Molecular mechanism of phosphopeptide neoantigen immunogenicity

doi:10.21203/rs.3.rs-2327641/v1

Download PDF

Research Article

Molecular mechanism of phosphopeptide neoantigen immunogenicity

https://doi.org/10.21203/rs.3.rs-2327641/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Altered protein phosphorylation in cancer cells often leads to surface presentation of phosphopeptide neoantigens. However, their role in cancer immunogenicity remains unclear. Here we describe a mechanism by which an HLA-B*0702-specific acute myeloid leukemia phosphoneoantigen pMLL_747-755 (EPR(pS)PSHSM) is recognized by cognate TCR27, which is a candidate for immunotherapy of AML. We show that the replacement of phosphoserine P₄ with serine or phosphomimetics does not affect the pMHC conformation or peptide-MHC affinity but abrogates the TCR27-dependent T cell activation and weakens binding between TCR27 and pMHC. We determined the crystal structures for TCR27 and cognate pMHC, mapped the pMHC-TCR interface by TROSY-NMR, generated a ternary pMHC-TCR complex using information-driven protein docking, and identified key polar interactions between phosphate group at P₄ and TCR27 that are crucial for ternary complex stability and TCR27 specificity. These data will support development of cancer immunotherapy through target expansion and TCR optimization.

*The authors would like to note that Yury Patskovsky and Aswin Natarajan contributed equally.

Structural Biology

Immunology

Cancer

immunity

neoantigen

phosphopeptide

acute myeloid leukemia

MHC

TCR

crystal structure

MHC-TCR interface

MHC-TCR complex

The magnitude of the T cell immune response to cancer typically correlates with neoantigen burden (Luksza et al., 2017). However, high tumor heterogeneity and the lack of shared neoepitopes between patients reduces the impact of neoantigen-based therapies as they are challenging to implement and difficult to generalize (Blass and Ott, 2021). This problem prompted the search for alternative shared neoantigens among the post-translational protein modifications (PTM). Among PTM, altered function of protein kinases or phosphatases may increase protein phosphorylation that in turn can elevate lifetimes of intracellular phosphopeptides. As a result, the phosphopeptides appear on the surface of cancer cells as pMHC complexes (Zarling et al., 2006, Bassani-Sternberg et al., 2016, Yen et al., 2021). The primary source of the dysregulation signature in tumor and virally infected cells is suppression of PP2A, a critical phosphatase that controls numerous signaling pathways (Mahoney et al., 2021). Moreover, SET proteins or CIP2A are the endogenous PP2A inhibitors, the high levels of which have been associated with cancer progression (Lucas et al., 2015). Protein phosphatase 1 demonstrated a similar effect on MHC presentation of phosphopeptides as did PP2A (Wu et al., 2009).

Phosphopeptides have been detected in pMHC preparations from the normal and cancer tissues, and some of them were confirmed as cancer specific and immunogenic, most prominently in leukemias and melanoma (Zarling et al., 2000, Zarling et al., 2006, Cobbold et al., 2013, Penny et al., 2021, Solleder et al., 2020). In support of phosphopeptides as potential targets in cancer therapy, Lin and colleagues demonstrated that transgenic mice immunized with a tumor necrosis factor receptor associated protein 1 (TRAP1) HLA-A2-specific phosphopeptide had delayed tumor growth and prolonged survival (Lin et al., 2019). Furthermore, clinical studies in melanoma patients demonstrated safety and immunogenicity of vaccines containing the cancer-associated HLA-A2 phosphopeptides pBCAR3_{126 − 134} and pIRS2_{1097 − 1105} (Engelhard et al., 2020).

However, the mechanism of phosphopeptides immunogenicity remains unexplored. To date, no pMHC-TCR structures involving phosphopeptide antigens were available. Several crystal structures have been published for the Class I pMHC HLA-A*0201 in complex with phosphopeptides (Mohammed et al., 2008, Mohammed et al., 2017). Nearly all of these peptides had phosphoserine in position P₄, and the majority had basic arginine or lysine residue in P₁ reflective of the phosphorylation pattern of a 1,4-basophilic protein kinase motif (e.g., Akt) and proline directed kinases (e.g., MAP kinases and CDKs) (Amanchy et al., 2007). Besides that, the crystal structures of HLA-B*4001 (Alpizar et al., 2017) and HLA-DR1 (Li et al., 2010) in complex with phosphopeptides have been reported.

Lack of mechanistic data has hampered the efforts to understand the role of phosphorylated peptides in immunity and the scope of their potential applications in cancer immunotherapy. To resolve this problem, we have determined the molecular mechanism by which the HLA-B*0702-specific phosphopeptide epitope pMLL_{747 − 755} (EPR(pS)PSHSM) is recognized by its cognate TCR, which we identified and named TCR27-LC13 (or TCR27). The pMLL_{747 − 755} epitope was previously found in acute myeloid leukemia (AML) cells but not in healthy tissues (Cobbold et al., 2013), and its presence in leukemia has been detected in several high throughput proteomics studies (Hornbeck et al., 2015). The wild-type (WT) MLL_{747 − 755} (EPRSPSHSM) peptide is encoded by a human gene histone-lysine N-methyl transferase 2A (KMT2A), commonly known as MLL.

We have determined the crystal structures for TCR27 and HLA-B*0702 in complex with pMLL_{747 − 755} or similar peptides (pMHC). Using these structures, TROSY-NMR protein interface mapping method and the information-driven Haddock docking, we have assembled a ternary TCR-pMHC complex, identified the key residues at its interface and demonstrated the key role of phosphoserine P₄ in complex stability and T cell activation. We expect that our data will be pivotal in optimizing cancer immunotherapy by utilizing shared phosphopeptide neoantigens in vaccination and T cell therapy. This work also provides a rationale for the development of AML immunotherapy by vaccination and TCR engineering.

Phosphoserine residue of pMLL_{747 − 755} is essential for activation of the TCR27 + T cells

The cognate TCR for pMLL_{747 − 755}, named TCR27, was identified in the CD8+, pMLL_{747 − 755}-activated T cells from a healthy donor, and it contains the TRAV27 and TRBV27 variable gene segments. The discovery of TCR27 will be described in a separate publication (manuscript in preparation), and here we will focus on the mechanistic characterization of TCR27. To determine epitope specificity, the CD8 + T cells from healthy donors were transduced with lentivirus expressing TCR27, cultured with HLA-B*0702 positive T2 cells pulsed with pMLL_{747 − 755}, and assessed for expression of activation markers CD25 and CD69, respectively (Fig. 1A). The TCR27⁺ T cells displayed a dose-dependent increase in activation markers in the presence of pMLL_{747 − 755}-treated T2 cells, whereas no activation was observed in the presence of MLL_{747 − 755} (Fig. 1B). Since phosphomimetics are resistant to phosphatases, we considered the possibility of replacing phosphoserine at P₄ with one of its mimetics for future therapeutic

application and for mechanistic studies. To determine the role of pSer-P₄ in T cell activation, we replaced it with acidic residues aspartate or glutamate leading to considerably diminished or no T cell activation, respectively (Fig. 1B). Similarly, replacement of phosphoserine with phosphonate analogue (E7P) resulted in low T cell activation, though it was the highest activation level among all the phosphomimetics tested (Fig. 1B and Fig.S1A). The lack or low T cell activation by peptides with substitutions at P₄ was attributed to the reduced affinity between pMHC and TCR27 and not the altered stability of the pMHC complex, as each peptide binds HLA-B*0702 with nearly the same affinity (Fig.S1B-C).

To further investigate the role of individual epitope residues in TCR27 recognition, we performed a peptide scan, in which a set of peptides synthesized with all possible amino acid replacements at all positions (except for the anchor residues Pro-P₂ or Met-P₉) were assessed for T cell activation potential. Residues Pro-P₅, His-P₇, and pSer-P₄ were intolerant to substitutions (Fig. 1C), while the P₁ position was relatively conserved, with only glutamic acid or leucine allowed. This indicates that the P₁ residue side chain is located close to TCR27 at the pMHC-TCR interface. Arg-P₃ is an anchor residue but not conserved, it was replaced with alanine, lysine or cysteine without a substantial decrease in T cell activation. Anchor Ser-P₆ could be substituted with asparagine or cysteine, but not with other residues, which correlated with the size of the corresponding polar cavity in pMHC structure (see below). Ser-P₈ substitutions did not affect the T cell activation.

The TSPRINT analysis (Karapetyan et al., 2019) provides a unique opportunity to determine if additional HLA-B*0702-restricted phosphopeptides encoded in human genome could be recognized by TCR27, the information which may be useful in mechanistic studies and clinical applications. In this analysis, we generated a 9-residue search template (E/L)PxSPxHx(A/L/V/I/M) to evaluate the possible 9-mer peptides that are present in human proteome (de Castro et al., 2006) and that match the structure activity relationship (SAR) requirements obtained from the scan data. Peptides identified in this manner represented a set of possible, but not confirmed TCR27-reactive epitopes. TSPRINT identified a total of 50 human proteome hits, including MLL_747-755 and DOT1L_998-1006. However, aside from pMLL_747-755, only pDOT1L_998-1006 activated the TCR27 + T cells when loaded onto HLA-B*0702 + T2 cells (Fig. 1D), while other 10 highest scored phosphopeptides tested were completely inactive. The amino acid sequence identity between pMLL_747-755 and pDOT1L_998-1006 is 45%. Both epitopes, pMLL_747-755 and pDOT1L_998-1006, activated the TCR27 + T cells to a similar degree, something that was not seen for their non-phosphorylated analogs (Fig. 1D). Similarly, the cytotoxic effect of the TCR27⁺ T cells was nearly the same or very similar for both pMLL_747-755 and pDOT1L_998-1006 (Fig.S1E,D).

Therefore, pSer-P₄ was crucial for the T cell activation as phosphate removal or its replacement with phosphomimetics either eliminated or sharply reduced the TCR27-dependent T cell activation. The scan data and the cross-reactivity between the two phospho-epitopes all suggested a shared pMHC-TCR27 recognition motif encompassing the residues P₄-P₇.

2. The phosphate group is essential for the TCR27-pMHC complex stability

To characterize the physical interactions between pMHC ligands and TCR27 we utilized biolayer interferometry (BLI) method. The data show that the binding affinity between pMHC and TCR27 is dependent on the nature of the amino acid residue at P₄ (Fig. 2A). The highest affinity was observed for pMHC in complex with phosphopeptides pMLL_{747 − 755} or pDOT1L_{998 − 1006}, with K_D

values of 1.9µM and 7.7µM, respectively. These values fall within the range typical for cancer-

specific and other TCRs (Merwe and Davis, 2003). By contrast, the pMHC complexes with non-phosphopeptides MLL_{747 − 755} and DOT1L_{998 − 1006} did not interact with TCR27 (K_D>2mM). Substitution of phosphate with phosphonate (E7P-MLL_{747 − 755}) or sulfate (OSE-MLL_{747 − 755}) sharply weakened the pMHC-TCR27 interactions (K_D=47µM and K_D>200µM, respectively). The binding data align well with the T cell activation data described above (Fig. 1B-D; S1D).

