Aberrant Notch Signaling in the Bone Marrow Microenvironment of Acute Lymphoid Leukemia Suppresses Osteoblast-Mediated Support of Hematopoietic Niche Function

Acute lymphoblastic leukemia (ALL) is the most common malignancy affecting children and young adolescence, with up to 20% to 25% of patients experiencing relapse with bone marrow being the most common site of all relapses (1). Patients who experience bone marrow relapse often have worse outcomes (2, 3). Compared with B-cell acute lymphoblastic leukemia (B-ALL), T-ALL patients have a higher relative risk of death, often showing diffuse infiltration of the bone marrow by immature T-cell lymphoblasts prior to T-ALL propagation, and a strong propensity to infiltrate the central nervous system (4). A better understanding of interplay between ALL and the marrow microenvironment during disease progression would help identify new therapy targets and improve patient outcome.

Recent studies have shown progress in our understanding of leukemia microenvironment. These have revealed contributions of both osteoblasts in B-ALL xenografts and the perivascular niche to the progression of acute myelogenous leukemia development in NOD-SCID/IL2Rγ^null mice (5, 6). Osteoblasts lining the endosteum play a critical role in hematopoietic stem cell (HSC) homing and maintenance (7–9). A direct interaction between HSCs and the sinusoidal endothelial cells in the perivascular niche also supports HSC proliferation (10–13), while CXCL12-abundant reticular (CAR) cells promote HSC retention and proliferation (12). However, little is known about alterations and regulations of T-ALL microenvironment.

Approximately 50% to 70% of T-ALL harbor “gain of function” mutations in the intracellular domain of Notch1 (ICN1; ref. 14). Notch1 is a member of a family of heterodimeric receptors. Ligand binding initiates successive proteolytic cleavage of Notch, which culminates in the release of the ICN1 and the formation of a transcriptional activation complex leading to the activation of downstream targets (15, 16). Notch1 mutations in T-ALL drive leukemia transformation characterized by the expansion of an immature population of T progenitors. A mouse model of T-ALL is achieved in marrow progenitors via retroviral transduction and expression of ICN1 that drives thymus-independent induction of T-cell leukemia (17–19). Using this model of T-ALL, we report here that when ICN1-expressing marrow progenitors and wild-type (WT) marrow cells were cotransplanted into WT mice, WT marrow cells displayed a disease stage-dependent reduction of the hematopoietic stem and progenitor cells (HSPC), a prominent suppression of B lymphopoiesis, and thrombocytopenia. Using this mouse model of T-ALL and studies of human T-ALL xenografts, we show that leukemia-induced suppression of hematopoiesis involves reconstruction of the bone marrow microenvironment by hijacking the proliferative vascular space and by repressing the endosteal/osteoblastic niche.

The animal research described in this article was approved by the Institutional Animal Care and Use Committee. Animals used in this study are 8 to 16 weeks old [C57BL/6 (Ly5.2) and B6.SJL-Ptprc^a Pep3^b/BoyJ (Ly5.1)] and were maintained as described (20, 21). NSG mice of 8 to 12 weeks old were from The Jackson Laboratory.

Retroviral (pMIG-eGFP-ICN1 provided by Drs. John Lowe, Jon Aster, and Andrew Weng) transfection of progenitors-enriched bone marrow cells after 5-fluorouracil (5-FU) treatment was performed as described (21). Forty-eight hours after infection, 2 to 5 × 10⁵ cells along with 2 × 10⁵ WT cells (Ly5.1) were transferred into lethally irradiated (10.5 Gy) WT recipients (Ly5.1). Recipients were monitored for leukemia development weekly. Secondary leukemia was generated in nonirradiated mice by i.v. injecting primary leukemia splenocytes (2–5 × 10⁵).

Isolation of HSPCs, analysis of megakaryocyte/erythroid progenitors, and Ki67 staining were performed as described (22). Osteoblast sorting and cell surface CXCL12 staining were performed using a modified procedure (22, 23). Megakaryocyte progenitors and erythroid progenitor markers were analyzed by PE-anti-CD41, PE-anti-TER119, and APC-anti-CD71 of LSK cells.

Lineage-depleted progenitors or LSKs were cultured with OP9 for 8 or 4 days in the presence of stem cell factor (SCF; 25 ng/mL), IL7 (5 ng/mL), and Flt3 ligand (5 ng/mL), and immunophenotyped as described previously (24). Confluent MC3T3 (provided by Dr. Yibin Kang, Princeton University, Princeton, NJ; retested for osteoblast differentiation) cells were cocultured with 0.5 to 1 × 10⁶ total marrow cells from ICN1 mice or pMIG mice in the presence or absence of 10 μmol/L dibenzazepine (DBZ; Selleckchem) for 72 hours followed by qRT-PCR, or 48 hours followed by coculture with LSK cells (0.1 × 10⁶).

Seven days after leukemia cell injection, mice received two intratibial injections of vehicle in one leg, and DBZ in the other leg, at a dose of 10 μL of 2 mmol/L, 4 days apart. In other experiments, DBZ was applied i.p. daily thrice starting on day 4 after cell transfer. LSK homing was then performed on day 8. For treating human T-ALL engrafted NSG mice, DBZ was given i.p. starting day 12 after mice receiving 1.6 to 2.0 × 10⁶ leukemia cells. Treatment continues for 3 consecutive days out of a 7-day cycle, at a dose of 10 mg/kg body weight, and repeated up to 4 cycles.

Study of human marrow specimens was approved by the Institutional Research Board of the University Hospitals Case Medical Center. Included in the study are 10 control bone marrow with patient age range from 1 to 17 years (medium age = 9.5), 2 T-ALL (age 3 and 27), and 8 B-ALL specimens from patients with age range from 2 to 23 years (medium age = 4.5). See details in Supplementary Information.

The human T-cell leukemia cell lines DND-41 and KOPT-K1 were a gift of Dr. Warren Pear and were retested by sequencing and γ-secretase inhibitor (GSI) response in vitro. Cells were plated at 0.5 to 2 × 10⁶/mL and cultured in RPMI1640 containing 10% FCS at 37°C with 5% CO_2.

Intravital 2-photon imaging preparation, data acquisition, and analysis were performed as previously described (20). See details in Supplementary Information.

Data are presented as mean ±SD, unless otherwise stated. Statistical significance was assessed by the Student t test or ANOVA analysis. For survival curve, P values were generated using the Mantel–Cox test.

To study the biologic influence of ALL leukemia cells on host hematopoiesis, we coinjected retroviral transfected progenitor-enriched marrow cells (Ly5.2) expressing either pMIG-ICN1-eGFP or the control plasmid (pMIG-eGFP), with WT marrow cells (Ly5.1) into lethally irradiated WT hosts (Ly5.1; Fig. 1A). While mice receiving control plasmids (referred hereafter as pMIG mice) showed no sign of disease, mice receiving ICN1-eGFP developed T-ALL within 4 to 12 weeks (referred hereafter as ICN1 mice), characterized by the expansion of an immature CD4⁺/CD8⁺ T progenitors in the marrow (Fig. 1B), thymus, and spleen (not shown). Full-blown leukemia was accompanied with a progressive alteration of the hematopoiesis in the nontransformed GFP⁻Ly5.1⁺ compartment, characterized by a suppressed B220⁺ population and an elevation of Gr-1⁺ cells (Fig. 1B; percentage and absolute numbers in Fig. 1C and D, respectively). Hematocrit was not changed (Supplementary Fig. S1A), while platelet number was decreased in mice developing full-blown leukemia (Fig. 1E). The GFP⁻Ly5.1⁺ marrow also showed an initial expansion of Lin⁻Sca-1⁺c-kit⁺ (LSK) at the earlier stage of reconstitution in both ICN1 and pMIG mice (0.27% and 0.21%, respectively, compared with 0.15% of WT LSK). At the later stage of disease, however, LSK cells were only present at 0.06 ± 0.02% in ICN1 mice but maintained at 0.22 ± 0.04% in pMIG mice (Fig. 1F and G, GFP⁻ panel). The frequency and number of megakaryocyte progenitor (MP; c-Kit⁺CD41⁺) were reduced (Supplementary Fig. S1B; ref. 25), whereas the frequency and number of megakaryocyte/erythrocyte progenitor (MEP) and erythroid progenitor (EP) showed a trend of decrease in ICN1 mice (Supplementary Fig. S1C and S1D). We did not observe altered maturation (data not shown) or increased apoptosis of megakaryocytes isolated from ICN1 mice (Supplementary Fig. S1E). Consistent with reports by others, we also observed a suppression of LSKs in transformed hematopoietic compartment (Fig. 1G, GFP⁺ panel; ref. 26). These findings indicate that lymphoblastic leukemia cells induce a disease stage-dependent reduction of HSPC (LSK and MP) and thrombocytopenia, a mild anemia, a prominent inhibition of B lymphopoiesis but an expansion of granulopoiesis.

