Inhibition of inflammatory signaling in Pax5 mutant cells mitigates B-cell leukemogenesis

PAX5 is one of the most frequently mutated genes in B-cell acute lymphoblastic leukemia (B-ALL), and children with inherited preleukemic PAX5 mutations are at a higher risk of developing the disease. Abnormal profiles of inflammatory markers have been detected in neonatal blood spot samples of children who later developed B-ALL. However, how inflammatory signals contribute to B-ALL development is unclear. Here, we demonstrate that Pax5 heterozygosis, in the presence of infections, results in the enhanced production of the inflammatory cytokine interleukin-6 (IL-6), which appears to act in an autocrine fashion to promote leukemia growth. Furthermore, in vivo genetic downregulation of IL-6 in these Pax5 heterozygous mice retards B-cell leukemogenesis, and in vivo pharmacologic inhibition of IL-6 with a neutralizing antibody in Pax5 mutant mice with B-ALL clears leukemic cells. Additionally, this novel IL–6 signaling paradigm identified in mice was also substantiated in humans. Altogether, our studies establish aberrant IL6 expression caused by Pax5 loss as a hallmark of Pax5-dependent B-ALL and the IL6 as a therapeutic vulnerability for B-ALL characterized by PAX5 loss.

Besides these somatically acquired PAX5 mutations, it has been shown that inherited hypomorphic variants of the gene predispose the carriers to the development of familiar B-ALL 7,8 . Studies both in affected human carriers and in animals carrying an heterozygous null mutation of Pax5 (Pax5 +/mice, henceforth also called "Pax5 mutant mice") have shown that Pax5 mutant mice tend to accumulate an expanded, aberrant, vulnerable, population of B cell progenitors prone to malignant transformation through the accumulation of secondary mutations in the presence of a selective pressure [7][8][9] . These accumulated cells are commonly defined as preleukemic cells 10 , and the selection and expansion of these preleukemic-B clones precede the development of B-ALL, both in the case of carriers of congenital PAX5 mutations, and also in almost all other B-ALL-predisposing somatically arising mutations studied to date [10][11][12] . Preleukemic cells remain latent until they progress to the development of full-blown B-ALL through acquisition of additional somatic mutations over time 1 . We have recently shown that natural exposure to infectious pathogens contribute to the ''switch'' from a preleukemic state to a leukemic state in cells bearing these PAX5 mutations 9,13 , and Pax5 mutant mice go on to develop B-ALL with modest penetration when they are exposed to natural infections 9 , usually associated with the inactivation of the other copy of Pax5. In this context, inflammation has been hypothesized to play an essential role [14][15][16] , but precisely how inflammatory signals influence the Pax5 mutant B-cell leukemogenesis process is poorly understood. Human populations with different ancestries have been naturally selected to present many allelic differences in the genes involved in immune system development and immune response 17 . All these variants allow specific responses (stronger or weaker) against certain types of infections 17 , but can also increase the chances of developing certain diseases; in the case of B-ALL, this can be caused by an imbalance in the immune system that helps disease progression. Given the limited genetic variation present in experimental mice, the results from these models suggest that the contribution of the rest of the individual genetic variations beyond the preleukemia-initiating one does not seem to play a major role in the transition from the preleukemic phase to B-ALL, a fact further supported by the possibility of triggering B-ALL conversion ex vivo with TLR ligands [18][19][20] . Considering that Pax5-mutant preleukemic cells give rise to infection-triggered B-ALL in both human and mice [7][8][9] , the relationship between inflammation and B-cell leukemogenesis is likely to be B-cell-dependent. Because preleukemic precursor B cells reside in the bone marrow, an inflammatory microenvironment can influence the growth of these cells in part by producing pro-inflammatory cytokines. In the present study, we asked whether the inflammatory signals contribute to B-ALL development triggered by environmental infection exposure as a result of Pax5-inherited susceptibility. The results from our experiments identify IL-6 as a key cytokine whose expression and secretion by mouse leukemic B-cells is induced when Pax5 is lost during the course of B-ALL development, and whose inhibition therapeutically targets leukemic cells.