The substitution of pSer-P₄ in pMLL_{747 − 755} with any other residue was deleterious for both, T cell activation (see Fig. 1) and the TCR-pMHC ternary complex stability (Fig. 2A). It is notable that among all the phosphomimetics tested, the phosphonate analogue E7P-MLL_{747 − 755} produced the highest T cell activation and supported the highest affinity between pMHC and TCR27 with K_D=47µM (K_D=1.9µM was for pMLL_{747 − 755}). Moreover, the phosphonate chemical structure is the most similar to that of phosphate, and it differs only by the replacement of the methylene group (E7P-MLL_{747 − 755}) with a bridging oxygen (O_γ in pMLL_{747 − 755}). Altogether, our data demonstrated that the TCR27 + T cell activation can be largely dependent on the strength of association between pMHC and TCR, with the phosphate group integrity essential for complex stability and T cell activation. To explain the TCR-pMHC binding data and to identify important molecular interactions within the ternary complex we used the structural biology methods.

3. The phosphoserine residue in pMLL _747-755 or pDOT1L_998-1006 adopts a non-anchor orientation when in complex with HLA-B*0702 and is accessible for TCR

According to the earlier data, a 9-residue epitope usually retains its overall conformation observed in pMHC following formation of the TCR-pMHC complex (Szeto et al., 2020). Therefore, comparison of pMHC structures with distinct bound peptides could reveal the possible effect of epitope sequence on the TCR-pMHC interactions. To exclude the influence of different crystallization conditions on protein conformation, we generated the fully isomorphous (having the same space group and very similar unit cell parameters) complexes between HLA-B*0702 and the following peptides, MLL_747-755, pMLL_747-755, DOT1L_998-1006, pDOT1L_998-1006, E7P-

MLL_747-755 and OSE-MLL_747-755 (Table S1). Initially, comparison of the two non-isomorphous pMLL_747-755/HLA-B*0702 structures (#1 and #2 in Table S1) revealed the conformational heterogeneity in the pMHC outer loops mainly associated with different crystal packing as seen in the corresponding RMSD plot (Fig.S2A). The pMLL_747-755 or MLL_747-755 conformations were also dependent on the presence of small molecules such as ethylene glycol or glycerol (cryoprotectants) that could bind inside the HLA pocket E (Fig.S2A and Fig.S4). To avoid that, we used sucrose for cryo-protection of crystals. In contrast, the isomorphous pMHC structures shared high similarity with core RMSD values of around or less than 0.1Å between each other (Fig.S2B), which indicated little or no effect of each individual peptide on the overall pMHC structure. This in turn allowed observing small conformational differences associated with binding of distinct epitopes.

Each solved pMHC complex had a typical MHC-I-fold, and every peptide amino acid sequence was validated during crystallographic refinement (R_free value for the ligand was around or below 0.3), the quality of which is reflected by the nearly perfect match between the refined structures and the corresponding 2F_oF_c electron density maps (Fig. 2B). Our data indicate that every peptide binds in the same position between the HLA helices α₁ and α₂ (Fig. 2C). In each peptide residue P₄ adopts a non-anchor orientation, which accounts for the minimal effect of amino acid substitutions at P₄ on the affinity between the corresponding peptides and HLA-B*0702 (Fig. 2B). The serine-phosphoserine replacement slightly decreased both pMLL_747-755 and pDOT1L_998-1006 K_D values, despite the salt bridge between the phosphate group of pSer-P₄ and the guanidine group of Arg₆₂ (Fig. 2C-D). This bond likely does not contribute much to the pMHC stability. In line with such a notion is the fact that pSer-P₄ in the pMLL_747-755 /HLA-B*0702 structure adopted two alternate conformations – one with phosphate H-bonded to Arg₆₂, and a second one (with occupancy refined to 0.3) in which phosphate is H-bonded to the Gln₁₅₅ side chain (Fig. 2E). A strong salt bridge between phosphate and more basic Arg₆₂ would have prevented the second conformation. The H-bond network between pMLL_747-755 and HLA-B*0702 (Fig. 2E and 2D) reflects a typical HLA-epitope bonding pattern, wherein epitope’s N- and C-termini and especially polar backbone atoms are important contributors to the binary complex stability.

The structures with phosphopeptides versus the corresponding WT peptides or phosphomimetics share high overall similarity between each other with minor conformational perturbations around Arg₆₂ and Ile₆₆. In the pMHC structures with WT MLL_747-755 or DOT1L_998-1006 peptides, no protein atoms interacted with the hydroxyl groups of Ser-P₄. However, in the structures with phosphoepitopes, the guanidine group of Arg₆₂ was H-bonded to the phosphate

moiety (Figs. 2E and 2F). Formation of this bond required a 0.5Å shift of the Arg₆₂ side chain towards phosphate, which occurs in concert with Ile₆₆ switching to another alternate conformation (as indicated by the arrows in Figs. 2E, 2F and in Fig. 3) to prevent inter-atomic clashes between Arg₆₂ and Ile₆₆. In contrast, neither phosphonate nor sulfate analogues maintain contacts with HLA atoms in the corresponding pMHC structures and are fully exposed to solvent (Fig. 3A-C). Moreover, the structures with E7P-MLL_747-755 and OSE-MLL_747-755 peptides were nearly identical to that with the wild-type peptide MLL_747-755, including exactly the same positions for the Arg₆₂ and Ile₆₆ side chains (Fig. 3D). Therefore, the presence of phosphoserine but not of any other residue at P₄ caused unique structural perturbations at the HLA-epitope interface. It is evident that the slight difference between the chemical structures of phosphate and phosphomimetics was sufficient to induce such perturbations. For instance, in the pMLL_747-755/HLA-B*0702 complex, the O_γ atom of pSer-P₄ is located within the hydrogen-bond distance (< 3.5Å) from the Arg₆₂ guanidine group. However, in the structure with phosphonate (E7P-MLL_747-755) the same atomic position is occupied by the nonpolar methylene group not supporting hydrogen bonds, and the guanidine group of Arg₆₂ remains about 4Å away from the C_γ atom of the E7P residue. In conclusion, replacement of serine P₄ with phosphoserine but not with phosphomimetics leads to the unique conformational changes at the ligand-HLA interface. As shown below, Arg₆₂ is also a part of the pMHC-TCR27 interface, and its orientation influences interactions between Arg₆₂ and pSer-P₄ or TCR27, which are important for ternary complex stability.

5. Molecular basis for TCR27 cross-reactivity with pDOT1L _{998 − 1006} and pMLL_{747 − 755} is the common pMHC recognition motif

We compared the pMHC crystal structures of HLA-B*0702 in complex with pDOT1L_{998 − 1006}, pMLL_{747 − 755} and the corresponding non-phosphopeptides to determine if there is a common recognition motif for TCR27. Both crystal structures were isomorphous and shared a similar fold (core RMSD value between C_α atoms was 0.22 Å) while displaying local conformational differences. For instance, the binding between pDOT1L_{998 − 1006} and HLA-B*0702 resulted in the N-terminus of the α₂ helix located approximately 0.5Å away from the α₁ helix as compared to the structure with pMLL_{747 − 755} epitope. In addition, the C-terminal segment of the pDOT1L_{998 − 1006} peptide was shifted along with the α₂ helix (Fig. 4A and Fig.S3B). By contrast, the N-terminus of pDOT1L_{998 − 1006}, including pSer-P₄, adopts a conformation very similar to that in pMLL_{747 − 755}. The helical “shift” was not driven by the epitope phosphorylation status, but was a result of amino acid discrepancies in positions P₃ and P₆, and the magnitude of the observed shifts was approximately the same with either pDOT1L_{998 − 1006} or DOT1L_{998 − 1006}, respectively (Fig.S3B). In the pMLL_{747 − 755}/HLA-B*0702 structure, the guanidine group of Arg-P₃ is H-bonded to the Asp₁₁₄ and Tyr₁₁₆ (both in strand E) side chain atoms, whereas the Arg₁₅₆ (α₂ helix) basic side chain is H-bonded to the Glu₁₅₂ and Gln₁₅₅ (both in α₂ helix) side chains. The substitution of Arg-P₃ and Ser-P₆ (pMLL_{747 − 755}) with alanine residues (pDOT1L_{998 − 1006}) leads to the distinct hydrogen

bonding pattern with pDOT1L_{998 − 1006} epitope, where Arg₁₅₆ is connected to the buried Asp₁₄₄ carboxyl group and the pSer-P₄ backbone oxygen instead (Fig. 4B). In the pMHC structures with DOT1L_{998 − 1006} and pDOT1L_{998 − 1006}, the N-termini of α₂ helices are positioned farther away from the α₁ helices to accommodate the bulky Arg₁₅₆ side chain inside the peptide binding cavity and to avoid clashes between the neighboring atoms (see Fig.S3B). The helical shift also affected the Gln₁₅₅ side chain orientation, which explains the lack of alternate epitope conformation #2 with pDOT1L_{998 − 1006} that is supported by the H-bonds between Gln₁₅₅ and pSer-P₄ of bound pMLL_{747 − 755} in the corresponding pMHC structure (Fig. 4B). Based on the similarities and differences between the crystal structures, we concluded that the common pMHC recognition motif for TCR27 includes the epitope residues from pSer-P₄ to His-P₇ and the adjacent to them HLA residues, but it excludes the N-terminal segment of the HLA α₂ helix and the penultimate C-terminal residues for both, pDOT1L_{998 − 1006} and pMLL_{747 − 755}, respectively.

The suggested recognition pattern is supported by the level of solvent exposure for epitopes residues. As a rule, exposed residues are involved in TCR recognition (Szeto et al., 2020). We have compared the solvent accessible surface area (SASA) for epitope residues in the corresponding pMHC structures. The results are presented in Fig. 4C-F. It is notable that pSer-P₄ remains exposed in both structures, as is His-P₇ and, to lesser extent, Pro-P₅. These residues were invariable in the peptide scan. Gln-P₈ in pDOT1L_{998 − 1006} is mostly solvent exposed, but the pMLL_{747 − 755} epitope has serine in this position, which was variable in the peptide scan, and thus P₈ was not essential for TCR binding.

In summary, our data confirmed the existence of a shared recognition motif that includes the pMLL_{747 − 755} and pDOT1L_{998 − 1006} invariant amino acid residues P₄, P₅, and P₇. Still, despite cross-reactivity to TCR27 and high conformational similarity, the structures with pMLL_{747 − 755} and pDOT1L_{998 − 1006} were not identical, which agreed with the fact that the affinity between TCR27 and pMHC was about 4-fold greater with pMLL_{747 − 755} epitope (see Fig. 2A). In contrast to pMLL_{747 − 755}, the structure with DOT1L_{998 − 1006} shows no alternate conformations for pSer-P₄. This observation allowed to identify the correct and common for both epitope conformations of pSer-P₄ (H-bonded to Arg₆₂) for TCR-pMHC docking.

6. TCR27 is structurally similar to other human TCRs but displays conformational heterogeneity

The crystal structure of the TCR27 ectodomain was determined by the X-ray crystallography. We obtained the two types of non-isomorphous TCR27 crystals using different sets of crystallization conditions, and the corresponding structures were named TCR_1 (space group P2₁) and TCR_2 (space group P2₁2₁2₁), respectively (Table S1). In every TCR, the complementarity determining regions (CDRs) responsible for epitope-MHC recognition are located at the extremities (CDR loops) of the variable (Vα and Vβ) domains (Rudolph et al.,

2006). The CDRs in TCR27 were defined based on similarity with other TCRs, and the CDR loops were well observed on the electron density maps (Fig. 5A). TCR27 structure is an αβ heterodimer, where each monomer consists of one constant and one variable domain, all arranged into the constant and the variable regions, respectively (Fig. 5B). Unexpectedly, the

quaternary conformation of TCR_1 was different from that of TCR_2, with an RMSD value between the coordinates of their C_α atoms of 3.1Å. To illustrate the difference between the two TCR27 conformations, only the coordinates of the V_α domains (AA residues 1-110) were aligned (RMSD = 0.4 Å), and the superimposed TCR27 structures are shown in Fig. 5C. RMSD plots between the variable regions and between the V_α (Fig. 5D-E) or V_β (Fig. 5F-G) domains demonstrate, that the main difference between the tertiary TCR structures is attributed to the distinct conformations of the CDR3 loops. The “hinge” areas around which the subunit “rotation” occurs (shown by the arrows in Fig. 5D and Fig. 5G) is located at or close to the dimer interface between the two variable TCR domains.

