Loss of Ikaros DNA-binding function confers integrin-dependent survival on pre-B cells and progression to acute lymphoblastic leukemia.

Deletion of the DNA-binding domain of the transcription factor Ikaros generates dominant-negative isoforms that interfere with its activity and correlate with poor prognosis in human precursor B cell acute lymphoblastic leukemia (B-ALL). Here we found that conditional inactivation of the Ikaros DNA-binding domain in early pre-B cells arrested their differentiation at a stage at which integrin-dependent adhesion to niches augmented signaling via mitogen-activated protein kinases, proliferation and self-renewal and attenuated signaling via the pre-B cell signaling complex (pre-BCR) and the differentiation of pre-B cells. Transplantation of polyclonal Ikaros-mutant pre-B cells resulted in long-latency oligoclonal pre-B-ALL, which demonstrates that loss of Ikaros contributes to multistep B cell leukemogenesis. Our results explain how normal pre-B cells transit from a highly proliferative and stroma-dependent phase to a stroma-independent phase during which differentiation is enabled, and suggest potential therapeutic strategies for Ikaros-mutant B-ALL.

B cell differentiation is characterized by stage-specific expression of cell surface markers and recombination of the immunoglobulin heavy chain (IgH) and light chain (IgL) genes. These events are responsible for the generation of a large pool of immature B cells from which selection based on antigen receptor specificity takes place[1,2]. Productive rearrangements at the Igh locus allow pairing of the expressed IgM with the surrogate light chains (SLC), VpreB and λ5, and the proximal signaling molecules Igα and Igβ to form a pre-B Cell Receptor (pre-BCR) signaling complex. Subsequent engagement of the protein tyrosine kinases (PTKs) Lyn, Fyn, Blk and Syk activates signaling cascades supporting pre-B cell proliferative expansion and differentiation[3]. Loss-of-function mutations in the pre-BCR signaling complex or in associated PTKs cause arrest at an early B cell precursor stage[4-10]. The pre-BCR, working in concert with the growth-promoting IL-7 cytokine receptor (IL-7R), activates the PI3K-Akt and Mitogen-Activated protein kinases (MAPK) Erk1 and Erk2, thereby providing pre-B cell survival and proliferation[11-14].

Pre-BCR signaling also induces differentiation through a distinct set of signaling effectors such as Btk, Slp65 (Blnk) and PLCγ2 (refs. 15-17). These inhibit the PI3K pathway while activating Ca2+ signaling and a network of transcription factors responsible for cell cycle withdrawal and immunoglobulin light chain (IgL) gene rearrangement[18-20]. Although the importance of pre-BCR signaling in proliferation and differentiation is well established, how the transition between these two disparate phases occurs remains unclear. Loss in IL-7R signaling as well as quantitative and qualitative changes in pre-BCR signaling have been proposed as possible mechanisms underlying this pre-B cell switch.

Human precursor B cell acute lymphoblastic leukemias (B-ALL) frequently display a pre-B cell phenotype, suggesting that a block at the pre-B cell proliferative stage may contribute to leukemogenesis[21]. Genome-wide studies in human leukemias have identified loss-of-function mutations in genes encoding regulators of B cell differentiation such as PAX5, TCF3, EBF1, and IKZF1 (IKAROS gene) in ~40% of samples from patients with precursor B-ALL[22]. Notably, IKZF1 mutations, including deletions in the Ikaros DNA-binding domain, were singled out as genetic lesions associated with B-ALL with poor prognosis[23-27]. Ikaros is required to induce transcription of lymphoid-specific genes in multi-potent progenitors, and its loss leads to developmental arrest prior to B cell lineage specification[28,29]. Ikaros, together with its family member Aiolos, which is induced after B cell lineage specification[30], have been implicated in promoting pre-BCR-mediated differentiation by repressing expression of the SLC of the pre-BCR complex[31].

Here, we provide new insight into how pre-B cells switch from proliferation to differentiation, a process that is vulnerable to leukemic transformation. We describe a stromal-adherent self-renewing phase in pre-B cell differentiation that expresses the pre-BCR signaling complex and shows strong activation of the Erk1 and Erk2 and PI3K-Akt proliferation and survival pathways, but which has no Ca2+ signaling potential, normally required for differentiation. Loss in pre-B cell stromal adhesion correlates with attenuation of proliferation, and an increase in the differentiation-inducing components of the pre-BCR signaling complex and the potential for Ca2+ signaling. Importantly, the transition of pre-B cells from a stromal-adherent proliferative to a non-adherent differentiation phase is dependent on Ikaros. Loss of Ikaros augments stromal adhesion in an integrin-dependent manner, locking pre-B cells in a highly proliferative and self-renewing phase from which BALL can arise. Importantly, the survival and proliferation of Ikaros-deficient pre-B cells is strictly dependent on the cooperation between integrin and growth factor receptor signaling, suggesting a new avenue for treatment of IKZF1 mutant, poor-prognosis B-ALL.

Results

The Ikaros family is required for pre-B cell differentiation

To determine the role of the Ikaros family during B cell differentiation, exon 5 of the Ikzf1 gene (defined hereafter as IkE5), encoding two Ikaros DNA binding zinc fingers, was floxed (IkE5fl/fl; ) and deleted from either the common lymphoid progenitor or the downstream definitive pro-B cell precursor using CD2-Cre or CD19-Cre transgenes, respectively (). Deletion of IkE5 generates Ikaros protein isoforms that lack DNA binding activity and are structurally similar to those encountered in human B-ALL (Ik6)[24] (, pre-B IkE5Δ/Δ. These mutant Ikaros isoforms act in a dominant-negative fashion by dimerizing with co-expressed family members, including Aiolos, and interfering with their DNA binding activity[30,32]. We confirmed the dominant-negative phenotype by combining the Ikzf3 (Aiolos gene) homozygous null and the Ikzf1 heterozygous null mutations (Ikzf3−/−Ikzf1+/−). Deletion of IkE5 or the combined Ikzf3−/− Ikzf1+/− mutations caused a similar block and expansion of large pre-B cells (CD19+CD43+BP1+; and ). These normally represent a minor population but were now found in numbers that were similar to those of all bone marrow (BM) B cells in wild type (WT) mice (). As in WT, the majority of mutant large pre-B cells were in cell cycle (). The few immature B (CD19+IgM+) cells detected in the IkE5fl/fl CD2-Cre mice () had not deleted IkE5fl/fl (), indicating that transition from the large to the small pre-B cell is absolutely dependent on the DNA binding activities of Ikaros gene family members expressed at this stage of differentiation.

