GVHD Is Sustained By T Cell Maintenance within Target Tissues: Insights from Clonal Tracking and Parabiosis

Abstract Graft-versus-host disease (GVHD) is a major cause of morbidity and mortality in allogeneic hematopoietic stem cell transplantation (alloSCT). In GVHD, donor T cells recognize recipient tissues as non-self and mount a broad attack that results in multi-organ damage. While there has been some success in diminishing the incidence of GVHD, less progress has been made in treating established or steroid-refractory disease without severe global immunosuppression. This is in part due to a lack of understanding of the underlining mechanisms that sustain GVHD despite chronic T cell antigen exposure. To address this, we developed a mouse GVHD model that allows us to track the progeny of single alloreactive T cell clones. We used this model to test hypotheses on GVHD maintenance: 1) GVHD is driven by the continuous output and trafficking of alloreactive T cells from secondary lymphoid tissues (SLT) into GVHD target organs; and 2) once tissues are seeded with alloreactive T cells from SLT, GVHD is maintained locally within affected tissues. We reasoned that if GVHD is maintained by the continuous SLT output of T cells, the progeny of single alloreactive clones in a target tissue would come into equilibrium with those in SLT and other target tissues. Alternatively, if there is a degree of local tissue GVHD maintenance, our model predicts that clonal progeny should be unequally distributed and not in equilibrium with SLT. We first tested these possibilities using a GVHD model wherein BALB/c RAG2-/- TCR transgenic (Tg) CD4 T cells (TS1) that recognize the S1 peptide derived from influenza hemagglutinin (HA) induce GVHD in BALB/c RAG2-/- mice that ubiquitously express HA (HA104 mice). We generated TS1 TCR Tg mice on 9 congenic backgrounds based on the expression of CD45.1/2, Thy1.1/2 and GFP. We transferred 500 naïve TS1 cells of 1 clonotype (to induce GVHD) along with single naïve TS1 cells from the remaining 8 clonotypes and BALB/c RAG2-/- bone marrow (BM) into lethally irradiated HA104 mice. We recovered a total of 432 single-cell derived TS1 clones (72% of input clones) from tissues of 79 mice, analyzed 7-35 days after transfer. We enumerated the TS1 clonal composition of each tissue (expressed as the % of all TS1 of that tissue) and found disparate clonal distribution across tissues within individual mice. For example, in a representative mouse analyzed at day 33 (Figure 1), the fractions of a single TS1 clone were relatively high in the colon (1.4%) and small intestine intraepithelial lymphocyte (IEL) (4.6%) when compared to the spleen (0.07%), BM (0.02%) and liver (0.03%). These data support that TS1 clones are not equally distributed among tissues and are not in equilibrium with SLT, suggesting that GVHD is at least in part maintained locally. We also analyzed TS1 clonal frequency distribution in a second model. BALB/c RAG2-/- BM and polyclonal T cells were transplanted into F1 (BALB/c HA104xB10.D2) recipients along with 8 single distinct TS1 cells. In this system, GVHD is induced by polyclonal BALB/c cells and TS1 cells are trackers of reactivity to HA. Preliminary experiments also indicate unequal distributions of TS1 clones across tissues in individual mice. In a second approach to test whether GVHD is maintained locally, irradiated HA104 mice were reconstituted with either 500 Thy1.1 or 500 Thy1.2 TS1 cells and CD45.1 or CD45.2 BALB/c RAG2-/- BM. One partner also received congenic TS1 single cells. We performed parabiosis of mice from one group to the other 21-28 days later. We analyzed 4 pairs 4 weeks post-joining, looking first at the blood to establish a baseline for TS1 crossover from the parabiotic partner. In blood, 18.9% ±1.9 of all TS1 were derived from the partner. Importantly, relative to equilibration in blood, there were far fewer partner-derived TS1 cells in all other tissues (Figure 2). Only a few single cell-derived TS1 clones were detected in the partner mouse at very low frequencies, even when they were dominant in tissues of the corresponding partner. Together, these data indicate that once GVHD is established, local maintenance dominates over new TS1 entry. Consistent with this, TS1 cells incorporate BrdU in vivo even at late time points. We are combining proliferation and clonality data at multiple timepoints to develop a mathematical model of GVHD establishment and maintenance. We are also extending our observations in a polyclonal GVHD model wherein the progeny from single alloreactive CD8 cells can be enumerated. Disclosures Shlomchik: NapaJen: Consultancy.