This result does not rule out the possibility that both, TCR_1 and TCR_2, structures can exist in conditions close to physiological, and their existence reflects conformational flexibility in TCR27. To date, however, there were no examples of the two highly divergent quaternary conformations for the same TCR, and this problem requires further study. The conformational heterogeneity in TCR27 could have impaired our many efforts to obtain the crystals of a ternary complex, the model of which was instead generated using the solved crystal structures of its components (pMHC and TCR), and the combination of the NMR-based interface mapping and information-driven molecular docking.

7. Trosy-nmr

7.1. The crystal structure of the pMLL _{747 − 755} /HLA-B*0702 binary complex correlates with its conformation in solution

To aid the information-driven docking and for generation of a ternary complex between pMLL_{747 − 755}/HLA-B*0702 and TCR27, a TROSY-NMR method was used to map pMHC residues at the pMHC-TCR interface. First, we have assigned NMR chemical shifts for the vast majority of HLA-B*0702 residues, including those located at the HLA-peptide interface, involved in epitope recognition and binding to TCR (Fig.S5). To determine the effect of serine-phosphoserine substitution on the NMR spectra, we analyzed the differences in chemical shifts between the pMLL_{747 − 755}/HLA-B*0702 and MLL_{747 − 755}/HLA-B*0702 complexes by calculating the scaled chemical shift perturbations (CSP) in the ¹⁵N-TROSY spectra (Fig. 6A-B). The highest

CSP values were observed for the α₁ helical residues – Glu₅₈, Tyr₅₉, Asp₆₁, Ile₆₆, Tyr₆₇, Lys₆₈, Ala₆₉, Gln₇₀, Thr₇₃, Arg₇₅, and Ser₇₇, and for the α₂ helical residues – Trp₁₄₇, Arg₁₅₁, Glu₁₅₂, Ala₁₅₃, Glu₁₅₄, Gln₁₅₅, Arg₁₅₆, Trp₁₆₇, Tyr₁₇₁, Leu₁₇₂, and Glu₁₇₃, respectively. The fact that CSPs are seen in residues that are distant from pSer-P₄ suggests subtle differences between the two pMHC complexes, including some that may be too small to be observed in the crystal

structures. In general, however, there was a good agreement between the NMR and crystallographic data. For instance, the highest CSP values (δ > 0.1_ppm) were detected among the HLA-B*0702 residues surrounding the phosphate moiety, such as Ile₆₆, Gln₇₀, Glu₁₅₂, Ala₁₅₃ and Gln₁₅₅. Arg₆₂ was excluded from CSP calculations due to spectral overlaps. The amino acid residues with the highest CSP values could be grouped into the 4 separate clusters (Fig. 6A). The location of each cluster in the crystal structure is shown in Fig. 6C. Two adjacent amino acid clusters (#1 and #2) are located on the α₁ helix, while two others (#3 and #4) are located on the N- and C-terminus of the α₂ helix, respectively. Cluster #1 incorporates the residues neighboring Arg₆₂, the guanidine group of which is hydrogen bonded to phosphate in the crystal structure (Fig. 6D). Cluster #2 includes Ile₆₆ and surrounding residues, chemical shifts of which could be directly or indirectly influenced by Ile₆₆. Cluster #3 includes residues close to Gln₁₅₅, the side chain of which in the crystal structure interacts with phosphate in the alternate conformation and is hydrogen bonded to the backbone oxygen atom of pSer-P₄. Finally, cluster #4 encompasses the residues close to Glu-P₁, the carboxyl group of which is within the H-bond distance from the guanidine group of Arg₆₂ in the X-ray structure. CSP clustering reflects the high sensitivity of TROSY-NMR in detecting small perturbations in residues surrounding the ligand-binding site (Williamson, 2013; Natarajan et al., 2016). Overall, the TROSY-NMR CSP data indicate that the presence of pSer-P₄ induces local structural perturbations around the same HLA-B*0702 amino acid residues that were identified in the crystal structure. The highest CSP values were observed for residues located predominantly in the vicinity of or in direct contact with pSer-P₄. The NMR study demonstrated high homology between the solution and crystal structures and validated the NMR-based approach that was further utilized in mapping the TCR-pMHC interface.

7.2. TROSY-NMR mapping of the TCR-pMHC interface

To identify the HLA residues that participate in the pMHC-TCR interface, we performed TROSY-NMR titration experiments by increasing the ratios between perdeuterated TCR27 and perdeuterated ¹⁵N-labeled HLA-B*0702 in complex with pMLL_{747 − 755}. Upon assembly, protein residues at the protein-protein interface experience positional changes in the chemical shifts (CSP), mostly because of a changing environment. In addition, strong protein-protein interactions usually result in reduced NMR peak intensities if an association-dissociation event occurs at the intermediate or slow NMR relaxation time scale (Purslow et al., 2020). Therefore, by monitoring both parameters, CSPs and peak intensity changes upon TCR-pMHC complexation (Fig.S6), the protein residues involved in the interactions can be identified. When perdeuterated ¹⁵N-labeled HLA-B*0702/ MLL_{747 − 755} complex was used as a negative control, no significant interactions with TCR27 were detected, as no changes were observed in CSPs or peak intensities in comparison with the NMR spectrum of the binary complex alone (Fig.S6B). By contrast, binding between TCR27 and pMLL_{747 − 755}/HLA-B*0702 was accompanied by an overall reduction in peak intensities in the corresponding ¹⁵N-TROSY spectra (Fig. S6C, average V/V₀ = 0.39, pMLL_{747 − 755}/HLA-B*0702: TCR27 (1:1)), indicating an increase in molecular tumbling time, that is evidence of a strong binding event. The localized CSPs and peak intensity changes were assigned to the individual HLA residues, as presented in Fig. 7A-C. The peptide-surrounding residues (indicated by the red arrows in Fig. 7A and Fig. 7C), such as Gln₇₀ and Thr₇₃ from the α₁ helix and Ala₁₄₉, Arg₁₅₆, Ala₁₅₈ from the α₂ helix, demonstrated increased CSP values (δ_ppm > 0.016) upon complexation. Significant peak intensity changes (V/V₀ < 0.1) were detected for residues Asp₆₁, Ile₆₆, Tyr₆₇, Lys₆₈, Ala₆₉, Gln₇₀, Thr₇₃ and Glu₇₆ from the α₁ helix and Trp₁₄₇, Arg₁₅₆ and Glu₁₆₃ from the α₂ helix (Fig. 7B). The distribution of the NMR-mapped residues is shown schematically using the crystal structure for the pMLL_{747 − 755} /HLA-B*0702 complex in Fig. 7D. Thus, the TROSY-NMR based CSP - intensity analysis confirmed direct binding between TCR27 and HLA-B*0702 in complex with pMLL_{747 − 755}, and identified a number of HLA residues located at or close to the TCR-pMHC interface.

8. The pMLL_{747 − 755}/HLA-B*0702/TCR27 ternary complex reveals the molecular mechanism of phosphopeptide immunogenicity

To understand the mechanism of phosphopeptide recognition by TCR27, we have generated a ternary pMHC-TCR complex using an information-driven Haddock docking algorithm (Dominguez et al., 2003). The crystal structures of TCR27 and the corresponding pMHC complexes with phosphopeptides served as starting models. Data obtained by NMR, peptide scan, mutagenesis and X-ray crystallography were utilized to define the probable interface residues and limit the docking bias. Haddock protein-protein data driven docking software was chosen based on previous reports that Haddock docking efficiency equals or exceeds that of other protein docking platforms (Peacock and Chain, 2021), and because it applies an advanced docking algorithm to improve docking efficiency by using information about protein interfaces obtained by other methods, like mutagenesis and NMR (Rodrigues et al., 2012). We utilized 4 docking scenarios, described in Methods and schematically shown in Fig. S7A. The pMHC interface residues were selected based on NMR mapping, peptide scan and crystallographic data, whereas the TCR residues were selected from the CDR loops based on homology with other pMHC-TCR structures. Selected residues and their location on the TCR_1 structure is schematically shown in Fig.S7B. It is noteworthy, that the key role of Arg96β (located in the CDR3β loop) in the TCR-pMHC interaction was confirmed by the site-directed

mutagenesis. When the side chain of Arg96β was substituted by alanine, a mutant TCR completely lost the ability to bind the pMLL_{747 − 755} /HLA-B*0702 complex (Fig. 2A).

TCR27 adopts at least two quaternary conformations, as described above. Both structures were used to generate the docking models. However, the model from a TCR_2 structure did not

yield a clear solution in any of the docking scenarios based on the (1) relatively low absolute Haddock docking score for the top cluster, (2) marginal difference in score values between the top scoring cluster and the next one, and (3) a poor match with an experimentally predicted TCR-pMHC interface. By contrast, the model based on the TCR_1 structure yielded a clear solution in each docking scenario, with the best solution set (cluster 1) having an outstanding Haddock score, the highest number of matching and common contacts within the same cluster, low interface RMSD value within the same cluster, the best geometry and the lowest clash scores. Moreover, the top solutions from each docking scenario with TCR_1 displayed very similar topology with an RMSD value between the two top solutions of 3.1 Å or less (Fig.S7C). Docking was also performed using the pDOT1L_{998 − 1006}/HLA-B*0702 crystal structure as a model. Its top scoring solution shared very similar topology with the solution obtained for pMLL_{747 − 755}, with an RMSD value of about 1.9Å between them (Fig.S7D). The best docking scores were obtained in scenario #4, an example of which is shown in Fig. S8A-B. This assembly is discussed in the more details below.