A hallmark of B cell differentiation is the successful recombination of the Igh locus, a prerequisite for transition to the pre-B cell stage. Both D-J and V-DJ proximal and distal recombination events at the Igh locus were detected at similar frequencies in WT and IkE5Δ/Δ pre-B cells ( and data not shown). However, the low-level Igk rearrangements detected in WT were not seen in mutant pre-B cells (), indicating either inability to undergo light chain recombination or a block in differentiation prior to Igk recombination. Consistent with recombination only at the Igh locus, the majority of mutant pre-B cells expressed intracellular IgM but not Igκ ().

Since Igk recombination is required for B cell maturation, we attempted to rescue the pre-B cell arrest by crossing the IkE5Δ/Δ mice to the D23 transgenic line that expresses a pre-rearranged Igκ chain[33]. IkE5Δ/Δ D23 pre-B cells were unable to differentiate past the large pre-B (CD19+CD43+ BP1+) stage ( and ), although both IgM and Igκ chains were expressed intracellularly. This indicates that lack of Igk recombination was not the cause of the maturation defect in Ikaros-deficient pre-B cells. Hence, the transition from large to small pre-B cell is regulated by the Ikaros gene family through a mechanism that is independent of recombination at the IgH or IgL gene loci.

Growth of IkE5Δ/Δ pre-B cells requires adhesion to stroma

The developmental defect in IkE5Δ/Δ large pre-B cells was further evaluated in in vitro cultures[11,34]. Under differentiation-inducing conditions (i.e. seven days of stromal-free culture in low concentrations of serum and IL-7), the majority of WT large pre-B cells exited the cell cycle and differentiated into small pre-B (CD19+CD2+IgM−) and immature B (CD19+IgM+CD2+) cells, whereas mutant large pre-B cells (CD19+CD43+BP1+) remained undifferentiated (). An increase in the concentration of IL-7 promoted the proliferative expansion of WT large pre-B cells but had little effect on their mutant counterparts. In the absence of stroma, survival of IkE5Δ/Δ pre-B cells was greatly compromised compared to WT pre-B cells even in the presence of high concentrations of IL-7, with high levels of apoptosis detected from early time points of culture (, left panel, and ).

Although pre-B cell precursors can proliferate and differentiate in the absence of stromal contact, they can only self-renew and undergo greater expansion on stroma[34-36]. Since IkE5Δ/Δ large pre-B cells proliferated and expanded in vivo in the BM, we tested whether they could grow in vitro on OP9 BM-derived stroma. Under these conditions, IkE5 large pre-B cells grew better than WT especially under limiting concentrations of IL-7 (, right panel), with ~2 to 10-fold more cells in cycle (). IkE5Δ/Δ pre-B cells also displayed an increase in cell cycle kinetics compared to WT. Labeling with BrdU showed increased incorporation by IkE5 pre-B cells during pulse and a faster decline during chase (), indicating shorter cell cycle transitions compared to WT pre-B cells.IkE5Δ/Δ pre-B cells are therefore dependent on stroma for survival and growth, with enhanced proliferation and more rapid cell cycling relative to WT pre-B cells.

Ikaros loss arrests pre-B cells in an adherent phase

A striking morphological difference was apparent between IkE5Δ/Δ and WT large pre-B cell OP9 stromal cultures. The majority of WT pre-B cells were round, light-refracting cells loosely attached to stroma, but the majority of mutant cells had a dark, flat morphology and appeared incorporated into the stromal layer (). Dark, stromal-adherent pre-B cells were also present in WT cultures, but at a much lower frequency (). The few IkE5Δ/Δ non-adherent cells displayed increased apoptosis (), indicating that in the absence of stromal contact, their survival was greatly compromised.

A progenitor-progeny relationship between adherent and non-adherent pre-B cells was next established in WT cultures. Comparison of transcriptional profiles revealed that small pre-B cell markers (e.g. Igκ, Rag1, Rag2, Irf4, Cd2, and Cd25) expressed at low levels in adherent pre-B cells were induced in the non-adherent fraction, whereas cell cycle-promoting genes, such as Ccnd2, Egr1, Pcna, Igfbp4 and Myc, displayed the opposite expression pattern (). The overall gene expression of mutant adherent pre-B cells was similar to that of their WT adherent counterparts, although a further reduction in small pre-B cell markers was seen in the mutant cells. The differential expression of small pre-B cell markers, such as CD25 and intracellular Igκ, between WT adherent and non-adherent pre-B cells, was also detected by flow cytometry. Intracellular IgM, a pan pre-B cell marker, was similarly expressed in both WT pre-B cell subsets (). In the mutant cultures, adherent pre-B cells expressed IgM but no Igκ or CD25, protein consistent with the ex vivo analysis of IkE5Δ/Δ pre-B cells ().

The cell cycle properties of WT adherent and non-adherent pre-B cells were evaluated. Whereas most WT adherent pre-B cells were in cycle, non-adherent WT pre-B cells consisted of large cycling and smaller non-cycling cells at a ratio that decreased with time in culture ( and data not shown). WT adherent pre-B cells could be serially passaged on stroma and gave rise to both adherent and non-adherent cells, whereas WT non-adherent pre-B cells gave rise to mostly non-adherent cells with limited proliferative expansion ().

Given the self-renewing potential of adherent pre-B cells, we compared the clonogenic properties of WT and mutant adherent pre-B cells in a limiting dilution colony-forming assay on stroma (). Even in the absence of IL-7, the colony-forming potential of IkE5Δ/Δ pre-B cells was high (~20%) and orders of magnitude greater than WT. Although addition of IL-7 had little effect on the ability of IkE5Δ/Δ adherent pre-B cells to form colonies on stroma, it did increase their size by increasing proliferation ( and data not shown). Evaluation of the ability of WT and IkE5Δ/Δ adherent pre-B cells to re-associate with stroma revealed another important difference. Within 3 hrs of replating, 68% of IkE5Δ/Δ pre-B cells rapidly re-bound to stroma, whereas only 15% of WT adherent cells did so even after overnight incubation ().

