Quantitative Assessment of the Impact of Ex Vivo Manipulation on the In Vivo Functionality of Human T Cells in RAG2^{−/− γ}c^−/− mice.

T cells retrovirally modified to express therapeutic genes encoding cytokines, exogenous TCRs or suicide molecules represent a novel class of immune therapeutics of great potency. However, recent clinical trials using retrovirally-modified T cells have indicated that T cells exhibit a diminished reactivity upon ex vivo manipulation. In addition, virus-specific memory T cells seem to be lost during gene transfer. In a BNML rat model we have shown that the culture procedure is one of the critical parameters. To preserve T cell reactivity, reliable models are required which permit readout of human T cell activity. We recently developed a huPBMC-RAG2^−/−γc^−/− mouse model for xenogeneic graft-versus-host disease (xGVHD), in which iv injection of 15 x 10⁶ human T cells into RAG2^−/−γc^−/− mice consistently leads to high level engraftment and lethal xGVHD within 3 weeks in 80% of mice (van Rijn et al, Blood 2003). We have now used this model to analyze in vivo functionality of human T cells following different ex vivo culture procedures. For this, we cultured human T cells for 7 days with either of the two currently available clinically applicable stimulation conditions: 1) via CD3 and 2) via CD3/CD28. In addition, we included CD3/CD28/4-1BB stimulation to explore the effect of extensive costimulation. Mice were injected with escalating doses T cells. HuCD45⁺ cells in peripheral blood were measured by FACS. Lethal xGVHD occurred at only 6 times (90.10⁶) the dose of fresh cells for CD3-stimulated T cells and 3 times for CD3/28- or CD3/28/4-1BB-stimulated cells. About 20% of surviving mice developed chronic xGVHD, independent of culture method. While lethal xGVHD was always associated with very high levels of engraftment (up to 95%) engraftment levels in chronic mice ranged from 1–75%. To compare the impact of the different culture conditions on in vivo T cell function, we analyzed engraftment potential. The fraction of huCD45⁺ cells was plotted against the time and the areas under the curves were compared. Based on a total of 68 mice, statistical analysis showed a 2-fold improvement of engraftment potential for C28-costimulated human T cells compared to CD3-stimulated cells (P<0.0001). Additional ligation of 4-1BB did not increase engraftment potential. In addition, different T cell subsets (naïve, memory, effector) were monitored based on the combined expression of CD45RA, CD27 and CCR7. For all primary T cells and variably cultured T cells, a strikingly similar pattern was observed in vivo. After 3 weeks mainly effector and memory effector T cells (both CD4⁺ and CD8⁺) could be detected, suggesting a (xeno-)antigen-driven survival and expansion. This was a very consistent observation independent of donor, culture condition, engraftment level or severity of disease. In conclusion, in vitro costimulation preserves in vivo functionality of human T cells and should therefore be included in future clinical protocols for ex vivo manipulation of T cells. These data show the feasibility to use the huPBMC-RAG2^−/−γc^−/− model for in vivo evaluation of in vitro effects on human T cells. This model is the most sensitive to date for in vivo evaluation of human T cells and will be a promising new tool for the study of human T cells in, for instance, autoimmune disease, cancer and infectious diseases like AIDS.