For the assembled pMLL_{747 − 755}/HLA-B*0702/TCR27 ternary complex, there was a good match between the predicted (Fig. 7D) and observed (Fig. 7E) TCR footprint over the pMHC, which included the epitope residues P₄-P₇ and the adjacent alpha helical segments of HLA-B*0702. A “truncated” version of this complex including pMLL_{747 − 755}, HLA-B*0702 (domain 1) and TCR27 (variable region) is presented in Fig. 7F as a cartoon model. The TCR heterodimer binds across the longest axis of the pMHC so that the N- (P₁-P₃) and C-terminal epitope residues (P₈-P₉) remain excluded from the interface, as defined by the interatomic distance cutoff > 4Å between the non-hydrogen atoms. The shape of both (TCR27 or pMHC) interacting faces is asymmetrical, but the relative orientation of the TCR with respect to the pMHC remained similar in all 4 docking scenarios, most likely being restrained by the epitope residue selections. Still, the NMR mapping and peptide scan data strongly suggested that epitope residues other than P₄-P₇ do not directly participate in recognition of TCR27. Overall, the pMLL_{747 − 755}/HLA-B*0702/TCR27 complex topology was very similar to the majority of other known TCR-pMHC-I structures (Gowthaman and Pierce, 2020). The TCR27 docking angle is approximately 60° and the incident angle is 8°. The surface area buried upon complexation is approximately 1800 A². The TCR27 footprint is typical, with the V_α domain positioned over the HLA α₂ helix and the V_β domain positioned over the HLA α₁ helix. The CDR3α loop is located over the epitope’s N-terminus, including pSer-P₄, whereas the contact area between the TCR CDR3β loop and the pMHC includes Pro-P₅ and Ser-P₆ backbone atoms and the His-P₇ side chain atoms (Fig. 7G). Besides Arg₆₂ from HLA, the TCR residues contacting or close to pSer-P₄ include Tyr91α and Tyr97α (CDR3α loop), Arg96β (CDR3β loop), and Ser29α (CDR1α loop) (Fig. 7H). Importantly, all the above residues are hydrogen bond donors, whereas all the phosphate oxygen atoms are the hydrogen bond acceptors, thus providing a perfect match. The large number of polar interactions involving all oxygen atoms of pSer-P₄ and its low solvent exposure upon complexation (less than 10% of ASA in the complex) indicates the importance of polar bonds in stabilizing the ternary complex. In overall, our data demonstrated that phosphate group integrity is vital in maintaining the pMHC-TCR complex stability, as any pSer-P₄ substitution (including phosphomimetics), that could alter the polar bond network, affected the complex stability resulting in the loss of T cell activation and diminished binding between TCR27 and pMHC.

Phosphopeptide antigens are post-translationally modified (PTM) antigens that represent a new class of “non-self” (not subject to central tolerance) neoantigens associated with a broad range of cancers and virally infected cells. Due to the underlying commonality of the transformed cellular pathways, many phosphopeptide neoantigens are prevalent across patients within particular cancers and, for some phosphopeptides, across various cancers, reflecting common modes of dysregulation (Paisley et al., 2018; Mahoney et al., 2021). Thus, phosphopeptides have potential as novel targets for new immune therapies, including vaccines and cell therapies. Immunogenicity of phospho-neoantigens has been demonstrated (Cobbold et al., 2013, Engelhard et al., 2020, Penny et al., 2021, Lin et al., 2019), but their anti-cancer potential has not been explored in extensive clinical trials. In this report we describe a comprehensive SAR analysis of TCR27, discovered from healthy human donors, which recognizes an AML-linked HLA-B*0702-restricted phosphopeptide epitope pMLL_{747 − 755}. After the cross-wide genome search we identified an additional TCR27-reactive phosphopeptide, pDOT1L_{998 − 1006}. Comparison of pMHC structures in complex with pMLL_{747 − 755} and pDOT1L_{998 − 1006} helped to determine the TCR footprint and the crucial role of phosphoserine in recognition of TCR27. The cross-reactivity of TCR27 may be advantageous, as DOT1L is reportedly overexpressed in AML, and thus is a potential therapeutic target (Mercher and Schwaller, 2019). As with pMLL_{747 − 755}, pDOT1L_{998 − 1006} was not detected in normal tissues (Hornbeck et al., 2019).

The TCR27 avidity correlated with the affinity between TCR27 and pMHC. Our structural analysis together with binding kinetics and the T cell activation data combined revealed that the electrostatic interactions, especially between the TCR CDR3 loops and the phosphoserine-P₄ side chain were crucial for the ternary pMHC-TCR complex stability. Substitution of the phosphoserine at P₄ with any other residue could affect geometry, size, charge distribution, and pK_a values of the corresponding peptide. For instance, at neutral pH the phosphonate moiety carries a total negative charge of between − 1 and − 2, the sulfate carries a charge of -1, and the phosphate’s carries a -2 charge (Elliott et al., 2012) (Fig.S9). Loss of charge would lead to the loss of electrostatic bonds that in turn could reduce stability of the complex. From the practical standpoint, it is evident that phosphomimetics may not serve as a good substitute for phosphorylated residues in immunotherapy. The incorporation of NMR data was integral to achieving successful pMHC-TCR docking. The method has previously been used to confirm complexation between various receptor/ligand pairs, though this approach differs greatly depending on the size of the complex (Mizukoshi et al., 2020), as increased molecular mass dramatically decreases the sensitivity and resolution of NMR spectra (Purslow et al., 2020). Such issue can be overcome by using the transverse relaxation-optimized spectroscopy (TROSY) method (Salzmann et al., 1998), though it has not previously been applied to TCR complexes due to their sheer size and complexity. In the earlier studies, the truncated versions of H-2L^d and 2C-TCR were used to map the TCR footprints on the MHC surface (using standard HSQC-NMR), and the results were further corroborated by the ternary complex crystal structure (Varani et al., 2007). A similar approach was used to determine the interface between HLA and β2M (Beerbaum et al., 2013), and to determine allosteric interactions in the CD3-TCR-pMHC complex (Natarajan et al., 2017, Rangarajan et al., 2018). NMR mapping is cumbersome due to spectral overlaps, but it was a key to our TCR-pMHC docking efforts and in validating its results. For instance, our high resolution TROSY-NMR spectral data were in a good agreement with the crystallographic data and correctly pinpointed the contact residues in HLA-B*0702 interacting with pSer-P₄. Moreover, we demonstrated that local structural perturbations extend beyond the HLA residues that were in the direct contact but include 4 main clusters of residues instead. The advantage of a cluster becomes obvious when the chemical shift for certain backbone amides cannot be assigned due to overlaps (e.g. Arg₆₂). Importantly, while NMR-based clustering introduced certain ambiguity in residue selection, the crystallographic data was used to more clearly define the pMHC-TCR interface by excluding from docking the HLA residues that were, for instance, solvent inaccessible or located at the HLA-β2M interface instead. As a result, TROSY-NMR in combination with crystallographic method, mutagenesis and molecular modeling greatly improved the odds of producing a less biased and more accurate model for a ternary pMHC-TCR complex.

In summary, we provided direct evidence of phosphopeptide immunogenicity by demonstrating the molecular mechanism of its recognition by a cognate TCR. These data can be utilized in further mechanistic studies and applied to the development of anti-cancer immunotherapies. Though immunogenicity of phosphopeptides has previously been demonstrated, our work was directed to characterize a phosphoepitope-specific TCR. This enabled us to determine the molecular mechanism of interaction between the phosphopeptide, HLA and the cognate TCR, highlighting the role of phosphoserine in the epitope immunogenicity. Further, using SAR and peptide scan data we discovered the epitope pDOT1L_{998 − 1006}, which also binds to HLA-B*0702 and engages the TCR27, driving T cell effector responses similar to pMLL_{747 − 755}. Determining the process by which MHC-bound phosphopeptide antigens are presented to T cells is critical to understanding how these novel neoantigens are involved in immune recognition and tumor control, and it will aid in development of cancer immunotherapy.

Cell Lines, Plasmids, Affinity and T cell activation assay

Tap-deficient T2 (174x CEM.T2) cells and HEK293T cells were purchased from ATCC and maintained in RPMI 1640 (BioConcept) supplemented with 10% fetal calf serum (FCS, HyClone), 2mM L-glutamine and 1% Penicillin/Streptomycin at 37°C and 5% CO₂. The EGFP reporter cell line was based on a murine TCR negative thymoma cell line derived from strain BW5417 (ATCC®TIB-47TM) and was stably transduced with (a) TCR27 in which the human constant regions were replaced with those of mouse, (b) a chimeric mouse/human CD8, and (c) an EGFP reporter construct linked to a minimal IL-2 promoter comprising three NFAT-binding sites (3xNFAT) (Rooney et al., 1995). The transduced cells were termed TCR27-AKD10R3 cells and will be called the effector cells. Mouse cell lines were cultured in SF-IMDM (BioConcept) supplemented with 3% FCS, 1% Penicillin/Streptomycin, and 50µM beta-mercaptoethanol at 37°C and 10% CO₂.

Peptides were dissolved in dimethyl sulfoxide (DMSO) at concentrations of 2–20 mg/mL, aliquoted and stored at − 80°C. Peptide-pulsing experiments were performed by pre-incubating T2 cells with 25µg peptide per 1 × 10⁶ cells in PBS for 2 h at 37°C and 5% CO₂, washed three times to remove unbound peptides and used for further downstream analysis. Peptide-pulsed T2 cells were incubated with effector cells in a 2:1 ratio overnight in SF-IMDM at 37°C and 10% CO₂, washed twice and stained with anti-mouse TCR-® chain clone H57-597 (BD Pharmingen, 1:500) for 30 min at RT. Cells were washed twice before FACS-analysis using a BD FACS Canto II. Data analysis was performed using FlowJo V10 Software. Binding affinity between each peptide and HLA-B*0702 was determined using a homogenous, proximity-based assay for detection of peptides binding to HLA class I molecules using a conformation-dependent anti-HLA class I antibody, W6/32, as one tag and a biotinylated recombinant HLA class I molecule as the other tag, and a proximity-based signal was generated through the luminescent oxygen channeling immunoassay technology (PerkinElmer) as described (Harndahl et al., 2009).

Cytotoxicity assay

Primary T cells isolated from PBMC (easysep, Stemcell) were transduced with a lentivirus encoding TCR27 (MOI 10:1) 24 to 48 hours post activation with anti-CD3/CD28 beads (Dynabeads, Thermofisher) as described (Karapetyan et al., 2019). Cells were expanded in complete RPMI supplemented with recombinant IL-2 (100 IU/ml) (Teceleukin, Roche). Cells were challenged in the functional assays 10 to 15 days post initial stimulation. TCR27 expression was verified by staining cells with pMLL_{747 − 755} -HLA-B*0702 pentamer (Proimmune) and analyzed by flow cytometry. Peptide-pulsed T2 cells were incubated with effector cells expressing or not TCR27 (1:1 ratio) in complete RPMI medium for 16 to 20 hours at 37°C and 5% CO₂. Cells were stained for dead cells (Live/Dead, Life Technologies), and then stained with anti-CD4, anti-CD8, anti-CD25, anti-CD69 monoclonal Abs (BD Pharmingen, 1:100 dilution) for 30 min at 4°C, washed twice and analyzed by flow cytometry (BD Fortessa LSR II). When stained for IFNγ, cells were pre-incubated for 4–6 hours with BFA and Monensin (Biolegend), fixed, permeabilized and stained with PE-anti-IFNγ antibody (Biolgend). Data analysis was performed using FlowJo V10 software.

Identification of the off-target phosphopeptide pDOT1L_{998 − 1006}

To identify the off-target peptides we have utilized an approach that we have developed previously (Karapetyan et al., 2019). In brief, we investigated which amino acids at each position in pMLL_{747 − 755} enable TCR interaction by sequentially replacing every amino acid position outside of the anchor positions 2 and 9 with all 19 alternative amino acids, resulting in 134 peptides (133 altered peptides plus original epitope). Each peptide was individually evaluated using cell-based in vitro cytotoxicity assay. To represent the TCR recognition kernel, we have defined Position Weight Matrices (PWMs) for each assay by assigning normalized measurements to each of the 20 amino acids in each position. To predict potential off-target peptides, we applied an algorithm projecting the PWM-defined kernel into the human proteome, scoring TCR27 recognition for the predicted HLA-B^*0702 binding 9-mer peptides. Of the 11 peptides with highest predicted cross-reactivity scores, we confirmed that only 2, pMLL_{747 − 75} and pDOT1L_{998 − 1006} strongly activated the TCR27-expressing effector T cells.