Together, these studies provide insight into pre-B cell differentiation by describing the transition from a stromal-adherent to a non-adherent phase. Stromal-adherent pre-B cells express pre-BCR are highly proliferative and have limited self-renewing potential. They are highly dependent on Ikaros for transition to a non-adherent phase where they exit the cell cycle, lose self-renewal capacity, and acquire expression of genes supporting B cell maturation. Loss of Ikaros augments stromal adhesion, self-renewal, and proliferation, pathways that most likely antagonize activation of the pre-B cell differentiation program.

Ikaros loss augments stromal-dependent proliferation

The survival and proliferative expansion of pre-B cells are supported by a combination of pre-BCR and IL-7R signaling that activates the PI3K-Akt and Erk1-2 MAPK pathways ( and ). Both PI3K-Akt and Erk1-2 were active in WT adherent but not in non-adherent pre-B cells (), which are in the process of exiting the cell cycle () and upregulating expression of small pre-B cell markers (). Activation of Akt was similar in IkE5Δ/Δ adherent pre-B cells compared to WT, but activation of Erk1 and Erk2 was greatly increased (). Consistent with a higher Erk1-2 MAPK activity, an increase in Cyclin D2 () and cell cycle () was observed in IkE5Δ/Δ compared to WT large pre-B cells.

Pre-BCR signaling also supports differentiation to the small pre-B cell stage by activating PLCγ and Ca2+ signaling (). These signaling events are required to switch cells from a proliferative to a quiescent state by inducing transcriptional responses that rely in part on the transcription factor Foxo1 (refs. 18,20). Notably, the baseline level of intracellular Ca2+ was low in both WT and IkE5Δ/Δ adherent pre-B cells but elevated in WT non-adherent pre-B cells, which were the only cells capable of fluxing Ca2+ either upon pre-BCR engagement or ionomycin treatment ( and ). Upstream and downstream effectors of Ca2+ signaling such as Blnk and Foxo1 proteins were expressed at low amounts in both WT and IkE5Δ/Δ adherent pre-B cells and were greatly induced in WT non-adherent cells, consistent with small pre-B cell differentiation (). The low amounts of Foxo1 expressed in adherent pre-B cells were phosphorylated, correlating with active PI3K-Akt in these cells. As previously reported for small pre-B cell differentiation[20], p38 MAPK activity was induced from WT adherent to non-adherent pre-B cells but was nearly undetectable in IkE5Δ/Δ adherent pre-B cells ().

Both the proliferation- and differentiation-inducing arms of pre-BCR signaling are dependent on activation of Fyn, Lyn, Blk and Syk. These PTKs were expressed at similar protein amounts in WT adherent and non-adherent pre-B cells (). In IkE5Δ/Δ compared to WT adherent pre-B cells, the protein amounts of these key proximal components of pre-BCR signaling (Fyn, Syk, Blk) were greatly reduced, while the amounts of activated (phosphorylated) Lyn were also diminished ( and ). The reduced protein expression or activation of these PTKs was unexpected, as it predicts not only a defect in differentiation but also in proliferation of IkE5Δ/Δ pre-B cells, contrary to what we observed both in vivo and in vitro ().

IL-7R signaling was examined as a possible mechanism of compensation for the loss in pre-BCR signaling. Phosphorylation of Stat5 (p-Stat5), a measure of IL-7R signaling, was comparable in WT and IkE5Δ/Δ adherent pre-B cells (). Furthermore and in contrast to WT pre-B cells, IL-7R signaling was unable to support the growth of mutant pre-B cells under stromal-free conditions, and the mutant pre-B cells were only partly dependent on IL-7 for growth on stroma ( and ). Thus, receptors other than IL-7R and pre-BCR must be responsible for activation of survival and proliferation signaling pathways in IkE5Δ/Δ large pre-B cells. Engagement of such receptors is likely to be mediated by interaction of the mutant pre-B cells with stroma.

Increased integrin signaling in IkE5Δ/Δ pre-B cells

A comparative genome-wide transcriptional analysis of primary and cultured WT and IkE5Δ/Δ pre-B cells was performed to reveal potential pathways that might support the aberrant adhesion and growth properties of the mutant population. A signature of genes was deduced that was differentially expressed between both freshly isolated IkE5Δ/Δ and WT large pre-B cells and between IkE5Δ/Δ and WT adherent pre-B cells cultured in vitro (). Up-regulated genes in IkE5Δ/Δ pre-B cells were highly enriched in pathways involved in focal adhesion and remodeling of the actin cytoskeleton (). Integrins (e.g. Itga1, Itga5, Itgb1) as well as other structural and signaling components of focal adhesions (e.g. Ptk2, Vcl, Actn1, Cttn, Dock1, Rogdi) were shared by many of these pathways (). The increase in integrin expression was validated at the protein level in both primary and cultured cells. Furthermore, expression of the active isoform of integrin β1 (detected with an activation-specific anti-integrin β1 antibody and elevated levels of phosphorylated focal adhesion kinase (p-FAK), a key downstream effector of integrin signaling, indicated that not only expression but also activation of integrin signaling were elevated in IkE5Δ/Δ pre-B cells (). Although not as pronounced as in IkE5Δ/Δ pre-B cells, significantly higher amounts of FAK and p-FAK were also observed in WT adherent relative to non-adherent pre-B cells, indicating that integrin signaling is also active in WT adherent pre-B cells ().

Further evidence for integrin-mediated adhesion was provided by measuring binding of adherent pre-B cells to integrin ligands in vitro. Notably, the frequency by which IkE5Δ/Δ (~80%) or WT (~20%) pre-B cells bound to fibronectin () was similar to that with which they bound to stroma (). A fibronectin peptide (RGD) that binds to integrin α4β1significantly inhibited the binding of both WT and IkE5Δ/Δ pre-B cells, implicating VLA-4 as one of the integrins participating in pre-B cell adhesion ().

