Advertisement

Deep Mutational Scans as a Guide to Engineering High Affinity T Cell Receptor Interactions with Peptide-bound Major Histocompatibility Complex*

  • Daniel T. Harris
    Affiliations
    From the Department of Biochemistry, University of Illinois, Urbana, Illinois 61801 and
    Search for articles by this author
  • Ningyan Wang
    Affiliations
    From the Department of Biochemistry, University of Illinois, Urbana, Illinois 61801 and
    Search for articles by this author
  • Author Footnotes
    1 Supported by a fellowship from the Indiana CTSI, funded in part by National Institutes of Health Grant UL1TR001108.
    Timothy P. Riley
    Footnotes
    1 Supported by a fellowship from the Indiana CTSI, funded in part by National Institutes of Health Grant UL1TR001108.
    Affiliations
    the Department of Chemistry and Biochemistry and the Harper Cancer Research Institute, University of Notre Dame, South Bend, Indiana 46557
    Search for articles by this author
  • Scott D. Anderson
    Affiliations
    From the Department of Biochemistry, University of Illinois, Urbana, Illinois 61801 and
    Search for articles by this author
  • Nishant K. Singh
    Affiliations
    the Department of Chemistry and Biochemistry and the Harper Cancer Research Institute, University of Notre Dame, South Bend, Indiana 46557
    Search for articles by this author
  • Erik Procko
    Affiliations
    From the Department of Biochemistry, University of Illinois, Urbana, Illinois 61801 and
    Search for articles by this author
  • Brian M. Baker
    Affiliations
    the Department of Chemistry and Biochemistry and the Harper Cancer Research Institute, University of Notre Dame, South Bend, Indiana 46557
    Search for articles by this author
  • David M. Kranz
    Correspondence
    To whom correspondence should be addressed: Dept. of Biochemistry, University of Illinois, 600 S. Matthews Ave., Urbana, IL 61801. Tel.: 217-244-2821; E-mail: .
    Affiliations
    From the Department of Biochemistry, University of Illinois, Urbana, Illinois 61801 and
    Search for articles by this author
  • Author Footnotes
    * This work was supported by National Institutes of Health Grants CA178844 and CA187592 (to D. M. K.), GM118166 (to B. M. B.), and CA180723 (to D. T. H.). The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
    1 Supported by a fellowship from the Indiana CTSI, funded in part by National Institutes of Health Grant UL1TR001108.
Open AccessPublished:September 28, 2016DOI:https://doi.org/10.1074/jbc.M116.748681

      Abstract

      Proteins are often engineered to have higher affinity for their ligands to achieve therapeutic benefit. For example, many studies have used phage or yeast display libraries of mutants within complementarity-determining regions to affinity mature antibodies and T cell receptors (TCRs). However, these approaches do not allow rapid assessment or evolution across the entire interface. By combining directed evolution with deep sequencing, it is now possible to generate sequence fitness landscapes that survey the impact of every amino acid substitution across the entire protein-protein interface. Here we used the results of deep mutational scans of a TCR-peptide-MHC interaction to guide mutational strategies. The approach yielded stable TCRs with affinity increases of >200-fold. The substitutions with the greatest enrichments based on the deep sequencing were validated to have higher affinity and could be combined to yield additional improvements. We also conducted in silico binding analyses for every substitution to compare them with the fitness landscape. Computational modeling did not effectively predict the impacts of mutations distal to the interface and did not account for yeast display results that depended on combinations of affinity and protein stability. However, computation accurately predicted affinity changes for mutations within or near the interface, highlighting the complementary strengths of computational modeling and yeast surface display coupled with deep mutational scanning for engineering high affinity TCRs.
      The process of increasing the affinity of a protein occurs naturally with antibodies, where somatic mutation within the variable region genes is followed by antigen-driven selection of B cells that express membrane-bound antibodies. In contrast, T cell receptors (TCRs)
      The abbreviations used are: TCR, T cell receptor; pepMHC, peptide-MHC; CDR, complementarity-determining regions; scFv, single-chain fragment variable; MFI, mean fluorescence intensity; SOE, splice overlap extension; RMSF, root mean square fluctuation.
      3The abbreviations used are: TCR, T cell receptor; pepMHC, peptide-MHC; CDR, complementarity-determining regions; scFv, single-chain fragment variable; MFI, mean fluorescence intensity; SOE, splice overlap extension; RMSF, root mean square fluctuation.
      do not undergo somatic mutations and bind to their antigen, a peptide-MHC (pepMHC), with low (micromolar) affinities. However, improvements in TCR affinity to the same levels of antibodies can be achieved by in vitro approaches involving the generation of mutant TCR libraries followed by antigen selection (
      • Foote J.
      • Eisen H.N.
      Breaking the affinity ceiling for antibodies and T cell receptors.
      ,
      • Holler P.D.
      • Holman P.O.
      • Shusta E.V.
      • O'Herrin S.
      • Wittrup K.D.
      • Kranz D.M.
      In vitro evolution of a T cell receptor with high affinity for peptide/MHC.
      • Li Y.
      • Moysey R.
      • Molloy P.E.
      • Vuidepot A.L.
      • Mahon T.
      • Baston E.
      • Dunn S.
      • Liddy N.
      • Jacob J.
      • Jakobsen B.K.
      • Boulter J.M.
      Directed evolution of human T-cell receptors with picomolar affinities by phage display.
      ).
      For therapeutic purposes, the affinity of a variety of protein-protein interactions, and especially antibody-antigen interactions, has been enhanced using in vitro directed evolution approaches, including phage, yeast, ribosomal, and mammalian display (e.g. see Refs.
      • Huse W.D.
      • Sastry L.
      • Iverson S.A.
      • Kang A.S.
      • Alting-Mees M.
      • Burton D.R.
      • Benkovic S.J.
