CLL Cells Can Diversify, Switch, and Differentiate in Response to Autologous T Cell Stimuli Present in a Murine Adoptive Transfer Model

Abstract 315

Adoptive transfer of primary patient CLL cells into NOD/SCID/γc^null(NSG) mice results in engraftment and proliferation of CLL cells if autologous T cells are present. Formation of splenic follicles consisting of B cells interspersed and surrounded by T cells indicates engraftment. However, ultimately these CD20+ cells are lost several weeks later. We describe one of the mechanisms for this apparent loss: differentiation to plasma cells.

Peripheral blood cells from 9 IgM+ CLL patients (6 U-CLL and 3 M-CLL) were adoptively transferred into NSG mice with enriched autologous CD3+ cells pre-activated with anti-CD3/28 beads. B and T cell engraftment and subset distributions were analyzed for 47 mice by immunohistochemistry (IHC) and flow cytometry (FC) at the time of sacrifice. The earliest and latest times of assessment were 12 and 124 days, respectively, after CLL cell injection. In some cases, CLL cells were labeled with CFSE to track cell division.

At sacrifice, 3 engraftment patterns were observed. Pattern 1 (observed up to day 56) showed small follicles of CD20+ cells with low-moderate numbers of surrounding T cells. Intensely positive CD38 cells were inconspicuous. FC showed CD19+CD5+ cells with no increase in CD38 and variable CFSE dilution indicating lower levels of proliferation. Pattern 2 (observed throughout the study period) showed much higher T and B cell numbers. CD20+ cells were interspersed with and surrounded by principally CD4+ cells which were activated and functional as indicated by expression of Ki-67, PD-1, CD57, and T cell derived cytokines IFNγ and IL5 in plasma. Follicles contained CD20 and cytoplasmic Ig+ (cIg+) cells that double stained for IRF-4 and Blimp-1, transcription factors required for B cell differentiation. While Bcl-6 staining in these cells was minimal or absent, follicles from all 9 patients contained activation-induced deaminase (AID)+ cells. Cells with dim IgM expression localized to follicles; however, cells with intense IgM, IgA, or IgG were present both within, surrounding, and outside follicles matched by similar CD38 staining. Smaller populations of CD138+ cells surrounded follicles and were interspersed throughout non-follicular splenic areas. FC showed a novel CD19+CD5-CFSE-CD38++ population containing a CD138+ subset. Pattern 3 (observed in a limited subset of cases not before day 63) had minimal CD20+ cells by IHC, but noticeable populations of cIg+CD38+ and CD138+ cells interspersed amongst plentiful T cells. Such cells corresponded with cells with plasma cell morphology.

Confirmation that differentiated cells were from the patient clone was achieved in 3 ways. First, in FACS sorted CD19+CD5+ and CD19+CD5-38++ cells from a subset of pattern 2 cases, RT-PCR revealed that all fractions contained both IGHC unswitched and switched clones identical to those found in the patients. Second, cases with pattern 3 engraftment generated CLL clonal switched and unswitched cDNA sequences. Finally, adoptive transfer of highly purified CD5+CD19+ patient cells generated IRF-4+Blimp-1+CD138+ cells. The generation of switched cells from all 9 patients indicated functional AID. In one U- CLL case, ultra-deep sequencing on pre-transfer and post-transfer human cells taken from mouse spleen revealed a significant number of new IGHVDJ mutations in spleen-derived cells. Such mutations targeted nucleotides typical for AID's action.

In conclusion, CLL cells can diversify, switch, and differentiate in NSG mice in response to autologous T cell signals. The extent of this maturation is a function of T cell numbers and activity and the duration of the experiment. Differentiation without significant Bcl-6 expression suggests that follicles in NSG mice are not recapitulating classic germinal center reactions, possibly giving clues to the origin of CLL. Several features of poor prognosis disease were demonstrated (e.g., increased CD38 and AID expression with the development of clonally related switched transcripts) that might mirror clinical disease features. AID expressed by CLL cells is fully functional as indicated by de novo somatic hypermutation and class switch recombination. Both U-CLL and M-CLL clones respond in a similar manner in this model, suggesting the importance of T– B cell interactions in all types of CLL. Finally, the demonstration that cells can differentiate when appropriately induced may lead to novel therapeutic options for CLL.