RESULTS
Pax5 +/mice developing B-ALL show enhanced expression of the proinflammatory cytokine IL-6. To determine whether the transformation of preleukemic B cells to full-blown B-ALL is in part due to dysregulated expression of inflammatory cytokines in Pax5 +/mice, we measured concentrations of 7 inflammatory cytokines (IL-2, IL-4, IL-6, IL-10, IL-17a, TNF and IFNγ) in the serum of Pax5 +/mice exposed to an infectious environment which developed B-ALL, exposed Pax5 +/mice without B-ALL, and age-matched control wild-type mice. We found that leukemic Pax5 +/mice had abnormal concentrations of IL-6 ( Fig. 1a and Supplementary Fig. 1). The emergence of this increase in IL-6 levels could further be linked to disease onset, since IL-6 in serum samples taken at routine intervals confirmed lack of IL-6 increase in the exposed Pax5 +/mice that never developed B-ALL ( Fig. 1a-b, Supplementary Table S1 and Supplementary Fig. 1). However, IL-6 increase was not detectable in serum samples taken at routine intervals in healthy (non-leukemic) Pax5 +/mice that later developed B-ALL, prior to the first phenotypic signs of illness ( Fig. 1b and Supplementary Fig. 2). In addition, this increase in IL-6 was not observed in mouse models where the appearance of B-ALL is triggered by infection exposure but is not linked to a congenital Pax5 alteration, like in Sca1-BCR-ABLp190 and Sca1-ETV6-RUNX1 mice where the second hit does not involve Pax5 inactivation (Fig. 1c) 21 . However, elevated levels of IL-6 protein could be measured at the time of B-ALL diagnosis in the serum of Pax5 +/-;Sca1-ETV6-RUNX1 leukemic mice and Pax5 +/-;Sca1-BCR-ABLp190 leukemic mice 22 (Fig. 1, Supplementary Table S1). These elevated levels of IL-6 protein could be recapitulated at the time of diagnosis in the serum of human B-ALL carrying PAX5 alterations (and lacking known JAK/STAT mutations) compared to Healthy Donors (HD) (Fig. 1d, Supplementary  Table S2). Thus, induction of a leukemogenic state by Pax5-loss is associated with an increase in IL-6 secretion in both human patients and mouse models, suggesting the IL-6 secretion may correspondingly be the product of oncogenic Pax5 inactivation.
Pax5 controls IL-6 expression in B-cells. We investigated the mechanisms controlling IL-6 production using leukemic proB cells and both control wild-type and Pax5 +/-proB-cells. Leukemic proB cells, lacking Pax5 activity due to secondary mutations of the WT Pax5 allele 9 , displayed high levels of IL-6 mRNA cells (Fig. 2a), which was not significant in Pax5 +/+ leukemic cells and it was not detectable neither in control wild-type proBcells nor in control Pax5 +/-proB-cells (Fig. 2a). However, high levels of IL-6 mRNA are present in healthy Pax5 -/precursor B cells ( Supplementary Fig. 3). A microarray analysis of gene expression confirmed enrichment of the IL-6 signaling pathway geneset only in leukemic proB cells ( Fig. 2b and Supplementary Fig. 4A-B) as well as an enrichment in inflammatory response and apoptosis gene sets ( Supplementary Fig. 4C-D). We then investigated leukemic proB cells lacking Pax5 activity for expression of a panel of effectors genes known to regulate IL-6 expression, and we found a significant downregulation of both Blnk (which is a Pax5 direct target 23 ) and Bcl6 expression ( Fig. 2c-d). BCL6 is a direct transcriptional repressor of the IL-6 gene 24 and recent work has shown that STAT5 activation inhibits BCL6 expression 25 . Therefore, Pax5-deficient leukemic proB cells had decreased Bcl6 levels and this, together with the STAT5 activation, could contribute to the upregulation of IL-6 expression Scientific Reports | (2020) 10:19189 | https://doi.org/10.1038/s41598-020-76206-y www.nature.com/scientificreports/ observed in Pax5-deficient leukemic proB cells. These results indicate that Pax5 activity controls IL-6 expression in proB cells, and suggest that both Bcl6 and STAT5 activation are involved in the complex molecular network mediating this effect. Taken together, these data suggest that IL-6 might be important for Pax5-deficient B-ALL development. Thus, we next examined whether IL-6 plays any role in Pax5-mediated B-cell leukemogenesis.
