A Novel Pre-Clinical Xenograft Model to Study TSLP-CRLF2 Interactions in B-ALL

Abstract 1498B-cell precursor acute lymphoblastic leukemia arising from overexpression of CRLF2 (CRLF2 B-ALL) is high-risk with poor prognosis. CRLF2 B-ALL occurs 5 times more frequently among children of Hispanic/Latino ethnicity than others and thus represents one of the most significant biological components of childhood cancer health disparity identified to date. CRLF2, together with the IL-7Rα, forms a receptor complex that is activated by the cytokine, TSLP. The JAK-STAT5 pathway is phosphorylated downstream of this receptor complex activation. Activating JAK mutations are found in some CRLF2 B-ALL and their presence has led to speculation that TSLP stimulation is not a factor in CRLF2 B-ALL. In preliminary studies to address this question we evaluated the effect of TSLP on CRLF2 B-ALL cell lines that have JAK defects and which have been reported to exhibit constitutive JAK-STAT5 activation. Our data show that TSLP increases STAT5 phosphorylation in these cell lines and also in primary B-ALL cells that overexpress CRLF2. Our next step was to evaluate the role of TSLP-CRLF2 interaction in vivo in the human-mouse xenograft model. However, mouse TSLP is different from most other cytokines produced in mice in that it is species-specific and does not activate the human TSLP receptor complex that contains CRLF2. Thus, traditional xenograft models do not provide the TSLP-CRLF2 interaction that our data implicate as a potential contributor to pathogenesis in CRLF2 B-ALL. To overcome this obstacle we engineered immune deficient NOD/SCID IL-2Rγ null (NSG) mice to express human TSLP (hTSLP+ mice), as well as control mice that lack the TSLP cytokine (hTSLP– mice). ELISA assays show plasma hTSLP levels in the hTSLP+ mice that approximate the normal range in human plasma. We used this hTSLP+/– xenograft model system to study the in vivo effects of TSLP on CRLF2 B-ALL cells harboring a JAK defect (MUTZ5 cell line) and on primary CRLF2 B-ALL cells from a Hispanic patient. Mice were euthanized at 5 weeks and BM disease was evaluated. In recipients of MUTZ5 B-ALL cells the percentage of viable leukemia cells in hTSLP– mice was half that observed in hTSLP+ mice. Similarly, in recipients of primary CRLF2 B-ALL cells, the percentage of viable leukemia cells was reduced in hTSLP– mice as compared to hTSLP+ mice. In addition, the viable primary B-ALL cells present in the BM of hTSLP+ mice showed higher expression levels of the TSLPR components (CRLF2 and IL-7Rα) than those in the hTSLP– mice. These data provide evidence that the TSLP produced in this model is active and that it impacts primary CRLF2 B-ALL cells. The hTSLP+/– xenograft model described here provides the first data on the in vivo role of TSLP-CRLF2 interactions in CRLF2 B-ALL. This preliminary data suggests that TSLP provides a signal that promotes in vivo survival of CRLF2 B-ALL cells and that it could play a role in selection of CRLF2-HI clones during in vivo leukemogenesis. This pre-clinical model will allow us to evaluate TSLP-CRLF2 interactions as a target for therapy and to perform translational studies that identify molecular mechanisms and additional targets downstream of TSLP-induced signaling in CRLF2 B-ALL. This model system will be be particularly important for assessing and identifying therapies, including drug and cellular delivery systems, to effectively target CRLF2 B-ALL and to reduce cancer health disparities in Hispanic childhood B-ALL.This work is supported by a National Institutes of Health R21CA162259, a Loma Linda University Grant to Promote Collaborative and Translational Research and a St. Baldrick's Foundation Research Award (to K.J.P.).