Peptides Derived from the CMV Proteome Mimic Unique Stem Cell Transplant Recipient Specific Peptides: Possible Role in Eliciting a Pro-Gvh T Cell Response

   Stem cell transplant (SCT) recipients who develop human cytomegalovirus (hCMV) reactivation are at risk of developing graft versus host disease (GVHD).  This may stem from immune cross reactivity (Figure 1) towards both pathogen-derived peptides and nearly identical recipient-derived alloreactive peptide minor histocompatibility antigens from SCT donor-recipient pairs (DRP). Whole exome sequencing was performed on 9 SCT DRP, and the resulting nucleotide sequences aligned and compared, identifying all the single nucleotide polymorphisms (SNPs) present in the recipient and absent in the donor. Alloreactive peptide libraries were compiled for each DRP (9-mer peptides) and HLA binding affinity calculated for relevant HLA (Front Immunol 2014). A library representing the hCMV proteome was generated from the NCBI Protein database and directly compared to libraries of recipient specific peptide-HLA complexes from the SCT D-R pairs. BLAST Protein sequence alignment was performed to interrogate each pair's peptide-HLA complex library for matches with hCMV peptides, identifying CMV peptides with a 6/6-9/9 amino acid sequence matching to recipient peptides bound to HLA (IC50<500). The resulting matched CMV peptide libraries were completed with flanking amino acids from the original hCMV protein sequences to generate a match library of 9-mer peptides, evaluated for binding to HLA Class I using the netMHCpan algorithm 2.8 (IC50<500). Class I HLA molecules were interrogated for both alloreactive and CMV-derived peptide binding affinity in each patient. 18 HLA molecules bound both types of peptides, of which 7 HLA had more than 5 matches, i.e. human peptide(s) with 6-9 consecutive amino acid sequence identity with CMV peptide(s), where multiple matches were possible. These included: HLA-C*03:03, C*03:04, A*02:01, B*15:03, B*27:05, B*07:02 and B*15:16. Upon peptide library comparison, a median 27 (Range: 3-40) unique CMV peptides/patient matched 22 (Range: 3-30) recipient peptides from 5 matched unrelated DRP (MUD) with corresponding HLA specific binding IC50<500. Matched related DRP (MRD; n=4) comparison yielded a median 7 (Range: 1-21) unique CMV peptides/patient matching 6.5 (Range: 1-19) recipient peptides. This pattern of CMV-matched alloreactive peptides in MUD demonstrated a 3-4 fold higher degree of potential cross reactivity (Median peptides/patient: p=0.048) than MRD. MUD patients with HLA-C*03:03 & C*03:04 specificity (n=4) produced a median 20 (Range: 19-22) unique CMV peptides matching 14.5 (Range: 12-19) potentially cross-reactive recipient peptides. Cross-reactive recipient peptides were capable of matching with up to 7 different CMV peptides (Range: 1-7) and alternatively matching CMV peptides with up to 5 different recipient peptides (Range: 1-5), indicating this phenomenon could promote various strength cross-reactive immune responses (within and outside the IC50<500 range) even from a single matched CMV peptide. Tissue distribution mapping (Genotype-Tissue Expression project, GTEx) of the corresponding DRP peptide source gene expression (human mRNAs) by major tissue group indicated an overlap with known GVHD target organs from CMV-matched alloreactive recipient peptides. CMV is known to infect the vascular endothelium where cross-reactive immune cells, and potentially cross-reactive memory T cell populations, interacting with the target cells of interest would have the greatest opportunity to impact baseline inflammation and potentially trigger graft-versus-host disease (GVHD). We propose that from in silico demonstration of sequence identity between DRP polymorphic peptides with CMV proteins that immune cross-reactivity may result in those DRP and cause recipient cells to be targeted by CMV specific T cells with pro-GVH implications.