitope-specific binders can be useful as inhibitors if they bind the target molecule at functionally important sites. Therefore, an efficient method of engineering epitope specificity should help with the engineering of inhibitors. We describe the use of yeast surface display to design single domain monobodies that bind and inhibit the activity of the kinase Erk-2 by targeting a conserved surface patch involved in protein–protein interaction. The designed binders can be used to disrupt signaling in the cell and investigate Erk-2 function in vivo. The described protocol is general and can be used to design epitope-specific binders of an arbitrary protein. Key words Yeast surface display, Epitope-specific interaction, Monobody, Negative design, Assay development
1
Introduction Yeast surface display (YSD) is a versatile protein engineering tool and has been used to engineer various molecular properties, including stability, affinity, binding kinetics, and catalysis [1–3]. Key advantages of yeast display compared to other display platforms, especially phage display, include (1) improved folding of synthesized proteins and (2) normalization of activity with respect to expression in order to achieve quantitative screening of the displayed library. Because yeast is genetically amenable, a large library of mutants can be assembled and maintained [4, 5], which is critical to find binders with desired properties. To engineer binding affinity, the yeast cells displaying potential binders are screened with the target molecule, from which the highest affinity binders are isolated based on fluorescence or direct physical association [6]. The end result of a typical sort cycle is a collection of binders with improved expression and affinity for the target molecule. While some engineered binders may bind the target in ways that
Bin Liu (ed.), Yeast Surface Display: Methods, Protocols, and Applications, Methods in Molecular Biology, vol. 1319, DOI 10.1007/978-1-4939-2748-7_7, © Springer Science+Business Media New York 2015
143
144
Jasdeep K. Mann and Sheldon Park
interfere with the target function, more often the selected binders need to be characterized through additional rounds of epitope mapping or a functional screen to identify genuine inhibitors of function [7, 8]. This is because the binding affinity is typically engineered independent of the molecular details of interaction, and many binders are just that—binders and not inhibitors. As the size of the target molecule increases, an increasingly small fraction of engineered binders is expected to be functionally relevant. We recently described a YSD-based procedure for directly engineering epitope selectivity in order to streamline the discovery of functionally useful binders [9]. Our protocol involves the use of two alternative forms of the target molecule: wild type and a mutant containing surface substitution(s) at the desired epitope. The surface displayed library is then alternatively screened for binding to the wild type protein and for a lack of binding to the mutant protein. If the two molecules are structurally similar and differ only at the targeted surface, then the combination of positive and negative selections with wild type and mutant proteins will identify high affinity binders that are also epitope specific (Fig. 1).
Fig. 1 Schematic of the sorting strategy to engineer epitope-specific binders. (a) Yeast surface display library is assembled and expressed. (b, c) Using the target molecule for labeling, perform a series of positive sorting to identify the clones with the highest target affinity. (d) Construct a mutant protein containing one or more mutations at the desired target epitope. (e) Using the mutant protein, perform a negative sort to identify the binders whose interaction with the target protein is abrogated by the mutations. (f) The selected clones correspond to high affinity, epitope-specific binders
Engineering Epitope-Specific Binders
145
Given that disruption of ligand binding may disrupt signaling or catalysis, blocking functionally important sites of a signaling molecule or an enzyme with engineered binders should potentially inhibit the target molecule function. This strategy was used to engineer monobody binders of Erk-2 that inhibited its kinase activity by blocking a binding site used in intermolecular interaction [10–12]. Engineering epitope specificity by YSD thus offers an efficient method of discovering useful inhibitors of protein function.
2
Materials It is important that all reagents are prepared using molecular biology grade reagents in deionized water. Once prepared and sterilized, liquid reagents may be stored at 4 °C for several months. Both selective and nonselective yeast plates may be stored at 4 °C up to 1–2 months.
2.1 Yeast Culture Media Components 2.1.1 Nonselective Medium
Nonselective medium: yeast peptone dextrose (YPD) is used to propagate untransformed yeast. Growing transformed yeast in YPD will result in plasmid loss over time. 1. 10 g of yeast extract. 2. 20 g of peptone. 3. 20 g of dextrose. 4. Dissolve in 1 L water. Autoclave for 35 min at 121 °C. Store at 4 °C.
2.1.2 Selective Growth Medium
Selective growth medium: synthetic dextrose with Trp and Ura dropout (SD/-Trp-Ura). 1. 20 g of dextrose. 2. 6.7 g of yeast nitrogen base (without amino acid or ammonium sulfate). 3. 5 g ammonium sulfate. 4. 1.4 g of yeast synthetic dropout media mix lacking tryptophan, uracil, histidine, and leucine (from Sigma). The use of a quadruple dropout (-Trp,-Ura,-His,-Leu) as a base is convenient in case yeast needs to be cultured in other combinations of selection markers. By supplementing the mix with different nutrients, additional media can be prepared using common reagent. 5. 380 mg of leucine. 6. 76 mg of histidine. 7. 100 ml of 10× pH 6.0 or pH 4.5 media buffer (see below).
146
Jasdeep K. Mann and Sheldon Park
8. Add water to 1 L. To prepare a 5× stock, which is used to make selection agar plates, add water to 200 ml instead. 9. Sterilize using a 0.2 μm filter. Store at 4 °C (see Note 1). 2.1.3 Selective Induction Medium
Selective induction medium: synthetic galactose with Trp and Ura dropout (SG/-Trp-Ura). The only difference between the growth and induction media is the use of galactose in the induction medium. 1. 20 g of galactose. 2.–7. Rest of the components are same as in selective growth medium (see above). 8. Sterilize using a 0.2 μm filter and store at 4 °C.
2.1.4 Media Buffer
1. 10× sodium phosphate buffer at pH 6.0: This buffer is used to prepare selective yeast medium used to grow and propagate transformed yeast. To prepare the stock, dissolve 102 g of Na2HPO4·7H2O and 86 g of NaH2PO4·H2O in 1 L water. Sterilize by autoclaving before storage. 2. 10× sodium citrate buffer at pH 4.5: This buffer is used during cell sorting, which exposes the yeast culture to possible bacterial contamination. Low pH retards bacterial growth and reduces possible contamination. To prepare the stock, dissolve 14.7 g tri sodium citrate dihydrate (HOC(COONa) (CH2COONa)2·2H2O) and 4.29 g citric acid monohydrate (HOC(COOH)(CH2COOH)2·H2O) in 1 L water. Autoclave and store at 4 °C (see Note 2).