Integrin signaling is involved in pre-B cell chemotaxis in response to CXCL12 (stromal cell-derived factor 1; SDF1)-CXCR4 receptor interactions[37]. We tested whether elevated integrin signaling in IkE5Δ/Δ pre-B cells affected their chemotactic properties. In sharp contrast to wild-type adherent pre-B cells, their mutant counterparts were unable to migrate in a transwell assay in response to SDF1 (), indicating that increased integrin signaling reduced chemokine-mediated chemotaxis of the mutant pre-B cells. Consistent with these in vitro data, circulating pre-B cells were not detected in IkE5fl/fl CD19-Cre mice although they were readily seen in WT mice (). Thus, the increase in integrin signaling manifested upon loss of Ikaros in pre-B cells is likely responsible for their stable adhesion to stroma, survival, and proliferative expansion.

IkE5Δ/Δ pre-B cell survival is dependent on FAK activity

The role of integrin signaling in supporting stromal adhesion and survival of IkE5Δ/Δ pre-B cells was validated by treatment with a small molecule inhibitor (PF-431396) that blocks the kinase activity of FAK (Ptk2) and the related kinase Ptk2b (ref. 38), which together serve as major signaling effectors of the integrin pathway. FAK-Ptk2b inhibitor treatment greatly reduced stromal adhesion not only in IkE5Δ/Δ but also in WT pre-B cells (). However, the loss in adhesion preceded an increase in apoptosis only in IkE5Δ/Δ pre-B cell and not in WT pre-B cells ().

The dependence of IkE5Δ/Δ pre-B cells on integrin signaling was also tested in vivo. IkE5fl/fl CD19-Cre mice and WT littermates were given 3-5 doses of an orally bioavailable FAK-Ptk2b inhibitor (PF-562271) or vehicle control and the number of BM pre-B/B cells and apoptotic index was quantified shortly afterwards (). IkE5Δ/Δ large pre-B cells constituted the great majority of BM B cells in vehicle-treated IkE5fl/fl CD19-Cre mice and showed rapid reduction following FAK inhibitor treatment (). This decrease correlated with an increase in apoptosis that was specific for large pre-B cells and correlated with a specific reduction in activated FAK (p-FAK) in IkE5Δ/Δ large pre-B cells ( and ). FAK inhibitor treatment had little effect on the cellularity of WT BM B cells that were mainly comprised of small pre-B and immature B cells. Given the small number of WT large pre-B cells present in the WT BM, it was difficult to discern the effect of the FAK inhibitor treatment on the WT large pre-B cell population.

Taken together, these studies indicate that increased integrin signaling mediated by FAK is responsible for increased stromal adhesion, survival, and accumulation of Ikaros-mutant large pre-B cells under both in vitro and in vivo settings.

Integrin and growth factor signaling cooperate in pre-B cells

We then tested whether integrin-mediated adhesion was sufficient to support stromal-dependent survival and proliferation of IkE5Δ/Δ pre-B cells. The majority of IkE5Δ/Δ pre-B cells plated on fibronectin and collagen died after overnight incubation, indicating that integrin signaling alone could not support their survival (). In sharp contrast, the majority of WT pre-B cells survived under these conditions.

Stromal niches provide adhesion and growth factor support. Growth factors such as the c-Kit ligand (Stem Cell Factor; SCF) and IL-7 are required for the growth of both early hematopoietic progenitors and lymphoid precursors[39,40]. In the absence of integrin ligand binding, IL-7 and/or SCF had little or no effect on the survival of IkE5Δ/Δ pre-B cells. However, the combination of integrin engagement and IL-7 or SCF greatly increased IkE5Δ/Δ pre-B cell survival (), and had a smaller but still significant stimulatory effect on the proliferation of IkE5Δ/Δ pre-B cells (). Thus, augmentation of integrin signaling in IkE5Δ/Δ compared to WT pre-B cells is not only important for maintaining cells in proximity to a stromal niche, but also for cooperating with growth factor signaling to support survival and proliferation, acting in lieu of pre-BCR signaling ().

High leukemogenic potential of IkE5Δ/Δ pre-B cells

The rapid development of precursor T-lymphoid neoplasms in IkE5Δ/Δ mice (data not shown and ref. 41) precludes the assessment of B-lymphoid leukemogenesis in these mutant mice. The leukemogenic potential of IkE5Δ/Δ pre-B cells was therefore evaluated by transplantation of this population into immunodeficient NOD-SCID-Il2rg−/− (NSG) recipient mice. Following transplantation with IkE5Δ/Δ pre-B cells isolated from either Cd19-Cre or Cd2-Cre donors, recipient NSG mice uniformly exhibited circulating immature CD19+BP1+CD2− B-lymphoid cells within 7 weeks (data not shown), and developed signs of disseminated leukemia/lymphoma by 3-4 months post-transplant, with weight loss, hyperventilation, and hepatosplenomegaly (mean spleen weight 668 ± 188 mg), whereas recipients of WT pre-B cells remained healthy (). The disease in recipients of IkE5Δ/Δ Cd19-Cre pre-B cells was somewhat more aggressive than in IkE5Δ/Δ Cd2-Cre recipients (; median survival 107d vs. 143d; P = 0.0021, Mantel-Cox test). At necropsy, recipients of IkE5Δ/Δ Cd19-Cre pre-B cells had pancytopenia with severe anemia (blood hemoglobin 4.8 ± 0.7 g/dL) that likely contributed to morbidity or death, while IkE5Δ/Δ Cd2-Cre recipients tended to develop hind-limb paralysis and malignant pleural effusions.

Histopathological analysis showed extensive invasion of spleen and liver and involvement of bone marrow with large lymphoid cells of high mitotic index (). Phenotypic analysis of the malignant IkE5Δ/Δ tumor cells revealed that they were similar to the initially transplanted population with regards to both large pre-B cell surface antigen expression (CD19+CD43+BP1+CD2−; ), expression of adhesion molecules (), and adherence to stroma (see below). However, in contrast to the polyclonal nature of the transplanted IkE5Δ/Δ pre-B cell population ( and ), the IkE5Δ/Δ leukemic cells were oligoclonal by Igh gene rearrangement ().

Given that the stromal-adhesion phenotype was maintained in IkE5Δ/Δ leukemic pre-B cells, we examined the status of integrin signaling and whether these cells were sensitive to FAK inhibition in vitro. Both integrin expression and FAK activation were elevated in the leukemic cells (). Similar to the pre-leukemic mutant pre-B cells, IkE5Δ/Δ leukemic pre-B cells were highly sensitive to FAK inhibition undergoing both loss of adhesion and dramatic increase in apoptosis () that correlated with suppression of p-FAK (). Thus, the arrest at the adherent large pre-B cell stage mediated by loss of Ikaros predisposes this population for transformation to a leukemic state, which however appears to be sensitive to inhibition of adhesion-based signaling pathways supported by FAK, thereby opening a new avenue for therapy of poor prognosis B cell precursor leukemias in humans.