Impairment of IL-6 signaling in Pax5 +/mice delays natural infection-driven B-ALL development. Given the observed upregulation of IL-6 in Pax5-deficient leukemic proB cells and mice, we next directly tested the requirement for IL-6 in Pax5-loss mediated leukemia growth in a system that recapitulates the   www.nature.com/scientificreports/ spontaneous process of leukemogenesis. To functionally demonstrate the role of IL-6, we impaired the expression of IL-6 by breeding Pax5 +/mice to IL-6 +/− mice, which have significantly lower IL-6 serum concentrations ( Fig. 1a), with the aim of testing whether IL-6 +/-/Pax5 +/mice are more resistant to infection-induced B-ALL than Pax5 +/mice. Accordingly, control IL-6 +/+ /Pax5 +/and experimental IL-6 +/-/Pax5 +/mice were exposed to natural infections, and B-ALL development was monitored as previously described 9 . The results showed that the reduction on IL-6 levels significantly delayed the emergency of leukemias in Pax5 +/mice upon exposure to natural infections (Fig. 3). B-ALL appeared between 6 and 16 months of age in IL-6 +/+ /Pax5 +/mice (mean = 11.29 months) and, at the end of the 24 months' experimental period, 22% of the mice had developed B-ALL ( Fig. 3a-b), in line with previously reported results. In sharp contrast, B-ALL was significantly delayed in the IL-6 +/-/Pax5 +/mice, appearing between 18 and 22 months of age (mean = 20.13 months) ( Fig. 3a-b), although by the termination of the experiment a similar percentage (20%) of the mice had developed B-ALL and IL-6 was elevated in their serum similarly to IL-6 +/+ /Pax5 +/leukemias (Fig. 3c). These IL-6 +/-/Pax5 +/-B-ALLs are histologically, phenotypically and genetically similar to IL-6 +/+ /Pax5 +/-B-ALLs, and FACS analyses revealed a CD19 +/-B220 + IgM -cKit +/-CD25 +/cell surface phenotype for tumor cells that extended through bone marrow (BM), peripheral blood (PB), spleen and lymph nodes ( Supplementary Fig. 5) and infiltrated non-lymphoid tissues like liver and intestine ( Supplementary Fig. 6). All IL-6 +/-/Pax5 +/-B-ALLs displayed clonal immature BCR rearrangements ( Supplementary Fig. 7). We then  Table S4). We next aimed to confirm that the specific significant delay in B-ALL development due to IL-6 +/heterozygosity does not modify the B-cell susceptibility and genetic characteristics of B-ALL in Pax5 +/mice; similar to IL-6 +/+ /Pax5 +/mice, preleukemic IL-6 +/-/Pax5 +/littermates presented a significantly reduced amount of total B-cells in the PB when compared to Pax5 +/+ (WT) littermates of the same breeding (Supplementary Fig. 9), but this PB B-cell decrease was similar to the one observed in IL-6 +/+ /Pax5 +/mice (Supplementary Fig. 9). In order to further identify somatically acquired 2nd hits leading to leukemia development, we next performed whole exome sequencing of 4 IL-6 +/-/Pax5 +/-B-ALLs and corresponding germline on a HiSeq 2500 (Illumina) platform. IL-6 +/-/Pax5 +/tumor DNA was derived from whole leukemic BM or lymph nodes, while tail DNA of the respective mouse was used as reference germline material. Similar to IL-6 +/+ /Pax5 +/leukemias 9 , IL-6 +/-/Pax5 +/tumors showed recurrent mutations affecting Pax5, Ss18, Jak1, and Jak3 (Fig. 4). Taken together, these data demonstrate that the decrease of IL-6 delays spontaneous formation of infection-driven B-ALL in Pax5 +/mice. Collectively, our data suggest that knocking down amplified IL6 levels represents a formidable barrier to Pax5-dependent leukemogenesis, a finding of clear clinical relevance.