2.1.5 Yeast Selection Agar Plates
These plates are used to grow transformed yeast to identify individual yeast clones. 1. 180 g of sorbitol. 2. 15 g of bacto-agar. 3. Add water approximately to 700 ml and resuspend the solutes by stirring with a magnetic stir bar. 4. Add 100 ml of 10× media buffer (typically pH 6.0 is used). 5. Autoclave for 35 min at 121 °C. 6. Cool the autoclaved mixture while stirring until the temperature falls below 60 °C (warm to the touch). 7. Add 200 ml of filter sterilized 5× SD/-Trp-Ura stock. Adjust the volume to 1 L with sterile deionized water. 8. Slowly pour into 15 mm petri dishes (each plate holds ~25 ml of agar). Stack the plates together while the agar sets in order to minimize condensation on the lid. 9. Store the plates in a sealed bag at 4 °C. The plates are good for up to 2 months.
Engineering Epitope-Specific Binders
2.2 Target Protein Preparation
147
The target construction will vary from project to project. Here we describe the preparation of the Erk-2 kinase. 1. pRSET-Erk-2: a bacterial expression vector based on pRSET that contains the full-length rat Erk-2 gene downstream of the T7 promoter. NheI and BamHI restriction enzymes (New England Biolab) were used to clone the full length of Erk-2 between an N-terminal 6×histidine affinity tag and a C-terminal FLAG tag. 2. BL21(DE3) pLysS (Invitrogen): the cell line suppresses premature expression of the target molecule, which may be toxic to the cells. 3. Talon resin (Clontech), or other comparable immobilized metal affinity chromatography resin. 4. Imidazole (Sigma). 5. SDS-PAGE gel (Biorad): prepare 10–15 % acrylamide gel as needed. 6. Amicon centrifugal filters (Millipore): the filters are used for buffer exchange and concentration.
2.3 Magnetic Activated Cell Sorting (MACS)
1. EZ-Link Sulfo-NHS-Biotin (Thermo Fisher Scientific): this is used to chemically biotinylate purified Erk-2 so that it can be used with magnetic streptavidin resin to isolate the yeast displayed clones that bind the target. 2. Magnetic streptavidin-microbeads (Miltenyi Biotec). 3. MidiMACS Separator magnet (Miltenyi Biotec). 4. PBSB: Phosphate buffer saline plus 0.5 % (w/v) bovine serum albumin. 5. Kanamycin (Sigma). 6. Penicillin-Streptomycin (10,000 U/ml) (Life Technologies).
2.4 Yeast Surface Labeling for Flow Cytometry
1. Mouse monoclonal anti-cMyc antibody, 9E10 (AbD Serotec). 2. Mouse monoclonal anti-FLAG antibody, M2 (Agilent technologies). 3. Rabbit monoclonal anti-FLAG antibody (Sigma). 4. Anti-mouse IgG (whole molecule)–FITC antibody (Sigma). 5. Anti-mouse IgG (whole molecule)–R-phycoerythrin (PE) antibody (Sigma). 6. Anti-rabbit IgG (whole molecule)–FITC antibody (Sigma). 7. Alkaline phosphatase (AP)-conjugated goat anti-mouse antibody (Sigma). 8. Streptavidin-R-phycoerythrin, SA-PE (Biolegend). 9. PBSS-Phosphate buffer saline plus 0.1 % bovine serum albumin.
148
3
Jasdeep K. Mann and Sheldon Park
Methods
3.1 Library Construction
3.2 Growth and Induction of the Yeast Library
The yeast displayed fibronectin type 3 (Fn3) monobody library G4 was a gift from Dane Wittrup lab (MIT). The construction and characterization of the library was previously described [13, 14]. The library is assembled in the yeast vector pYD1 as a fusion with Aga2 and transformed into the yeast strain EBY100 (Invitrogen). 1. Thaw a frozen aliquot containing 2.5 × 109 cells or ~10-fold excess of the estimated library diversity, and inoculate into 500 ml of SD/-Trp-Ura culture medium. The starting OD600 ~ 0.5 (see Note 3). 2. Grow the cells at 30 °C with constant orbital shaking (250 rpm) overnight. The following morning, dilute the cells in 500 ml of fresh SD/-Trp-Ura medium to the starting OD600 = 0.5. 3. Continue to grow the cells in SD/-Trp-Ura at 30 °C until OD600 = 3. Centrifuge 100 ml of cells and remove the medium. Resuspend the cells in 300 ml SG/-Trp-Ura. Grow at 30 °C for 18 h with constant shaking to induce synthesis of the library (see Note 4).
3.3 Erk-2 Protein Purification
The protein purification steps will depend on the target protein. The following protocol describes the preparation of the Erk-2 kinase. 1. Transform pRSET-Erk-2 into BL21(DE3) pLysS and induce protein synthesis at the OD600 = 0.8 with 0.4 mM IPTG. Grow for 4 h at 37 °C. 2. Lyse the cells by sonication in a buffer containing 50 mM sodium phosphate (pH 8.0), 300 mM NaCl, 1 % triton X-100, 0.1 % 2-mercaptoethanol, 2 mM PMSF, and 10 % w/v glycerol. Mix the lysate with 2 ml of Talon resin. 3. Incubate the lysate and resin at 4 °C for 1 h with constant shaking. Wash the resin with 50 mM sodium phosphate (pH 7.2), 300 mM NaCl, and 20 mM imidazole. Repeat. 4. Elute the bound protein in three fractions of 1 ml of 50 mM sodium phosphate (pH 7.4), 300 mM NaCl, and 300 mM imidazole. Keep the elution volume as small as possible since the yield is low. 5. Concentrate the eluates using an Amicon centrifugal filter with the nominal cutoff of 10 kDa. Analyze the elution on a 12 % SDS-PAGE gel. 6. For biotinylation of Erk-2, incubate purified protein with tenfold excess of biotinylation reagent (EZ-Link Sulfo-NHSBiotin from Thermo Fisher Scientific) at 4 °C for 2 h. Remove unreacted biotin by buffer exchanging to PBS with a 30 kDa Amicon centrifugal filter.