To differentiate if inhibition of HSPC and B lymphopoiesis is a result of direct contact-mediated suppression by leukemia cells, we performed in vitro OP9 coculture, in which either WT lineage-depleted bone marrow (Lin⁻) or LSKs were cocultured with leukemia or control cells on OP9 cells (Fig. 2A) in the presence of SCF, IL7, and Flt3L (24). In coculture with Lin⁻ cells, either pMIG mice marrow cells or WT thymocytes representing immature T cells were used as controls. We found that after 8 days of coculture, B-cell production from Lin⁻ progenitors cocultured with leukemia cells was significantly reduced compared with those from coculture with pMIG cells (98.7 ± 6.9 × 10³ vs. 157.2 ± 18.2 × 10³, P < 0.05; Fig. 2B), but not from coculture with WT thymocytes, suggesting that B-cell inhibition was caused by immature T cells rather than leukemia cells. We then examined whether coculture with leukemia cells could affect the expansion and differentiation of more primitive LSK cells. To avoid excessive LSK differentiation, LSK cells were cocultured on OP9 with either WT marrow cells (LSK/control), pMIG (LSK/pMIG), or ICN1 (LSK/ICN1) marrow cells for only 4 days (Fig. 2A). We found that total numbers of expanded cells (Fig. 2C) and expanded progenitors (Lin⁻Sca-1⁺) were similar among three culture conditions (Fig. 2D). Although there was a mild reduction of B220⁺ cells (31.5 ± 8.1% and 21 ± 2.8%, P = 0.05), and an increase of Gr-1⁺ cells generated from LSKs cocultured with ICN1 cells (25.8 ± 9.6% and 47.0 ± 6.4%, P < 0.05) when comparing LSK/control with LSK/ICN1 conditions, there was no significant difference in any cell types generated when comparing LSK/ICN1 with LSK/pMIG condition (Fig. 2D). In addition, there was no apparent apoptotic alteration in the emerging B220⁺ cells (data not shown). These results indicate that direct contact with leukemia cells mildly interferes with progenitor cell differentiation along the B and myeloid lineages, but has no obvious effect on the expansion of the progenitor cells.

A mild effect by direct leukemia contact suggests that other mechanism such as changes in the leukemic microenvironment may play a more significant role in hematopoietic suppression. To explore this possibility, we first examined the homing and niche localization of ICN1-transformed preleukemia cells using 2-photon imaging analysis. Twenty-four hours after transfer of ICN1- or pMIG-expressing cells into lethally irradiated WT mice, we were able to locate individual GFP⁺ cells in the calvarium and long bone (data not shown) marrow. Most pMIG-expressing cells were positioned >5 μm from vessel walls (Fig. 3A, top), whereas all of the ICN1-expressing GFP⁺ cells were found either directly attached to or within 5 μm of vessels (Fig. 3A, bottom; Fig. 3B). On day 8 after cell transfer, duplex and clusters of ICN1-expressing cells were found at perivascular regions (Fig. 3C, bottom), whereas pMIG-expressing cells were seen throughout the marrow without significant clustering (Fig. 3C, top). These ICN1-expressing cell clusters represent proliferating preleukemia cells after homing to the marrow as they were not seen on day 1 (Fig. 3D). These cells continued to accumulate at perivascular regions (Fig. 3E, bottom) and populated the entire marrow space in full-blown leukemia mice (Fig. 3F), whereas pMIG-expressing cells were seen mainly as isolated cells (Fig. 3E, top).