Peptides

The following peptides were utilized: MLL_{747 − 755} (EPRSPSHSM), pMLL_{747 − 755} (EPR(pS)PSHSM), E7P-MLL_{747 − 755} (EPR(E7P)PSHSM, where E7P is the phosphono-serine residue analogue, in which the serine hydroxyl group was replaced by phosphonate: (2S)-2-amino-4-phosphonobutanoic acid); OSE-MLL_{747 − 755} (EPR(OSE)PSHSM, where OSE is the sulfoserine, in which the serine hydroxyl group was substituted with a sulfoxy moiety: (2S)-2-amino-3-sulfooxy-propanoic acid), EPRDPSHSM, EPREPSHSM; DOT1L_{998 − 1006} (LPASPAHQL) and pDOT1L_{998 − 1006} (LPA(pS)PAHQL) peptides with serine or phosphorylated serine in P₄, respectively. The protein sequences were derived from the following UniProt IDs: Q8TEK3 (DOT1L), and Q03164 (KMT2A).

Peptides were synthesized using a standard Fmoc solid-phase chemical synthesis with pre-loaded polystyrene Wang (PS-Wang) resin in a Symphony-X automatic synthesizer (Gyros Protein Technologies Inc). Standard Fmoc protected L-amino acids were activated using HCTU/NMM chemistry. For the standard phosphopeptide synthesis, the phosphate group was incorporated using N-α-Fmoc-O-benzyl-L-phosphoserine as a building block. With respect to resin substitution, a 5-fold excess of amino acid, 5-fold excess of activating reagent (HCTU) and a 10-fold excess of N-methyl morpholine were used for coupling each amino acid. The reaction was performed for 6 min with a double coupling cycle for any incomplete coupling throughout the synthesis. These steps were repeated until the desired peptide was obtained. At the end of synthesis, the resin was washed with dichloromethane (DCM) and dried.

The sulfo-serine was produced by on-resin sulfation using dicyclohexylcarbodiimide (DCC) and H₂SO₄. Peptide resin, DCC, and H₂SO₄ were allowed to react in a fixed molar ratio of 1:5:1, respectively, in DMF at 37°C for 10 minutes. The reaction vessel was drained and washed 5 times with DCM and dried. Phosphono-serine peptide was synthesized by incorporating the Fmoc-protected 2-amino- 4-phosphonobutanoic acid using the solid phase synthesis (Findeis, 2017). Upon completion of peptide assembly, the resin was transferred to another vessel for cleaving peptide from the resin. A cleavage cocktail reagent TFA: DTT: Water: TIS (88:5:5:2 v/w/v/v) was mixed with the resin and stirred for 4 hours at 25°C. The filtrate was evaporated with N₂ gas followed by precipitation with chilled diethyl ether and stored at -20°C for 12 hours. The precipitated peptides were centrifuged and washed twice with ether, dried, dissolved in the choice of solvent, and lyophilized to produce a crude dry powder.

The peptides were purified by HPLC on a Luna C₁₈ column (Phenomenex, Inc) using a water (0.1% TFA) vs acetonitrile (0.1% TFA) gradient. Peptide purity and stability was tested using analytical Luna C₁₈-column (Phenomenex, Inc). Peptide sequences were confirmed by LC/MS (6550 Q-TOF, Agilent Technologies).

Protein sequences

Amino acid sequences for human HLA-B*0702 (fragment, amino acid residues 25–299, Genebank ID AZU90178) and Human β2-microglobulin (hβ2M), a fragment with residues 21–119, UniProt ID P61769) were selected for expression in E.coli. Another version of recombinant HLA-B*0702 also included an AVI-tag sequence (GLNDIFEAQKIEWHE) (Beckett et al., 1999) added at the protein C terminus. TCRα and TCRβ chains included variable domains for TRAV27 (Genebank ID MZ701715, AA residue 1-110) and TRBV27 (Genebank ID MZ701716, AA residues 1-110), respectively, that were linked to the corresponding human LC13 constant domains (Boulter et al., 2003). To further stabilize recombinant TCRαβ heterodimer and improve protein yield during expression in E.coli, an inter-chain disulfide bridge was introduced between the constant chains, as described previously (Boulter et al., 2003).

Gene cloning, protein expression in Escherichia coli, refolding and purification

Ectodomains for human HLA-B*0702, hβ2M, or TCRα and TCRβ chains were cloned and expressed individually as inclusion bodies according to the previously described protocols (Natarajan et al., 2016) with minor modifications. In brief, a codon-optimized cDNA for every gene was synthesized by Genewiz, cloned between the NdeI and HindIII restriction sites of a plasmid pET30a, and verified by DNA sequencing. Mutagenesis was performed using Quick Change II Site Directed Mutagenesis kit (Agilent) according to the manufacturer’s instruction, and the results were verified by DNA sequencing. After transformation into E.coli BL21(DE3), the exponentially grown bacteria were placed in LB + Kanamycin (50µg/ml) medium, protein expression was induced by adding IPTG (Sigma) to the final concentration of 1mM, and the cells were incubated by shaking for 4–5 hours at + 37°C. The cell pellet was harvested by centrifugation, and the proteins as inclusion bodies were purified as described earlier (Natarajan et al., 2016), then solubilized in a buffer containing 6M Guanidine-HCl, 50mM Tris-HCl (pH 8.0), cleared by centrifugation (15,000g, 10min), aliquoted and kept frozen at -80°C until use.

Refolding of denatured proteins was performed as described (Boulter et al., 2003, Natarajan et al., 2016). Each pMHC complex with a different peptide was refolded separately. Purified proteins were dialyzed against 10mM HEPES-NaOH (pH 7.5), protein purity was validated by non-reducing SDS-PAGE (4–20%). Proteins were concentrated to 5-10mg/ml, and either flash-frozen in liquid nitrogen and stored at -80°C until use or subjected directly to crystallization and other analyses. TCRα and TCRβ chains were cloned, expressed separately, and refolded as pMHC, but the corresponding refolding buffer was supplemented with 3M urea. For long-term storage all the proteins were frozen in liquid nitrogen and kept at -80°C.

Biolayer interferometry

Binding between pMHC and TCR27 was measured using an Octet RED96 instrument and the streptavidin-conjugated SA chips (Pall ForteBio, Menlo Park, CA) according to the manufacturer’s instructions. HST buffer (10mM HEPES, pH 7.5, 100mM NaCl, 0.01% Tween-20, 1% BSA) was used in all analyses. Biotinylated pMHC was diluted to 2–10µg/ml, immobilized on a sensor chip with approximately the same signal magnitude for each peptide and then followed by the association step (30 or 60 sec) in 200µL of TCR27 solution (concentrations were ranging between 0 and 20–100µM), and the dissociation step in 200µL of HST buffer for 3–20 min. Assays were performed at 25°C. Data were analyzed using Octet® 9.1 System Data Analysis software to determine the steady-state binding parameters.

Protein crystallization, structure solution and refinement

The proteins were crystallized using sitting drop vapor diffusion methods and the 1:1 (vol:vol) protein to precipitant ratio. To identify initial conditions, the MCSG crystallization screens have been utilized (Anatrace, OH, USA). The crystallization drops were set up using the 96-well Intelli-plates (Art Robbins Instruments) and a Mosquito crystallization Robot (SPT Labtech), and the plates were kept at + 18°C. The crystals appeared between 3–10 days and up to 6 months from the beginning of the incubation. Crystallization conditions were optimized manually. After optimization, each pMHC complex was crystallized by mixing 0.5µl of protein solution (8-10mg/ml) with 0.5µl of precipitant A (18–26% PEG 4000, 0.1M sodium citrate, pH 8.0, and 20% isopropanol) in the 24-well Intelli-plates (Art Robbins Instruments) and equilibrated against the same precipitant at + 18°C. The rod-like crystals appeared 2–10 days after beginning of incubation and grew up in length of up to 1.5mm. The crystals were flash-frozen in liquid nitrogen (LN₂) after brief soak in the precipitant solution supplemented with 20–30% (vol/vol) of a cryoprotectant such as glycerol, ethylene glycol or sucrose, and stored frozen in LN₂ until data collection. In addition, the pMLL_{747 − 755}/HLA-B*0702 pMHC complex was crystallized using a precipitant B, such as 30–35% PEG 4000, 0.1M Tris-HCl, pH 8.5, 0.2M sodium acetate. In these conditions the prism-shaped crystals appeared in 3–6 months and grew up to the average size of around 100 microns; they were directly flash-frozen and stored in LN₂ until collection of the X-ray diffraction data. The diffraction quality crystals for TCR27 were obtained by the crystallization method described above and using the following set of precipitants, (1) 20–25% PEG 1000, 50mM HEPES-NaOH (pH 7.5), crystal morphology – plates; and (2) 20–25% PEG 4000, 0.2M ammonium sulfate, 0.1M sodium citrate (pH 5.6), crystal morphology – prisms and plates. In conditions #1 the crystals grew up in 5–10 days, but in conditions #2 they appeared after 3–4 months. The crystals were frozen directly in LN₂ and stored frozen until further analysis.

The X-ray diffraction data were collected from the frozen crystals (100°K) at the APS beamline 19BM or at the NSLS-II beamline 17-ID-2 (FMX). The data were integrated and processed by HKL3000 (19BM) or XDS (FMX), respectively, and scaled by AIMLESS/CCP4 package (Kabsch, 2010, Minor et al., 2006, Winn et al., 2011). The crystal structure for HLA-B*07:02 in complex with pMLL_{747 − 755} peptide was solved by molecular replacement using PHASER (McCoy et al., 2007) and the PDB structure 5WMN as a search model, and then refined by REFMAC (Murshudov et al., 2011) and finally, fixed manually using COOT visualization, refinement and validation software (Emsley et al., 2010). The refined structure for pMLL_{747 − 755} /HLA-B*0702 complex served as a search model to solve other pMHC structures described in this report. The TCR27 crystal structure was solved by MR using the PDB file 1OGA as a search model and refined as described above. All the models and the corresponding structure factor files have been validated using a validation server at the PDB. The X-ray data collection and refinement statistics are presented in Table S1 altogether with PDB IDs for all the deposited structures. When the precipitant A was utilized in crystallization of pMHC complexes with different peptides, it resulted in isomorphous crystals (Table S1).

The structure-based pairwise alignment between sets of atomic coordinates was carried out using a PYMOL align algorithm or “LSQ Superpose” command in COOT. Sequence alignment was performed by NCBI BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi). Output data from PYMOL alignment were utilized to produce RMSD plots in MS Excel. The “core RMSD” values were calculated using 5 refinement cycles for each alignment in PYMOL, and it was carried out upon condition that less than 20% of the aligned atoms were excluded as outliers. Solvent exposed surface area (SASA, A²) for individual amino acid residues was calculated using an EBI PISA software (Krissinel and Henrick, 2007), and it was expressed as a percentage out of all residue surface area (ASA). Electron density maps (in CCP4 format) were generated in COOT. Figures were produced using PYMOL, Microsoft Excel and Biorender (Perkel, 2020).

Triple-resonance experiments for NMR backbone assignments

NMR analysis of large protein molecules is always confounded by its size, asymmetry and dynamical nature (Purslow et al., 2020). To overcome such complexities and resulting signal relaxation that leads to poor spectra, HLA-B*0702 molecules were prepared in a 100% deuterated minimal media with either ¹⁵N or ¹⁵N + ¹³C labels and refolded with unlabeled hβ2M and peptide as follows.