Engineering Epitope-Specific Binders
3.4 Magnetic Bead Enrichment
149
1. Pellet 2.5 × 109 yeast cells (tenfold coverage of the library diversity) at 3,000 × g for 3 min. Wash the cells twice with 25 ml of PBSB and resuspend in 10 ml PBSB. 2. Incubate the cells with 1 μM biotinylated Erk-2 for 1 h at 20 °C (see Note 5). 3. Spin down the cells and wash twice with 25 ml PBSB. 4. Resuspend the pellet in 5 ml PBSB and incubate with 100 μl magnetic streptavidin coated microbeads for 15 min on ice. The yeast clones that display Erk-2 binding monobodies will now be decorated with streptavidin beads. 5. Separate the cells bound to the beads from the rest by using an LS column filled with ferromagnetic particles and a MidiMACS Separator magnet. Because the yeast particles covered with the beads are attracted to the magnet, they are selectively retained in the column compared to uncoated cells (see Note 6). 6. Wash the column with PBSB to remove the yeast cells that are not firmly bound to the beads. 7. Elute the streptavidin coated cells by removing the column from the magnet and washing the column in SD/-Trp-Ura (pH 4.5). 8. Grow the selected cells in citrate-based SD/-Trp-Ura (pH 4.5), supplemented with 50 μg/ml Kanamycin and penstrep (1:100) to retard the growth of bacteria. 9. Estimate the number of cells by plating serial dilutions of collected cells on SD/-Trp-Ura agar plates. Around ~0.1–1 % of the starting cells are typically retained in the fraction. For example, we obtained ~2 × 107 cells from the initial 2.5 × 109 cells corresponding to an estimate of 2 × 106 unique sequences.
3.5 High-Throughput Screen by Fluorescence Activated Cell Sorting (FACS)
1. Grow the cells until they have reached OD600 = 3. Dilute to OD600 = 0.5 and grow back to OD600 = 3. Induce protein synthesis by switching the medium to SG/-Trp-Ura. 2. Pellet the number of yeast cells corresponding to three- to tenfold excess of the estimated sequence diversity (e.g., 107 cells) and wash them with PBSS. 3. Label the cells with 500 nM biotinylated Erk-2 and anti-cMyc antibody (1:100 dilutions) at room temperature (i.e., around 20 °C) for 1 h (see Notes 7 and 8). 4. Wash the cells and incubate with SA-PE (1:100 dilution) and anti-mouse-FITC (1:50 dilution) on ice for 30 min in the dark. 5. Wash the cells and proceed to sorting by FACS. Because the average binders have weak affinity and likely dissociate quickly, it is important that the cells are analyzed as soon as possible after the final wash. Each FACS sorting cycle should not last more than 15 min. If additional cells need to be screened than can be analyzed in 15 min, a new batch of cells needs to be
150
Jasdeep K. Mann and Sheldon Park
washed and prepared every 15 min. Collect the clones with the most intense SA-PE labeling for a given expression level (e.g., roughly corresponding to 1–2 % of the analyzed cells) (see Notes 9 and 10). 6. Collect the selected cells in SD/-Trp-Ura (pH 4.5) containing kanamycin (50 μg/ml) and 1× pen-strep to prevent bacterial contamination. 7. Repeat the FACS analysis while gradually increasing the stringency of selection. For example, systematically decrease the concentration of biotinylated Erk-2 from 500 to 200, 100, 50, and 10 nM. To enrich the binding population, collect the top 1–2 % of the cells each time (see Note 11). 3.6 Epitope-Specific Selection with Mutant Protein
To identify the Fn3 clones with desired epitope specificity, perform negative selection with a mutant Erk-2 (see Note 12). The following factors need to be considered to design a mutant protein appropriate for the negative screen. First, the structure of the target protein should be consulted to make rational mutations. This is not as difficult a requirement to satisfy as it seems at first, since inhibitor design is typically attempted on proteins for which abundant biochemical and structural information is available. Second, the mutations should not interfere with protein folding or otherwise create significant structural perturbation to the protein. Because the mutated residues are usually located on the surface, it should be possible to introduce a mutation that does not cause significant structural perturbation. Finally, the engineered mutation should alter the chemical or physical properties at the mutated site. For example, the mutation may replace a hydrophobic residue with a charged residue or replace a small residue with a large residue. The substitutions should thus affect potential interaction at the mutated surface so that it can be used during negative selection. Keeping the above requirements in mind, we describe the design of the Erk-2 mutant used in our study. 1. Using the D-peptide bound Erk-2 structure (2GPH), design three mutations within the conserved docking site (“CD” domain), H123N, Y126H, and D319N, to disrupt the binding of a D-peptide. Introduce the mutations using mutagenic primers. Purify the mutant similarly as wild type. Verify that the overall structure of the protein remains the same by performing circular dichroism spectroscopy (Fig. 2a). 2. Incubate affinity optimized Fn3 clones on the yeast surface with 500 nM mutant Erk-2. A high concentration of Erk-2 is used to obtain clear separation between binding and nonbinding populations. 3. Collect the cells that do not bind mutant Erk-2 in order to enrich the Fn3 clones whose epitopes include mutated residues and are therefore likely to competitively inhibit the binding of
Engineering Epitope-Specific Binders
151
Fig. 2 Construction of mutant Erk-2 to engineer epitope specificity. (a) The circular dichroism spectra of wild type and mutant Erk-2 were compared to ensure that the mutations introduced at the CD domain do not affect the folding of Erk-2. (b) Functional testing to demonstrate that the engineered mutations prevent the binding of a D-peptide (bait). (c) An example of an engineered, epitope-specific monobody that binds wild type Erk-2 (left) but not mutant Erk-2 (right)
a D-peptide, whose binding is also disrupted by the mutation (Fig. 2b, c) (see Note 13). 4. Perform a final round of positive selection with wild type Erk-2 to identify the highest affinity clones with engineered epitope specificity (see Notes 14 and 15).