      • Lerner R.A.
      Generation of a large combinatorial library of the immunoglobulin repertoire in phage λ.
      • Hanes J.
      • Plückthun A.
      In vitro selection and evolution of functional proteins by using ribosome display.
      ,
      • Boder E.T.
      • Wittrup K.D.
      Yeast surface display for directed evolution of protein expression, affinity, and stability.
      • Chervin A.S.
      • Aggen D.H.
      • Raseman J.M.
      • Kranz D.M.
      Engineering higher affinity T cell receptors using a T cell display system.
      ). These methods rely on the generation of large libraries of mutants at residues within the protein-protein interface, followed by several rounds of selection for desired parameters (such as affinity, stability, and expression levels) (
      • Hoogenboom H.R.
      Selecting and screening recombinant antibody libraries.
      ,
      • Van Deventer J.A.
      • Wittrup K.D.
      Yeast surface display for antibody isolation: library construction, library screening, and affinity maturation.
      ).
      Although directed evolution using larger degenerate libraries has been successful, the most recent techniques involving deep sequencing of single-codon libraries have the potential both to provide mechanistic structural information about a binding site and at the same time to provide leads for affinity improvements. Sequence fitness landscapes have successfully been utilized to map protein-DNA interactions (
      • Shultzaberger R.K.
      • Malashock D.S.
      • Kirsch J.F.
      • Eisen M.B.
      The fitness landscapes of cis-acting binding sites in different promoter and environmental contexts.
      ), protein-peptide interactions (
      • Fowler D.M.
      • Araya C.L.
      • Fleishman S.J.
      • Kellogg E.H.
      • Stephany J.J.
      • Baker D.
      • Fields S.
      High-resolution mapping of protein sequence-function relationships.
      ), and protein-protein interactions (
      • Procko E.
      • Hedman R.
      • Hamilton K.
      • Seetharaman J.
      • Fleishman S.J.
      • Su M.
      • Aramini J.
      • Kornhaber G.
      • Hunt J.F.
      • Tong L.
      • Montelione G.T.
      • Baker D.
      Computational design of a protein-based enzyme inhibitor.
      ). Furthermore, using a PDZ protein domain as a model system, McLaughlin and colleagues were able to manipulate ligand-binding specificity through key mutations identified using sequence fitness landscapes (
      • McLaughlin Jr, R.N.
      • Poelwijk F.J.
      • Raman A.
      • Gosal W.S.
      • Ranganathan R.
      The spatial architecture of protein function and adaptation.
      ). Additionally, a sequence fitness landscape of an influenza-binding protein inhibitor was used to enhance affinity and redirect specificity toward a single H1 hemagglutinin subtype (
      • Whitehead T.A.
      • Chevalier A.
      • Song Y.
      • Dreyfus C.
      • Fleishman S.J.
      • De Mattos C.
      • Myers C.A.
      • Kamisetty H.
      • Blair P.
      • Wilson I.A.
      • Baker D.
      Optimization of affinity, specificity and function of designed influenza inhibitors using deep sequencing.
      ).
      A major goal in protein engineering is to be able to accurately identify mutations that yield improvements in stability or affinity. In addition to directed evolution and sequence fitness landscapes, when the protein structures are known, structure-based computational design has also been used to achieve these goals. Although there have been inspiring successes (reviewed in Ref.
      • Schreiber G.
      • Fleishman S.J.
      Computational design of protein-protein interactions.
      ), advances in computational approaches require a thorough understanding of the relationships between protein structural and physical properties as well as increases in the ability to rapidly and accurately sample different conformational and configurational states (
      • Das R.
      Four small puzzles that Rosetta doesn't solve.
      ).
      In the present study, we focused on TCRs because they have evolved to bind to a diverse repertoire of clinically relevant targets and thus represent a class of molecules with significant therapeutic potential. In addition, because of their naturally low affinities, they represent protein engineering targets for both stability and affinity. Previously, we reported successful affinity engineering of TCRs by directed evolution using yeast display (
      • Holler P.D.
      • Holman P.O.
      • Shusta E.V.
      • O'Herrin S.
      • Wittrup K.D.
      • Kranz D.M.
      In vitro evolution of a T cell receptor with high affinity for peptide/MHC.
      ,
      • Holler P.D.
      • Chlewicki L.K.
      • Kranz D.M.
      TCRs with high affinity for foreign pMHC show self-reactivity.
      • Weber K.S.
      • Donermeyer D.L.
      • Allen P.M.
      • Kranz D.M.
      Class II-restricted T cell receptor engineered in vitro for higher affinity retains peptide specificity and function.
      ,
      • Aggen D.H.
      • Chervin A.S.
      • Insaidoo F.K.
      • Piepenbrink K.H.
      • Baker B.M.
      • Kranz D.M.
      Identification and engineering of human variable regions that allow expression of stable single-chain T cell receptors.
      ,
      • Smith S.N.
      • Sommermeyer D.
      • Piepenbrink K.H.
      • Blevins S.J.
      • Bernhard H.
      • Uckert W.
      • Baker B.M.
      • Kranz D.M.
      Plasticity in the contribution of T cell receptor variable region residues to binding of peptide-HLA-A2 complexes.
      • Smith S.N.
      • Wang Y.
      • Baylon J.L.
      • Singh N.K.
      • Baker B.M.
      • Tajkhorshid E.
      • Kranz D.M.
      Changing the peptide specificity of a human T-cell receptor by directed evolution.
      ) and mammalian cell display (
      • Chervin A.S.
      • Aggen D.H.
      • Raseman J.M.
      • Kranz D.M.
      Engineering higher affinity T cell receptors using a T cell display system.
      ,
      • Schmitt T.M.
      • Aggen D.H.
      • Stromnes I.M.
      • Dossett M.L.
      • Richman S.A.
