Wnt signaling induces transcription, spatial proximity, and translocation of fusion gene partners in human hematopoietic cells

One of the most common chromosomal abnormalities in acute myeloid leukemia (AML) is the reciprocal translocation t(8;21)(q22;q22), which involves the RUNX1 (AML1) and the ETO (RUNX1T1) genes and produces a Runx1-ETO fusion transcript that inhibits Runx1-dependent transcriptional regulation.^1,2  Although RUNX1-ETO translocations could act as initiating events in hematopoietic stem cells (HSCs), after which leukemia clonally evolves through the acquisition of secondary mutations,^3,4  the molecular mechanism and cellular signals that drive the generation of the t(8;21) translocation remain to be elucidated.

Although much has been learned in recent years about the onset or development of AML from studies examining Runx1 expression or function,^5-8  we are still far from a complete understanding of the cellular mechanisms controlling the transcription of its translocation partner, ETO.^9-11  Considering that Wnt/β-catenin plays essential roles during the proliferation or differentiation of HSCs and that reactivation of β-catenin signaling is important for self-renewal of leukemia stem cells,^12-14  we studied whether Wnt/β-catenin signaling was involved in ETO expression and RUNX1-ETO fusion.

CD34⁺ cells were isolated from umbilical cord blood samples after normal full-term delivery from the Maternity Service at Complejo Asistencial Barros Luco (Santiago, Chile). Informed consent was obtained in accordance with the Declaration of Helsinki and the ethics committee of the Universidad Andres Bello. KG1, HEK293, and L cells secreting Wnt3a were obtained from ATCC (Rockville, MD).

Cell culture conditions, Wnt/β-catenin signaling activity, semiquantitative and quantitative polymerase chain reaction, ETO promoter activity, plasmids and site-directed mutagenesis, β-catenin and RNAPII immunofluorescence, RNA and DNA fluorescent in situ hybridization (FISH) experiments, image acquisition and processing, and statistical analyses are described in supplemental Methods (available on the Blood Web site).

According to the current assembly of the human genome (GRCh38), the ETO gene consists of 17 exons distributed over 148 kb (Figure 1A), and its expression generates multiple messenger RNA (mRNA) isoforms by alternative splicing or the use of different promoters.^9-11  We searched for core TCF/LEF-binding elements (TBE: CTTTG) upstream and downstream of the proposed transcriptional start sites of the human gene and found multiple TBE sites in these genomic regions. We designed primers to detect common alternative mRNA transcripts (HGNC: ETO-001, 002, 004, and 009; supplemental Table 1). Although endogenous ETO mRNA levels were readily detected in isolated CD34⁺ cells under control conditions,⁶  we observed that a differential enhancement in the transcription of ETO-004 variant occurred in CD34⁺ cells after short-term treatment (4 hours) with purified Wnt3a (Figure 1B). Similarly, whereas endogenous levels of ETO mRNA were barely detected in the KG1 cells¹⁵  under control conditions, a clear increase in the expression of ETO-004 variant occurred in these cells upon Wnt3a treatment. ETO variants were also detected in HEK293 cells under control or Wnt3a conditions, suggesting that this embryonic cell line could be a useful model to study alternative ETO expression.

Quantitative determination of mRNA levels in CD34⁺ cells showed that Wnt/β-catenin induced a rapid (4-hour) and significant increase in ETO-004 variant (3.7-fold, P = .0028, n = 6) or total ETO mRNA levels (3.6-fold, P = .0028, n = 6; Figure 1C). Wnt3a effects on total ETO levels were further confirmed by TaqMan analyses (2.9-fold, P = .0011, n = 7; Figure 1C and supplemental Figure 1). Finally, the effect of the signaling cascade on ETO expression was paralleled by enhanced transcription of Runx1^16,17  (2.6-fold, P = .0075, n = 5) and classical target genes Axin2 and Myc (3.5- and 2.8-fold, respectively; P = .0075, n = 5).^18,19 