4
Notes 1. Synthetic yeast medium cannot be autoclaved because some components, e.g., glutamine, are heat labile and decompose during autoclaving. 2. When sorting a yeast library, the use of a low pH medium is strongly recommended to retard potential bacterial contamination. However, a low pH buffer alone does not guarantee contamination-free growth, and it is essential to practice good
152
Jasdeep K. Mann and Sheldon Park
sterile techniques at all time. We have observed that even with the addition of kanamycin and/or chloramphenicol the library can still get contaminated during a sort. Contaminated cultures are easy to spot based on visual inspection. Whereas yeast cells quickly settle to the bottom of a culture tube if left undisturbed, contaminated culture remains cloudy even after centrifugation. If the yeast culture becomes contaminated with bacteria and repeated sorting is not an option, it is possible to rid the culture of contamination by passing the cells through multiple centrifugation-wash cycles and propagating the cells in a low pH medium with antibiotics. 3. When working with a frozen yeast library, check cell viability by plating serial dilutions on SD agar plates. A plate containing a few hundred colonies gives the most reliable estimate of cell density. If the cells were frozen according to the established protocol (see ref. 4, for example), cell viability should be high and greater than 90 %. Low cell viability may reduce the sequence diversity in the library. 4. The cells should be freshly passaged just before induction in order to optimize protein expression. 5. Yeast cells are heavy and precipitate during long incubation. Keep the cells in suspension through gentle rocking. 6. Pipette the cells to avoid clumps. Magnetic separation should be done promptly at 4 °C to avoid dissociation of bound Erk-2. 7. The exact temperature of incubation is not important. For convenience, we leave the cells on the bench during the first labeling step. Subsequent labeling is done on ice to minimize dissociation of bound ligand. 8. Make sure that Erk-2 is used in stoichiometric excess compared to the number of monobody molecules on the yeast surface so that the binding does not significantly change the free protein concentration. The number of displayed molecules is estimated to be ~50,000 per cell. 9. Calibrate the compensation using cells labeled with 9E10 and secondary antibody conjugated with FITC or PE. For a double-labeled control, we express an engineered monobody that binds maltose binding protein (MBP) and label the cells with biotinylated MBP and SA-PE as well as 9E10 and secondary antibody. For negative controls, Fn3 displaying cells are labeled with secondary antibodies only. 10. Selecting a much smaller than 1 % of the population may collapse the diversity of the library too quickly and run the risk of losing useful binders that may not have the highest affinity. 11. The cells are initially labeled with SA-PE. However, the use of biotinylated Erk-2 introduces a bias. Some of the selected clones may also bind the FLAG tag rather than Erk-2. To
Engineering Epitope-Specific Binders
153
identify the clones that bind Erk-2, label bound Erk-2 with anti-FLAG antibody (M2) starting from Round 5. M2 labeling is not used until a later round because the Erk-2 affinity of the Fn3 clones is on average low at first and many monobodies may dissociate from Erk-2 during the time it takes for antibody labeling. 12. Negative selection should preferably be done after several rounds of affinity maturation, since it is difficult to differentiate between weak binders and nonbinders. 13. A gate should be drawn during the negative sort to minimize contamination of the clones that also bind the mutant protein. Because a high concentration of the mutant protein is used for labeling (e.g., 500 nM of mutant compared to 10 nM wild type during the preceding positive selection), the gate can be somewhat generous and still bias the clones based on different binding to wild type and mutant proteins. 14. If necessary, perform one or more rounds of positive sort following the negative sort to optimize the binding affinity. It is recommended that the negative sort is introduced toward the end of the selection cycle but not as the last selection step. 15. The individual clones must be tested against wild type and mutant proteins to demonstrate epitope-specific binding before functional testing is attempted. Only a small percentage of the selected clones are expected to bind both variants.
Acknowledgements This work was supported by the NSF grant (1053608) to S.P. References 1. Feldhaus MJ, Siegel RW (2004) Yeast display of antibody fragments: a discovery and characterization platform. J Immunol Methods 290:69–80 2. Gai SA, Wittrup KD (2007) Yeast surface display for protein engineering and characterization. Curr Opin Struct Biol 17:467–473. doi:10.1016/j.sbi.2007.08.012, S0959440X(07)00119-4 [pii] 3. Pepper LR, Cho YK, Boder ET et al (2008) A decade of yeast surface display technology: where are we now? Comb Chem High Throughput Screen 11:127–134 4. Chao G, Lau WL, Hackel BJ et al (2006) Isolating and engineering human antibodies using yeast surface display. Nat Protoc 1:
755–768. doi:10.1038/nprot.2006.94, nprot. 2006.94 [pii] 5. Benatuil L, Perez JM, Belk J et al (2010) An improved yeast transformation method for the generation of very large human antibody libraries. Protein Eng Des Sel 23:155–159. doi:10.1093/protein/gzq002, gzq002 [pii] 6. Boder ET, Wittrup KD (2000) Yeast surface display for directed evolution of protein expression, affinity, and stability. Methods Enzymol 328:430–444 7. Cochran JR, Kim YS, Olsen MJ et al (2004) Domain-level antibody epitope mapping through yeast surface display of epidermal growth factor receptor fragments. J Immunol Methods 287:147–158
154
Jasdeep K. Mann and Sheldon Park
8. Levy R, Forsyth CM, LaPorte SL et al (2007) Fine and domain-level epitope mapping of botulinum neurotoxin type A neutralizing antibodies by yeast surface display. J Mol Biol 365:196– 210. doi:10.1016/j.jmb.2006.09.084, S00222836(06)01308-8 [pii] 9. Mann JK, Wood JF, Stephan AF et al (2013) Epitope-guided engineering of monobody binders for in vivo inhibition of Erk-2 signaling. ACS Chem Biol 8:608–616. doi:10.1021/ cb300579e 10. Bardwell AJ, Flatauer LJ, Matsukuma K et al (2001) A conserved docking site in MEKs mediates high-affinity binding to MAP kinases and cooperates with a scaffold protein to enhance signal transmission. J Biol Chem 276: 10374–10386. doi:10.1074/jbc.M010271200, M010271200 [pii] 11. Dimitri CA, Dowdle W, MacKeigan JP et al (2005) Spatially separate docking sites on