Pax5 mutant B-ALL is not sensitive to IL-6 inhibition in transplant-based mouse models.
To test the hypothesis that IL-6 inhibition may represent a preferential target for Pax5-dependent B-ALLs, and to expand our studies to more clinically relevant settings, we first tested whether IL-6 contributes to Pax5-dependent B-ALL maintenance, since IL-6 is a secreted protein, and as such it is amenable to targeting by neutralizing antibodies 26 and, moreover, such an antibody treatment has already been shown to be tolerated in humans 27 .
In the past, the efficiency of experimental therapies against B-ALL has been routinely measured by transplanting leukemic B cells into recipient mice that had been preconditioned by irradiation 28,29 . However, a potential limitation of this approach to studies focused on how inflammatory signaling affects B-ALL, resides in the fact that irradiation triggers the production of pro-inflammatory cytokines such as IL-1, IL-6 and TNF 30,31 . Nevertheless, to be consistent with previous studies in the literature, we first tested the efficacy of the neutralizing IL-6 antibody in mice infused with leukemic Pax5 +/− proB cells harboring the Jak3 V670A mutation 9 ( Supplementary  Fig. 10A). These leukemic proB cells were isolated from the BM of diseased mice, as B220 + , by MACS-sorting, then cultured in medium containing IL-7 and, later, propagated in IL-7-independent culture conditions 9 . These leukemic Pax5 +/− pro-B cells harboring Jak3 V670A cells were injected into syngenic mice (n = 10), and the animals were randomized to treatment at day 14 when the disease was confirmed by the presence of blast cells in the PB ( Supplementary Fig. 10). Mice were treated with two doses of the anti-IL-6 antibody at 10 mg/kg, 3 days apart. FACS analysis of the PB was used to verify disease remission during therapy. Anti-IL-6 antibody therapy provided no significant survival advantage and none of all mice treated (n = 5) showed a decrease in the percentage of blast cells ( Supplementary Fig. 10B), even though the anti-IL-6 treatment significantly reduced the levels of IL-6 in the serum of treated mice (Supplementary Fig. 10C and Supplementary Table S5). However, an important caveat to take into account when interpreting this result is the fact that we found that the standard culture conditions with IL-7 used to isolate leukemic proB cells from mice abolished the IL-6 production both in vitro (Supplementary Fig. 11 and Supplementary Table S5) and in vivo (Supplementary Fig. 10C and Supplementary  Table S5). These findings are in agreement with previous observations showing that cytokine-mediated survival signals do not compete between them 32 . These observations explain why the use of anti-IL-6 antibody therapy was not able of modifying the course of the disease, and underscore the limitations of the use of ex vivo functional studies involving leukemic pro-B cell in vitro expansion and bone marrow transplantation to identify how inflammatory signaling affects B-ALL development. Therefore, as previously suggested 33 , transplant-based mouse models need to be viewed with particular caution when trying to dissect the role of cytokine factors during oncogenic transformation.   Fig. 12A). The IgG controltreatment did not modify the course of the disease with independence of the percentage of blasts at diagnosis. However, we found that the anti-IL-6 antibodies abolished the increase in IL-6 serum levels characteristic of Pax5-dependent B-ALL (Fig. 5a). Blockade of IL-6 in vivo reduced disease progression in 100% of the anti-IL-6-treated mice when the percentage of blast cells in PB was lower than 40% at the time of treatment (Fig. 5b-c and Supplementary Fig. 12B-D). However, disease progression could not be modified by IL-6 inhibition when www.nature.com/scientificreports/ the percentage of blast cells in PB was higher than 70% at the time of treatment (Fig. 5b-c and Supplementary  Fig. 12E-G). In order to further clarify if the relapse (the reoccurrence of disease in the responders) after IL-6 inhibition is driven by pre-existing leukemic cells or by the selection of IL6-resistant clones through the acquisition of new mutations, we next performed whole exome sequencing of the 6 IL-6 +/+ /Pax5 +/tumors before anti-IL-6-treatment and after relapse. Tumor DNA was derived from whole leukemic PB, while tail DNA of the respective mouse was used as reference germline material. Leukemias at relapse showed similar recurrent mutations than leukemias before anti-IL-6-treatment (Supplementary Fig. 13 and Supplementary Table S6). Thus, these data further suggest that IL-6 retains driver functions in established leukemia and demonstrate the application of this in vivo native assay to identify the importance of players relevant for B-ALL development.
