Introduction: Our understanding of T cell-mediated alloresponses has been limited by the lack of information about allogeneic pMHC epitopes. Recently, our group identified over 40 H-2Kb-peptide epitopes that are directly recognised by CD8+ T cells from allogeneic B10.BR (H-2k) or Balb/c (H-2d) mice. Here, we aimed to characterise the TCR repertoire responding to these epitopes during skin graft rejection or tolerance induction.
Methods: B10.BR mice underwent secondary skin grafts from transgenic donors mismatched for H-2Kb, with or without prior treatment with a liver-specific AAV vector encoding Kb heavy chain. AAV-Kb treatment induced indefinite graft acceptance.  Activated CD8+ T cells were isolated from the liver at intervals from d3-d100 post-inoculation, while infiltrating lymphocytes were isolated from rejecting or accepted skin grafts on d14. Single cells were captured and TCR/transcriptome/surface marker libraries prepared. Specificity was assessed using a panel of 12 Kb-peptide dextramers.
Results: Within the liver CD8+ T cell compartment, cycling cells and cells in the deletional tolerance clusters became progressively less frequent with increasing time post AAV-Kb inoculation, while the proportion of tissue resident cells expressing an “exhausted” gene signature increased. By d14, tissue-resident cells outnumber other populations. Previously-identified public clonotypes were present early during tolerance induction, but declined as a proportion of liver alloreactive T cells as the duration of tolerance increased, suggesting that these clones may be preferentially deleted. CD4+ Tregs were over-represented in accepted skin grafts while the infiltrate of rejecting grafts featured cytotoxic effectors (NK and CD8+ T cells).

Rejecting skin infiltrate is characterized by presence of tetramer-bright (high avidity) cells, and by a high proportion of clonotypes recognizing H-2Kb in complex with RTYTYEKL, VGPRYTNL or INFDFPKL. In accepted skin grafts, cells specific for these epitopes are supplanted by cells recognizing Kb-VAFDFTKV, Kb-VIVRFLTV or Kb-SSYTFPKM. To determine the overlap between the skin and liver repertoires, we calculated the Morisita-Horn index for all samples. Minimal overlap was noted between the T cell infiltrate in accepted skin and any other repertoires. Abundant public clonotypes were absent from accepted skin, even when present among the liver T cell population from the same individuals, suggesting that sequestration of alloreactive T cells with a tissue-resident phenotype within the liver may contribute to acceptance of remote tissues.
Conclusions: Multiple mechanisms contribute to the induction of tolerance to allogeneic skin grafts following expression of donor MHC I in recipient hepatocytes. In addition to deletion and exhaustion, prevention of recirculation by liver-resident T cells may impede skin graft rejection. CD4+ regulatory T cells may also contribute to mediating skin graft acceptance.