ERK2 regulate distinct signaling events in vivo. Curr Biol 15:1319–1324. doi:10.1016/j. cub.2005.06.037, S0960-9822(05)00672-X [pii] 12. Zhou T, Sun L, Humphreys J et al (2006) Docking interactions induce exposure of activation loop in the MAP kinase ERK2. Structure 14:1011–1019. doi:10.1016/j.str.2006.04.006, S0969-2126(06)00222-X [pii] 13. Hackel BJ, Ackerman ME, Howland SW et al (2010) Stability and CDR composition biases enrich binder functionality landscapes. J Mol Biol 401:84–96. doi:10.1016/j.jmb.2010. 06.004, S0022-2836(10)00604-2 [pii] 14. Hackel BJ, Kapila A, Wittrup KD (2008) Picomolar affinity fibronectin domains engineered utilizing loop length diversity, recursive mutagenesis, and loop shuffling. J Mol Biol 381:1238–1252. doi:10.1016/j.jmb.2008.06. 051, S0022-2836(08)00767-5 [pii]
1
Introduction Yeast surface display (YSD) is a versatile protein engineering tool and has been used to engineer various molecular properties, including stability, affinity, binding kinetics, and catalysis [1–3]. Key advantages of yeast display compared to other display platforms, especially phage display, include (1) improved folding of synthesized proteins and (2) normalization of activity with respect to expression in order to achieve quantitative screening of the displayed library. Because yeast is genetically amenable, a large library of mutants can be assembled and maintained [4, 5], which is critical to find binders with desired properties. To engineer binding affinity, the yeast cells displaying potential binders are screened with the target molecule, from which the highest affinity binders are isolated based on fluorescence or direct physical association [6]. The end result of a typical sort cycle is a collection of binders with improved expression and affinity for the target molecule. While some engineered binders may bind the target in ways that
Bin Liu (ed.), Yeast Surface Display: Methods, Protocols, and Applications, Methods in Molecular Biology, vol. 1319, DOI 10.1007/978-1-4939-2748-7_7, © Springer Science+Business Media New York 2015
143
144
Jasdeep K. Mann and Sheldon Park
interfere with the target function, more often the selected binders need to be characterized through additional rounds of epitope mapping or a functional screen to identify genuine inhibitors of function [7, 8]. This is because the binding affinity is typically engineered independent of the molecular details of interaction, and many binders are just that—binders and not inhibitors. As the size of the target molecule increases, an increasingly small fraction of engineered binders is expected to be functionally relevant. We recently described a YSD-based procedure for directly engineering epitope selectivity in order to streamline the discovery of functionally useful binders [9]. Our protocol involves the use of two alternative forms of the target molecule: wild type and a mutant containing surface substitution(s) at the desired epitope. The surface displayed library is then alternatively screened for binding to the wild type protein and for a lack of binding to the mutant protein. If the two molecules are structurally similar and differ only at the targeted surface, then the combination of positive and negative selections with wild type and mutant proteins will identify high affinity binders that are also epitope specific (Fig. 1).
Fig. 1 Schematic of the sorting strategy to engineer epitope-specific binders. (a) Yeast surface display library is assembled and expressed. (b, c) Using the target molecule for labeling, perform a series of positive sorting to identify the clones with the highest target affinity. (d) Construct a mutant protein containing one or more mutations at the desired target epitope. (e) Using the mutant protein, perform a negative sort to identify the binders whose interaction with the target protein is abrogated by the mutations. (f) The selected clones correspond to high affinity, epitope-specific binders
Engineering Epitope-Specific Binders
145
Given that disruption of ligand binding may disrupt signaling or catalysis, blocking functionally important sites of a signaling molecule or an enzyme with engineered binders should potentially inhibit the target molecule function. This strategy was used to engineer monobody binders of Erk-2 that inhibited its kinase activity by blocking a binding site used in intermolecular interaction [10–12]. Engineering epitope specificity by YSD thus offers an efficient method of discovering useful inhibitors of protein function.
2
Materials It is important that all reagents are prepared using molecular biology grade reagents in deionized water. Once prepared and sterilized, liquid reagents may be stored at 4 °C for several months. Both selective and nonselective yeast plates may be stored at 4 °C up to 1–2 months.
2.1 Yeast Culture Media Components 2.1.1 Nonselective Medium
Nonselective medium: yeast peptone dextrose (YPD) is used to propagate untransformed yeast. Growing transformed yeast in YPD will result in plasmid loss over time. 1. 10 g of yeast extract. 2. 20 g of peptone. 3. 20 g of dextrose. 4. Dissolve in 1 L water. Autoclave for 35 min at 121 °C. Store at 4 °C.
2.1.2 Selective Growth Medium
Selective growth medium: synthetic dextrose with Trp and Ura dropout (SD/-Trp-Ura). 1. 20 g of dextrose. 2. 6.7 g of yeast nitrogen base (without amino acid or ammonium sulfate). 3. 5 g ammonium sulfate. 4. 1.4 g of yeast synthetic dropout media mix lacking tryptophan, uracil, histidine, and leucine (from Sigma). The use of a quadruple dropout (-Trp,-Ura,-His,-Leu) as a base is convenient in case yeast needs to be cultured in other combinations of selection markers. By supplementing the mix with different nutrients, additional media can be prepared using common reagent. 5. 380 mg of leucine. 6. 76 mg of histidine. 7. 100 ml of 10× pH 6.0 or pH 4.5 media buffer (see below).
146
Jasdeep K. Mann and Sheldon Park
8. Add water to 1 L. To prepare a 5× stock, which is used to make selection agar plates, add water to 200 ml instead. 9. Sterilize using a 0.2 μm filter. Store at 4 °C (see Note 1). 2.1.3 Selective Induction Medium
Selective induction medium: synthetic galactose with Trp and Ura dropout (SG/-Trp-Ura). The only difference between the growth and induction media is the use of galactose in the induction medium. 1. 20 g of galactose. 2.–7. Rest of the components are same as in selective growth medium (see above). 8. Sterilize using a 0.2 μm filter and store at 4 °C.
2.1.4 Media Buffer
1. 10× sodium phosphate buffer at pH 6.0: This buffer is used to prepare selective yeast medium used to grow and propagate transformed yeast. To prepare the stock, dissolve 102 g of Na2HPO4·7H2O and 86 g of NaH2PO4·H2O in 1 L water. Sterilize by autoclaving before storage. 2. 10× sodium citrate buffer at pH 4.5: This buffer is used during cell sorting, which exposes the yeast culture to possible bacterial contamination. Low pH retards bacterial growth and reduces possible contamination. To prepare the stock, dissolve 14.7 g tri sodium citrate dihydrate (HOC(COONa) (CH2COONa)2·2H2O) and 4.29 g citric acid monohydrate (HOC(COOH)(CH2COOH)2·H2O) in 1 L water. Autoclave and store at 4 °C (see Note 2).