Figure 8

Leukemogenic potential of IkE5Δ/Δ pre-B cells

a, Kaplan-Meier survival curve of NSG mice transplanted with WT or IkE5Δ/Δ pre-B cells. The survival of both cohorts of recipients of IkE5Δ/Δ pre-B cells was significantly shorter than recipients of WT pre-B cells (P = 0.013, Mantel-Cox tests). b, Histopathology of precursor B-cell acute lymphoblastic leukemia/lymphoma derived from IkE5Δ/Δ pre-B cells. (i–iii): Hematoxylin & eosin-stained sections of spleen (i), liver (ii), and BM (iii) from a premorbid NSG recipient (sacrificed day 63 post-transplant) of IkE5Δ/Δ pre-B cells from a CD19-Cre donor. Note the extensive infiltration of all organs with large cells with moderate cytoplasm and prominent nucleoli, and frequent mitotic figures (arrows). Scale bars, 50 μm. (iv) Wright-Giemsa stain of cytospin of BM from this recipient (scale bar, 20 μm). Note predominant population of large lymphoblasts with immature nuclei and basophilic cytoplasm (arrows). c, Integrin expression is elevated in both IkE5Δ/Δ pre-leukemic and leukemic pre-B cells. Percentage of WT, IkE5Δ/Δ pre-leukemic and leukemic pre-B cells expressing integrins α5 (CD49e), α6 (CD49f) and β1 (CD29). d, FAK activation (pFAK) measured by flow cytometry in the presence and absence of FAK inhibitor in WT and mutant pre-B cells. e, FAK inhibition interferes with stromal adhesion of IkE5Δ/Δ preleukemic and leukemic pre-B cells. Inhibitor-treated, closed symbols; vehicle-treated, open symbols. (n=2 each). f, FAK inhibition induces cell death in IkE5Δ/Δ pre-leukemic and leukemic pre-B cells (n=4; *P <10-6, **P <10−7 two-tailed Student's t-test).

Discussion

Our studies define a key step in pre-B cell differentiation that is characterized by adherence to bone marrow stroma, self-renewal and proliferative expansion[36]. Normal pre-B cells transit rapidly through this stromal adherent phase and enter into a non-adherent phase in which self-renewal is lost, proliferation is diminished and differentiation into an Igκ-expressing cell is induced. Loss of Ikaros activity arrests pre-B cells in the adherent, self-renewing, pro-proliferative phase and promotes their transformation to a malignant state.

Ikaros-deficient large pre-B cells show increased expression of structural and signaling components of the focal adhesion and actin cytoskeleton pathways at both the transcription and protein level. Integrins (Itgb1, Itgb3, Itga9, Itga5), vinculin (Vcl), α-actinin (Actn1), myosins (Myo1b, Myl12b), FAK (Ptk2), and the Rac activating GEFRho guanine exchange factors (Arhgef12, Arhgef5), Rho GTPase activating protein 5 (Arhgap5), and Dedicator of cytokinesis (Dock1) were such examples. Extracellular matrix (ECM) components such as laminin (Lamb1), secreted phospho protein (Spp1) and matrix metallopeptidase 14 (Mmp14) were also induced in mutant pre-B cells. These transcriptional changes underscore a strong integrin-signaling environment that is demarcated by high levels of activated FAK, and a stable adhesion and re-adhesion phenotype to both stroma and integrin ligands. WT pre-B cells also demonstrate integrin-dependent adhesion, however, this is transient as the majority of adherent cells rapidly switch to a non-adherent phase and fail to re-adhere upon replating.

Integrins, engaged by the ECM, serve as signaling centers that control actin filament polymerization required for the formation and maturation of focal adhesions[42,43]. Recent studies suggest that actin also plays a role in organizing adhesion sites and the actinintegrin linkage composition can determine adhesion stability[44,45]. Cells that lack highly bundled actin structures, such as lymphocytes, have less prominent adhesions[46]. The actin cross-linker α-actinin, required for the formation of actin filaments[47], is upregulated in Ikaros deficient pre-B cells. Actin polymerization and disassembly are regulated by the opposing activities of the Rho and Rac small GTPases[48]. A potential increase in the Rac activating GEF Dock1 in Ikaros-deficient pre-B cells may contribute to the establishment of an actin environment that is conducive to integrin signaling. Given the low levels of Ca2+ in adherent pre-B cells, recycling of adhesion structures, for example through calpain cleavage of talin, may be ineffective. Consistent with a stable focal adhesion environment, Ikaros-deficient pre-B cells failed to undergo SDF1-mediated chemotaxis in spite of normal or elevated expression of CXCR4 and do not exit the BM microenvironment. In this regard, a recent study reported increased chemotaxis of FAK-deficient pre-B cell precursors from the BM to the periphery[49] that is consistent with our observations herein of increased FAK activity in Ikaros-deficient pre-B cells and their inability to migrate from the BM to the periphery.

Notably, in WT pre-B cells, the pro-proliferative and differentiation-inducing arms of pre-BCR signaling are segregated away from one other and into the stromal-adherent and non-adherent phases of pre-B cell differentiation. The presence of IL-7R and pre-BCR together with active Erk1-2 MAPK and PI3K-Akt pathways in WT adherent pre-B cells indicates that receptor signaling is actively contributing to survival and proliferation. As WT pre-B cells detach from stroma, they rapidly turn off Erk1-2 and Akt signaling although expression of pre-BCR and IL-7R persists. An increase in p38 MAPK, Blnk and intracellular Ca2+ is observed, together with a transcriptional induction of pre-B cell differentiation markers. A working model supported by these findings is that in addition to pre-BCR and IL-7R signaling, integrin-mediated adhesion and FAK signaling contribute to the proliferative expansion of early pre-B cells and provide limited self-renewal by keeping them engaged on stroma. Another important effect of integrin signaling is to shut down the differentiation-inducing pathways in pre-B cells either directly or indirectly by promoting proliferation. As adhesion is lost, possibly due to recycling of focal adhesions, the negative effects on differentiation signaling are reduced. Increase in FAK activity has been implicated in the pathogenesis of a variety of cancers by engaging pro-proliferative signaling. Increased integrin and FAK signaling, as in Ikaros-deficient pre-B cells, may be responsible for augmenting self-renewal and proliferation and for further repressing differentiation at this critical developmental stage. A recent study has shown that loss of FAK during B cell differentiation causes a reduction in pre-B and immature B cells[49] and is consistent with the above-proposed model of integrin signaling in pre-B cell differentiation.