Triple labeled HLA-B*0702 (¹⁵N, ¹³C, ²H) was produced as inclusion bodies in E.coli by supplementing M9 minimal media with ¹⁵NH₄Cl, ¹³C,²H-glucose, ²H,¹⁵N,¹³C-celtone (Cambridge Isotopes) and 100% D₂O (Gardner and Kay, 1998). Protein production was induced with 1mM IPTG at culture OD₆₆₀ = 0.6 and then continued for 4 hours at 37 °C after that. Bacteria were pelleted by centrifugation, the inclusion bodies were purified and used in pMHC refolding along with unlabeled hβ2M and peptide, respectively. Refolded pMHC was purified by SEC in PBS, pH 7.4, as mentioned above. The buffer in a protein sample was exchanged with 25mM MES, 150mM NaCl, pH 6.5. A final pMHC concentration of 200µM was used in NMR experiments.

TROSY-based triple resonance experiments were performed as suggested (Salzmann et al., 1998) on the perdeuterated,¹⁵N,¹³C pMLL_{747 − 755}/HLA-B*0702 sample at 30°C on 600 MHz (Bruker AVANCE III, New York University) and 800 MHz (Bruker AVANCE III HD, New York Structural Biology Center) spectrometers, equipped with cryoprobes to obtain the backbone atoms N, HN, Cα, C’ assignments. The NMR data were processed by NMRPipe (Delaglio et al., 2017) and analyzed using NMRViewJ (Johnson, 2004). Each peak in the ¹⁵N-TROSY spectra corresponded to one amide (N-H) pair belonging to a protein amino acid. About 86% of all assignable (excluding proline residues) backbone N and HN atoms were assigned. Further, about 84% of the alpha-helical residues that interact with pMLL_{747 − 755} peptide and TCR27 were assigned. The assignments were transferred without ambiguity to the ¹⁵N-TROSY spectra of pMLL_{747 − 755}/¹⁵N,²H-HLA-B*0702 obtained in HST buffer containing 10mM HEPES, 100mM NaCl, 0.1% P-20 (surfactant), pH 7.5. The same buffer was used in the BLI experiments.

Production of pMLL_{747 − 755}/HLA-B*0702 with labeled individual amino acids

The pMLL_{747 − 755}/HLA-B*0702 complexes with ¹⁵N-Arg, ¹⁵N-Leu or ¹⁵N-Tyr labels were prepared by growing E.coli expressing HLA-B*0702 in M9 minimal media containing unlabeled ingredients including unlabeled amino acids except for individual ¹⁵N-labeled amino acids listed above. Each complex was refolded with unlabeled hβ2M and unlabeled pMLL_{747 − 755} peptide and purified as mentioned above. These protein samples were used to resolve NMR assignment ambiguities and peak overlaps.

Production and chemical shift perturbation (CSP) analysis of the isotopically labeled pMLL_{747 − 755}/HLA-B0702 and MLL_{747 − 755}/HLA-B0702 complexes

Perdeuterated (²H), ¹⁵N-labeled HLA-B*0702 was produced as inclusion bodies in E.coli by supplementing M9 minimal media with ¹⁵NH₄Cl, ²H-glucose, ²H, ¹⁵N-celtone (Cambridge Isotopes) and 100% D₂O. Protein production was induced with 1mM IPTG when culture OD₆₆₀ was 0.6 and incubation continued for 4 hours at 37°C after induction. Each complex was produced by refolding labeled HLA-B*0702 in the presence of unlabeled hβ2M and unlabeled pMLL_{747 − 755} or unlabeled MLL_{747 − 755} peptide, respectively, and purified as mentioned above. The buffer in protein samples was exchanged into HST buffer, and the final protein concentrations were of 216.4µM (pMLL₇₄₇/755-HLA-B*0702) and 200µM (MLL_{747 − 755}/HLA-B*0702), respectively. ¹⁵N-TROSY data were collected using a 600 MHz NMR spectrometer at 25 °C. Assignments were transferred from the ¹⁵N-TROSY spectra of pMLL_{747 − 755}/HLA-B*0702 to ¹⁵N-TROSY spectra of MLL_{747 − 755}/HLA-B*0702. The scaled (normalized) chemical shift differences between the HLA residues in both spectra were calculated using the formula, δ_ppm=((δH)²+0.11(δN)²)^1/2 (Natarajan et al., 2016). Residues with significant CSP values were defined to have the chemical shift difference (δ_ppm) greater than 1σ above the mean value.

Production of isotopically labeled TCR27 and mapping the pMHC-TCR interface

Perdeuterated α and β subunits of TCR27 were produced separately as inclusion bodies in E.coli by using M9 minimal media containing ²H-glucose, ²H-celtone (Cambridge isotopes) and 100% D₂O. Protein expression was induced with 1mM IPTG when culture OD₆₆₀ was 0.6, and continued for 4–5 hours at 37°C after that. Cells were collected and inclusions bodies isolated, the TCRαβ heterodimer was refolded and purified as mentioned above. The deuterated TCR27 sample was exchanged into HST buffer for chemical shift perturbation experiments.

Titrations of perdeuterated TCR27 with perdeuterated pMLL_{747 − 755}/¹⁵N-HLA-B*0702 was performed on a 600 MHz spectrometer (Bruker AVANCE III) at 25 °C. ¹⁵N-TROSY was collected for samples with the following pMLL_{747 − 755}/HLA-B*0702 and TCR27 ratios, 100µM:0µM (1:0), 93µM:23.4µM (1:0.25), 88µM:44µM (1:0.5) and 78.5µM:78.5µM (1:1), respectively. Another ¹⁵N-TROSY spectrum was collected for the binary pMLL_{747 − 755}/HLA-B*0702 complex alone at 78.5µM concentration for CSP and peaks intensity analyses. Residues with overlapping spectral peaks were excluded from calculations and subsequent analyses. Assignments were carefully transferred from the pMHC (pMLL_{747 − 755}/HLA-B*0702, 78.5µM) spectrum to the TCR27 plus pMHC (78.5µM:78.5µM) spectrum. The scaled chemical shift differences (δ_ppm) were calculated using the same formula as in the previous section. Residues with significant CSPs were defined to have the chemical shift difference (δ_ppm) greater that 1σ above the mean value. Residues with significant signal intensity change were with a peak relative volume (V/V₀) value higher than 1σ below the mean value.

TCR-pMHC docking using Haddock 2.4

The protein-protein information-driven docking between pMHC and TCR27 was performed using HADDOCK software (Rodrigues JP, 2012). Models were generated using the HADDOCK webserver (https://wenmr.science.uu.nl) (van Zundert G, 2016) version 2.4, using docking scenarios which differed by selection of interacting amino acid residues at the TCR-pMHC interface. In each docking experiment the following settings were used: top scored 1000 models were selected at the rigid-body (it0) stage and the top scored 200 models for the flexible (it1) and water refinement steps. The random removal of restraints was set to 50% for each docking run. The interacting residues were specified as either active or passive residues (such as directly involved in TCR-MHC interface or indirectly – as neighboring residues, respectively) to estimate the docking bias. The pMHC residues were selected using the data obtained by site-directed mutagenesis, epitope mapping, NMR and crystallographic data described in this report. The TCR residues were selected at the extremities of the CDR loops, and they have been identified using sequence and structural similarities between TCR27 and the TCR 3D database (https://tcr3d.ibbr.umd.edu/). In every Haddock run, TCR27 was assigned as a receptor, and pMHC as a ligand. The models for docking were generated from the crystal structures produced in this study, including the two TCR27 distinct conformations TCR_1 and TCR_2, respectively (see Table S1). To reduce running time, every pMHC model was truncated to include the HLA-B*0702 domain 1 (AA residues 1-180) in complex with corresponding epitope (pMLL_{747 − 755} or pDOT1L_{998 − 1006}, respectively). Each truncated TCR model included the variable Vα and Vβ domains only (AA residues 1-110 from each TCR chain). The following interface residues were selected in the docking scenarios 1–4: #1 – the epitope residues P₄-P₇ and the CDR3 loops, #2 – the epitope residues P₄-P₇ and all the CDR loops, #3 – the epitope residues P₄-P₇, NMR-mapped HLA residues (with SASA > 20%) and the CDR3 loops, #4 – the epitope residues P₄-P₇, the NMR-mapped pMHC residues and all 3 CDR loops. Further details are provided in the results section. Top 4 clusters with highest HADDOCK scores from each run were analyzed, and the top scoring clusters from every scenario were compared between each other to validate the docking strategy and identify the correct TCR-pMHC model.

ACKNOWLEDGEMENTS

We thank Duane Moogk (McMaster University) for critical reading of the manuscript. This work was supported by the NIH grant NIGMS R01 GM124489 (to M.K.), R01 CA243486 (to M.K) and a Sponsored Research Agreement from Agenus to M.K.

Results shown in this report are partially derived from work performed at Argonne National Laboratory, Structural Biology Center at the Advanced Photon Source. SBC is operated by UChicago Argonne, LLC, for the U.S. Department of Energy, Office of Biological and Environmental Research under contract DE-AC02-06CH11357.

Results in this report are partially derived from work performed at The Center for BioMolecular Structure (CBMS) primarily supported by the National Institutes of Health, National Institute of General Medical Sciences (NIGMS) through a Center Core P30 Grant (P30GM133893), and by the DOE Office of Biological and Environmental Research (KP1607011). As part of NSLS-II, a national user facility at Brookhaven National Laboratory, work performed at the CBMS is supported in part by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences Program under contract number and DE-SC0012704.

CONFLICT OF INTEREST

All authors have directly participated in the planning, execution, or analysis of the study and drafting or revising of the manuscript, and have read and approved the final version of the manuscript. YP, LP, SN, AN and MK have no financial interests to disclose. DU, BM, XM, BJ, CB, OH, SG, FS, RBS, MAF, AH, MvD, JSB have ownership of equity securities and/or are currently or previously employed by Agenus. This does not alter adherence to policy of the journal on sharing the data and materials.

AUTHORS CONTRIBUTIONS

Experiments were conceptualized by YP, AN, LP, XM DU and MK. DU and MK supervised the whole project. X-Ray crystallography sample preparation, data collection, structure refinement and analysis were performed by YP, LP, and SN. NMR sample preparation, data collection and analysis were performed by AN and YP. Biolayer interferometry sample preparation, data collection and analysis were performed by YP, LP and SN. Site-directed mutagenesis was performed by SN. Haddock docking and analysis were performed by YP and LP. The T-cell related studies and peptide synthesis: BJ, BM, CB, OH, SG, FS (investigation); BJ, BM, CB, OH, SG ( data analysis and validation); BJ, BM, CB, FS, MAF, JB, AH, MvD (resources); BM, CB (visualization); BJ, BM and MAF (supervision); BS provided critical insight and advice during the studies; XM conceived, designed and performed the cell-based experiments. The draft was written and edited by YP, AN, LP, BJ, BM, DU and MK.

ALPIZAR, A., MARINO, F., RAMOS-FERNANDEZ, A., LOMBARDIA, M., JEKO, A., PAZOS, F., PARADELA, A., SANTIAGO, C., HECK, A. J. & MARCILLA, M. 2017. A Molecular Basis for the Presentation of Phosphorylated Peptides by HLA-B Antigens. Mol Cell Proteomics, 16, 181-193.