Putative promoters for ETO-002 and -004 variants are located downstream of common RUNX1-ETO translocation hotspots (Figure 1A). To study if these promoters contained functional TBE sites, we PCR-cloned 890 and 2459 nucleotides of the 5′ upstream/exon 1 sequences of ETO-002 and -004, containing 6 and 7 potential TBE sites, respectively (Figure 1D). We also analyzed 1535 nucleotides with 5 TBE sites upstream of the transcriptional start sites described for variant ETO-009.¹⁰  We observed a significant increase in the promoter activity of pETO-004 (2.5 kb) in HEK293 cells cotransfected for 24 hours with increasing concentrations of constitutive active β-catenin (S33Y) (Figure 1D). Although basal transcription was observed for pETO-002 and pETO-009 promoters, no significant effects were found in cells cotransfected with these constructs and β-catenin S33Y (24 hours). Then, we serially deleted potential TBE sites in pETO-004 (pETO-004-0.9 and pETO-004-04, respectively) and observed that these constructs responded accordingly to β-catenin stabilization or its inhibition by a dominant-negative ΔTCF4 construct (Figure 1E). Next, because pETO-004-0.4 contains a single TBE site (−457 nucleotides upstream of the ATG of ETO-004 variant), we mutated the core consensus sequence (CTTTG → CCTCG) and found that the activity of the mutant promoter was decreased to basal levels (supplemental Figure 2). Collectively, our findings indicate that Wnt/β-catenin signaling rapidly induces ETO-004 variant and that this effect is probably accomplished through ETO-004 TBE site I.

Translocation-prone genes are preferentially recruited into transcription factories with active RNA polymerase II (RNAPII-Ser5)^20,21  and need to be positioned in close spatial proximity relative to each other prior to translocation.^22,23  Because β-catenin participates in recruiting basal transcription components at the promoter of its target genes, we studied if activation of the cascade induced β-catenin relocalization into these transcriptional structures. We found that transcription factories labeled with active RNAPII-Ser5 were clearly distinguished in resting or Wnt3a-treated CD34⁺ cells (4 hours) (Figure 2A-B). Under control conditions, the number and diameter of RNAPII speckles within the nucleus were in agreement with previous observations²²  (32.1 ± 1.13 and 130.7 ± 3.5 nm, respectively; supplemental Figure 3A). Remarkably, Wnt3a induced a significant increase in the number of endogenous β-catenin speckles within the nucleus of CD34⁺ cells (control: 1.5 ± 0.3; Wnt3a: 19.5 ± 2.7; P < .001; supplemental Figure 3A), some of which clearly colocalized with active RNAPII foci (supplemental Figure 3B-C). Similar effects of β-catenin colocalization with RNAPII-Ser5 were observed in Wnt3a-treated KG1 cells (supplemental Figure 4).

Our data indicate that β-catenin rapidly associates with RNAPII transcription factories in response to Wnt/β-catenin activation, further implying that expression of ETO and RUNX1 translocation partners could occur in these chromosomal structures. To address this idea, we performed RNA-FISH experiments in CD34⁺ cells using fluorescent probes targeting nascent sequences within ETO and RUNX1 genes (introns 11 and 4, respectively) (Figure 2C). We noted that the total number of ETO and RUNX1 alleles expressed in control CD34⁺ cells increased from 25 to 54 after 4-hour Wnt3a-treatment (supplemental Figure 5 and supplemental Table 2). Wnt/β-catenin induction was similarly observed for both genes (ETO alleles: from 9 to 21; RUNX1 alleles: from 16 to 33) and turned on the expression of a second allele in a greater number of cells. We determined the distance between RUNX1 and ETO alleles being simultaneously transcribed and observed that from a total of 105 Wnt3a-treated cells, 7 cells had distances <300 nm between alleles (there were no events in the control setting that showed a proximity <300 nm) and that the RNA signals in 3 of these CD34⁺ cells localized into the same chromosomal foci following activation of the cascade (Figure 2C and supplemental Figure 5). Altogether, these observations confirm that Wnt/β-catenin enhances ETO and Runx1 expression in hematopoietic precursors and indicate that transcription of these genes is likely occurring into specialized nuclear factories. To further examine the hypothesis that Wnt signaling facilitates genomic proximity, we performed DNA-FISH experiments in CD34⁺ cells and observed that RUNX1 and ETO genes are indeed brought up in close proximity (<1 μm) after short-term Wnt3a treatment (4 hours) (23% vs 10%; Wnt3a vs control conditions, respectively; Figure 2D and supplemental Table 3). No significant differences in proximity were found between RUNX1 and ETO at 48 hours, suggesting that an early transcriptional program commanded by the signaling cascade is involved in chromosome spatial organization and gene regulation.

