This study provides the first evidence that autoreactive CD8 T-cells cloned from a T1D individual can participate in islet cell destruction in-vivo using a novel humanized mouse model. IGRP265–273-specific CD8 T-cells producing IFNγ and expressing CD107a upon antigen-specific activation are capable of lysing murine HLA-A2 expressing islets in-vitro. In-vivo, these T-cells lysed target cells pulsed with islet peptide. Following intra-pancreatic injection into HLA-A2 transgenic NOD-scid IL2rγnull HHD mice, IGRP-specific CTLs localized to the islets and caused beta-cell destruction. However, no changes in blood glucose-levels were observed. No islet cell destruction was detected when mice were injected with equal numbers of control HLA-A2-restricted cloned CD8 T-cells.
The assumption that IGRP-specific CD8 T-cells possess diabetogenic potential in-vivo was based on the observation that these cloned T-cells were highly cytotoxic for HLA-A2 expressing islets in-vitro. However, the extent of lysis observed was low in comparison to lysis of peptide-pulsed EBV-LCL. This difference might be explained by lower expression of HLA-A2 molecule (HHD) on murine islets compared with HLA-A2 levels on human cells, which consequently results in decreased presentation of the IGRP epitope. In addition, the IGRP-specific T-cells possess low-avidity T-cell receptors as revealed by low staining with specific tetramer and CD8 dependency of tetramer staining. This may be inherent to autoreactive T-cells, since these should have been eliminated during thymic education due to high-avidity interactions with their cognate self-antigen. Thus, besides weak recognition of the epitope by the low-avidity TCR of the IGRP-specific T-cells, these observations are also consistent with low amount of IGRP epitope presentation on the surface of the islets.
Using an adapted in-vivo
cytotoxicity assay, we obtained evidence that IGRP T-cells are capable of lysing targets in-vivo
. Mobilization of effector CD8 T-cells to infected tissue requires CD4 T-help 
, however, CD4 T-cells only provide this help when specifically activated by their cognate antigen at this site 
. Given that human IGRP-specific CD4 T-cells were not available, we opted to inject the IGRP-specific CD8 T-cells intra-pancreatically and to provide the optimal cytokine milieu for survival of the transferred T-cells by systemic injection of HLA-A2+
PBMC from non-diabetic donors prior,for generation of the necessary cytokines (i.e.
IL2, IL7, IL15) needed for T-cell homeostasis.
Immunohistochemical analysis of the pancreata four weeks after injection of the human CD8 T-cells revealed clear differences in localization between the IGRP-specific T-cells and control T-cells. IGRP-specific CTLs infiltrated into the islets, whereas control T-cells were retained in the peri-vascular areas. Our observation confirms previous findings that islet antigen expression is a key factor in governing the ability of the autoantigen-specific T-cells to accumulate in the islets 
. Interestingly, both CD4 and CD8 T-cells migrating into pancreatic tissue expressed the chemokine receptor CXCR3, while distressed beta-cells produce its ligand CXCL-10 (IP-10) 
. This feature may present a master switch for migration of islet autoreactive T-cells to the beta-cell environment, as suggested by preclinical studies 
. Some human leukocytes were noted in the exocrine pancreatic tissue in both mice injected with IGRP-specific T-cells and control T-cells, which may be partly due to their intrapancreatic injection causing tissue damage and chemokine production leading to inflammatory infiltrates. The CD45 staining will target the injected T-clones as well as co-injected PBMC. Yet, since the same PBMC donor was used for all three T cells clones, the insulitis lesions could be attributed to the particular humans T-cell clones injected, rather than the PBMC that were not as feeders of the CD8 T cells.
Injection of the IGRP-specific T-cells resulted in a destructive inflammation, but not when a high number of cells from a control HLA-A2-restricted CD8 T-cell clone were injected. Since no changes in blood glucose-levels of the recipient mice were observed upon injection of the IGRP-specific T-cells in spite of clear islet cell destruction, we speculate that adoptive transfer of islet-specific CD4 T-cells may be required for propagation and sustained CD8 T-cell expansion and cytotoxic activity 
. A recent study employing the same HLA-A2 NOD-scidIL2rγnull
mouse model showed that high frequencies of CD4 T-cells were present when mice were engrafted with PBMC from a T1D patient 
. This was paralleled by the presence of diabetogenic epitope-specific CD8 T-cells. However, in this study no islet cell destruction was demonstrated, which may be due to lower frequency of islet-specific T-cells in the pancreas compared to our study. Yet, the fact that IGRP265–273
-reactive CD8 T-cells were present among the islet-specific CD8 T-cells, as well as the recent finding that specific targeting of IGRP-reactive CD8 T-cells effectively inhibited diabetes development in NOD.β2mnull
.HHD mice 
corroborate our current findings that this epitope is pathogenic in human disease.
T-cell autoreactivity may require more than 4 weeks for complete beta cell destruction, and require CD4 T-cells acting in concert. Indeed, the limited survival of CD8 T-cell clones in mice in the absence of CD4 T-cells underscores this notion. Mouse CD4 T-cells specific for a low-affinity insulin-derived peptide required 20 weeks before hyperglycemia developed in 50% of recipient mice 
. Despite the great value of immunocompetent NOD mice, there are discrepancies between the etiopathogenesis of T1D in NOD mice versus humans. These differences include the pattern of insulitis. The intra-islet infiltration pattern of human islet-specific T-cells and islet destruction observed in our study closely resembles the pathogenesis found in human inflamed pancreas 
rather than inflammatory lesion typically observed in NOD mice 
. The relevance of islet autoreactive T-cells circulating in the blood in the pathogenesis of insulitis in humans remains unresolved. Nonetheless, our current study on human peripheral blood CD8 T-cell clones, as well as our earlier studies on homing of human anti-islet CD4 T-cells clones to pancreatic tissue, imply that circulating human T-cells can contribute to insulitis. Recent studies on islet autoreactivity in an explanted human pancreas graft indicated that circulating islet-antigen autoreactive CD8 T-cells in peripheral blood were also present in pancreatic inflammatory lesions, albeit at a lower frequency. Moreover, additional islet autoreactivities that were below detection levels in the circulation could be detected in the inflamed pancreas 
In conclusion, our data provide first evidence that human autoreactive CD8 T-cells have the potential to target HLA-A2-matched beta-cells in-vivo, demonstrating the translational value of humanized mice to study mechanisms of disease and interventional strategies.