To the editor

Leukemia usually requires multiple genetic events, as exemplified in ETV6-RUNX1 B-cell precursor acute lymphoblastic leukemia (BCP-ALL), one of the most frequent pediatric BCP-ALL [1]. The translocation t(12;21)(p13;q22) resulting in ETV6-RUNX1 fusion gene arises predominantly in utero and produces a preleukemic clone. Additional mutations occur years after the translocation t(12;21) and give rise ultimately to leukemia [1]. Those additional genetic alterations observed in ETV6-RUNX1 BCP-ALL are predominantly caused by illegitimate genomic rearrangements mediated by aberrant RAG recombinase activity [2].

The RAG recombinase consists of two subunits, RAG1 and RAG2. It recognizes and cleaves DNA at recombination signal sequence (RSS), and is responsible for the V(D)J rearrangement of immunoglobulin genes during differentiation of B and T lymphoid lineages. Illegitimate off-target RAG cleavages can be pathological. High incidence of recombination events, RAG recombinase aberrant activity and high RAG1 gene expression have been repeatedly reported in ETV6-RUNX1 leukemia or equivalent mouse models [3,4,5,6,7,8]. Consistent with epidemiological findings on childhood BCP-ALL etiology [1], this aberrant RAG recombinase activity can be explained by an excessive immune response or repeated exposure to inflammatory stimuli (chronic infection) [7, 1]. However, a genetic cause of RAG increased activity related to the presence of the fusion gene ETV6-RUNX1 can also be proposed [2, 3, 7].

A proper regulation of RAG1 and RAG2 gene expression is crucial for the integrity of lymphocyte development. This regulation is complex and tightly controlled by promoters, and proximal and distal cis-regulatory elements. In B-cells, Rag1 and Rag2 genes are controlled by the strong −22kb Erag enhancer, the Irag2 enhancer, Rag2 distal and proximal enhancers (Ed and Ep), Rag1 and Rag2 promoters and Irag1 located about 15 kb upstream of the Rag1 promoter. Runx1 is described to be an essential regulator of the Rag1 promoter and Rag1-Rag2 silencer and antisilencer in mouse T-cells.

We aimed to delineate the causative link between the presence of RUNX1, the ETV6-RUNX1 fusion protein and RAG1 upregulation in ETV6-RUNX1 BCP-ALL. Our findings fulfill a missing step in the multi-hit model of ETV6-RUNX1-related leukemogenesis between the ETV6-RUNX1 fusion gene and RAG1 aberrant activity.

RAG1 and RUNX1 transcripts levels are positively and significantly correlated exclusively in ETV6-RUNX1 BCP-ALL compared to other childhood BCP-ALL (Fig. 1A, Supplementary Fig. S1A), suggesting either a common regulator for RAG1 and RUNX1 (or ETV6-RUNX1) expression, or a dependency between them.

Fig. 1: ETV6-RUNX1 and RUNX1 upregulate the expression of RAG1 mRNA and protein and bind promoter and enhancer of RAG1 gene.
figure 1