Next, to examine the effect of perivascular accumulation of preleukemia cells on normal HSPC homing and niche locations, we cotransferred WT LSK cells with ICN1-expressing or pMIG cells. When cotransferred with ICN1-expressing preleukemia cells, WT LSK cells were found located more distal from the perivascular regions (Fig. 3G; Fig. 3I, right) but closer to the endosteum surface than when cotransferred with pMIG cells (Fig. 3H; Fig. 3I, left). Numbers of WT LSK homing to the marrow were decreased by 79% when cotransferred with ICN1-expressing cells (n = 15) than with control cells (n = 72; Fig. 3G and H). Further, Ki67 index of LSKs was decreased in the presence of ICN cells (Fig. 3J). These findings indicate that leukemia cells outcompete normal HSPC in the perivascular distribution and suppress normal HSPC proliferation.

To examine cellular alterations mediated by leukemia cells, we quantified numbers of marrow stromal cells, including osteoblastic lineage cells (OB) and multipotent stroma cells (MSC; Fig. 4A) in age-matched pMIG mice (top) and ICN1 mice (bottom). Consistent with histologic examination showing decreases of OBs in ICN1 mice (Fig. 4B), we found a remarkable decrease of OBs in ICN1 mice (Fig. 4C), while MSC (Fig. 4D) numbers were variable. Isolated OBs from pMIG and ICN1 mice were then examined for their RNA expression of osteoblast transcription factors. Indeed, we found that RNA levels of Runx2, osterix, osteocalcin, ostegrin, and CXCL12 were all decreased in ICN1 mice OBs (Fig. 4E), while overall stroma cells of ICN1 mice (by selecting CD45⁻TER119⁻CD31⁻ population) showed increased expression of IL6, SCF, HIF1α, VEGFα, and Notch ligand Jagged1 (JAG1; Fig. 4F).

Because JAG1 RNA expression was upregulated in leukemia marrow, we examined whether Notch activation was indeed responsible for the observed phenotype in ICN1 mice. We found that Notch1 and Notch2 receptors as well as several Notch target expressions in ICN1 mice OBs and MSCs were all increased compared with those of pMIG mice (Fig. 5A). Possible contamination by leukemia cells was excluded (Supplementary Fig. S2). We then cocultured osteoblastic progenitors, MC3T3, with pMIG or ICN1 cells (Fig. 5B). We confirmed that JAG1 protein expression was elevated in leukemia stroma (not shown) as well as in MC3T3 cells exposed to ICN1 cells (Fig. 5C). Further, we observed an ICN1-induced upregulation of Notch1, Hes1, and Hey1 in MC3T3 cells (Fig. 5D), while coculture of MC3T3 with ICN cells in the presence of GSI, DBZ, completely reversed the induction of Notch1 and Hes1 expression (Fig. 5C and D). To investigate if exposure to leukemia cells interferes with osteoblast hematopoietic supportive function, we first pretreated MC3T3 cells with leukemia cells or controls for 2 days, and then cocultured MC3T3 with WT LSKs (Ly5.2) for 4 days after removal of leukemia or control cells (Fig. 5E). Expanded LSK cells (Fig. 5F) were then cotransplanted with WT marrow cells (Ly5.1) to lethally irradiated mice. Three months after transplantation, cells (Ly5.2) derived from ICN1-pretreatment culture displayed similar chimerism and proliferation as cells from control-treated culture (not shown), but showed a modestly skewed myeloid differentiation at the expense of B lymphocytes (Fig. 5G). In combination, these findings are reminiscent of de novo leukemia mice showing suppression of B lymphopoiesis and expansion of granulocytes. However, we suspect that a more prominent HSPC suppression requires either other stromal components, or more likely, a sustained in vivo leukemia milieu.

Because CXCL12-expressing stroma cells are essential components of the stem cell niche (27, 28), we then investigated whether Notch activation in osteoblasts inhibits CXCL12 expression (Fig. 4E). Notch regulates gene expression by forming transcriptional complex with the DNA binding protein RBPJ/CSL. We found that approximately 1.1 kb upstream of CXCL12 promoter contains the RBPJ/CSL sequence (TGGGAA; Supplementary Fig. S3A; ref. 29), and confirmed this by chromatin immunoprecipitation analysis (Supplementary Fig. S3B). In luciferase reporter assay, cotransfection of the CXCL12 promoter construct including the RBPJ/CSL binding sequence with RBPJ siRNA into MC3T3 resulted in approximately 2.5-fold increase of luciferase activity (Fig. 5H), whereas its cotransfection with either activated Notch1 (ICN1)- or Notch2 (ICN2)-expressing plasmid reduced activity by 60% and 53%, respectively (Fig. 5I and J). These findings suggest that Notch activation in osteoblasts negatively regulates CXCL12 expression. However, the mechanism whereby a transcriptional activator formed by Notch/RBPJ negatively regulates CXCL12 remains unclear and warrants further investigation.

We then tested whether blocking Notch activation could rescue the suppression of osteoblast and hematopoiesis. Nonirradiated WT recipients receiving leukemia cells from primary ICN1 mice readily develop secondary leukemia within 2 weeks (26). Mice then received intratibial injection of GSI (DBZ), or vehicle, ipsilaterally. Analysis of tibia morrow 1 week later revealed that, compared with vehicle-treated marrow, DBZ-treated marrow displayed a partial restoration of OB (Fig. 6A and B) and LSK cells (Fig. 6C and D), and modestly increased MP numbers (Supplementary Fig. S4A). DBZ treatment of pMIG mice (control) showed no significant effect. DBZ treatment also modestly corrected expanded Gr-1 and suppressed B220 cells (Fig. 6E) in leukemia marrow, but had no obvious effect in the control marrow. These effects mediated by DBZ are unlikely results of leukemia burden reduction, as cleaved ICN1 cells are not direct target of GSI; and indeed, we do not see a significant reduction of GFP⁺ cells in DBZ-treated tibia (35 ± 8.2%) compared with vehicle-treated tibia (39 ± 5.6%). We then treated secondary leukemia mice systemically with DBZ. Compared with control-treated mice that had 100% mortality by day 15 (day 11–14 peripheral GFP⁺ level: 37.8 ± 6.5%), DBZ treatment of ICN1 mice (day 11–14 peripheral GFP⁺ level: 34.2±8.1%) improved survival such that 40% of mice lived longer than 20 days (Fig. 6F; P < 0.05). Homing of WT LSK cells to DBZ-treated leukemia marrow also improved by 40% (Fig. 6G), with more LSK cells identified in the vicinity of perivascular region in DBZ-treated marrow (Supplementary Fig. S4B, bottom) compared with nontreated marrow (Supplementary Fig. S4B, top). In addition, DBZ treatment increased ki67 index of engrafted LSKs (Fig. 6H) and osteoblast cell-surface expression of CXCL12 in leukemia mice (Fig. 6I and J). Further, systemic Notch blockade suppressed elevated expression of IL6 and some of Notch targets, but reversed stromal expression of CXCL12, Runx2, and osterix (Fig. 6K). Systemic Notch blockade also improved platelet numbers in secondary leukemia mice (Supplementary Fig. S4C); however, it had no significant effect on hematocrit or MEP (Supplementary Fig. S4D and S4E). In summary, these findings suggest that leukemia-mediated osteoblastic alteration and suppression of marrow niche function are substantially improved by blocking Notch activation.

To examine if human T-ALL mediate similar hematopoietic and stromal suppression, we engrafted NSG mice with human T-ALL cells (14). We found that NSG mice engrafted with DND41 (Fig. 7A) or KOPT-K1 (data not shown) showed suppression of non-human LSK populations that was partially reversed by DBZ treatment. Both OB and MSC frequencies (Fig. 7B) and their cell surface expressions of CXCL12 (Fig. 7C) were decreased in leukemia-developing mice, but were much improved in mice receiving DBZ treatment. Further, nonhematopoietic stromal cells in DND41-engrafted NSG mice displayed downregulation of Runx2 and CXCL12, and upregulation of Deltex, JAG1, and Hey2 (Fig. 7D) that were reversed by DBZ treatment. Therefore, human T-ALL cells also show GSI-reversible OB and HSPC suppression as mouse T-ALL cells, although such responses are likely also contributed by decreased leukemia burden following treatment (84 ± 8% in nontreated mice vs. 65 ± 10% in treated mice, P < 0.05) because these T-ALL cells still retain γ-secretase cleavage sites (14).

To further examine the relevance of our findings in human ALL, we examined sampled (n = 10) nonleukemic human bone marrow sections and age-matched ALL marrow sections from 2 T-ALL and 8 B-ALL patients. Bone-lining cuboid shaped or flattened osteoblasts are readily identified in nonleukemic bones (Fig. 7E), but are notably decreased in 2 T-ALL (Fig. 7F) specimens, and are also modestly decreased in B-ALL bone marrow (Fig. 7G). In addition, compared with 2 nonleukemia tissues (Fig. 7H), Hes1 expression was increased showing stronger nuclear staining in the remaining osteoblasts in 2 T-ALL (Fig. 7I) and also in 6 of 8 B-ALL specimens. These findings of aberrant Notch activation in osteoblasts and a general loss of osteoblasts are in agreement with findings from animal models and human T-ALL grafts.

Proliferation of leukemia cells in acute leukemia patients is often associated with anemia and cytopenia. This condition is generally believed to reflect a “crowding-out” effect by the rapidly proliferating leukemia cells. Emerging evidences, however, indicate that leukemia cells modulate the marrow microenvironment to disrupt the communication between the stem cell niche and the residing HSPCs that is essential for hematopoietic homeostasis. In this study, using a mouse model of T-ALL, we found that lymphoblast leukemia cells induce a marked suppression of normal hematopoiesis by harnessing two critical niche components, the osteoblastic lineage cells and the perivascular region. We identified that aberrant Notch activation plays an important role in these processes because blocking Notch activation can largely recover suppressed osteoblast numbers and increase HSPC proliferation, and improves animal survival as well as thrombocytopenia. We show that, at least in ICN1 T-ALL mouse model, an improved hematopoiesis and survival are a direct result of leukemia microenvironment change rather than a consequence of decreased leukemia burden. Notably, we confirmed that osteoblast loss and associated Notch activation were present in human T-ALL xenografts as well as in human ALL specimens.

Osteoblasts are defined as MSC progeny committed to the osteoblastic lineage expressing many cell-signaling molecules, such as JAG1, CXCL12, membrane-bound SCF, angiopoietin-1, osteocalcin, and osteopontin, some of which are critical to support HSC self-renewal and survival (9, 23, 30). Mature osteoblasts also directly regulate B lymphopoiesis (31, 32). Recent studies have linked dysfunctional osteoblast expansion with disease pathophysiology in acute and chronic myeloid leukemia (33, 34). However, the significance of osteoblast in modulating lymphoblastic leukemia microenvironment has not been studied. Here, we show that unlike in myeloproliferative neoplasia, osteoblast population is strikingly suppressed in mice developing T-ALL. We also show that osteoblast suppression is associated with a loss of normal HSPC population, and suppressed B lymphocytes, consistent with the dual function of these cells in supporting HSPC homeostasis and B lymphopoiesis. Further, we identified a critical role of Notch activation in osteoblast suppression. The implication of Notch as the major pathway to suppress osteoblast differentiation is not surprising as Notch activation has been shown to inhibit osteoblastic progenitor terminal differentiation through Hes1 repressing Runx2 transcriptional activity (35–37). Our findings further suggest that osteoblast CXCL12 expression is negatively regulated by Notch activation. Finally, variable degree of osteoblast loss is observed in human T-ALL as well as B-ALL specimens. Although we show that blocking Notch activation could partially account for the recovery of dysregulated osteoblastic niche functions, neither have we excluded contributions by other pathways, such as Wnt or TGF-β, in the suppression of HSPC, nor have we examined other molecular mechanisms contributing to disease-specific suppression of osteoblasts and their niche function in various forms of human ALL.

Besides osteoblasts, perivascular niche and related cells are another major focus of research in leukemia microenvironment. Perivascular homing provides leukemia cells proliferative and survival cues, and also promotes dysregulated angiogenesis in the marrow, linking marrow to be a common site of relapse for hematologic malignancies, such as T-ALL (38, 39). Marrow could serve as sanctuary sites for the integration of leukemia cells and contributes to the leukemic potential of transformed endothelial cells (40). In a pre-B ALL xenograft model, CXCL12/CXCR4 axis signaling was found to support leukemia metastasis to the specialized microvascular domain (5). Consistent with other reports (41, 42), CXCR4 expression was found increased in leukemic T cells. However, GSI did not inhibit increased CXCR4 expression nor affect leukemia homing to the perivascular region (data not shown), indicating that either Notch activation is not responsible for the increased CXCR4 expression and preferred perivascular homing of leukemia cells, or that other pathways or molecules also regulate CXCR4 expression. Nevertheless, lymphoblastic leukemia cells out-compete normal HSPC and displace HSPC from the perivascular region, and contribute to the suppression of HSPC proliferation. In addition, disruption of endothelial niche where megakaryocytes are normally in close interaction with HSC may explain platelet abnormality observed in ICN1 mice (43). Although MEP and EP were mildly decreased in ICN1 leukemia marrow, prominent anemia seen in acute leukemia patients was not found in these leukemia mice. It is possible that severe anemia may take longer to develop in mice. Alternatively, splenomegaly-associated increase of EPs and extramedullary hematopoiesis could compensate for the reduction of marrow red cell production (data not shown).

Leukemia-mediated environmental change also implicates abnormal angiogenesis (44), hypoxia (45), or the inflammatory cytokines, such as SCF and IL6 (33, 46). Increased inflammatory cytokines were found responsible for supporting reinforced leukemia-niche for CML cells (23). Notably, inflammatory cytokines, including IL6, were found increased in the marrow stroma of T-ALL driven by ICN1. Increased IL6 is associated with poor response to GSI in cancer cell lines (47). We found that IL6 was suppressed after GSI treatment, suggesting that alterations of inflammatory cytokines are likely a secondary effect downstream of Notch activation in T-ALL. Stroma Notch activation could be induced by upregulated Notch ligand (Fig. 5C), hypoxia, or inflammation. Further studies are required to elucidate the exact molecular mechanism underlying stromal NOTCH activation in ALL leukemic marrow, and the molecular links between aberrant Notch activation and inflammatory cytokine increase implicated in leukemia progression and therapy resistance.

In summary, our findings reveal that during disease progression, lymphoblastic leukemia employs disease-specific cellular and molecular machinery to dysregulate hematopoietic niche function. It is known that in relapsed ALL, reinduction with aggressive chemotherapy or stem cell transplantation is associated with long-term sequelae and overall poor outcomes (48). Knowledge gained from our studies thus provides rationale of developing niche-targeted approach in the future to improve clinical outcome in ALL patients.

No potential conflicts of interest were disclosed.

Conception and design: L. Zhou

Development of methodology: W. Wang, G. Zimmerman, X. Huang, S. Moreton, J. Nthale, A. Awadallah, R. Beck, D. Wald, A.Y. Huang, L. Zhou

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Zimmerman, X. Huang, S. Yu, J. Myers, Y. Wang, S. Moreton, J. Nthale, R. Beck, W. Xin, A.Y. Huang, L. Zhou

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W. Wang, G. Zimmerman, X. Huang, S. Yu, W. Xin, D. Wald, L. Zhou

Writing, review, and/or revision of the manuscript: W. Wang, X. Huang, A.Y. Huang, L. Zhou

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Moreton, W. Xin, D. Wald, L. Zhou

Study supervision: L. Zhou

This work was supported by grants from American Cancer Society LIB-125064 (L. Zhou), NIH HL103827 (L. Zhou), Hyundai Hope-on-Wheels Program (A.Y. Huang), and the Keira Kilbane Cancer Innovation Fund (A.Y. Huang).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.