While integrin-mediated adhesion and activation of the downstream signaling effector FAK are augmented in Ikaros-deficient pre-B cells, all pre-BCR-affiliated PTKs are reduced. As transcription of these PTKs is relatively unperturbed, this effect is probably a consequence of altered protein stability. Feedback mechanisms originating from hyperactive FAK or MAPKs may stimulate degradation of these pre-BCR signaling effectors, thereby limiting the number of pro-proliferative signaling pathways operating in mutant pre-B cells. Loss in proximal pre-BCR signaling does not affect proliferation as this is also supported by other receptor signaling pathways such as growth factor receptors and integrins, but inhibits differentiation, which is solely dependent on the pre-BCR complex.

The switch from the pre-BCR-IL-7R signaling axis in normal pre-B cells to a more ‘progenitor-like’ integrin-growth factor signaling paradigm supports the survival and proliferative expansion of Ikaros-deficient pre-B cells. In fact, mutant pre-B cells cycle more rapidly, possibly due to a niche-mediated increase in MAPK and its downstream targets Cyclin D2 and CDK6. The ability of Ikaros-deficient pre-B cells to respond to different growth factors also highlights a potential to survive in different micro-environments. Aberrant expansion of mutant pre-B cells to non-physiological numbers represents the first step in a leukemic transformation process that evolves rapidly when the Ikaros-deficient pre-B cells are adoptively transferred to an immune-compromised bone marrow environment. However, the fact that the malignant precursor B-ALLs that develop are oligoclonal with respect to Igh gene rearrangement suggests that additional events are necessary for full malignant transformation. Whereas a subset of Igh rearrangements appears to be shared between leukemias arising in different recipients, this further suggests that some steps toward malignant transformation have occurred in the primary mice. Nonetheless, the Ikaros-deficient leukemic pre-B cell clones retain stromal-adherent properties and are still dependent for their survival on FAK activation.

The insights into the mechanisms that support normal pre-B cell differentiation and its aberrant manifestations described herein illuminate potential new strategies for the therapy of B-ALL that are linked to the underlying biology of the pre-B cell. Humans with IKZF1 mutant B-ALL have an inferior prognosis despite intensive treatment, which correlates with persistent residual disease following induction chemotherapy[23, 50]. Whereas inhibiting FAK causes detachment and death of Ikaros-deficient pre-B cells by depriving them from anchorage-dependent survival, it has little effect on WT pre-B cells. Hence, these and other pathways activated in Ikaros-deficient pre-B cells may provide additional targets for therapeutic intervention in poor-prognosis B-ALL.

Methods

Mice

The IkE5fl/fl mouse line was generated by inserting loxP sites flanking the Ikzf1 exon 5 by a standard gene targeting method. CD2-Cre and CD19-Cre transgenic lines were obtained from Drs. D. Kioussis and K. Rajewsky, respectively. All mice were bred and maintained under pathogen-free conditions in the animal facility at Massachusetts General Hospital, Bldg. 149-8. At the time of analysis, mice were 5–9wk of age. All animal experiments were done according to protocols approved by the Subcommittee on Research Animal Care at Massachusetts General Hospital (Charlestown, MA) and in accordance with the guidelines set forth by the National Institutes of Health.

Antibodies

Antibodies for bone marrow lineage depletion and flow cytometry were purchased from BD Pharmingen, Southern Biotech and eBiosciences. In some cases, hybridoma supernatant containing antibodies against Mac-1,Gr-1, TER119, c-Kit and CD3ε were used. Antibodies and the specific clones used were: CD3 (17A2), CD8α (53-6.7), TCRβ (H57−597), Mac-1 (M1/70), DX5, Gr-1 (RB6-8C5), Ter119 and IgM (R6.60-2 or 11/41), FcγRII/III (2.4G2), CD19 (1D3), c-Kit (ACK2 or 2B8) BP1 (FG35.4), CD43 (S7), CD2 (RM2-5), Ig kappa (187.D), Integrin α5 (HMa5-1), Integrin α6-PE (GoH3) and Integrin β1 (9EG7). For immunoblotting and immunofluorescence, antibodies from Cell signaling technologies raised against the phosphorylated and total protein for Akt (4060/4685), Erk (4377/4695), p38(4511/8690), Stat5 (9351), Lyn (2731/2796), Syk (12358), FAK (3283/8556/3285), Foxo1 (2880/9454), Fyn (4023), Blnk (12168), Btk (8547), and Cyclin D2 (3741) were utilized. Antibodies against total Stat5 (sc-835X), β-tubulin (sc-9104) and Blk (K-23) were purchased from Santa Cruz Biotechnology.

Flow cytometry and cell sorting

. Bone marrow (BM) cells were isolated as previously described[51]. Briefly, BM cells were harvested from femurs and tibias and subjected to red blood cell (RBC) lysis using ACK buffer (0.15 M ammonium chloride, 10 mM potassium bicarbonate, 0.1 mM EDTA). For large pre-B cell isolation, bone marrow cells were depleted with antibodies against Ter119, Mac-1, Gr-1, IgM, CD3, CD8a, TCRβ, DX5 and positive cells removed with magnetic beads conjugated to goat anti-rat IgG (QIAGEN). Remaining cells after depletion were labeled with various fluorochrome-conjugated mAbs against B-cell markers for phenotypic analysis, and CD19+CD43+BP1+cells were sorted as large-preB cells, for in vitro cultures and RNA-seq analysis and CD19+CD43- CD25+ as small pre-B cells for RNA-seq analysis. The BP1+ fraction of the CD19+CD43+ population expressed lower levels of c-Kit in both WT and IkE5Δ/Δpre-B cells. For analysis of immature IgM+ cells, undepleted BM cells were stained with mAbs against CD19 and IgM and analyzed within the lymphoid population by electronic gating based on size and granularity. For flow cytometry of integrins, cells were stained with either phycoerythrin (PE) conjugates or primary antibody followed by PE-conjugated secondary antibody. Antibody to P-FAK 925 was used for flow cytometry (Cell Signaling, 3284). Flow cytometric analysis was performed using a two-laser FACSCanto™ (BD) or a three-laser MoFlo® (Dako Cytomation). Cell sorting was performed using a three-laser MoFlo®. The resulting files were uploaded to FlowJo (Tree Star) for further analysis.