2.1.5 Yeast Selection Agar Plates
These plates are used to grow transformed yeast to identify individual yeast clones. 1. 180 g of sorbitol. 2. 15 g of bacto-agar. 3. Add water approximately to 700 ml and resuspend the solutes by stirring with a magnetic stir bar. 4. Add 100 ml of 10× media buffer (typically pH 6.0 is used). 5. Autoclave for 35 min at 121 °C. 6. Cool the autoclaved mixture while stirring until the temperature falls below 60 °C (warm to the touch). 7. Add 200 ml of filter sterilized 5× SD/-Trp-Ura stock. Adjust the volume to 1 L with sterile deionized water. 8. Slowly pour into 15 mm petri dishes (each plate holds ~25 ml of agar). Stack the plates together while the agar sets in order to minimize condensation on the lid. 9. Store the plates in a sealed bag at 4 °C. The plates are good for up to 2 months.
Engineering Epitope-Specific Binders
2.2 Target Protein Preparation
147
The target construction will vary from project to project. Here we describe the preparation of the Erk-2 kinase. 1. pRSET-Erk-2: a bacterial expression vector based on pRSET that contains the full-length rat Erk-2 gene downstream of the T7 promoter. NheI and BamHI restriction enzymes (New England Biolab) were used to clone the full length of Erk-2 between an N-terminal 6×histidine affinity tag and a C-terminal FLAG tag. 2. BL21(DE3) pLysS (Invitrogen): the cell line suppresses premature expression of the target molecule, which may be toxic to the cells. 3. Talon resin (Clontech), or other comparable immobilized metal affinity chromatography resin. 4. Imidazole (Sigma). 5. SDS-PAGE gel (Biorad): prepare 10–15 % acrylamide gel as needed. 6. Amicon centrifugal filters (Millipore): the filters are used for buffer exchange and concentration.
2.3 Magnetic Activated Cell Sorting (MACS)
1. EZ-Link Sulfo-NHS-Biotin (Thermo Fisher Scientific): this is used to chemically biotinylate purified Erk-2 so that it can be used with magnetic streptavidin resin to isolate the yeast displayed clones that bind the target. 2. Magnetic streptavidin-microbeads (Miltenyi Biotec). 3. MidiMACS Separator magnet (Miltenyi Biotec). 4. PBSB: Phosphate buffer saline plus 0.5 % (w/v) bovine serum albumin. 5. Kanamycin (Sigma). 6. Penicillin-Streptomycin (10,000 U/ml) (Life Technologies).
2.4 Yeast Surface Labeling for Flow Cytometry
1. Mouse monoclonal anti-cMyc antibody, 9E10 (AbD Serotec). 2. Mouse monoclonal anti-FLAG antibody, M2 (Agilent technologies). 3. Rabbit monoclonal anti-FLAG antibody (Sigma). 4. Anti-mouse IgG (whole molecule)–FITC antibody (Sigma). 5. Anti-mouse IgG (whole molecule)–R-phycoerythrin (PE) antibody (Sigma). 6. Anti-rabbit IgG (whole molecule)–FITC antibody (Sigma). 7. Alkaline phosphatase (AP)-conjugated goat anti-mouse antibody (Sigma). 8. Streptavidin-R-phycoerythrin, SA-PE (Biolegend). 9. PBSS-Phosphate buffer saline plus 0.1 % bovine serum albumin.
148
3
Jasdeep K. Mann and Sheldon Park
Methods
3.1 Library Construction
3.2 Growth and Induction of the Yeast Library
The yeast displayed fibronectin type 3 (Fn3) monobody library G4 was a gift from Dane Wittrup lab (MIT). The construction and characterization of the library was previously described [13, 14]. The library is assembled in the yeast vector pYD1 as a fusion with Aga2 and transformed into the yeast strain EBY100 (Invitrogen). 1. Thaw a frozen aliquot containing 2.5 × 109 cells or ~10-fold excess of the estimated library diversity, and inoculate into 500 ml of SD/-Trp-Ura culture medium. The starting OD600 ~ 0.5 (see Note 3). 2. Grow the cells at 30 °C with constant orbital shaking (250 rpm) overnight. The following morning, dilute the cells in 500 ml of fresh SD/-Trp-Ura medium to the starting OD600 = 0.5. 3. Continue to grow the cells in SD/-Trp-Ura at 30 °C until OD600 = 3. Centrifuge 100 ml of cells and remove the medium. Resuspend the cells in 300 ml SG/-Trp-Ura. Grow at 30 °C for 18 h with constant shaking to induce synthesis of the library (see Note 4).
3.3 Erk-2 Protein Purification
The protein purification steps will depend on the target protein. The following protocol describes the preparation of the Erk-2 kinase. 1. Transform pRSET-Erk-2 into BL21(DE3) pLysS and induce protein synthesis at the OD600 = 0.8 with 0.4 mM IPTG. Grow for 4 h at 37 °C. 2. Lyse the cells by sonication in a buffer containing 50 mM sodium phosphate (pH 8.0), 300 mM NaCl, 1 % triton X-100, 0.1 % 2-mercaptoethanol, 2 mM PMSF, and 10 % w/v glycerol. Mix the lysate with 2 ml of Talon resin. 3. Incubate the lysate and resin at 4 °C for 1 h with constant shaking. Wash the resin with 50 mM sodium phosphate (pH 7.2), 300 mM NaCl, and 20 mM imidazole. Repeat. 4. Elute the bound protein in three fractions of 1 ml of 50 mM sodium phosphate (pH 7.4), 300 mM NaCl, and 300 mM imidazole. Keep the elution volume as small as possible since the yield is low. 5. Concentrate the eluates using an Amicon centrifugal filter with the nominal cutoff of 10 kDa. Analyze the elution on a 12 % SDS-PAGE gel. 6. For biotinylation of Erk-2, incubate purified protein with tenfold excess of biotinylation reagent (EZ-Link Sulfo-NHSBiotin from Thermo Fisher Scientific) at 4 °C for 2 h. Remove unreacted biotin by buffer exchanging to PBS with a 30 kDa Amicon centrifugal filter.