      • Kranz D.M.
      • Greenberg P.D.
      Enhanced-affinity murine T-cell receptors for tumor/self-antigens can be safe in gene therapy despite surpassing the threshold for thymic selection.
      ). We also described structure-guided design strategies that estimated the binding energies of both favorable and unfavorable mutations and led to the design of additional high affinity TCRs (
      • Pierce B.G.
      • Hellman L.M.
      • Hossain M.
      • Singh N.K.
      • Vander Kooi C.W.
      • Weng Z.
      • Baker B.M.
      Computational design of the affinity and specificity of a therapeutic T cell receptor.
      ,
      • Riley T.P.
      • Ayres C.M.
      • Hellman L.M.
      • Singh N.K.
      • Cosiano M.
      • Cimons J.M.
      • Anderson M.J.
      • Piepenbrink K.H.
      • Pierce B.G.
      • Weng Z.
      • Baker B.M.
      A generalized framework for computational design and mutational scanning of T cell receptor binding interfaces.
      ).
      More recently, we reported the use of single-codon libraries with two different TCRs to generate sequence fitness landscapes that allowed analysis of the impact of each residue on binding to their cognate peptide·HLA-A2 complexes (
      • Harris D.T.
      • Singh N.K.
      • Cai Q.
      • Smith S.N.
      • Vander Kooi C.W.
      • Procko E.
      • Kranz D.M.
      • Baker B.M.
      An engineered switch in T cell receptor specificity leads to an unusual but functional binding geometry.
      ). Sequence fitness landscapes offer a powerful perspective on protein-protein interactions not available from structural data alone by experimentally determining, on a residue-by-residue basis, which amino acids contribute to binding as well as the optimal amino acids at each position (
      • Fowler D.M.
      • Araya C.L.
      • Fleishman S.J.
      • Kellogg E.H.
      • Stephany J.J.
      • Baker D.
      • Fields S.
      High-resolution mapping of protein sequence-function relationships.
      ,
      • Araya C.L.
      • Fowler D.M.
      Deep mutational scanning: assessing protein function on a massive scale.
      ). Accordingly, the two higher affinity TCRs, A6-c134 and RD1-MART1HIGH, that are specific for Tax·HLA-A2 and MART1·HLA-A2, respectively, were examined both structurally and by deep scanning mutagenesis to determine the basis of specificity and binding (
      • Harris D.T.
      • Singh N.K.
      • Cai Q.
      • Smith S.N.
      • Vander Kooi C.W.
      • Procko E.
      • Kranz D.M.
      • Baker B.M.
      An engineered switch in T cell receptor specificity leads to an unusual but functional binding geometry.
      ).
      Here we further studied the mutations that were highly enriched in the sequence fitness landscape of the cancer antigen-specific TCR, RD1-MART1HIGH, interacting with its target peptide·MHC by yeast surface display. We demonstrate that the mutations that exhibited the highest levels of enrichment acted both individually and in synergy to significantly increase the affinity and yeast surface levels of the RD1-MART1HIGH TCR. We also compared strategies that involved individual combinations of mutations versus construction and selection of multicodon libraries, both based on the sequence fitness landscape. These approaches were further compared with computationally predicted affinity-enhancing mutations using a previously described in silico approach (
      • Riley T.P.
      • Ayres C.M.
      • Hellman L.M.
      • Singh N.K.
      • Cosiano M.
      • Cimons J.M.
      • Anderson M.J.
      • Piepenbrink K.H.
      • Pierce B.G.
      • Weng Z.
      • Baker B.M.
      A generalized framework for computational design and mutational scanning of T cell receptor binding interfaces.
      ,
      • Riley T.P.
      • Singh N.K.
      • Pierce B.G.
      • Weng Z.
      • Baker B.M.
      Computational modeling of T cell receptor complexes.
      ). Some of the most highly enriched residues identified in the sequence fitness landscape were successfully identified by the in silico approach, but these were not distinguished from many other substitutions that did not yield substantial enrichment. This is probably due in part to the sensitivity of yeast surface display to not only changes in binding but also changes in protein stability. The in silico approach nonetheless performed well when focused on mutations experimentally shown to improve binding and proved advantageous for providing structural interpretations. Overall, we show that deep sequencing methods combined with yeast display provides for significant opportunities for enhancing TCR affinity while also controlling for impacts on stability and demonstrate the utility of computational modeling in providing structural interpretations for affinity gains.

      Discussion

      Proteins often require affinity and stability enhancements for therapeutic purposes. To achieve the desired characteristics, various protein-engineering techniques that use in vitro directed evolution have been employed. These include ribosomal, phage, bacterial, yeast, and mammalian display of degenerate mutational libraries, followed by various selection schemes. More recently, it has been possible to conduct deep mutational scans of every residue in a protein's binding site to construct a sequence fitness landscape. Here we used information from the sequence fitness landscape of a TCR (RD1-MART1HIGH) specific for the cancer antigen MART1·HLA-A2 as a guide to compare various approaches to engineer improvements in the affinity and/or stability. We were able to identify specific mutants that enhanced the affinity >200-fold and increased the yeast surface level expression by 6-fold. Furthermore, we compared these site-specific mutants with structure-based predictions using an in silico approach that modeled the effects of the mutations and predicted the impacts on binding.
      To determine the robustness of identifying higher affinity mutations strictly from the results of the sequence fitness landscape, we generated single site-specific mutations with enrichment values ranging from 10- to 51-fold. These mutations were tested as individual and combined mutants. Site-specific mutagenesis was able to increase the TCR affinity by 100-fold when three of the highest affinity mutants were combined. A previous study that aimed to improve the affinity of an influenza inhibitor reported a maximum affinity improvement of 28-fold using a combination of eight mutations identified from a sequence fitness landscape (
      • Whitehead T.A.