AMANCHY, R., PERIASWAMY, B., MATHIVANAN, S., REDDY, R., TATTIKOTA, S.G. AND PANDEY, A., 2007. A curated compendium of phosphorylation motifs. Nature biotechnology, 25(3), pp.285-286.

BASSANI-STERNBERG, M., BRAUNLEIN, E., KLAR, R., ENGLEITNER, T., SINITCYN, P., AUDEHM, S., STRAUB, M., WEBER, J., SLOTTA-HUSPENINA, J., SPECHT, K., MARTIGNONI, M. E., WERNER, A., HEIN, R., D, H. B., PESCHEL, C., RAD, R., COX, J., MANN, M. & KRACKHARDT, A. M. 2016. Direct identification of clinically relevant neoepitopes presented on native human melanoma tissue by mass spectrometry. Nat Commun, 7, 13404.

BEERBAUM, M., BALLASCHK, M., ERDMANN, N., SCHNICK, C., DIEHL, A., UCHANSKA-ZIEGLER, B., ZIEGLER, A. & SCHMIEDER, P. 2013. NMR spectroscopy reveals unexpected structural variation at the protein-protein interface in MHC class I molecules. J Biomol NMR, 57, 167-78.

BECKETT D, KOVALEVA E, SCHATZ PJ. A minimal peptide substrate in biotin holoenzyme synthetase-catalyzed biotinylation. Protein Sci. 1999 Apr;8(4):921-9. doi: 10.1110/ps.8.4.921. PMID: 10211839; PMCID: PMC2144313.

BLASS, E. & OTT, P. A. 2021. Advances in the development of personalized neoantigen-based therapeutic cancer vaccines. Nat Rev Clin Oncol, 18, 215-229.

BOULTER, J. M., GLICK, M., TODOROV, P. T., BASTON, E., SAMI, M., RIZKALLAH, P. & JAKOBSEN, B. K. 2003. Stable, soluble T-cell receptor molecules for crystallization and therapeutics. Protein Eng, 16, 707-11.

COBBOLD, M., DE LA PENA, H., NORRIS, A., POLEFRONE, J. M., QIAN, J., ENGLISH, A. M., CUMMINGS, K. L., PENNY, S., TURNER, J. E., COTTINE, J., ABELIN, J. G., MALAKER, S. A., ZARLING, A. L., HUANG, H. W., GOODYEAR, O., FREEMAN, S. D., SHABANOWITZ, J., PRATT, G., CRADDOCK, C., WILLIAMS, M. E., HUNT, D. F. & ENGELHARD, V. H. 2013. MHC class I-associated phosphopeptides are the targets of memory-like immunity in leukemia. Sci Transl Med, 5, 203ra125.

DE CASTRO, E., SIGRIST, C. J., GATTIKER, A., BULLIARD, V., LANGENDIJK-GENEVAUX, P. S., GASTEIGER, E., BAIROCH, A. & HULO, N. 2006. ScanProsite: detection of PROSITE signature matches and ProRule-associated functional and structural residues in proteins. Nucleic Acids Res, 34, W362-5.

DELAGLIO, F., WALKER, G. S., FARLEY, K. A., SHARMA, R., HOCH, J. C., ARBOGAST, L. W., BRINSON, R. G. & MARINO, J. P. 2017. Non-Uniform Sampling for All: More NMR Spectral Quality, Less Measurement Time. Am Pharm Rev, 20.

DOMINGUEZ, C., BOELENS, R. AND BONVIN, A.M. 2003. HADDOCK: a protein− protein docking approach based on biochemical or biophysical information. Journal of the American Chemical Society, 125(7), pp.1731-1737.

ELLIOTT, T. S., SLOWEY, A., YE, Y. & CONWAY, S. J. 2012. The use of phosphate bioisosteres in medicinal chemistry and chemical biology. MedChemComm, 3, 735-751.

EMSLEY, P., LOHKAMP, B., SCOTT, W. G. & COWTAN, K. 2010. Features and development of Coot. Acta Crystallogr D Biol Crystallogr, 66, 486-501.

ENGELHARD, V. H., OBENG, R. C., CUMMINGS, K. L., PETRONI, G. R., AMBAKHUTWALA, A. L., CHIANESE-BULLOCK, K. A., SMITH, K. T., LULU, A., VARHEGYI, N., SMOLKIN, M. E., MYERS, P., MAHONEY, K. E., SHABANOWITZ, J., BUETTNER, N., HALL, E. H., HADEN, K., COBBOLD, M., HUNT, D. F., WEISS, G., GAUGHAN, E. & SLINGLUFF, C. L., JR. 2020. MHC-restricted phosphopeptide antigens: preclinical validation and first-in-humans clinical trial in participants with high-risk melanoma. J Immunother Cancer, 8.

FINDEIS, M. A. 2017. Protected derivatives of 2-amino-4-phosphobutanoic acid.

GARDNER, K. H. & KAY, L. E. 1998. The use of 2H, 13C, 15N multidimensional NMR to study the structure and dynamics of proteins. Annu Rev Biophys Biomol Struct, 27, 357-406.

GOWTHAMAN, R. & PIERCE, B. G. 2019. TCR3d: The T cell receptor structural repertoire database. Bioinformatics, 35, 5323-5325.

GOWTHAMAN, R. & PIERCE, B. G. 2020. Modeling and Viewing T Cell Receptors Using TCRmodel and TCR3d. Methods Mol Biol, 2120, 197-212.

HARNDAHL, M., JUSTESEN, S., LAMBERTH, K., RODER, G., NIELSEN, M. & BUUS, S. 2009. Peptide binding to HLA class I molecules: homogenous, high-throughput screening, and affinity assays. J Biomol Screen, 14, 173-80.

HORNBECK, P. V., KORNHAUSER, J. M., LATHAM, V., MURRAY, B., NANDHIKONDA, V., NORD, A., SKRZYPEK, E., WHEELER, T., ZHANG, B. & GNAD, F. 2019. 15 years of PhosphoSitePlus(R): integrating post-translationally modified sites, disease variants and isoforms. Nucleic Acids Res, 47, D433-D441.

HORNBECK, P. V., ZHANG, B., MURRAY, B., KORNHAUSER, J. M., LATHAM, V. & SKRZYPEK, E. 2015. PhosphoSitePlus, 2014: mutations, PTMs and recalibrations. Nucleic Acids Res, 43, D512-20.

JOHNSON, B. A. 2004. Using NMRView to visualize and analyze the NMR spectra of macromolecules. Methods Mol Biol, 278, 313-52.

KABSCH, W. 2010. Integration, scaling, space-group assignment and post-refinement. Acta Crystallogr D Biol Crystallogr, 66, 133-44.

KARAPETYAN, A. R., CHAIPAN, C., WINKELBACH, K., WIMBERGER, S., JEONG, J. S., JOSHI, B., STEIN, R. B., UNDERWOOD, D., CASTLE, J. C., VAN DIJK, M. & SEIBERT, V. 2019. TCR Fingerprinting and Off-Target Peptide Identification. Front Immunol, 10, 2501.

KRISSINEL, E. & HENRICK, K. 2007. Inference of macromolecular assemblies from crystalline state. J Mol Biol, 372, 774-97.

LI, Y., DEPONTIEU, F. R., SIDNEY, J., SALAY, T. M., ENGELHARD, V. H., HUNT, D. F., SETTE, A., TOPALIAN, S. L. & MARIUZZA, R. A. 2010. Structural basis for the presentation of tumor-associated MHC class II-restricted phosphopeptides to CD4+ T cells. J Mol Biol, 399, 596-603.

LIN, M. H., SHEN, K. Y., LIU, B. S., CHEN, I. H., SHER, Y. P., TSENG, G. C., LIU, S. J. & SUNG, W. C. 2019. Immunological evaluation of a novel HLA-A2 restricted phosphopeptide of tumor associated Antigen, TRAP1, on cancer therapy. Vaccine X, 1, 100017.

LUCAS, C. M., HARRIS, R. J., HOLCROFT, A. K., SCOTT, L. J., CARMELL, N., MCDONALD, E., POLYDOROS, F. & CLARK, R. E. 2015. Second generation tyrosine kinase inhibitors prevent disease progression in high-risk (high CIP2A) chronic myeloid leukaemia patients. Leukemia, 29, 1514-23.

LUKSZA, M., RIAZ, N., MAKAROV, V., BALACHANDRAN, V. P., HELLMANN, M. D., SOLOVYOV, A., RIZVI, N. A., MERGHOUB, T., LEVINE, A. J., CHAN, T. A., WOLCHOK, J. D. & GREENBAUM, B. D. 2017. A neoantigen fitness model predicts tumour response to checkpoint blockade immunotherapy. Nature, 551, 517-520.

MAHONEY, K. E., SHABANOWITZ, J. & HUNT, D. F. 2021. MHC Phosphopeptides: Promising Targets for Immunotherapy of Cancer and Other Chronic Diseases. Mol Cell Proteomics, 20, 100112.

MCCOY, A. J., GROSSE-KUNSTLEVE, R. W., ADAMS, P. D., WINN, M. D., STORONI, L. C. & READ, R. J. 2007. Phaser crystallographic software. J Appl Crystallogr, 40, 658-674.

MERCHER, T. & SCHWALLER, J. 2019. Pediatric Acute Myeloid Leukemia (AML): From Genes to Models Toward Targeted Therapeutic Intervention. Front Pediatr, 7, 401.

MERWE, P. ANTON VAN DER, AND SIMON J. DAVIS. "Molecular interactions mediating T cell antigen recognition." Annual review of immunology 21, no. 1 (2003): 659-684.

MINOR, W., CYMBOROWSKI, M., OTWINOWSKI, Z. & CHRUSZCZ, M. 2006. HKL-3000: the integration of data reduction and structure solution--from diffraction images to an initial model in minutes. Acta Crystallogr D Biol Crystallogr, 62, 859-66.

MIZUKOSHI, Y., TAKEUCHI, K., TOKUNAGA, Y., MATSUO, H., IMAI, M., FUJISAKI, M., KAMOSHIDA, H., TAKIZAWA, T., HANZAWA, H. & SHIMADA, I. 2020. Targeting the cryptic sites: NMR-based strategy to improve protein druggability by controlling the conformational equilibrium. Sci Adv, 6.

MOHAMMED, F., COBBOLD, M., ZARLING, A. L., SALIM, M., BARRETT-WILT, G. A., SHABANOWITZ, J., HUNT, D. F., ENGELHARD, V. H. & WILLCOX, B. E. 2008. Phosphorylation-dependent interaction between antigenic peptides and MHC class I: a molecular basis for the presentation of transformed self. Nat Immunol, 9, 1236-43.

MOHAMMED, F., STONES, D. H., ZARLING, A. L., WILLCOX, C. R., SHABANOWITZ, J., CUMMINGS, K. L., HUNT, D. F., COBBOLD, M., ENGELHARD, V. H. & WILLCOX, B. E. 2017. The antigenic identity of human class I MHC phosphopeptides is critically dependent upon phosphorylation status. Oncotarget, 8, 54160-54172.

MURSHUDOV, G. N., SKUBAK, P., LEBEDEV, A. A., PANNU, N. S., STEINER, R. A., NICHOLLS, R. A., WINN, M. D., LONG, F. & VAGIN, A. A. 2011. REFMAC5 for the refinement of macromolecular crystal structures. Acta Crystallogr D Biol Crystallogr, 67, 355-67.