Intracellular staining

Large Pre B cells were fixed with 2% paraformaldehyde (Electron Microscopy Sciences, PA, USA) in PBS at room temperature for 20 min. After two washes with PBS, the cells were subsequently permeabilized with 0.5% saponinin 2% FCS/PBS for 20 minutes at 4°C. The cells were stained with FITC-conjugated anti-μ or anti-κ antibody for additional 30 min at 4°C, then washed twice in 2% FCS-PBS before analysis on a FACSCanto™ (BD). For intracellular staining for p-FAK, fixed cells were permeablized with 90% methanol for 2 hours and washed. The cells were then incubated with primary antibody for 30 minutes. Cells were washed and incubated with FITC-conjugated secondary antibody for an additional 30 minutes at 4°C washed and analyzed by flow cytometry.

Immunoglobulin gene rearrangement analysis

DNA was isolated from sorted large pre-B cells and analyzed for immunoglobulin heavy and light chain gene rearrangements by PCR approach using primers specific for D-J and V-DJ or V-J rearrangements as described before[52, 53]. Briefly, For D-J rearrangement, the DH sense primer was used with a JH3 antisense primer. V- DJ rearrangements were evaluated using a mixture of three different degenerate (at three positions) oligonucleotides homologous to sequences in the conserved framework region 3 (FR3) of the indicated VH gene families and the JH3 antisense primer. For PCR, serial dilutions (1x and 1:3) of the samples were heated to 94°C for 5 min and then subjected to amplification for 35 cycles of 1 min at 94°C, 1 min at 60°C, and 1 min 30 sec at 72°C. After the last cycle, a final extension step at 72°C for 10 min was carried out. PCR products were run on 1% agarose gels, transferred and hybridized with probe upstream of the JH3 primer region. For analysis of the Ig V-J rearrangement, PCR amplification was performed with Vκ sense primer mixture that is degenerate at four positions and Jκ5 antisense oligo. Southern hybridization of PCR-products for V-J rearrangement was performed with probes binding upstream of the Jκ5 region.

Stromal- free cultures

Differentiation in stromal-free cultures was performed as previously described[54]. Briefly, 2×103 sorted large pre-B cells were plated in opti-MEM media (Gibco) for 4 days (d4) supplemented with 10% FCS, 50 μM 2-ME, 2.4g/L NaHCO3, 100 μg/ml penicillin, 100 μg/ml streptomycin and 5ng/ml of IL-7 (Peprotech). At day 4, cells were washed and re-plated in opti-MEM with 2% FCS and 0.05ng/ml IL-7. After 3 days, cells were harvested, counted and stained for cell surface expression of CD19, BP1, IgM and CD2. For addressing survival and proliferation, large pre-B cells were cultured without stroma for 1-3 days in 0.05-5ng/ml of IL-7. Cells were harvested and analyzed for counts, cell cycle and apoptosis.

Stromal cultures

WT and IkE5Δ/Δsorted large pre-B cells were co-cultured with OP9 stroma in DMEM media (Sigma, D-5671) supplemented with 10% FBS (Sigma, 2442), 50μM, 100 μg/ml penicillin, 100 μg/ml streptomycin, 1X Glutamax (Gibco 35050-062), 10mM HEPES (Gibco, #156-30-80) and 1X Sodium Pyruvate (Gibco, 11360-070) in the presence of indicated amounts of IL-7 as previously described[51]. Equal number of WT and IkE5Δ/Δ pre-B cells plated in presence of IL-7 were harvested for counts, cell cycle, proliferation and apoptosis at indicated time points. All analysis on cultured large pre-B cells was performed after removal of the OP9–GFP by flow cytometry. Exclusion by electronic gating based on size and granularity was performed.

For calculation of adherent/non-adherent ratios, 5×104 cells were plated on stroma in 0.05-5ng/ml of IL-7 for 1-3 days. The non-adherent cells were harvested followed by PBS wash. The adherent cells were detached with limited trypsinization treatment. Cells from each fraction were counted under a bright-field microscope and ratios were calculated. For re-adhesion assay, equal number of adherent WT and IkE5Δ/Δ cells were allowed to re-attach onto stroma and at 3 hours cells were enumerated as described above.

Limiting dilution analysis

Adherent WT and IkE5Δ/Δadherent large pre-B cells were sorted on stroma in a 96 well plate in step-wise three-fold serial limiting dilution (10 replicates per dilution) from 300 to 1 cells with 0-5ng/ml of IL-7. Colonies were scored visually after 6 days. The mean frequency of colony forming cells was calculated by L-Calc software (Stem Cell Technologies) based on Poisson distribution of the probability of wells scoring positive.

Intracellular calcium and flux measurements

For measurement of intracellular calcium, non-adherent and adherent WT and IkE5Δ/Δ pre-B cells were stained with Fura-red (Life technologies) as per manufacturer's protocol. For calcium flux, cells were harvested into staining buffer that contained 25mM Hepes (pH 7.2), 125 mM NaCl, 5mMKCl, 1mM Na2HPO4, 0.1% glucose and 0.5mM MgCl2, 1 mMCaCl2 and 0.1g BSA just prior to use. Calcium green (Life technologies) and Fura-red were added for 30 minutes at 37°C. Cells were washed twice and re-suspended in staining buffer and placed on ice. Just prior to analysis on FACSCanto™ (BD), the cells were equilibrated to 37°C. Data was acquired for 30 seconds and then pulsed with anti-IgM antibody or ionomycin and acquired for additional indicated time points. Data was analyzed using kinetics platform on FlowJo software (Tree Star).

Apoptosis assay

Cells were stained for apoptosis using Annexin V: Apoptosis detection kit I (BD) according to manufacturer's protocol.