      • Chevalier A.
      • Song Y.
      • Dreyfus C.
      • Fleishman S.J.
      • De Mattos C.
      • Myers C.A.
      • Kamisetty H.
      • Blair P.
      • Wilson I.A.
      • Baker D.
      Optimization of affinity, specificity and function of designed influenza inhibitors using deep sequencing.
      ). In our study, when enrichment values were >12, we observed a direct correlation with enrichment value and affinity enhancement. In addition to affinity increases, the yeast display approach also selects for TCRs that are expressed at higher levels due to the increased stability of the mutant (
      • Weber K.S.
      • Donermeyer D.L.
      • Allen P.M.
      • Kranz D.M.
      Class II-restricted T cell receptor engineered in vitro for higher affinity retains peptide specificity and function.
      ,
      • Kieke M.C.
      • Shusta E.V.
      • Boder E.T.
      • Teyton L.
      • Wittrup K.D.
      • Kranz D.M.
      Selection of functional T cell receptor mutants from a yeast surface-display library.
      ,
      • Shusta E.V.
      • Kieke M.C.
      • Parke E.
      • Kranz D.M.
      • Wittrup K.D.
      Yeast polypeptide fusion surface display levels predict thermal stability and soluble secretion efficiency.
      • Orr B.A.
      • Carr L.M.
      • Wittrup K.D.
      • Roy E.J.
      • Kranz D.M.
      Rapid method for measuring ScFv thermal stability by yeast surface display.
      ). Accordingly, the present study identified mutations that increased surface levels up to about 6-fold. Similar observations have been made about the role of residues in the thermal stability of antibody scFv fragments (
      • Orr B.A.
      • Carr L.M.
      • Wittrup K.D.
      • Roy E.J.
      • Kranz D.M.
      Rapid method for measuring ScFv thermal stability by yeast surface display.
      ,
      • Honegger A.
      • Malebranche A.D.
      • Röthlisberger D.
      • Plückthun A.
      The influence of the framework core residues on the biophysical properties of immunoglobulin heavy chain variable domains.
      ,
      • Miller B.R.
      • Demarest S.J.
      • Lugovskoy A.
      • Huang F.
      • Wu X.
      • Snyder W.B.
      • Croner L.J.
      • Wang N.
      • Amatucci A.
      • Michaelson J.S.
      • Glaser S.M.
      Stability engineering of scFvs for the development of bispecific and multivalent antibodies.
      ).
      Although the individual mutations with the highest enhancement values in the sequence fitness landscape yielded higher affinity, and their combinations provided even higher affinity (100-fold), we were interested in determining whether a directed evolution approach based on the same data could yield combined mutations with even higher affinity or stability. This combinatorial approach would include variants with multiple mutations that have additivity or positive cooperativity in binding. To test this, we generated a combinatorial library at the four most highly enriched residues (Gly-28α, Thr-91α, Val-50β, and Ala-99β). Three successive selections of this library yielded multiple clones that bound with higher affinity than RD1-MART1HIGH to the MART1·HLA-A2 ligand. One of these clones (S3-2) had the highest affinity (700 pm) and greatest surface level (6-fold above RD1-MART1HIGH) among all of the mutants examined in the present study. The sequences of this and the other two unique clones revealed that three of the four positions in the library had mutations that were highly enriched in the sequence fitness landscape. Unexpectedly, the Gly-28α position had two clones, including S3-2, with an arginine substitution, whereas the G28Rα substitution was not enriched in the sequence fitness landscape. Because G28Rα is predicted to interact strongly with Glu-1 of the peptide as well as Glu-58 of HLA-A2, this mutation alone may destabilize the TCR, a consequence offset by the other mutations. Accordingly, the use of sequence fitness landscapes to generate higher affinity mutants can benefit by subsequent selections of combinatorial libraries in the codons of enriched amino acid positions.
      Our recently described structure-guided computational design approach for engineering TCRs (
      • Riley T.P.
      • Ayres C.M.
      • Hellman L.M.
      • Singh N.K.
      • Cosiano M.
      • Cimons J.M.
      • Anderson M.J.
      • Piepenbrink K.H.
      • Pierce B.G.
      • Weng Z.
      • Baker B.M.
      A generalized framework for computational design and mutational scanning of T cell receptor binding interfaces.
      ,
      • Riley T.P.
      • Singh N.K.
      • Pierce B.G.
      • Baker B.M.
      • Weng Z.
      Computational reprogramming of T cell antigen receptor binding properties.
      ) provided an opportunity to evaluate the sequence fitness landscapes computationally and compare enrichment with predicted impacts on binding. Although there was qualitative agreement between those mutations predicted to weaken binding and sequence depletion, the in silico approach did not distinguish well between mutations predicted to improve binding and sequence enrichment, because 25% of mutations selected against were predicted to enhance binding. Much of this may reflect the sensitivity of yeast surface display to changes in protein stability, resulting in some substitutions that are either reduced or enhanced in surface levels. This is supported by our analysis of the modeled structures of each mutant for changes in exposed hydrophobic surface area; sequences enriched showed a much stronger trend to reduce hydrophobic solvent-accessible surface area, and vice versa. There can be a tendency for structure-guided design to select for hydrophobic mutations (
      • Procko E.
      • Hedman R.
      • Hamilton K.
      • Seetharaman J.
      • Fleishman S.J.
      • Su M.
      • Aramini J.
      • Kornhaber G.
      • Hunt J.F.
      • Tong L.
      • Montelione G.T.
      • Baker D.
      Computational design of a protein-based enzyme inhibitor.
      ,
      • Riley T.P.
      • Ayres C.M.
      • Hellman L.M.
      • Singh N.K.
      • Cosiano M.
      • Cimons J.M.
      • Anderson M.J.
      • Piepenbrink K.H.
      • Pierce B.G.