Engineering Epitope-Specific Binders
3.4 Magnetic Bead Enrichment
149
1. Pellet 2.5 × 109 yeast cells (tenfold coverage of the library diversity) at 3,000 × g for 3 min. Wash the cells twice with 25 ml of PBSB and resuspend in 10 ml PBSB. 2. Incubate the cells with 1 μM biotinylated Erk-2 for 1 h at 20 °C (see Note 5). 3. Spin down the cells and wash twice with 25 ml PBSB. 4. Resuspend the pellet in 5 ml PBSB and incubate with 100 μl magnetic streptavidin coated microbeads for 15 min on ice. The yeast clones that display Erk-2 binding monobodies will now be decorated with streptavidin beads. 5. Separate the cells bound to the beads from the rest by using an LS column filled with ferromagnetic particles and a MidiMACS Separator magnet. Because the yeast particles covered with the beads are attracted to the magnet, they are selectively retained in the column compared to uncoated cells (see Note 6). 6. Wash the column with PBSB to remove the yeast cells that are not firmly bound to the beads. 7. Elute the streptavidin coated cells by removing the column from the magnet and washing the column in SD/-Trp-Ura (pH 4.5). 8. Grow the selected cells in citrate-based SD/-Trp-Ura (pH 4.5), supplemented with 50 μg/ml Kanamycin and penstrep (1:100) to retard the growth of bacteria. 9. Estimate the number of cells by plating serial dilutions of collected cells on SD/-Trp-Ura agar plates. Around ~0.1–1 % of the starting cells are typically retained in the fraction. For example, we obtained ~2 × 107 cells from the initial 2.5 × 109 cells corresponding to an estimate of 2 × 106 unique sequences.
3.5 High-Throughput Screen by Fluorescence Activated Cell Sorting (FACS)
1. Grow the cells until they have reached OD600 = 3. Dilute to OD600 = 0.5 and grow back to OD600 = 3. Induce protein synthesis by switching the medium to SG/-Trp-Ura. 2. Pellet the number of yeast cells corresponding to three- to tenfold excess of the estimated sequence diversity (e.g., 107 cells) and wash them with PBSS. 3. Label the cells with 500 nM biotinylated Erk-2 and anti-cMyc antibody (1:100 dilutions) at room temperature (i.e., around 20 °C) for 1 h (see Notes 7 and 8). 4. Wash the cells and incubate with SA-PE (1:100 dilution) and anti-mouse-FITC (1:50 dilution) on ice for 30 min in the dark. 5. Wash the cells and proceed to sorting by FACS. Because the average binders have weak affinity and likely dissociate quickly, it is important that the cells are analyzed as soon as possible after the final wash. Each FACS sorting cycle should not last more than 15 min. If additional cells need to be screened than can be analyzed in 15 min, a new batch of cells needs to be
150
Jasdeep K. Mann and Sheldon Park
washed and prepared every 15 min. Collect the clones with the most intense SA-PE labeling for a given expression level (e.g., roughly corresponding to 1–2 % of the analyzed cells) (see Notes 9 and 10). 6. Collect the selected cells in SD/-Trp-Ura (pH 4.5) containing kanamycin (50 μg/ml) and 1× pen-strep to prevent bacterial contamination. 7. Repeat the FACS analysis while gradually increasing the stringency of selection. For example, systematically decrease the concentration of biotinylated Erk-2 from 500 to 200, 100, 50, and 10 nM. To enrich the binding population, collect the top 1–2 % of the cells each time (see Note 11). 3.6 Epitope-Specific Selection with Mutant Protein
To identify the Fn3 clones with desired epitope specificity, perform negative selection with a mutant Erk-2 (see Note 12). The following factors need to be considered to design a mutant protein appropriate for the negative screen. First, the structure of the target protein should be consulted to make rational mutations. This is not as difficult a requirement to satisfy as it seems at first, since inhibitor design is typically attempted on proteins for which abundant biochemical and structural information is available. Second, the mutations should not interfere with protein folding or otherwise create significant structural perturbation to the protein. Because the mutated residues are usually located on the surface, it should be possible to introduce a mutation that does not cause significant structural perturbation. Finally, the engineered mutation should alter the chemical or physical properties at the mutated site. For example, the mutation may replace a hydrophobic residue with a charged residue or replace a small residue with a large residue. The substitutions should thus affect potential interaction at the mutated surface so that it can be used during negative selection. Keeping the above requirements in mind, we describe the design of the Erk-2 mutant used in our study. 1. Using the D-peptide bound Erk-2 structure (2GPH), design three mutations within the conserved docking site (“CD” domain), H123N, Y126H, and D319N, to disrupt the binding of a D-peptide. Introduce the mutations using mutagenic primers. Purify the mutant similarly as wild type. Verify that the overall structure of the protein remains the same by performing circular dichroism spectroscopy (Fig. 2a). 2. Incubate affinity optimized Fn3 clones on the yeast surface with 500 nM mutant Erk-2. A high concentration of Erk-2 is used to obtain clear separation between binding and nonbinding populations. 3. Collect the cells that do not bind mutant Erk-2 in order to enrich the Fn3 clones whose epitopes include mutated residues and are therefore likely to competitively inhibit the binding of
Engineering Epitope-Specific Binders
151
Fig. 2 Construction of mutant Erk-2 to engineer epitope specificity. (a) The circular dichroism spectra of wild type and mutant Erk-2 were compared to ensure that the mutations introduced at the CD domain do not affect the folding of Erk-2. (b) Functional testing to demonstrate that the engineered mutations prevent the binding of a D-peptide (bait). (c) An example of an engineered, epitope-specific monobody that binds wild type Erk-2 (left) but not mutant Erk-2 (right)
a D-peptide, whose binding is also disrupted by the mutation (Fig. 2b, c) (see Note 13). 4. Perform a final round of positive selection with wild type Erk-2 to identify the highest affinity clones with engineered epitope specificity (see Notes 14 and 15).