Recent experiments indicate that β-catenin promotes genomic instability and T-cell transformation in mice by compromising DNA repair and enhancing illegitimate recombination.²⁴  Because most RUNX1-ETO translocations occur within RUNX1 intron 5 and ETO intron 7 we designed primers in RUNX1 exon 5 and ETO exon 8 to detect the product of the fusion of these genes. Notably, in CD34⁺ cells incubated with Wnt3a for 48 hours, we observed amplification of the expected RUNX1-ETO chimeric product (Figure 2E and supplemental Figure 6). Finally, Wnt/β-catenin effects were further detected in HEK293 cells transfected for 48 hours with increasing doses of β-catenin S33Y (supplemental Figure 7). Our data thus indicate that sustained transcriptional activation induced by the signaling cascade represents an early event that triggers the translocation of RUNX1 and ETO genes.

Deregulation of Wnt/β-catenin signaling as a causative factor in leukemogenesis has become increasingly apparent.^12-14  Here, we studied the transcriptional program directed by Wnt/β-catenin signaling in CD34⁺ and KG1 cells because they resemble normal human HSCs¹⁵  and it has been suggested that immature cells, rather than committed progenitors, are the targets for leukemic transformation.²⁵  We observed that gain of function of Wnt/β-catenin signaling enhances transcription of ETO and RUNX1 translocation partners in genome organization centers within the nucleus, thereby increasing the potential for a chromosomal translocation event. To understand the role of Wnt/β-catenin signaling in other recurrent translocations remains a goal of further investigations.

The authors are grateful to the Maternity Service at Complejo Asistencial Barros Luco. The authors thank Dr Martín Montecino for sharing the pCMV5-AML1-ETO plasmid and Flavia Román and Dr Miguel Avila for technical assistance or preliminary work.

This work was supported by Comisión Nacional de Investigación Científica y Tecnológica (CONICYT) grants Anillo ACT-1119 (G.V.D.F. and A.L.), FONDECYT Postdoctorado 3130509 (G.D.U.) and 3150612 (M.F.V.), Fondo Nacional de Desarrollo Científico y Tecnológico (FONDECYT) Regular 1140353 (G.V.D.F.) and 1130697 (S.E.G.), Chilean Copper Commission (COCHILCO)-FONDECYT 1100995 and IMII P09-016-F (A.E.), and FONDAP-CRG 15090007 (G.V.D.F.). R.T.M. is supported as an Investigator of the Howard Hughes Medical Institute.

Contribution: G.V.D.F. and G.D.U. designed the study; G.D.U., M.F.V., M.A.M., P.L., and D.N. performed experiments, analysis, and interpretation of the data; S.E.G., A.A.E., R.T.M., and A.L. helped in drafting the manuscript and contributed cell or molecular tools; and G.D.U., M.F.V., M.A.M., and G.V.D.F. wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Giancarlo V. De Ferrari, Center for Biomedical Research, FONDAP Center for Genome Regulation, Faculty of Biological Sciences, Universidad Andrés Bello, República 239, Santiago, Chile; e-mail: gdeferrari@unab.cl.