Cell cycle analysis

Cells were harvested at the indicated time points and fixed in 70% cold ethanol overnight at 4°C. Fixed cells were stained with propidium iodide (PI) staining buffer (250 μg/ml RNaseA, 50 μg/ml PI) for 30 min at 37°C and the DNA content was detected by FACS canto. The resulting files were analyzed with FlowJo (Tree Star).

BrdU pulse-chase assay

Cells were labeled for 45min with BrdU washed and then incubated in growth media for up to 48hrs. Cells were harvested at the indicated time points for proliferation analysis using the BrdU flow Kit (BD) per the manufacturer's protocol.

Phase Contrast Microscopy

Phase-contrast pictures of large-preB cells were taken with a Zeiss Axiovert 200M microscope. Prior to microscopy, cells were cultured on OP9-GFP for 24 hours in 5ng/ml of IL-7.

Immuno-blotting

Cells were harvested and whole cell extracts were prepared using RIPA buffer containing 10mM Tris-HCl (pH 8.0), 1mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS and140 mM NaCl. Protease and phosphatase inhibitors (Roche) were added to extraction buffer just prior to use. Equal amounts of protein lysates were separated by SDS-PAGE and transferred to PVDF membranes (Millipore) and probed with indicated antibodies as per manufacturer's protocol.

Immunofluorescence

Cells grown on stroma on Lab-Tek® Chamber Slide (Electron Microscopy Sciences) were fixed with 3% paraformaldehyde for 20 min at room temperature and then permeabilized with phosphate-buffered saline (PBS) plus 0.1% Tween-20 for 45 min. After blocking with PBS containing 2% bovine serum albumin (BSA) for 10 min, the cells were stained with anti-phospho FAK antibody for 45 min, followed by TRITC-conjugated secondary antibody for 30 min. The slides were treated with Vectashield containing DAPI (Vector Laboratories) and mounted. Images were collected using Nikon A1SiR Confocal Microscope and processed using NIS element confocal imaging software.

Adhesion to integrin ligands and in vitro FAK inhibition assay

Adhesion assay were performed in non-TC treated plates (BD). The plates were coated with 10ug/ml fibronectin (FN), collagen (Col.) (Invitrogen) or BSA alone overnight at 4°C. After blocking the plates with 2% BSA for 1 h, equal number of cells were plated and incubated for the indicated time points. At the end of the assay, unbound and bound cells were harvested and enumerated. Percent adhesion was calculated by taking ratio of bound cells over total cells used in assay.

For inhibition assay, 1 μm FAK inhibitor (PF-431396 or PF- 562271, Sigma) or DMSO control was used for treatment and cells were harvested as bound and unbound fractions at 4 or 24 hour post-treatment for calculation of percent adhesion and inhibition of adhesion, apoptosis and cell cycle assays. For peptide mediated blocking of adhesion[55], equal number of cells were pretreated for 45 minutes with either 400 μg/ml G-R-G-D-S-P or the control peptide G-R-G-E-S (American Peptide Company) and plated on FN-coated TC dishes. Cells were enumerated for calculation of percent adhesion and percent inhibition.

The effect of growth factors and adhesion was evaluated by plating 2×105adherent pre-B cells on FN and Col. or BSA coated plates in the presence of 5 ng/ml IL-7, 100 ng/ml SCF, both (IL-7 + SCF) or no cytokines for 24 hours. At 24 hours, cells were counted and analyzed for cell cycle and apoptosis. For integrin and growth factor signaling assay, equal number (2×105) cells were plated in FN and Col. or BSA coated plates in presence of 5ng/ml IL-7, 100 ng/ml SCF, both (IL-7 + SCF) or no cytokines for 24 hours. At 24 hours, cells were harvested and counted. Mutant cells were analyzed for cell cycle and apoptosis.

In vivo FAK inhibition assay

WT and IkE5fl/fl CD19cre mice were used for treatment with FAK inhibitor PF-562271 or vehicle (50% DMSO/50%PEG-400). Both cohorts were given a dosage of 25mg/kg/mouse of inhibitor or equal volume of vehicle by oral gavage. Dosage regimens of either 3 or 5 doses were given at ~12 hours apart. At 3 hours after the final dose, mice were sacrificed by CO2 asphyxiation. Bones were flushed and cells were collected and total bone marrow cellularities were estimated. Cells were stained for pre-B cell cell surface markers and PI/Annexin V staining was performed to estimate apoptotic cell frequency.

Transwell migration assay

Transwell plates with (Corning, 3422) were coated with fibronectin (10μg/ml). The wells were washed and blocked with BSA. Serum-free medium with 1% BSA containing the CXCL12 / SDF-1α, (100 ng/ml) was added to the bottom well of the transwell plate. Equal number of cells in 100 μl of serum-free media were added to the upper chambers and incubated for 2 hours at 37°C. At the end of the time point, inserts were removed and migrated cells in bottom wells were counted. Percent migration was calculated by taking ratios of migrated cells over total cells plated in inserts.

RNA-seq, gene expression and pathway analysis

RNA isolated with Trizol was purified using the PureLink RNA mini kit (Ambion). The Truseq RNA sample prep kit was used for construction of cDNA libraries for RNA-sequencing (Illumina). The cDNA libraries were ligated with indexed primers and amplified by 15 cycles of PCR. The amplified libraries were multiplexed and sequenced by the Genome Analyzer at Systems Biology Lab, Harvard University. Read alignment on mouse mm9 assembly was conducted by the BWA algorithm implemented by the DNA Nexus suite. The Deseq algorithm implemented by the R platform was used to determine differential gene expression in freshly sorted WT and IkE5Δ/Δ large pre-B, WT small pre-B as well as in adherent and non-adherent fractions of sorted large pre-B cells after limited propagation on OP9 stroma. Pathway analysis of upregulated genes in IkE5Δ/Δ relative to WT large pre-B was conducted using the Ingenuity software. Heatmaps of normalized tags for gene subsets across WT and IkE5Δ/Δpre-B cell populations were generated with the Avadis software.

Adoptive transfer of purified pre-B cell populations to NSG mice

NOD/SCID/Il2rg−/− (NSG) mice (Jackson Laboratory) were conditioned by 300 cGy gamma irradiation and injected via lateral tail vein with 3× 106 sorted large pre-B (CD19+CD43+BP1+) cells. Diseased mice were characterized by histopathological analysis as previously described[56].