The corepressors BCOR and BCORL1: two novel players in acute myeloid leukemia

Acute myeloid leukemia (AML) is a disease characterized by high molecular heterogeneity. About one-third of AML patients harbor well-defined chromosomal translocations and are included as a distinct entity named “AML with recurrent genetic abnormalities” in the 2008 World Health Organization (WHO) classification of myeloid neoplasms.¹ During the past decade, there have been major efforts to better define the genetic lesions underlying the large group of AML with normal cytogenetics (CN-AML) that accounts for 40–50% of all AML cases. Mutations associated with CN-AML include those affecting NPM1, FLT3, MLL, CEBPA, TET2, ASXL1, IDH1, IDH2, DNMT3A, RUNX1 and other genes.² Because of their distinct biological and clinical features, “AML with mutated NPM1” and “AML with mutated CEBPA” are already recognized as provisional entities in the 2008 WHO classification.^1,3,4

The increasing use of next-generation sequencing technologies applied to the study of cancer genomes is making a remarkable contribution to our knowledge of the molecular landscape of CN-AML. Following the identification of IDH1 and DNMT3A mutations,^5–7 this strategy has recently led to the discovery that two homologous genes, i.e. BCOR and BCORL1, are recurrently mutated in AML.^8,9 Their involvement in AML emerged from in-solution whole-exome capture followed by massively parallel sequencing of paired leukemic cells at diagnosis and normal hematopoietic cells at the time of complete remission from patients with AML. However, the strategies leading to the discovery of BCOR and BCORL1 mutations in AML differed. We identified mutated BCOR through whole-exome sequencing of a CN-AML patient who was selected for analysis because of lack of any known mutations.⁸ Instead, Li et al.⁹ found BCORL1 to be mutated by searching for recurrent mutations in a discovery cohort of 8 patients with secondary AML (s-AML) that were subjected to whole-exome sequencing. In both studies, these mutations were subsequently studied in larger series of AML patients.^8,9

The structures of BCOR and BCORL1 are shown in Figure 1 and their main features are summarized in Table 1. Both BCOR and BCORL1 genes are located on the X-chromosome and encode for large nuclear proteins that are ubiquitously expressed in human tissues.^10,11 The BCOR protein acts as corepressor of BCL6,¹⁰ it can bind to other transcriptional factors^12–14 and appears to play a key role in the regulation of early embryonic development,¹⁵ mesenchymal stem cell function¹⁶ and hematopiesis.¹⁵ BCORL1 is also a transcriptional corepressor and functional studies have shown that it can bind to class II histone deacetylases (HDAC4, HDAC5, HDAC7), to interact with the CTBP1 corepressor, and to affect the repression of E-cadherin.¹¹ In spite of their similarities, the BCOR and BCORL1 proteins show a number of different features, including their capability to interact with BCL6, their expression levels in human tissues, and their localization within subnuclear structures (Table 1). These findings strongly suggest that, although BCOR and BCORL1 are homologous genes, they encode for proteins that are likely to play distinctly different roles in normal human cells.

Figure 1.

Schematic representation of the BCOR (top) and BCORL1 (bottom) transcriptional corepressors with their respective domains and the location and type of mutations occurring in AML patients. The numbers above the arrows indicate the presence, in more than one case, of the same type of mutation occurring at the same exonic location or involving the splicing of the same intron. BCORL1 mutations also include five mutations that were found in four AML cell lines and in the Jurkat T-lymphoblastic leukemia cell line. The specific type (i.e. frameshift, nonsense, missense or splice-site) of BCORL1 mutations was inferred from Li et al.⁹

Table 1.

Features of the BCOR and BCORL1 transcriptional corepressors.