      • Weng Z.
      • Baker B.M.
      A generalized framework for computational design and mutational scanning of T cell receptor binding interfaces.
      ), and our findings reiterate that improvements in in silico approaches may be found from continued attention to electrostatic features, such as polar solvation, hydrogen bonding, and interactions with formal charges (
      • Das R.
      Four small puzzles that Rosetta doesn't solve.
      ,
      • Nielsen J.E.
      • Gunner M.R.
      • García-Moreno B.E.
      The pKa Cooperative: a collaborative effort to advance structure-based calculations of pKa values and electrostatic effects in proteins.
      • Li Z.
      • Yang Y.
      • Zhan J.
      • Dai L.
      • Zhou Y.
      Energy functions in de novo protein design: current challenges and future prospects.
      ,
      • Fleishman S.J.
      • Whitehead T.A.
      • Strauch E.M.
      • Corn J.E.
      • Qin S.
      • Zhou H.X.
      • Mitchell J.C.
      • Demerdash O.N.
      • Takeda-Shitaka M.
      • Terashi G.
      • Moal I.H.
      • Li X.
      • Bates P.A.
      • Zacharias M.
      • Park H.
      • et al.
      Community-wide assessment of protein-interface modeling suggests improvements to design methodology.
      • Stranges P.B.
      • Kuhlman B.
      A comparison of successful and failed protein interface designs highlights the challenges of designing buried hydrogen bonds.
      ).
      For those mutants studied directly, our in silico analysis showed very good agreement between predicted and impacted effects on binding, permitting the use of the structural models in helping to assess how the mutations acted to improve binding. A key observation was that multiple “second shell” mutations appeared to be enriched, and these most likely act via indirect or long range effects (exemplified by the V50Dβ mutation). Such sites are not always considered in structure-guided computational design, highlighting a strength of sequence fitness landscapes and yeast surface display as well as further suggesting how in silico methods might be extended.
      The single-codon approach used with sequence fitness landscapes also improves upon the many early studies using directed evolution and error-prone PCR techniques. Although the latter can sample the entire protein interface, it is limited to single-site mutations that are generated by a single base substitution. This is because the probability of having two substitutions in the same codon or in two codons with beneficial substitutions is quite low and often beyond the size of the libraries (or detrimental mutations at a higher error rate would obscure these mutations). As an example, the A99Yα mutant was one of the most highly enriched mutations of RD1-MART1HIGH, but it would require two nucleotide substitutions to mutate from alanine to tyrosine.
      Finally, it is worth considering whether information about binding affinities of mutants gained in the yeast display system could be extrapolated to these TCRs expressed in their normal context, T cells. The TCRs are in single-chain form (Vβ-linker-Vα) as Aga2 fusions for yeast display, whereas in T cells, the Vα and Vβ each contain transmembrane-spanning constant regions. Despite these differences, we have invariably observed with other TCRs that the mutations yielding higher affinity in the yeast display system also yielded higher affinity when used as full-length TCRs transferred into T cells (
      • Weber K.S.
      • Donermeyer D.L.
      • Allen P.M.
      • Kranz D.M.
      Class II-restricted T cell receptor engineered in vitro for higher affinity retains peptide specificity and function.
      ,
      • Stone J.D.
      • Artyomov M.N.
      • Chervin A.S.
      • Chakraborty A.K.
      • Eisen H.N.
      • Kranz D.M.
      Interaction of streptavidin-based peptide-MHC oligomers (tetramers) with cell-surface TCRs.
      ,
      • Holler P.D.
      • Kranz D.M.
      Quantitative analysis of the contribution of TCR/pepMHC affinity and CD8 to T cell activation.
      ,
      • Chervin A.S.
      • Stone J.D.
      • Holler P.D.
      • Bai A.
      • Chen J.
      • Eisen H.N.
      • Kranz D.M.
      The impact of TCR-binding properties and antigen presentation format on T cell responsiveness.
      ). Thus, although not formally tested, we have reason to believe that the mutations of RD1-MARTHIGH TCRs identified in the present work would exhibit binding with enhanced affinities when transferred to the full-length TCR context.
      In conclusion, deep mutational scans of the entire protein-protein interface provide physical insights into binding, and they will allow improvements in structure-guided in silico analyses. Sequence fitness landscapes also serve as a robust guide for conducting site-specific mutagenesis to enhance affinity. In the future, one might be able to simply combine all of the top enriched mutations into a single mutant and thus bypass the need for any selections. This might be especially true where the residues are located at different locations in the interface. However, the combinatorial library/selection approach, based on the results of the sequence fitness landscape, yielded an improved mutant compared with the single-site mutant strategy. Thus, there may still be advantages to using a directed evolution, selection-based strategy for optimizing improvements.

      Author Contributions

      D. T. H., N. W., T. P. R., S. D. A., and N. K. S. performed experiments and analyzed data. D. M. K., E. P., and B. M. B. conceived and organized the project. D. T. H., N. W., T. P. R., S. D. A., N. K. S., E. P., B. M. B., and D. M. K. wrote the manuscript. All authors reviewed the results and approved the final version of the manuscript.

      Acknowledgments

      We thank Barbara Pilas and Alvaro Hernandez for assistance with flow cytometry and deep sequencing, respectively, and Cory Ayres for performing molecular dynamics simulations.

      References

        • Foote J.
        • Eisen H.N.
        Breaking the affinity ceiling for antibodies and T cell receptors.
        Proc. Natl. Acad. Sci. U.S.A. 2000; 97: 10679-10681
        • Holler P.D.
        • Holman P.O.
        • Shusta E.V.
        • O'Herrin S.
        • Wittrup K.D.
        • Kranz D.M.
        In vitro evolution of a T cell receptor with high affinity for peptide/MHC.
