Large deletions of the 5′ region of <i><scp>IKZF</scp>1</i> lead to haploinsufficiency in B‐cell precursor acute lymphoblastic leukaemia

Deletions or mutations of IKZF1, encoding the IKAROS transcription factor, are associated with poor prognosis in some subtypes of B-cell precursor acute lymphoblastic leukaemia (BCP-ALL) (Mullighan et al, 2009; Clappier et al, 2015). IKZF1 status is therefore important for risk stratification and defining therapies (Stanulla et al, 2018; Yeoh et al, 2018). Most IKZF1 abnormalities are intragenic deletions, or deletions of the entire locus (Mullighan et al, 2008; Iacobucci et al, 2009; Marke et al, 2018), which lead to loss of IKAROS expression and/or function. However, deletions affecting only the sequences 5′ of the coding region, including the non-coding exon 1, have also been detected by comparative genomic hybridisation-arrays (Iacobucci et al, 2009, 2012; Dupuis et al, 2013) and multiplex ligation-dependent probe amplification (MLPA), which is routinely used to assess IKZF1 status in patients (Fig 1A) (Caye et al, 2013). Because the sequences surrounding exon 1 are rich in G/C nucleotides, their detection by MLPA may sometimes be artefactual. Furthermore, the breakpoints of the 5′ deletions, as well as their consequence on mRNA and protein expression, are uncertain. There are currently no clear guidelines for labelling “exon 1 deletions” as true IKZF1 mutations. Understanding the molecular nature and impact of IKZF1 5′ deletions is therefore important. To determine if the exon 1 deletions (Δex1) designated by MLPA are bona-fide deletions, exon 1 genomic (g)DNA level was assessed for 13 MLPA-detected Δex1 BCP-ALL samples by quantitative polymerase chain reaction (qPCR) (Table SI). See Appendix S1 for Material and Methods. The 5′ and 3′ ends of exon 1 were analysed (Fig 1B). DNA from healthy donors (2 copies of IKZF1) and BCP-ALL samples with monosomy 7 or an IKZF1 monoallelic deletion (both with one copy of IKZF1), were used as controls (Fig 1C). Nine of 13 (70%) MLPA-detected Δex1 samples displayed exon 1 levels at half those of healthy donors, similar to samples with single IKZF1 copies. Similar results were obtained using two reference genes (GAPDH, HBB), confirming that exon 1 was deleted in the majority, though not all, MLPA-detected samples. To map the genomic breakpoints of the 5′ deletions, we screened gDNA from the 9 PCR-verified samples, from −500 kb to +14 kb of exon 1 by qPCR (Fig 1D and Table SII). Eight patients had 3′ breakpoints near +1·5 kb, and 1 near +5 kb, suggesting a shared mechanism. In contrast, the 5′ breakpoints were variable and mapped near −15 kb, −29 kb, −79 kb, −110 kb, −115 kb, −120 kb and −165 kb. One deletion extended beyond −500 kb. We sequenced the breakpoint region of 3 patients, and found VDJ recombination signal sequences at both the 5′ and 3′ breakpoints, indicating a RAG enzyme-mediated mechanism (Fig 1E). Indeed, Patients 2 and 4 had nearly identical 3′ breakpoints. These results suggested that large 5′ deletions occur upstream of the IKZF1 coding region, probably during VDJ recombination of IGH in B-cell precursors. To investigate their impact of 5′ deletions on protein expression, IKAROS levels were assessed by intracellular flow cytometry in 5 Δex1 samples (Fig 2A). Patient cells with normal IKZF1 status were used as controls. IKAROS levels were reduced in all samples with a monoallelic IKZF1 5′ deletion. Both alleles were mutated (a complete deletion and a 5′ deletion) in Patient 6; here, IKAROS expression was completely lost in most cells, though some cells retained intermediate IKAROS levels, suggesting clonal heterogeneity. We also evaluated IKAROS expression by Western blot (Fig 2B). Full-length IK1 and the smaller IK2 isoforms were both reduced in the tested samples compared with control, and no novel isoforms were detected, indicating that 5′ deletions result in loss of IKAROS protein expression. To determine if the 5′ deletions affect mRNA expression, IKZF1 mRNA levels from 7 Δex1-only patients were compared with those of patients