A Statistical analysis of the expression between RUNX1 and RAG1 mRNA originating from ETV6-RUNX1 BCP-ALL cells using Pearson correlation. Data of mRNA levels (expressed in Fragments Per Kilobase Million – FPKM) have been extracted from the St. Jude Children’s Research Hospital RNA-Seq Pediatric Cancer Data Portal. B Relative mRNA expression of ETV6-RUNX1, RAG1 and RUNX1 measured by RT-qPCR in Nalm6control, Nalm6ETV6-RUNX1 and Nalm6RUNX1 cells. Results are presented in-terms of a fold change in log2 scale after normalizing with ABL mRNA. Each value represents the mean ± S.D. of four independent experiments (i.e. independent stable cell lines). C Representative images of western blot showing endogenous RAG1 protein and HSC70 protein (for normalization) in Nalm6control, Nalm6ETV6-RUNX1 and Nalm6RUNX1 cells. The western blot also shows the presence of ETV6-RUNX1 and RUNX1 revealed with Halotag antibody. D ChIP-Seq profiles across the human RAG1 gene. Genomic tracks display ChIP-Seq profiles for RUNX1, ETV6-RUNX1 and the histones H3K4me3, H3K27ac and H3K4me1 from REH cells (2 replicates for RUNX1 and ETV6-RUNX1). RUNX1 ChIP-Seq (2 replicates) for Nalm6 cells and bone marrow mononuclear cells isolated from three pre-B acute lymphoblastic leukemia patients (BCP-ALL) are also displayed. ChIP-Seq data were acquired by Illumina sequencing and visualized with Integrated Genome Browser 9.0.0. ChIP-Seq reads were aligned to the reference human genome version GRCh37 (hg19). Both genomic regions of RAG1 gene that were associated with an overlap of RUNX1 and ETV6-RUNX1 peaks are indicated by boxes: one region is an enhancer at −1200 bp from TSS, and the other is a promoter at −80 bp from TSS. E ChIP-qPCR on the −1200 bp enhancer and the −80 bp promoter with IgG and Halotag antibodies in Nalm6RUNX1 and Nalm6ETV6-RUNX1 cells. Results are expressed as percentage of input (n = 3). TSS: transcription starting site.

To investigate the effect of RUNX1 and ETV6-RUNX1 fusion protein on RAG1 expression, we used human B-cell precursor Nalm6 cells (see supplemental data for detailed materials and methods, Supplementary Tables S1S2), which do not normally express the ETV6-RUNX1 fusion gene. Enforced expression of ETV6-RUNX1 or RUNX1 in Nalm6 cells (named Nalm6ETV6-RUNX1 and Nalm6RUNX1) induced high levels of endogenous RAG1 transcript and protein (Fig. 1B–C, Supplementary Fig. S1B). Expression of an inactive RUNX1 (due to a truncation in the RUNT DNA-binding domain) decreases the expression of RAG1 transcript and protein compared to wild-type RUNX1 (Supplementary Fig. S1C–D). Those results demonstrated that both ETV6-RUNX1 and RUNX1 upregulate, directly or indirectly, the expression of RAG1 and, additionally, that the DNA-binding domain of RUNX1 is involved in this regulation.

To investigate whether ETV6-RUNX1 and RUNX1 could be recruited to the RAG locus control region in human pre-B lymphocytes, we performed chromatin immunoprecipitation followed by sequencing (ChIP-Seq) with RUNX1 and ETV6 antibodies in different cells: bone marrow mononuclear cells from 3 childhood BCP-ALL patients negative for ETV6-RUNX1, Nalm6 cells, and REH cells that express ETV6-RUNX1 fusion protein (Supplementary Fig. S2). Of note, REH cells are deleted for the normal ETV6 allele; ChIP-Seq with ETV6 antibody in REH cells is specific for ETV6-RUNX1. We have already described the genomic occupancy of RUNX1 in BCP-ALL patients, Nalm6 and REH cells [9, 10]. Additionally, RUNX1 and ETV6-RUNX1 share 5,377 peaks in REH cells, and about 2000 peaks are uniquely identified for ETV6-RUNX1 (Supplementary Fig. S2A). About 25% of the regions occupied by RUNX1 or ETV6-RUNX1 are transcriptionally active (H3K4me1, markers of active enhancers; H3K4me3, active promoters; H3K27ac, transcriptionally active chromatin) (Supplementary Fig. S2B). Several peaks for RUNX1 and ETV6-RUNX1 are observed within the + /−100kb region overlapping the RAG locus control region (Supplementary Fig. S2C). Two peaks were clearly identified at −1200 bp (referred to enhancer because of its negativity for H3K4me3 and positivity for H3K4me1) and −80 bp (promoter, H3K4me3 positive, H3K4me1 negative) from RAG1 transcription starting site (TSS), in all the samples (BCP-ALL patients, Nalm6 and REH cells) and shared for RUNX1 and ETV6-RUNX1 (Fig. 1D, Supplementary Fig. S2C).