        Proc. Natl. Acad. Sci. U.S.A. 2000; 97: 5387-5392
        • Li Y.
        • Moysey R.
        • Molloy P.E.
        • Vuidepot A.L.
        • Mahon T.
        • Baston E.
        • Dunn S.
        • Liddy N.
        • Jacob J.
        • Jakobsen B.K.
        • Boulter J.M.
        Directed evolution of human T-cell receptors with picomolar affinities by phage display.
        Nat. Biotechnol. 2005; 23: 349-354
        • Huse W.D.
        • Sastry L.
        • Iverson S.A.
        • Kang A.S.
        • Alting-Mees M.
        • Burton D.R.
        • Benkovic S.J.
        • Lerner R.A.
        Generation of a large combinatorial library of the immunoglobulin repertoire in phage λ.
        Science. 1989; 246: 1275-1281
        • Hanes J.
        • Plückthun A.
        In vitro selection and evolution of functional proteins by using ribosome display.
        Proc. Natl. Acad. Sci. U.S.A. 1997; 94: 4937-4942
        • Boder E.T.
        • Wittrup K.D.
        Yeast surface display for directed evolution of protein expression, affinity, and stability.
        Methods Enzymol. 2000; 328: 430-444
        • Chervin A.S.
        • Aggen D.H.
        • Raseman J.M.
        • Kranz D.M.
        Engineering higher affinity T cell receptors using a T cell display system.
        J. Immunol. Methods. 2008; 339: 175-184
        • Hoogenboom H.R.
        Selecting and screening recombinant antibody libraries.
        Nat. Biotechnol. 2005; 23: 1105-1116
        • Van Deventer J.A.
        • Wittrup K.D.
        Yeast surface display for antibody isolation: library construction, library screening, and affinity maturation.
        Methods Mol. Biol. 2014; 1131: 151-181
        • Shultzaberger R.K.
        • Malashock D.S.
        • Kirsch J.F.
        • Eisen M.B.
        The fitness landscapes of cis-acting binding sites in different promoter and environmental contexts.
        PLoS Genet. 2010; 6: e1001042
        • Fowler D.M.
        • Araya C.L.
        • Fleishman S.J.
        • Kellogg E.H.
        • Stephany J.J.
        • Baker D.
        • Fields S.
        High-resolution mapping of protein sequence-function relationships.
        Nat. Methods. 2010; 7: 741-746
        • Procko E.
        • Hedman R.
        • Hamilton K.
        • Seetharaman J.
        • Fleishman S.J.
        • Su M.
        • Aramini J.
        • Kornhaber G.
        • Hunt J.F.
        • Tong L.
        • Montelione G.T.
        • Baker D.
        Computational design of a protein-based enzyme inhibitor.
        J. Mol. Biol. 2013; 425: 3563-3575
        • McLaughlin Jr, R.N.
        • Poelwijk F.J.
        • Raman A.
        • Gosal W.S.
        • Ranganathan R.
        The spatial architecture of protein function and adaptation.
        Nature. 2012; 491: 138-142
        • Whitehead T.A.
        • Chevalier A.
        • Song Y.
        • Dreyfus C.
        • Fleishman S.J.
        • De Mattos C.
        • Myers C.A.
        • Kamisetty H.
        • Blair P.
        • Wilson I.A.
        • Baker D.
        Optimization of affinity, specificity and function of designed influenza inhibitors using deep sequencing.
        Nat. Biotechnol. 2012; 30: 543-548
        • Schreiber G.
        • Fleishman S.J.
        Computational design of protein-protein interactions.
        Curr. Opin. Struct. Biol. 2013; 23: 903-910
        • Das R.
        Four small puzzles that Rosetta doesn't solve.
        PLoS One. 2011; 6: e20044
        • Holler P.D.
        • Chlewicki L.K.
        • Kranz D.M.
        TCRs with high affinity for foreign pMHC show self-reactivity.
        Nat. Immunol. 2003; 4: 55-62
        • Weber K.S.
        • Donermeyer D.L.
        • Allen P.M.
        • Kranz D.M.
        Class II-restricted T cell receptor engineered in vitro for higher affinity retains peptide specificity and function.
        Proc. Natl. Acad. Sci. U.S.A. 2005; 102: 19033-19038
        • Aggen D.H.
        • Chervin A.S.
        • Insaidoo F.K.
        • Piepenbrink K.H.
        • Baker B.M.
        • Kranz D.M.
        Identification and engineering of human variable regions that allow expression of stable single-chain T cell receptors.
        Protein Eng. Des. Sel. 2011; 24: 361-372
        • Smith S.N.
        • Sommermeyer D.
        • Piepenbrink K.H.
        • Blevins S.J.
        • Bernhard H.
        • Uckert W.
        • Baker B.M.
        • Kranz D.M.
        Plasticity in the contribution of T cell receptor variable region residues to binding of peptide-HLA-A2 complexes.
        J. Mol. Biol. 2013; 425: 4496-4507
        • Smith S.N.
        • Wang Y.
        • Baylon J.L.
        • Singh N.K.
        • Baker B.M.
        • Tajkhorshid E.
        • Kranz D.M.
        Changing the peptide specificity of a human T-cell receptor by directed evolution.
        Nat. Commun. 2014; 5: 5223
        • Schmitt T.M.
        • Aggen D.H.
        • Stromnes I.M.
        • Dossett M.L.
        • Richman S.A.
        • Kranz D.M.
        • Greenberg P.D.
        Enhanced-affinity murine T-cell receptors for tumor/self-antigens can be safe in gene therapy despite surpassing the threshold for thymic selection.
        Blood. 2013; 122: 348-356
        • Pierce B.G.
        • Hellman L.M.
        • Hossain M.
        • Singh N.K.
        • Vander Kooi C.W.
        • Weng Z.
        • Baker B.M.
