Upregulation of β-selection checkpoint factors in thymic transplantation model of T-ALL
Murine T-ALL was generated by transplanting single thymic lobes derived from newborn wild type (immunocompetent) mice under the kidney capsule of adult immunodeficient Rag2−/−IL2Rγc−/− hosts21. Mice developed T-ALL with high penetrance and a median time to disease of approximately 25 weeks, consistent with published results20,21. Leukemic disease in this model broadly recapitulated key clinical features of human T-ALL, including splenomegaly, infiltration of blast cells into bone marrow, and a high frequency of gain-of-function Notch1 mutations20,21.
We analyzed transcriptomes of these tumors by comparing T-ALL tumor-infiltrated spleens (n = 7) and early stage local pre-T-ALL tumors isolated before the emergence of peripheral leukemic disease (n = 14) to the spleens of non-tumor bearing mice (n = 5) and to CD4 and CD8 T-cells isolated from the spleens of non-tumor bearing mice (n = 4). 7,485 genes that were differentially expressed between T-ALL samples and control spleens at an FDR < 0.001, p < 0.01 and an absolute fold change difference of 1.5 (Supplemental Table 1).
Consistent with the developmental origins of T-ALL, we observed strong differential expression of multiple markers of thymocyte development, including Dntt, Rag1, Rag2 and Thy1 in T-ALL cases compared to control splenic tissue (Fig. 1A) or purified peripheral CD4+ and CD8+ T cells (Supplemental Fig. 1A). Consistent with the molecular biology of T-ALL, multiple Notch target genes, including Heyl, Dtx1 and Myc were upregulated (Fig. 1A). Strikingly, we also observed strong upregulation of genes associated with the β-selection checkpoint, notably Ptcra (524-fold), Notch3 (58-fold) and Notch1 (7.5-fold) (Fig. 1A-D, Supplemental Fig. 1). Along with Ptcra, other components of the pre-TCR were also higly-expressed in T-ALL, including Cd3 complex molecules, Lck, Lat, and Zap70. Consistent with potential driver function, expression of Ptcra was elevated in disseminated T-ALL vs pre-invasive thymic graft lesions, whereas expression of thymic lineage markers remained constant (Fig. 1E). Cell surface expression of Ptcra and other β-selection factors in T-ALL blasts was confirmed by flow cytometry (Fig. 1F-H).
Pre-tcr Signaling Is Required For Efficient Leukemogenesis And Sustained Proliferation Of T-all Cells
To interrogate the contribution of pre-TCR signaling to leukemogenesis in this model, we generated Ptcra KO mice and used thymic lobes from these animals in transplantation studies. Consistent with previous reports22,23, neonatal Ptcra KO mice developed thymic hypoplasia (Supplementary Fig. 2A). Immunophenotyping revealed a thymocyte compartment largely devoid of CD4+CD8+ DP cells, with most thymocytes arrested in the DN3 stage (Supplementary Fig. 2B, 1C). Consistent with this stage of developmental arrest, RT-PCR analysis revealed strong expression of canonical DN3 genes in Ptcra KO thymi relative to wild type controls, including Notch3, Hes1, Dtx1, Hes5 and Notch1. Similarly, we verified low expression of genes associated with post-β selection populations, including Cd4 and Cd8 (Supplementary Fig. 2D). This Ptcra KO thymic phenotype was partially reversed by adulthood (Supplementary Fig. 3A, 3B).
To account for the reduced thymic cellularity in Ptcra KO neonates, multiple lobes were used in each transplant vs. one thymic lobe from wild type mice. Ptcra KO thymi engrafted efficiently in Rag2−/−IL2Rγ−/− hosts, and T cell output was similar to that from wild type thymus when multiple (4–8) lobes were transplanted, as assessed by longitudinal monitoring of peripheral T cell levels (Supplementary Fig. 4). Compared to wild type controls, however, Ptcra-deficient thymi exhibited markedly reduced capacity to induce leukemogenesis, even when multiple Ptcra KO thymi were transplanted (Fig. 2A). Overall disease penetrance was significantly reduced, with median time to leukemic disease not reached within 52-weeks (vs. 25 weeks for wild type controls, HR = 0.249, p < 0.001). These differences in T-ALL development were not driven by differences in cellularity between wild type and Ptcra KO donor thymi, since increasing the number of Ptcra KO thymic lobes used per transplant did not change the efficiency of leukemogenesis (Fig. 2A).
These thymic transplant experiments indicate that pre-TCR signaling is required for efficient leukemogenesis in this model. To assess whether pre-TCR signaling more broadly participates in sustaining leukemic cell proliferation in T-ALL, we used CRISPR/Cas9 to knock out PTCRA in human T-ALL cell lines. Knock out of PTCRA in SupT1 and HPB-ALL cells, which have robust PTCRA expression (Suppl. Figure 5), resulted in significant reduction of cell proliferation (Fig. 2B, C). When subcutaneously implanted in the flank of NSG mice, wild type SupT1 cells developed tumors, whereas SupT1PTCRA − KO cells did not (Fig. 2D). The observation that PTCRA is required for proliferation in human T-ALL cell lines both in vitro and in vivo is consistent with a sustained dependence on pre-TCR signaling in these leukemic cells. Thymocyte pre-TCR signaling is mediated by SRC-family protein tyrosine kinases, with a pivotal role for lymphocyte-specific protein tyrosine kinase, LCK24. Consistent with the notion that signaling, per se, through the pre-TCR is required for proliferation in PTCRA-dependent T-ALL cell lines, treating these cell lines with the SRC-family kinase inhibitor, PP1, induced a dose-dependent anti-proliferative response (Fig. 2E). Furthermore, CRISPR/Cas9 KO of LCK markedly reduced SupT1 cell proliferation (Fig. 2F).