NATARAJAN, A., NADARAJAH, V., FELSOVALYI, K., WANG, W., JEYACHANDRAN, V. R., WASSON, R. A., CARDOZO, T., BRACKEN, C. & KROGSGAARD, M. 2016. Structural Model of the Extracellular Assembly of the TCR-CD3 Complex. Cell Rep, 14, 2833-45.

NATARAJAN, K., MCSHAN, A. C., JIANG, J., KUMIROV, V. K., WANG, R., ZHAO, H., SCHUCK, P., TILAHUN, M. E., BOYD, L. F., YING, J., BAX, A., MARGULIES, D. H. & SGOURAKIS, N. G. 2017. An allosteric site in the T-cell receptor Cbeta domain plays a critical signalling role. Nat Commun, 8, 15260.

PAISLEY T. MYERS, ERIN D. JEFFERY, BENJAMIN MORIN, CHRISTINE BRITTSAN, BISHNU JOSHI, ANTOINE TANNE KAREN SMITH, ALBERT HURWITZ, KARA L. CUMMINGS, AMANDA M. LULU, DANIEL L. LEVEY, LINH MACH, MARK A. FINDEIS, GAYLE C. HUNT, JEFFREY SHABANOWITZ, DONALD F. HUNT, ARTHUR A. HURWITZ5 ROBERT N. STEIN, ALEX DUNCAN, DENNIS UNDERWOOD. 2018. 33rd Annual Meeting & Pre-Conference Programs of the Society for Immunotherapy of Cancer (SITC 2018). Journal for ImmunoTherapy of Cancer, 6(1), p.114.

PEACOCK, T. & CHAIN, B. 2021. Information-Driven Docking for TCR-pMHC Complex Prediction. Front Immunol, 12, 686127.

PENNY, S. A., ABELIN, J. G., MALAKER, S. A., MYERS, P. T., SAEED, A. Z., STEADMAN, L. G., BAI, D. L., WARD, S. T., SHABANOWITZ, J., HUNT, D. F. & COBBOLD, M. 2021. Tumor Infiltrating Lymphocytes Target HLA-I Phosphopeptides Derived From Cancer Signaling in Colorectal Cancer. Front Immunol, 12, 723566.

PERKEL, J. M. 2020. The software that powers scientific illustration. Nature, 582, 137-138.

PETERSEN, J., WURZBACHER, S. J., WILLIAMSON, N. A., RAMARATHINAM, S. H., REID, H. H., NAIR, A. K., ZHAO, A. Y., NASTOVSKA, R., RUDGE, G., ROSSJOHN, J. & PURCELL, A. W. 2009. Phosphorylated self-peptides alter human leukocyte antigen class I-restricted antigen presentation and generate tumor-specific epitopes. Proc Natl Acad Sci U S A, 106, 2776-81.

PURSLOW, J. A., KHATIWADA, B., BAYRO, M. J. & VENDITTI, V. 2020. NMR Methods for Structural Characterization of Protein-Protein Complexes. Front Mol Biosci, 7, 9.

RANGARAJAN, S., HE, Y., CHEN, Y., KERZIC, M. C., MA, B., GOWTHAMAN, R., PIERCE, B. G., NUSSINOV, R., MARIUZZA, R. A. & ORBAN, J. 2018. Peptide-MHC (pMHC) binding to a human antiviral T cell receptor induces long-range allosteric communication between pMHC- and CD3-binding sites. J Biol Chem, 293, 15991-16005.

RODRIGUES, J. P., TRELLET, M., SCHMITZ, C., KASTRITIS, P., KARACA, E., MELQUIOND, A. S. & BONVIN, A. M. 2012. Clustering biomolecular complexes by residue contacts similarity. Proteins, 80, 1810-7.

ROONEY, J. W., SUN, Y. L., GLIMCHER, L. H. & HOEY, T. 1995. Novel NFAT sites that mediate activation of the interleukin-2 promoter in response to T-cell receptor stimulation. Mol Cell Biol, 15, 6299-310.

RUDOLPH, M. G., STANFIELD, R. L. & WILSON, I. A. 2006. HOW TCRS BIND MHCS, PEPTIDES, AND CORECEPTORS. Annual Review of Immunology, 24, 419-466.

SALZMANN, M., PERVUSHIN, K., WIDER, G., SENN, H. & WUTHRICH, K. 1998. TROSY in triple-resonance experiments: new perspectives for sequential NMR assignment of large proteins. Proc Natl Acad Sci U S A, 95, 13585-90.

SZETO, C., LOBOS, C.A., NGUYEN, A.T. AND GRAS, S. 2020. TCR recognition of peptide–MHC-I: Rule makers and breakers. International journal of molecular sciences, 22(1), p.68.

SOLLEDER, M., GUILLAUME, P., RACLE, J., MICHAUX, J., PAK, H. S., MULLER, M., COUKOS, G., BASSANI-STERNBERG, M. & GFELLER, D. 2020. Mass Spectrometry Based Immunopeptidomics Leads to Robust Predictions of Phosphorylated HLA Class I Ligands. Mol Cell Proteomics, 19, 390-404.

VAN ZUNDERT G, RODRIGUES J, TRELLET M, SCHMITZ C, KASTRITIS P, KARACA E, ET AL. The HADDOCK2.2 Web Server: User-Friendly Integrative Modeling of Biomolecular Complexes. J Mol Biol (2016) 428:720–5. doi: 10.1016/j.jmb.2015.09.014

VARANI, L., BANKOVICH, A. J., LIU, C. W., COLF, L. A., JONES, L. L., KRANZ, D. M., PUGLISI, J. D. & GARCIA, K. C. 2007. Solution mapping of T cell receptor docking footprints on peptide-MHC. Proc Natl Acad Sci U S A, 104, 13080-5.

WILLIAMSON, M.P., 2013. Using chemical shift perturbation to characterize ligand binding. Progress in nuclear magnetic resonance spectroscopy, 73, pp.1-16.

WINN, M. D., BALLARD, C. C., COWTAN, K. D., DODSON, E. J., EMSLEY, P., EVANS, P. R., KEEGAN, R. M., KRISSINEL, E. B., LESLIE, A. G., MCCOY, A., MCNICHOLAS, S. J., MURSHUDOV, G. N., PANNU, N. S., POTTERTON, E. A., POWELL, H. R., READ, R. J., VAGIN, A. & WILSON, K. S. 2011. Overview of the CCP4 suite and current developments. Acta Crystallogr D Biol Crystallogr, 67, 235-42.

WU, J. Q., GUO, J. Y., TANG, W., YANG, C. S., FREEL, C. D., CHEN, C., NAIRN, A. C. & KORNBLUTH, S. 2009. PP1-mediated dephosphorylation of phosphoproteins at mitotic exit is controlled by inhibitor-1 and PP1 phosphorylation. Nat Cell Biol, 11, 644-51.

YEN, Y. T., CHIEN, M., WU, P. Y., HO, C. C., HO, C. T., HUANG, K. C., CHIANG, S. F., CHAO, K. S. C., CHEN, W. T. & HUNG, S. C. 2021. Protein phosphatase 2A inactivation induces microsatellite instability, neoantigen production and immune response. Nat Commun, 12, 7297.

ZARLING, A. L., FICARRO, S. B., WHITE, F. M., SHABANOWITZ, J., HUNT, D. F. & ENGELHARD, V. H. 2000. Phosphorylated peptides are naturally processed and presented by major histocompatibility complex class I molecules in vivo. J Exp Med, 192, 1755-62.

ZARLING, A. L., POLEFRONE, J. M., EVANS, A. M., MIKESH, L. M., SHABANOWITZ, J., LEWIS, S. T., ENGELHARD, V. H. & HUNT, D. F. 2006. Identification of class I MHC-associated phosphopeptides as targets for cancer immunotherapy. Proc Natl Acad Sci U S A, 103, 14889-94.

Supplementalinformation.docx
Supplemental figures and figure legends
TableS1.docx
Supplemental Table S1

Download PDF

Version 1

posted

You are reading this latest preprint version

Molecular mechanism of phosphopeptide neoantigen immunogenicity

Status:

Version 1

Abstract

Figures

Introduction

Main

Phosphoserine residue of pMLL_{747 − 755} is essential for activation of the TCR27 + T cells

2. The phosphate group is essential for the TCR27-pMHC complex stability

6. TCR27 is structurally similar to other human TCRs but displays conformational heterogeneity

7. Trosy-nmr

7.2. TROSY-NMR mapping of the TCR-pMHC interface

8. The pMLL_{747 − 755}/HLA-B*0702/TCR27 ternary complex reveals the molecular mechanism of phosphopeptide immunogenicity

Discussion

Methods

Cell Lines, Plasmids, Affinity and T cell activation assay

Cytotoxicity assay

Identification of the off-target phosphopeptide pDOT1L_{998 − 1006}

Peptides

Protein sequences

Biolayer interferometry

Protein crystallization, structure solution and refinement

Triple-resonance experiments for NMR backbone assignments

Production of pMLL_{747 − 755}/HLA-B*0702 with labeled individual amino acids

Production and chemical shift perturbation (CSP) analysis of the isotopically labeled pMLL_{747 − 755}/HLA-B0702 and MLL_{747 − 755}/HLA-B0702 complexes

Production of isotopically labeled TCR27 and mapping the pMHC-TCR interface

TCR-pMHC docking using Haddock 2.4

Declarations

References

Supplementary Files

Status:

Version 1

Molecular mechanism of phosphopeptide neoantigen immunogenicity

Status:

Version 1

Abstract

Figures

Introduction

Main

Phosphoserine residue of pMLL747 − 755 is essential for activation of the TCR27 + T cells

2. The phosphate group is essential for the TCR27-pMHC complex stability

6. TCR27 is structurally similar to other human TCRs but displays conformational heterogeneity

7. Trosy-nmr

7.2. TROSY-NMR mapping of the TCR-pMHC interface

8. The pMLL747 − 755/HLA-B*0702/TCR27 ternary complex reveals the molecular mechanism of phosphopeptide immunogenicity

Discussion

Methods

Cell Lines, Plasmids, Affinity and T cell activation assay

Cytotoxicity assay

Identification of the off-target phosphopeptide pDOT1L998 − 1006

Peptides

Protein sequences

Biolayer interferometry

Protein crystallization, structure solution and refinement

Triple-resonance experiments for NMR backbone assignments

Production of pMLL747 − 755/HLA-B*0702 with labeled individual amino acids

Production and chemical shift perturbation (CSP) analysis of the isotopically labeled pMLL747 − 755/HLA-B*0702 and MLL747 − 755/HLA-B*0702 complexes

Production of isotopically labeled TCR27 and mapping the pMHC-TCR interface

TCR-pMHC docking using Haddock 2.4

Declarations

References

Supplementary Files

Status:

Version 1

Phosphoserine residue of pMLL_{747 − 755} is essential for activation of the TCR27 + T cells

8. The pMLL_{747 − 755}/HLA-B*0702/TCR27 ternary complex reveals the molecular mechanism of phosphopeptide immunogenicity

Identification of the off-target phosphopeptide pDOT1L_{998 − 1006}

Production of pMLL_{747 − 755}/HLA-B*0702 with labeled individual amino acids

Production and chemical shift perturbation (CSP) analysis of the isotopically labeled pMLL_{747 − 755}/HLA-B0702 and MLL_{747 − 755}/HLA-B0702 complexes