These results show that IL-6-neutralizing antibodies may be useful therapeutic options in the treatment of Pax5dependent leukemias.

Discussion
Proleukemic inflammatory environment in Pax5-mutant B-cell leukemogenesis. Our results uncover an essential mechanism involving the proinflammatory cytokine IL-6 in sustaining Pax5-dependent B-ALL development. Abnormal profiles of inflammatory markers, including higher concentrations of IL-6, IL-17, and IL-18, have already been detected in neonatal blood spot samples of children who later developed B-cell precursor ALL 34 , and in vitro studies have also shown that pro-inflammatory cytokines like IL-6/IL-1β/ TNFα 35 or TGF-β dependent signaling 36,37 , can predispose pre-leukemic B cells to malignant transformation. However, the real contribution of these inflammatory pathways to B-ALL development remains unclear. In this study, we examined the impact of inflammation on the conversion of Pax5 +/preleukemic cells. PAX5 is frequently mutated in a large percentage of B-ALL cases 3-6 , but these mutations can also be present in normal children who will never develop B-ALL 7,8 , very much like the ETV6-RUNX1 translocation only causes B-ALL in a small percentage of carriers 38 . It is known that infection exposure pushes the progression from preleukemic state to full-blown B-ALL in a significant fraction of Pax5 +/or ETV6-RUNX1 + cases 9,13,21 . Here, we have examined how the inflammatory profile changes the behavior of preleukemic cells in a Pax5 +/mouse model. We identify IL-6 as a key cytokine whose expression and secretion by mouse leukemic B-cells is induced when Pax5 is lost during the course of B-ALL development. Serum IL-6 levels and the expression of IL-6 are significantly upregulated in both Pax5-deficient leukemic and healthy B-cells. The presence of IL-6 was further confirmed in serum of B-ALL patients where PAX5 function is lost. Mechanistically, we show unambiguously by a genetic approach in the Pax5 +/mouse that impairment of IL-6 signaling delays B-ALL development, hence confirming the essential contribution of this pathway as a feedback loop supporting B-ALL development. An important aspect that should be taken into consideration is that, although our data show that Pax5 indeed regulates IL6 expression, it cannot be rouled out that the microenvironment, following infections and/or leukemia, might also contribute to the observed IL6 increase. Nevertheless, our findings, taken together, demonstrate that Pax5dependent B-ALLs are profoundly affected by the proleukemic inflammatory environment in which leukemic progenitor cells reside, and also support the view that increased levels of the pro-inflammatory cytokine IL-6 are an essential trigger of the B-ALL disease observed in Pax5 +/mice.  45 , and pancreatic cancer 46 . Likewise, it has been suggested that increased autocrine IL-6 expression in PAX5 − mantle cell lymphoma (MCL) cells may contribute to the reduction in TP53 gene expression and could provide survival advantages to lymphoma cells 47 . In this study, we identify IL-6 as a major player in Pax5-dependent B-ALL pathogenesis, and we demonstrate that Pax5 activity dictates the level of IL-6 produced by both mouse and human leukemic B-cells, regulating IL-6 mRNA levels through BCL6 and the STAT5 pathway, which are two well known transcriptional regulators of the IL-6 gene 24,48 . Furthermore, we find that disruption of the IL-6 loop through genetic downregulation of the IL-6 gene in Pax5 +/mice significantly delays B-ALL onset, and that blocking IL-6 signaling kills Pax5-dependent B-ALL in unmanipulated animals. All these results show that the IL-6 signaling pathway represents a therapeutic vulnerability in Pax5-dependent B-ALL, and that its targeting could be a promising therapy for this disease, which could also be extended to other hematological diseases, taking into account the high frequency of somatic PAX5 losses-offunction in different types of B cell leukemias.