In RUNX1 ChIP-seq profiles from Nalm6 and BCP-ALL patients’ cells, RUNX1 seems to preferentially bind the −80 bp proximal region compared to the −1200 bp. We confirmed this preferential binding of RUNX1 on the −80 bp proximal region compared to the −1200 bp enhancer by ChIP-qPCR (Supplementary Fig. S3A). We also observed that ETV6-RUNX1 binds to the −1200 bp enhancer similarly to RUNX1 (Fig. 1E, left), but less than RUNX1 on the −80 bp promoter (Fig. 1E, right). Competition assays between tagged RUNX1 and ETV6-RUNX1 on each of these two regulatory regions showed that ETV6-RUNX1 is a potent competitor for the binding to −1200 bp RAG1 enhancer while RUNX1 is the major binding protein for −80 bp RAG1 promoter (Supplementary Fig. S3B–E).

To ascertain the physiological role of these regions on the regulation of RAG1 expression, we used a CRISPR-mediated activation system dCas9-VP64, where dCas9 is dead and VP64 induces transcription [11]. In HEK cells, the three gRNAs targeting the −1200 bp enhancer did not significantly affect RAG1 transcript level, probably due to some limitation of this technique to achieve a long-range action [11, 12] (Supplementary Fig. S4A). On the contrary, a strong activation of RAG1 mRNA expression is observed with gRNAs targeting the −80 bp RAG1 promoter in HEK cells and Nalm6 cells, an appropriate model for BCP-lymphoblasts (Fig. S4A, Supplementary Fig. 2A). We also observed a slight increase in RAG1 protein level (Fig. 2B). Altogether, those results demonstrate that the −80 bp RAG1 promoter is a physiologically active site of transcription in pre-B cells and controls RAG1 expression.

Fig. 2: ETV6-RUNX1 and RUNX1 physiologically activate the transcription of RAG1 and increase RAG-mediated recombination.
figure 2

A Relative mRNA expression of RAG1 measured by RT-qPCR in Nalm6control and Nalm6−80gRNA cells for the CRISPR dCas9-VP64 experiments. The gRNA used correspond to gRNA#1 for the −80 bp promoter. Results are presented in-terms of a fold change in log2 scale after normalizing with GAPDH mRNA. Each value represents the mean ± S.D. of three independent experiments (i.e. independent stable cell lines). B Representative images of western blot (left panel) and densitometric analysis (right panel) showing the quantitation of endogenous RAG1 protein in Nalm6control and Nalm6−80gRNA cells. Results are presented after normalizing with β-actin protein levels. Each value represents the mean ± S.D. of three independent experiments. C Partial sequences of the −1200 bp enhancer and the −80 bp promoter with indication of putative RUNX1 binding sequence (identified in silico by JASPAR). The relative score is indicated below each sequence. D Luciferase assays with plasmids containing either the −1200 bp enhancer (left panel) or the −80 bp promoter (right panel) of RAG1 upstream a minimal promoter and a luciferase ORF in Nalm6control, Nalm6RUNX1 and Nalm6ETV6-RUNX1 cells. Luciferase levels are represented using a scatter dot plot indicating the means and S.D. of at least 4 experiments. ns: non-significant. E Quantification, by flow cytometry, of RAG-mediated recombination using the reporter assay from [15] in Nalm6control, Nalm6RUNX1 and Nalm6ETV6-RUNX1 cells, 7 days after transduction with GFPi vectors. Absence of RAG activity results in only RFP (red fluorescence protein) production, while active RAGs mediate the inversion of the GFP (green fluorescence protein) gene allowing for GFP and RFP production. Results are expressed as the percentage of GFP-positive cells in RFP-positive cells (n = 4). F Proposed schematic representation of the ETV6-RUNX1 “multi-hit” leukemogenesis model with a causative transcriptional link between the first and second steps. The * indicates the steps demonstrated in this work. AID: activation-induced cytidine deaminase.