4
Notes 1. Synthetic yeast medium cannot be autoclaved because some components, e.g., glutamine, are heat labile and decompose during autoclaving. 2. When sorting a yeast library, the use of a low pH medium is strongly recommended to retard potential bacterial contamination. However, a low pH buffer alone does not guarantee contamination-free growth, and it is essential to practice good
152
Jasdeep K. Mann and Sheldon Park
sterile techniques at all time. We have observed that even with the addition of kanamycin and/or chloramphenicol the library can still get contaminated during a sort. Contaminated cultures are easy to spot based on visual inspection. Whereas yeast cells quickly settle to the bottom of a culture tube if left undisturbed, contaminated culture remains cloudy even after centrifugation. If the yeast culture becomes contaminated with bacteria and repeated sorting is not an option, it is possible to rid the culture of contamination by passing the cells through multiple centrifugation-wash cycles and propagating the cells in a low pH medium with antibiotics. 3. When working with a frozen yeast library, check cell viability by plating serial dilutions on SD agar plates. A plate containing a few hundred colonies gives the most reliable estimate of cell density. If the cells were frozen according to the established protocol (see ref. 4, for example), cell viability should be high and greater than 90 %. Low cell viability may reduce the sequence diversity in the library. 4. The cells should be freshly passaged just before induction in order to optimize protein expression. 5. Yeast cells are heavy and precipitate during long incubation. Keep the cells in suspension through gentle rocking. 6. Pipette the cells to avoid clumps. Magnetic separation should be done promptly at 4 °C to avoid dissociation of bound Erk-2. 7. The exact temperature of incubation is not important. For convenience, we leave the cells on the bench during the first labeling step. Subsequent labeling is done on ice to minimize dissociation of bound ligand. 8. Make sure that Erk-2 is used in stoichiometric excess compared to the number of monobody molecules on the yeast surface so that the binding does not significantly change the free protein concentration. The number of displayed molecules is estimated to be ~50,000 per cell. 9. Calibrate the compensation using cells labeled with 9E10 and secondary antibody conjugated with FITC or PE. For a double-labeled control, we express an engineered monobody that binds maltose binding protein (MBP) and label the cells with biotinylated MBP and SA-PE as well as 9E10 and secondary antibody. For negative controls, Fn3 displaying cells are labeled with secondary antibodies only. 10. Selecting a much smaller than 1 % of the population may collapse the diversity of the library too quickly and run the risk of losing useful binders that may not have the highest affinity. 11. The cells are initially labeled with SA-PE. However, the use of biotinylated Erk-2 introduces a bias. Some of the selected clones may also bind the FLAG tag rather than Erk-2. To
Engineering Epitope-Specific Binders
153
identify the clones that bind Erk-2, label bound Erk-2 with anti-FLAG antibody (M2) starting from Round 5. M2 labeling is not used until a later round because the Erk-2 affinity of the Fn3 clones is on average low at first and many monobodies may dissociate from Erk-2 during the time it takes for antibody labeling. 12. Negative selection should preferably be done after several rounds of affinity maturation, since it is difficult to differentiate between weak binders and nonbinders. 13. A gate should be drawn during the negative sort to minimize contamination of the clones that also bind the mutant protein. Because a high concentration of the mutant protein is used for labeling (e.g., 500 nM of mutant compared to 10 nM wild type during the preceding positive selection), the gate can be somewhat generous and still bias the clones based on different binding to wild type and mutant proteins. 14. If necessary, perform one or more rounds of positive sort following the negative sort to optimize the binding affinity. It is recommended that the negative sort is introduced toward the end of the selection cycle but not as the last selection step. 15. The individual clones must be tested against wild type and mutant proteins to demonstrate epitope-specific binding before functional testing is attempted. Only a small percentage of the selected clones are expected to bind both variants.
Acknowledgements This work was supported by the NSF grant (1053608) to S.P. References 1. Feldhaus MJ, Siegel RW (2004) Yeast display of antibody fragments: a discovery and characterization platform. J Immunol Methods 290:69–80 2. Gai SA, Wittrup KD (2007) Yeast surface display for protein engineering and characterization. Curr Opin Struct Biol 17:467–473. doi:10.1016/j.sbi.2007.08.012, S0959440X(07)00119-4 [pii] 3. Pepper LR, Cho YK, Boder ET et al (2008) A decade of yeast surface display technology: where are we now? Comb Chem High Throughput Screen 11:127–134 4. Chao G, Lau WL, Hackel BJ et al (2006) Isolating and engineering human antibodies using yeast surface display. Nat Protoc 1:
755–768. doi:10.1038/nprot.2006.94, nprot. 2006.94 [pii] 5. Benatuil L, Perez JM, Belk J et al (2010) An improved yeast transformation method for the generation of very large human antibody libraries. Protein Eng Des Sel 23:155–159. doi:10.1093/protein/gzq002, gzq002 [pii] 6. Boder ET, Wittrup KD (2000) Yeast surface display for directed evolution of protein expression, affinity, and stability. Methods Enzymol 328:430–444 7. Cochran JR, Kim YS, Olsen MJ et al (2004) Domain-level antibody epitope mapping through yeast surface display of epidermal growth factor receptor fragments. J Immunol Methods 287:147–158
154
Jasdeep K. Mann and Sheldon Park
8. Levy R, Forsyth CM, LaPorte SL et al (2007) Fine and domain-level epitope mapping of botulinum neurotoxin type A neutralizing antibodies by yeast surface display. J Mol Biol 365:196– 210. doi:10.1016/j.jmb.2006.09.084, S00222836(06)01308-8 [pii] 9. Mann JK, Wood JF, Stephan AF et al (2013) Epitope-guided engineering of monobody binders for in vivo inhibition of Erk-2 signaling. ACS Chem Biol 8:608–616. doi:10.1021/ cb300579e 10. Bardwell AJ, Flatauer LJ, Matsukuma K et al (2001) A conserved docking site in MEKs mediates high-affinity binding to MAP kinases and cooperates with a scaffold protein to enhance signal transmission. J Biol Chem 276: 10374–10386. doi:10.1074/jbc.M010271200, M010271200 [pii] 11. Dimitri CA, Dowdle W, MacKeigan JP et al (2005) Spatially separate docking sites on
ERK2 regulate distinct signaling events in vivo. Curr Biol 15:1319–1324. doi:10.1016/j. cub.2005.06.037, S0960-9822(05)00672-X [pii] 12. Zhou T, Sun L, Humphreys J et al (2006) Docking interactions induce exposure of activation loop in the MAP kinase ERK2. Structure 14:1011–1019. doi:10.1016/j.str.2006.04.006, S0969-2126(06)00222-X [pii] 13. Hackel BJ, Ackerman ME, Howland SW et al (2010) Stability and CDR composition biases enrich binder functionality landscapes. J Mol Biol 401:84–96. doi:10.1016/j.jmb.2010. 06.004, S0022-2836(10)00604-2 [pii] 14. Hackel BJ, Kapila A, Wittrup KD (2008) Picomolar affinity fibronectin domains engineered utilizing loop length diversity, recursive mutagenesis, and loop shuffling. J Mol Biol 381:1238–1252. doi:10.1016/j.jmb.2008.06. 051, S0022-2836(08)00767-5 [pii]