In AML, the BCOR gene is targeted by both translocations and mutations. One patient with acute promyelocytic leukemia (APL) has been reported to carry the 45,-Y,t (X;17)(p11;q12) karyotype, leading to the formation of a BCOR-RAR alpha fusion gene.¹⁷ Compared to typical APL cases with t(15;17), this patient showed distinct morphological features, i.e. rectangular body inclusions in the cytoplasm of leukemic cells and an atypical clinical course characterized by multiple relapses following chemotherapy plus all-transretinoic acid.¹⁷ BCOR mutations were detected in about 4% (10 of 262) of an unselected cohort of de novo CN-AML. Importantly, they occurred at a higher frequency (about 17%; 14 of 82 cases) in the subset of CN-AML devoid of NPM1, CEBPA, FLT3-ITD, and MLL-PTD mutations, and mostly also lacking IDH1 mutations.⁸ Interestingly, the latter genotype mimicked that of the AML index patient initially used for whole-exome sequencing. BCOR mutations were demonstrated to be clonal, somatic, disruptive events involving the only functional allele, not only in male but also in female AML patients.⁸ Notably, the features of BCOR mutations in AML closely resembled those of the germline BCOR disruptive mutations causing the oculo-facio-cardio-dental (OFCD) X-linked syndrome¹⁸ that is characterized by microphtalmia, congenital cataracts, dysmorphic appearance, radiculomegaly, and digital and cardiac defects. Finally, BCOR mutations in AML were often associated with a decrease in BCOR mRNA levels, absence of full-length BCOR protein, and lack or low expression of a truncated BCOR protein.⁸ Taken together, the above features conform to those of loss-of-function mutations in a tumor-suppressor gene.

BCORL1 somatic mutations were found in about 6% of a series of 173 AML patients which included cases with secondary leukemia (showing myelodysplasia-related changes or being therapy-related) and a high incidence (approx. 65%) of abnormal karyotypes.⁹ Similarly to mutations of BCOR, most of the alterations affecting BCORL1 were nonsense mutations, out-of-frame insertions/deletions or splice site mutations that, although not resulting in clearly diminished mRNA levels of BCORL1, were predicted to encode truncated proteins lacking the last C-terminal LXXLL nuclear receptor recruitment motif.⁹ Although targeting of BCORL1 mutations to the expressed allele in female AML patients was not formally demonstrated by Li et al.⁹, the above findings again point to BCORL1 as a tumor-suppressor gene that is inactivated by mutations in a subset of AML.

At present, only limited information is available about the distribution of BCOR and BCORL1 mutations across the main cytogenetic/molecular AML categories. Mutations of BCOR mainly clustered with CN-AML, were virtually mutually exclusive with NPM1 and FLT3-ITD mutations and showed an association with DNMT3A and RUNX1 mutations.⁸ BCORL1 mutations mostly occurred in patients with s-AML and/or with abnormal karyotype, and carried germline TP53, CEBPA and NPM1 genes.⁹ Because a relatively small number of CN-AML cases has been analyzed for BCORL1 mutations⁹ and cases of s-AML have not been investigated for BCOR mutations,⁸ it is not yet clear whether or not mutations of these two homologous genes may occur in the same clinical-genetic AML backgrounds.

All these findings point to mutational targeting of corepressors as a new mechanism of leukemogenesis. However, the way in which BCOR and BCORL1 mutations contribute to AML development is still not clear. The observation that about half the BCOR-mutated cases also carry mutations of the DNMT3A gene suggests that these two mutations may act synergistically to induce AML, possibly interfering with epigenetic mechanisms. Indeed, DNMT3A encodes a methyltransferase enzyme catalyzing the addition of methyl groups to CpG dinucleotides.⁶ Moreover, BCOR increases the repression of transcriptional activity by interacting with class I and II HDACs, the polycomb group protein PCGF1/NSPC1 and the histone demethylase FBXL10,^13,14,16 which implies BCOR may suppress gene transcription by epigenetic mechanisms.^16,19 Because BCOR (and BCORL1) are ubiquitously expressed, it will be crucial to define the key targets of their corepressive transcriptional activity specifically in AML cells, in order to gain more precise insights into mechanistic role of BCOR and BCORL1 mutations in leukemogenesis.

Do BCOR and BCORL1 mutations have any clinical impact in AML? Analysis of a large series of CN-AML patients suggests that BCOR mutations may confer a poorer prognosis.⁸ However, due to the low frequency of the mutation, additional studies are necessary to confirm these findings. For the moment, no prognostic information is available concerning BCORL1 mutations⁹ and this needs to be clarified.

In conclusion, BCOR and BCORL1 add to the growing list of genes recurrently mutated in AML. These findings also highlight the cost and the huge amount of time involved to investigate the functional and clinical aspects of genetic lesions that occur at a relatively low frequency in AML. Researchers should be prepared to take on this difficult task, since in the future, novel low frequency mutations in other genes are likely to emerge from the sequencing of additional AML genomes.