Next, we validated the specific binding sites of RUNX1 predicted in silico by JASPAR in HEK cells by deleting those sites in luciferase assays (Fig. 2C, Supplementary Fig. S4B). As expected in non-hematopoietic lineages ETV6-RUNX1 is not active on those luciferase assays [13, 14]. Importantly, we demonstrated the responsiveness of those two regulatory regions on RAG1 expression in pre-B cells using the corresponding retroviral-luciferase reporter assay in Nalm6 cells (Fig. 2D). Stable expression of ETV6-RUNX1 in Nalm6 cells induces luciferase activity with the −1200 bp enhancer, whereas stable overexpression of RUNX1 activates the −80 bp proximal promoter. This result demonstrates that, in a proper B-cell lineage, the −1200 bp enhancer is activated by ETV6-RUNX1, and that RUNX1 is a major activator of the −80 bp promoter.

We next tested the link of causality between the expression of ETV6-RUNX1 or RUNX1 and the increase in RAG recombinase activity, making use of the quantitative GFPi reporter assay [15]. In this assay, the higher the GFP signal, the higher the RAG recombinase activity. When applied to Nalm6RUNX1 and Nalm6ETV6-RUNX1 cells, the GFPi assay shows value significantly higher than for Nalm6control, demonstrating that enforced expression of RUNX1 as well as ETV6-RUNX1 causes an increase in RAG activity in pre-B lymphoblasts (Fig. 2E, Supplementary Fig. S4C).

Taken together, these data contribute to complete the multi-hit model of ETV6-RUNX1 BCP-ALL leukemogenesis. We propose the following model that involves components from genetics, gene expression and activity, and immunity (Fig. 2F). The first step consists in the t(12;21)(p13;q22) translocation that usually occurs in utero and produces the ETV6-RUNX1 fusion gene. For step 2, the abnormal ETV6-RUNX1 transcription factor and RUNX1 directly induce RAG1 overexpression by binding mainly to the −1200 bp enhancer and the −80 bp promoter in the RAG1 locus, respectively. For step 3, a dysregulated immune response occurs during infections [1]. The binding of ETV6-RUNX1 and RUNX1 to the RAG1 locus (step 2) results in a RAG aberrant increased activity (step 4), and participates, together with additional stimuli such as inflammation and abnormal immune response (step 3) in the generation of inappropriate genomic rearrangements (step 5) [2, 7, 1]. Those additional mutations will promote conversion of the ETV6-RUNX1 preleukemic clone into overt leukemia.

The involvement of ETV6-RUNX1 in the upregulation of RAG1 has been previously examined in vitro and in murine preleukemic ETV6-RUNX1 pro/pre B cells [7, 8]. However, our data, while concordant with those results, go beyond. We demonstrate a direct causal hierarchy between the presence of ETV6-RUNX1 and RUNX1 proteins and RAG1 upregulation. Our findings clearly demonstrate that RAG1 transcripts are directly upregulated by ETV6-RUNX1 from the –1200 bp RAG1 enhancer and by RUNX1 from the –80 bp RAG1 promoter in human pre-B cells. Our findings are complementary to previous studies unraveling the infectious/immune component of ETV6-RUNX1 BCP-ALL. Those reports demonstrated that the activation-induced cytidine deaminase (AID) and/or RAG recombinase drove leukemia with repeated exposure to inflammatory stimuli (step 3 of our model), paralleling chronic infections in childhood [7, 8]. We demonstrated that ETV6-RUNX1 and RUNX1 directly induce RAG1 overexpression (step 2) and a direct link between RAG1 overexpression and RAG aberrant increased activity (step 2 and step 4).

We propose a convincing model directly linking the leukemia-initiating event (i.e., the t(12;21) ETV6-RUNX1 translocation) with upregulation of RAG1 as well as with a stronger activity of RAG recombinase as observed in ETV6-RUNX1 BCP-ALL leukemogenesis.