        Computational design of the affinity and specificity of a therapeutic T cell receptor.
        PLoS Comput. Biol. 2014; 10: e1003478
        • Riley T.P.
        • Ayres C.M.
        • Hellman L.M.
        • Singh N.K.
        • Cosiano M.
        • Cimons J.M.
        • Anderson M.J.
        • Piepenbrink K.H.
        • Pierce B.G.
        • Weng Z.
        • Baker B.M.
        A generalized framework for computational design and mutational scanning of T cell receptor binding interfaces.
        Protein Eng. Des. Sel. 2016; (10.1093/protein/gzw050)
        • Harris D.T.
        • Singh N.K.
        • Cai Q.
        • Smith S.N.
        • Vander Kooi C.W.
        • Procko E.
        • Kranz D.M.
        • Baker B.M.
        An engineered switch in T cell receptor specificity leads to an unusual but functional binding geometry.
        Structure. 2016; 24: 1142-1154
        • Araya C.L.
        • Fowler D.M.
        Deep mutational scanning: assessing protein function on a massive scale.
        Trends Biotechnol. 2011; 29: 435-442
        • Riley T.P.
        • Singh N.K.
        • Pierce B.G.
        • Weng Z.
        • Baker B.M.
        Computational modeling of T cell receptor complexes.
        Methods Mol. Biol. 2016; 1414: 319-340
        • Feldhaus M.J.
        • Siegel R.W.
        • Opresko L.K.
        • Coleman J.R.
        • Feldhaus J.M.
        • Yeung Y.A.
        • Cochran J.R.
        • Heinzelman P.
        • Colby D.
        • Swers J.
        • Graff C.
        • Wiley H.S.
        • Wittrup K.D.
        Flow-cytometric isolation of human antibodies from a nonimmune Saccharomyces cerevisiae surface display library.
        Nat. Biotechnol. 2003; 21: 163-170
        • Kieke M.C.
        • Sundberg E.
        • Shusta E.V.
        • Mariuzza R.A.
        • Wittrup K.D.
        • Kranz D.M.
        High affinity T cell receptors from yeast display libraries block T cell activation by superantigens.
        J. Mol. Biol. 2001; 307: 1305-1315
        • Kieke M.C.
        • Shusta E.V.
        • Boder E.T.
        • Teyton L.
        • Wittrup K.D.
        • Kranz D.M.
        Selection of functional T cell receptor mutants from a yeast surface-display library.
        Proc. Natl. Acad. Sci. U.S.A. 1999; 96: 5651-5656
        • Shusta E.V.
        • Kieke M.C.
        • Parke E.
        • Kranz D.M.
        • Wittrup K.D.
        Yeast polypeptide fusion surface display levels predict thermal stability and soluble secretion efficiency.
        J. Mol. Biol. 1999; 292: 949-956
        • Orr B.A.
        • Carr L.M.
        • Wittrup K.D.
        • Roy E.J.
        • Kranz D.M.
        Rapid method for measuring ScFv thermal stability by yeast surface display.
        Biotechnol. Prog. 2003; 19: 631-638
        • Starwalt S.E.
        • Masteller E.L.
        • Bluestone J.A.
        • Kranz D.M.
        Directed evolution of a single-chain class II MHC product by yeast display.
        Protein Eng. 2003; 16: 147-156
        • Jones L.L.
        • Brophy S.E.
        • Bankovich A.J.
        • Colf L.A.
        • Hanick N.A.
        • Garcia K.C.
        • Kranz D.M.
        Engineering and characterization of a stabilized α1/α2 module of the class I major histocompatibility complex product Ld.
        J. Biol. Chem. 2006; 281: 25734-25744
        • Traxlmayr M.W.
        • Hasenhindl C.
        • Hackl M.
        • Stadlmayr G.
        • Rybka J.D.
        • Borth N.
        • Grillari J.
        • Rüker F.
        • Obinger C.
        Construction of a stability landscape of the CH3 domain of human IgG1 by combining directed evolution with high throughput sequencing.
        J. Mol. Biol. 2012; 423: 397-412
        • Traxlmayr M.W.
        • Obinger C.
        Directed evolution of proteins for increased stability and expression using yeast display.
        Arch. Biochem. Biophys. 2012; 526: 174-180
        • Stone J.D.
        • Artyomov M.N.
        • Chervin A.S.
        • Chakraborty A.K.
        • Eisen H.N.
        • Kranz D.M.
        Interaction of streptavidin-based peptide-MHC oligomers (tetramers) with cell-surface TCRs.
        J. Immunol. 2011; 187: 6281-6290
        • Park S.
        • Xu Y.
        • Stowell X.F.
        • Gai F.
        • Saven J.G.
        • Boder E.T.
        Limitations of yeast surface display in engineering proteins of high thermostability.
        Protein Eng. Des. Sel. 2006; 19: 211-217
        • Riley T.P.
        • Singh N.K.
        • Pierce B.G.
        • Baker B.M.
        • Weng Z.
        Computational reprogramming of T cell antigen receptor binding properties.
        Methods Mol. Biol. 2016; 1414: 305-318
        • Chothia C.
        Hydrophobic bonding and accessible surface area in proteins.
        Nature. 1974; 248: 338-339
        • Reynolds J.A.
        • Gilbert D.B.
        • Tanford C.
        Empirical correlation between hydrophobic free energy and aqueous cavity surface area.
        Proc. Natl. Acad. Sci. U.S.A. 1974; 71: 2925-2927
        • Thiltgen G.
        • Goldstein R.A.
        Assessing predictors of changes in protein stability upon mutation using self-consistency.
        PLoS One. 2012; 7: e46084
        • Baker D.
        • Agard D.A.
        Kinetics versus thermodynamics in protein folding.
        Biochemistry. 1994; 33: 7505-7509
        • Honegger A.
        • Malebranche A.D.
        • Röthlisberger D.
        • Plückthun A.
        The influence of the framework core residues on the biophysical properties of immunoglobulin heavy chain variable domains.
        Protein Eng. Des. Sel. 2009; 22: 121-134
        • Miller B.R.
        • Demarest S.J.
        • Lugovskoy A.
        • Huang F.
        • Wu X.
        • Snyder W.B.
        • Croner L.J.
        • Wang N.
        • Amatucci A.
        • Michaelson J.S.
        • Glaser S.M.
        Stability engineering of scFvs for the development of bispecific and multivalent antibodies.
        Protein Eng. Des. Sel. 2010; 23: 549-557
        • Nielsen J.E.
        • Gunner M.R.
        • García-Moreno B.E.
        The pKa Cooperative: a collaborative effort to advance structure-based calculations of pKa values and electrostatic effects in proteins.
        Proteins. 2011; 79: 3249-3259
        • Li Z.
        • Yang Y.
        • Zhan J.
        • Dai L.
        • Zhou Y.
        Energy functions in de novo protein design: current challenges and future prospects.
        Annu. Rev. Biophys. 2013; 42: 315-335
        • Fleishman S.J.
        • Whitehead T.A.
        • Strauch E.M.
        • Corn J.E.
        • Qin S.
        • Zhou H.X.
        • Mitchell J.C.
        • Demerdash O.N.
        • Takeda-Shitaka M.
        • Terashi G.
        • Moal I.H.
        • Li X.
        • Bates P.A.
        • Zacharias M.
        • Park H.
        • et al.
        Community-wide assessment of protein-interface modeling suggests improvements to design methodology.
        J. Mol. Biol. 2011; 414: 289-302
        • Stranges P.B.
        • Kuhlman B.
        A comparison of successful and failed protein interface designs highlights the challenges of designing buried hydrogen bonds.
        Protein Sci. 2013; 22: 74-82
        • Holler P.D.
        • Kranz D.M.
        Quantitative analysis of the contribution of TCR/pepMHC affinity and CD8 to T cell activation.
        Immunity. 2003; 18: 255-264
        • Chervin A.S.
        • Stone J.D.
        • Holler P.D.
        • Bai A.
        • Chen J.
        • Eisen H.N.
        • Kranz D.M.
        The impact of TCR-binding properties and antigen presentation format on T cell responsiveness.
        J. Immunol. 2009; 183: 1166-1178
        • Rodenko B.
        • Toebes M.
        • Hadrup S.R.
        • van Esch W.J.
        • Molenaar A.M.
        • Schumacher T.N.
        • Ovaa H.
        Generation of peptide-MHC class I complexes through UV-mediated ligand exchange.
        Nat. Protoc. 2006; 1: 1120-1132
        • Gietz R.D.
        • Woods R.A.
        Transformation of yeast by lithium acetate/single-stranded carrier DNA/polyethylene glycol method.
        Methods Enzymol. 2002; 350: 87-96
        • Smith S.N.
        • Harris D.T.
        • Kranz D.M.
        T cell receptor engineering and analysis using the yeast display platform.
        Methods Mol. Biol. 2015; 1319: 95-141
        • Kaufmann K.W.
        • Lemmon G.H.
        • Deluca S.L.
        • Sheehan J.H.
        • Meiler J.
        Practically useful: what the Rosetta protein modeling suite can do for you.
        Biochemistry. 2010; 49: 2987-2998
        • Leaver-Fay A.
        • O'Meara M.J.
        • Tyka M.
        • Jacak R.
        • Song Y.
        • Kellogg E.H.
        • Thompson J.
        • Davis I.W.
        • Pache R.A.
        • Lyskov S.
        • Gray J.J.
        • Kortemme T.
        • Richardson J.S.
        • Havranek J.J.
        • Snoeyink J.
        • et al.
        Scientific benchmarks for guiding macromolecular energy function improvement.
        Methods Enzymol. 2013; 523: 109-143
        • Moretti R.
        • Fleishman S.J.
        • Agius R.
        • Torchala M.
        • Bates P.A.
        • Kastritis P.L.
        • Rodrigues J.P.
        • Trellet M.
        • Bonvin A.M.
        • Cui M.
        • Rooman M.
        • Gillis D.
        • Dehouck Y.
        • Moal I.
        • Romero-Durana M.
        • et al.
        Community-wide evaluation of methods for predicting the effect of mutations on protein-protein interactions.
        Proteins. 2013; 81: 1980-1987
        • Chaudhury S.
        • Lyskov S.
        • Gray J.J.
        PyRosetta: a script-based interface for implementing molecular modeling algorithms using Rosetta.
        Bioinformatics. 2010; 26: 689-691
        • Mandell D.J.
        • Coutsias E.A.
        • Kortemme T.
        Sub-angstrom accuracy in protein loop reconstruction by robotics-inspired conformational sampling.
        Nat. Methods. 2009; 6: 551-552
        • Bradley P.
        • Misura K.M.
        • Baker D.
        Toward high-resolution de novo structure prediction for small proteins.
        Science. 2005; 309: 1868-1871
        • Canutescu A.A.
        • Dunbrack Jr., R.L.
        Cyclic coordinate descent: a robotics algorithm for protein loop closure.
        Protein Sci. 2003; 12: 963-972
        • Kellogg E.H.
        • Leaver-Fay A.
        • Baker D.
        Role of conformational sampling in computing mutation-induced changes in protein structure and stability.
        Proteins. 2011; 79: 830-838
        • Götz A.W.
        • Williamson M.J.
        • Xu D.
        • Poole D.
        • Le Grand S.
        • Walker R.C.
        Routine microsecond molecular dynamics simulations with AMBER on GPUs. 1. Generalized Born.
        J. Chem. Theory Comput. 2012; 8: 1542-1555