with different IKZF1 status (wildtype [WT], mono- and biallelic deletions) by reverse transcription-qPCR, using primers to amplify exons 4–5 (Fig 2C). The 5′ deletions were associated with significant reduction in IKZF1 mRNA, similar to that observed in other types of monoallelic IKZF1 deletions. Conversely, 2 MLPA-detected Δex1 samples not validated by PCR (non-Δex1) showed WT IKZF1 mRNA levels. To confirm that the 5′ deletions lead to a loss of transcription from the affected allele, we took advantage of polymorphisms in the IKZF1 gene. One common single nucleotide polymorphism (SNP) (rs61731355; C→A at position Chr7:50,400,069 within exon 8) was detected in the gDNA and mRNA (cDNA) of 2 control patients, showing that IKZF1 is biallelically expressed in BCP-ALL cells (Fig 2D). Patient 4 showed a novel SNP in their gDNA (C→A at position Chr7:50,400,071, also within exon 8) (Fig 2D). This polymorphism was not detected in the patient mRNA, indicating that only one allele was transcriptionally active (Fig 2D). These results suggested that deletion of the 5′ sequences transcriptionally inactivates the IKZF1 gene in Δex1 patients. Ascertaining the frequency of exon 1 deletions in BCP-ALL is complicated by the high G/C content in this region, which results in false-positive results using routine techniques. Interestingly, by improving the MLPA protocol (i.e., using a denaturation step of 20 min instead of 5, and controlling dubious cases with MLPA kits that contain more 5′ probes), we reduced the frequency of exon 1 deletions from 1·5% to 0·5% [data from patients from the EORTC 58951 (1189 patients) and 58081 (388 patients) studies, respectively]. A higher frequency was reported with SNP microarrays (3%; data from the P9906 cohort (221 patients) (Mullighan et al, 2009), although it is unclear how the 5′ deletions were defined in this study. A thorough assessment of the frequency of 5′ deletions in adult and paediatric BCP-ALL is therefore warranted. In conclusion, our results indicate that IKZF1 5′ deletions are recurrent alterations. They comprise large stretches of DNA that vary in length by hundreds of kb, and end at their 3′ end near Chr7:50,306,420 in most cases. We further show that the mutant alleles lead to loss of IKZF1 mRNA and protein. IKZF1 mutations and deletions are often associated with poor prognosis in BCP-ALL (Mullighan et al, 2009; Clappier et al, 2015), and it will be important to determine if the present 5′ deletions negatively impact prognosis. Our study, however, aimed at characterizing exon 1 deletions in molecular terms, and our cohort was too small and heterogeneous (Table SI) to correlate with clinical outcome. Nonetheless, since the 5′ deletions are null alleles, like Δex2-8 and entire IKZF1 deletions, we suggest that they be considered as such during risk stratification (Yeoh et al, 2018). This work was supported by grants from the Conférence de Coordination Inter-Régionale du Grand-Est of the Ligue Contre le Cancer to B. Heizmann (01Q-2018), the Institut National du Cancer to P. Kastner and H. Cavé (2015-114), and by funds from ANR-10-LABX-0030-INRT. G. Morel received an “Année Recherche” fellowship from the Faculté de Médecine of the Université de Strasbourg. M-C. Deau received an IGBMC International PhD Programme fellowship (from ANR-10-LABX-0030-INRT). C. Simand received a PhD fellowship from the Ligue Contre le Cancer. G.M. and MC.D. performed experiments, interpreted the data and revised the manuscript; C.S. collected and organized the patients’ data; A.CE., C.A., A.I. and L.Mi. prepared and analyzed patients’ samples at diagnosis; L.Ma., R.H., C.P. and M.S. provided patients’ samples and clinical data; H.C. provided patients’ samples and clinical data and contributed to the study with fruitful discussions. S.C. and P.K. designed the research, interpreted the data and wrote the manuscript; B.H. performed experiments, designed and supervised the research, interpreted the data and wrote the manuscript. The authors declare no conflict of interest. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.