Collectively, these results from a thymic transplantation-based mouse model of T-ALL and human T-ALL cell lines reveal a critical role for pre-TCR signaling in driving and sustaining leukemogenesis of T-ALL.
Ptcra Is Frequently Upregulated In Human T-all
Examination of patient sample datasets indicate that PTCRA is highly and selectively expressed in T-ALL vs. other hematologic malignancies (Fig. 3A). Approximately 77% of T-AL samples express PTCRA at a threshold greater than 4 FPKM. Expression of PTCRA is higher in NOTCH1-mutated T-ALL, consistent with PTCRA being a target gene for NOTCH125. PTCRA was found to be expressed broadly across various molecular T-ALL subtypes, except the ETP-ALL subtype (Fig. 3C). To evaluate cell surface expression of PTCRA in human samples, we first confirmed the capacity of the commercially available antibody clone 2F5, raised against murine Ptcra, to detect human PTCRA in engineered and endogenously expressing cell lines (Supplementary Fig. 6). We then used the 2F5 reagent to detect cell surface expression of PTCRA in a series of prospectively acquired T cell acute leukemia patient samples. PTCRA cell surface expression was readily detected cell in T-ALL samples, but not in B-ALL or AML samples. (Fig. 3D). Further, consistent the transient role of pre-TCR in T-cell development, mature T-cells from normal donors (n = 4) did not express PTCRA (Fig. 3E). Cumulatively, the selective expression of PTCRA on leukemic cells but not normal T-cells coupled with the functional requirement of PTCRA to sustain leukemogenesis highlight PTCRA as an attractive target for the treatment of T-ALL.
Targeting PTCRA with cytotoxic antibody-drug conjugates promotes specific killing of T-ALL cells in vitro and in vivo
While several mechanisms for targeting cell surface antigens have demonstrated compelling clinical activity in recent years, including antibody-drug conjugates (ADCs), T-cell engaging bispecific antibodies and CAR-T cell therapy, we had observed that pre-TCR is rapidly and constitutively internalized in SupT1 T-ALL cells (Fig. 4A). Further, treating SupT1 cells with a with the translation inhibitor, cycloheximide, revealed that PTCRA is rapidly degraded (Fig. 4B). These receptor properties led us to test an ADC-based modality for targeting pre-TCR. As such, we conjugated the 2F5 antibody to a potent cell permeable (PTCRA-MAYT-P) or non-cell permeable (PTCRA-MAYT-NP) maytansinoid microtubule inhibitor via a cleavable linker at a drug-antibody ratio (DAR) of ~ 3.526. Both these PTCR-ADCs, but not control-ADCs, promoted dose-dependent killing of SupT1 cells with an IC50 in the low nanomolar range (Fig. 4C). PTCRA-ADC selectively induced killing of human SupT1 T-ALL cells but did not impact viability of B-ALL (NALM6) and AML (K562) cell lines or normal, peripheral T cells, consistent with the selective expression pattern of PTCRA (Fig. 4D).
The cytotoxic activity of these PTCRA-ADCs in vitro prompted their evaluation in vivo; we prioritized the more potent PCTRA-MAYT-C for testing. SupT1 cells were implanted subcutaneously into the flanks of NSG mice, and when tumors became palpable mice were treated with 3 mg/kg of PTCRA-MAYT-P or Control-MAYT-P on days 15, 18 and 22. Treatment with PTCRA-MAYT-P, but not the control ADC, strongly suppressed SupT1 tumor growth (Fig. 4E). In a more stringent model, C57Bl/6 mice were i.v. injected with 100,000 Ptcra + murine T-ALL cells to induce disseminated disease. Mice were randomized according to tumor burden on day 5 post-implantation and treated with PTCRA-MAYT-P or Control-MAYT-P on days 5, 8 and 12. Tumor burden was assessed longitudinally throughout the study by quantifying the number of blast cells in peripheral blood and by quantifying splenic mass at the end of study. T-ALL-bearing mice treated with PTCRA-ADC had significantly reduced tumor burden in both spleen and peripheral blood, relative to control-ADC controls (Fig. 4F and 4G). PTCRA-ADC treatment was well tolerated, with no signs of distress or differences in body weight evident in the treatment group relative to control mice. Importantly, no evidence of PTCRA-ADC directed cytotoxicity towards non-malignant T-cells was observed (Fig. 4H). Collectively, these results indicate that targeting pre-TCR in T-ALL with a cytotoxic ADC represents a promising therapeutic approach to specifically eradicate tumor cells while sparing normal T cells.