Material and methods
Mouse model for natural infection-driven leukemia. Pax5 +/mice 49 were crossed with IL-6 -/mice 50 to generate Pax5 +/-/IL-6 +/mice. These Pax5 +/-/ IL-6 +/and Pax5 +/mice were bred and maintained in the SPF area of the animal house until the moment when they were relocated to an environment where natural infectious agents were present, as previously described 9 . Animal studies were performed in accordance with relevant guidelines and regulations and approved by the pertinent institutional committees of both University of Salamanca and Spanish Research Council (CSIC). IL-6 +/-, IL-6 -/-, Pax5 +/-/IL-6 +/-, and Pax5 +/mice of a mixed C57BL/6 × CBA background were used in this study, with approximately equal representation of both males and females. For the experiments, IL-6 +/-, IL-6 -/-, Pax5 +/-/IL-6 +/-, and Pax5 +/of the same litter were used. When the animals showed evidences of illness, they were humanely killed, and the main organs were extracted by standard dissection. All major organs were macroscopically inspected under the stereo microscope, and then representative samples of tissue were cut from the freshly dissected organs, and were immediately fixed. Differences in Kaplan-Meier survival plots of transgenic and WT mice were analyzed using the log-rank (Mantel-Cox) test.
Flow cytometry and cell sorting. Flow cytometry and cell sorting was carried out as previously described 9,13 . Total mouse BM cells were obtained by washing the long bones with PBS with 1% FCS, using a 27-G needle. Cells were as well collected from peripheral blood, and also from the thymus and spleen after disrupting these organs by passing them through a 70-μm cell strainer. Erythrocytes were osmotically lysed using RCLB buffer, and nucleated cells were then washed with PBS-1% FCS. Cells were stained with the appropriate antibodies against the indicated cellular markers for 20 min at 4 °C, washed once with PBS-1% FCS, and finally they were resuspended in PBS-1% FCS with 10 μg/mL propidium iodide (PI) to exclude dead cells during data acquisition; this was performed in an AccuriC6 Flow Cytometer, and data files were analyzed using Flowjo software. For this analysis, gates were set by employing the commonly used forward and perpendicular lightscattering properties of mouse hematopoietic cells, and the specific fluorescence of the staining dyes used [FITC, PE, PI, and APC excited at 488 nm (0.4 W) and 633 nm (30 mW), respectively]; an example of such a gating strategy is shown in Supplementary Fig. 14

Real-time PCR quantification (Q-PCR) of mIL-6, mBcl6 and mBlnk. Real-time PCR quantification
(Q-PCR) was carried out as previously described 9,13 . We analyzed the expression of mIL-6, mBcl6 and mBlnk in leukemic Pax5 +/cells as well as in healthy Pax5 +/and WT proB cells by Q-PCR as follows: cDNA was synthesized using reverse transcriptase (Access RT-PCR System; Promega, Madison, WI) and genomic DNA was Histology. Histology was carried out as previously described 9,13 . Animals were sacrificed by cervical dislocation; tissue samples were formalin-fixed and included in paraffin. Pathology assessment was performed on hematoxylin-eosin stained sections under the supervision of Dr. Oscar Blanco, an expert pathologist at the Salamanca University Hospital.
Quantification of cytokine levels in serum. Quantification of cytokine levels in serum was carried out as previously described 41 . Serum cytokine levels were analysed using The Cytometric Bead Array immunoassay system (CBA) (BD Biosciences) which assesses simultaneously IL-2, IL-4, IL-6, IL-10, IL-17A, TNF alpha and IFN gamma in serum from the mice (Mouse Th1 Th2 Th17 Cytokine Kit #560,485; BDB); and serum from the patients and controls (Human Th1 Th2 Th17 Cytokine Kit #560,484; BDB). Human studies were performed in accordance with relevant guidelines and regulations and approved by the Comisión de donación de muestras del Institute Josep Carreras (IJC). Serum was obtained from children and adults and all participants or their legal guardians provided written informed consent to take part in the study. Data acquisition was performed on a FACSCanto II flow cytometer (BDB) using the FACSDiva™ software program (BDB). For the evaluation of cytokine serum levels or cytokine secretion into the culture supernatants, 50 µl of serum was collected. Briefly, 50 µl of the serum was incubated at room temperature for 2 h at room temperature (RT) with 50 µl of anticytokine MAb-coated beads and with 50 µl of the appropriate phycoerythrin (PE)-conjugated anticytokine antibody detector. After this incubation period, samples were washed once (5 min at 200 g) in order to remove the Scientific Reports | (2020) 10:19189 | https://doi.org/10.1038/s41598-020-76206-y www.nature.com/scientificreports/ excess of detector antibodies. Immediately afterwards, data acquisition was performed on a FACSCanto II flow cytometer (BDB) using the FACSDiva™ software program (BDB). During acquisition, information was stored for 3,000 events corresponding to each bead population analysed per sample (total number of beads > 9,000). For data analysis, FCAP Array Software v3.0 program (BDB) was used.
Anti-IL-6R treatment. Anti-IL-6R antibody (Tocilizumab) was obtained from Chugai Pharmaceuticals Co. Ltd. (Shizuoka, Japan). Anti-IL-6R antibody was intraperitoneally administered at 10 mg/kg twice a week. Treatments were started after the establishment of B-ALL.
Transplantation. Transplantation was carried out as previously described 9,13 . IL-7-independent leukemic pro-B cells were intravenously injected into 12-week-old male syngenic mice (C57BL/6 × CBA) that had previously been sublethally irradiated (4 Gy). Leukemia development in the injected mice was followed by regular analysis of peripheral blood, until the moment when leukemic blasts were detected in the blood; at this point, animals were treated with anti-IL-6R antibody.
V(D)J recombination. V(D)J recombination analysis was carried out as previously described 9,13 . Immunoglobulin rearrangements were amplified by PCR using the primers below. Cycling conditions consisted of an initial heat-activation at 95 °C followed by 31-37 cycles of denaturation for 1 min at 95 °C, annealing for 1 min at 65 °C, and elongation for 1 min 45 s at 72 °C. This was followed by a final elongation for 10 min at 72 °C. The following primer pairs were used: Microarray data analysis. Microarray data analysis was carried out as previously described 9,13 . The total RNA was first isolated using TRIzol (Life Technologies), and then it was subjected to purification with the RNeasy Mini Kit (Qiagen) using also the On-Column DNase treatment option. Quality and quantification of RNA samples were determined by electrophoresis. Determination of the expression of the different genes in the RNA samples was performed using Affymetrix Mouse Gene 1.0 ST arrays. All bioinformatic analyses of the array data were performed using R 51 and Bioconductor 52 . First, we applied background correction, intra-and inter-microarray normalization, and expression signal calculation using the microarray analysis algorithm [53][54][55] , in order to determine the absolute expression signal for each gene in each array. Then, we used the significance analysis of microarray (SAM) 56 method to identify the gene probe sets with differential expression between experimental and control samples, SAM uses a permutation algorithm to allow to statistically infer the significance of the differential expression, and it provides P-values adjusted to correct for the multiple testing problem, by using FDR 57 . An FDR cutoff of < 0.05 was used as a threshold to determine differential expression. The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus (GEO) 58 and are accessible through GEO Series accession number GSE154589. Enrichment analysis. Enrichment analysis was carried out as previously described 9,13 . In order to identify potential signatures of gene expression associated with different biological processes, gene set enrichment analysis (GSEA) was performed using the MSigDB databases from the Broad Institute (GSEAv2.2.2) 59 and hallmark collection of gene sets 60,61 . Mouse exome library preparation and NGS. Mouse exome library preparation and NGS was carried out as previously described 9,13 . DNA was purified from samples using the AllPrep DNA/RNA Mini Kit (Qiagen) according to the manufacturer's instructions. The exome library was prepared using the Agilent SureSelectXT Scientific Reports | (2020) 10:19189 | https://doi.org/10.1038/s41598-020-76206-y www.nature.com/scientificreports/