Philadelphia chromosome (Ph)-like B cell acute lymphoblastic leukemia (B-ALL) is a high-risk leukemia with a gene expression profile similar to BCR-ABL1+ B-ALL. Approximately 50% of all Ph-like B-ALL is characterized by genetic alterations leading to overexpression of CRLF2 (CRLF2 B-ALL). CRLF2 B-ALL occurs 5 times more often in Hispanic and Native American children than others and is prevalent in adolescents and young adults. The poor outcomes associated with CRLF2 B-ALL represent a major clinical challenge and an important component of pediatric cancer health disparities. Biologically, CRLF2 acts as a receptor component for the cytokine, TSLP, which induces JAK2-STAT5 and PI3/AKT/mTOR pathway activation downstream of binding to CRLF2. Activating JAK mutations are associated with CRLF2 B-ALL, but overall data indicate that JAK mutations are present in 50% or less of CRLF2 B-ALL. Our data show that normal primary human bone marrow (BM) stromal cells express TSLP, suggesting that TSLP-induced CRLF2 signals could play a role in the initiation, maintenance and progression of CRLF2 B-ALL, particularly in cases without JAK mutations. Consistent with this, TSLP has been reported to increase in vitro production of human fetal B cell precursors. However studies of TSLP in B lymphopoiesis have been conducted almost exclusively in mice which show low homology (~40%) to human TSLP and CRLF2. Further, using phospho flow cytometry we show that mouse TSLP is unable to induce increases in pSTAT5, pAKT and pS6 observed in CRLF2 B-ALL cells stimulated with human TSLP, confirming the species specificity of mouse TSLP. These findings underscore the importance and challenge of developing in vivo systems that can model human TSLP-CRLF2 interactions for evaluating therapies and studying leukemogenesis of CRRLF2 B-ALL. To address this challenge we engineered patient-derived xenograft (PDX) mice to produce human TSLP (hTSLP) by transplanting them with stromal cells transduced to express hTSLP (+T mice). Control (-T) mice were produced by transplanting with stroma transduced with a control vector. Supernatant from engineered +T stroma, but not -T stroma, induced JAK/STAT5 and PI3K/AKT/mTOR pathway activation in CRLF2 B-ALL cells. ELISA assays showed normal serum levels of hTSLP (12-32 pg/ml) in +T mice, while hTSLP was undetectable in -T mice. Since TSLP has been shown to increase in vitro production of human B cell precursors, we evaluated the in vivo functionality of our model by comparing the production of normal B cell precursors in the BM of +T and -T PDX mice generated with human umbilical cord blood CD34+ cells. Data from 3 different cord blood donors showed that production of B cell precursors is 3-5 fold increased in +T as compared to -T mice. TSLP-induced increases were specific to B lineage cells, initiated in the earliest CD19+ B cell precursors, and maintained through later stages of B cell development. Next we evaluate the in vivo functionality of our model using primary leukemia cells. +T and -T PDX mice were produced using primary CRLF2 B-ALL cells. BM was harvested and whole genome microarray was performed on isolated CRLF2 B-ALL cells. Evaluation of microarray data by Gene Set Enrichment Analysis (GSEA) and Ingenuity Pathway Analysis showed that genes downstream of mTOR pathway activation were upregulated in +T as compared to -T PDX mice, confirming hTSLP activity in the +T PDX mice. Next we tested whether +T PDX mice provide an in vivo model of B-ALL that more closely mirrors patients than -T PDX mice. +T and -T PDX mice were generated from primary high risk B-ALL. RNAseq gene expression profiles from primary patient B-ALL cells were compared to those of the same patient sample expanded in +T and -T PDX mice. The gene expression pattern in +T mice was significantly closer to the primary patient sample than those from -T mice. The +T and -T PDX mice described here provide a novel preclinical model for studying the role of TSLP in the initiation, progression and maintenance of CRLF2 B-ALL and for evaluating drug efficacy in an in vivo model that more closely mirrors the in vivo environment present in patients.

Disclosures

No relevant conflicts of interest to declare.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution