PML‐RARα interferes with erythropoiesis by repressing LMO2 in acute promyelocytic leukaemia

Abstract The PML‐RARα fusion gene, generated by the t(15;17) chromosome translocation, is regarded as the initiating factor of acute promyelocytic leukaemia (APL). In addition to the well‐known effects on blocking myeloid differentiation at the promyelocytic stage, promyelocytic leukaemia‐retinoic acid receptor α (PML‐RARα) has also been reported to interfere with multiple differentiation processes, including erythroid differentiation. However, the detailed molecular mechanism by which PML‐RARα impairs erythropoiesis has not yet been fully addressed. By chromatin immunoprecipitation‐PCR assay, we found that PML‐RARα bound to the distal promoter region of LMO2 (LIM‐only protein 2), a critical erythroid‐specific transcription factor. Luciferase reporter assays and qRT‐PCR results demonstrated that PML‐RARα down‐regulated the expression of the LMO2 distal transcript through transrepressing its promoter activity. Analysis of gene expression profiling data from large cohorts of acute myeloid leukaemia (AML) patients confirmed that LMO2 expressed at a markedly lower level in APL patients in comparison to non‐APL AML patients. Further flow cytometry analysis demonstrated that PML‐RARα inhibited erythropoietin‐induced erythroid differentiation by down‐regulating LMO2 expression. Our findings reveal a previously unidentified mechanism, by which PML‐RARα interferes with erythropoiesis through directly targeting and transrepressing LMO2 expression in the development of APL.

expressed at the early stage of hematopoietic hierarchy such as multipotent progenitors rather than committed myeloid progenitors and promyelocytes only, [6][7][8] indicating that the influence of PML-RARα may not be limited to myeloid cells but other lineages of blood cells as well. Furthermore, PML is consistent with the previous finding in early hematopoiesis and erythropoiesis, 9 suggesting that the disrupted expression pattern of PML by PML-RARα may affect normal erythropoiesis. Indeed, it has been reported that PML-RARα can interfere with hemin-induced erythroid differentiation in K562 cells, 10 further supporting the idea that PML-RARα may impair erythropoiesis. However, the molecular mechanism by which PML-RARα influences erythroid differentiation is not yet clear.
LMO2 (LIM-only protein 2, also known as RBNT2), is an important regulator of hematopoietic stem cell development and erythropoiesis, as mice deficient in Lmo2 show a complete lack of blood cells and defects in the formation of foetal erythrocytes. 11 LMO2 has been demonstrated to function as a bridge molecule and assist in the assembly of multimeric transcription factor complexes. LMO2 is capable of inducing erythroid differentiation through the interaction with transcription factors, including SCL, E2A, LDB1 and GATA-1. 12,13 Such a transcriptional complex regulates the expression of erythroid-specific genes, such as the α-globin genes, 14 EKLF 15 and glycophorin A (GPA). 16 Knockdown of LMO2 results in the disassembly of this transcriptional complex and thereby attenuates the chromatin occupancy of GATA-1 and LDB1, 17 ultimately leading to the dysregulated expression of erythroid-specific genes. Moreover, forced expression of LMO2 is able to rescue the defective erythroid differentiation caused by c-myb silencing in CD34 positive cells. 18 The above findings indicate the important role of LMO2 in erythropoiesis.
In the present work, we found that PML-RARα but not wild-type RARα bound to the distal promoter of LMO2 and thereby down-regulated the expression of LMO2 through decreasing the promoter activity. We showed that LMO2 expression was significantly lower in APL patients than that in non-APL AML patients.

| Chromatin immunoprecipitation analysis
Chromatin immunoprecipitation (ChIP) assays were performed according to the Affymetrix protocol as described, 19 with the fol-

| Gene expression analysis
Three transcriptome data sets of AML patients, including TCGA, 20 GSE10358 21 and GSE1159, 22 were used to compare the expression of LMO2 between APL and non-APL patient samples. To perform interarray comparison, the CEL files were analysed by Affymetrix MAS 5.0 software (Affymetrix, Santa Clara, CA, USA).
Two-tailed t-tests were used to validate the significance of the observed differences, which were considered statistically significant when P < 0.05.

| Gene Ontology analysis
ChIP-Seq data set GSM552237 23  is considered statistically significant.

LMO2
To identify the potential genes that might be involved in the inhibition of erythroid differentiation in the pathogenesis of APL, we screened the PML-RARα targets that we previously discovered from genome-wide studies. 24 Interestingly, we found that PML-RARα was significantly enriched in the distal promoter region of LMO2 ( As illustrated in Figure 1B, the positive bands were only amplified in the ChIPed region in ZnSO 4 -treated PR9 cells and NB4 cells but not in untreated PR9 cells. These results indicate that PML-RARα rather than wild-type RARα binds to the distal promoter of LMO2 in APL cells.

LMO2 through transcriptional repression of the LMO2 distal transcript
The next question we asked was whether such binding affected the transcription of LMO2. To answer this question, we first scanned the enriched motifs within the LMO2 distal promoter. As shown in Figure 2A

| LMO2 is expressed at a lower level in APL than in non-APL AML subtypes
The above observations demonstrated that PML-RARα repressed the LMO2 expression via targeting the LMO2 distal promoter, which indicated a negative correlation between PML-RARα and LMO2 in APL.
To further verify the correlation between PML-RARα and LMO2 in a large population, we retrieved three data sets (TCGA, GSE10358 and GSE1159) on the expression profiling of 743 AML patients, [20][21][22] including 76 APL patients and 667 patients with other AML subtypes.
Using these data sets, we compared the LMO2 expression values between APL patients and non-APL AML patients. As shown in Figure 3, the large-scale gene expression revealed that LMO2 was expressed at a lower level in APL patients as compared with non-APL AML patients, further confirming that LMO2 expression was specifically down-regulated with the expression of PML-RARα in APL.

| PML-RARα interferes with erythroid differentiation through repressing LMO2 in APL
Since  33 These observations strongly suggest a global differentiation F I G U R E 3 LMO2 is expressed at a lower level in APL than in non-APL AML subtypes. Three gene expression profiling data sets were retrieved, including TCGA, 20 GSE10358 21 and GSE1159. 22 The difference in LMO2 expression between APL and non-APL AML subtypes was assessed using the two-tailed t-test. in between. 24 The two RAREh sites within the LMO2 promoter were around 300 bp apart, raising the possibility that the two RAREh sites could be spatially close due to the high-order structure of chromatin.
In addition to LMO2 per se, we also looked at LMO2 target genes and focused on the genes with differential expression between APL and non-APL patients. Interestingly, we found that these genes were enriched in several signalling pathways critical for erythropoiesis. For instance, activation of PI3-kinase is crucial for cell proliferation of erythroid progenitors. 28 Moreover, PI3-kinase/AKT signalling pathway is regarded as a mediator in EPO-induced erythropoiesis through favoring GATA-1 transcription. 27 Ras signalling pathway negatively regulates erythroid maturation by observing that overexpression of RAS blocks the differentiation of erythroid progenitor cells. 26 39 In our study, we observed different changes in LMO2 expression upon ATRA or ATO treatment in NB4 cells ( Figure S1A and B). A similar observation has been found in our previous findings, in which PSMB8, PSMB9 and PSMB10 show response to ATRA but not ATO. 40 LMO2 expression showed no change or even further down-regulation in ATRA-treated NB4 cells ( Figure S1B). The result observed upon ATRA treatment was reasonable, since miR-223 is reported to repress LMO2 expression, which is up-regulated during the ATRA-induced differentiation process from promyelocytes to neutrophils. 41,42 In contrast, ATO treatment could restore LMO2 expression in NB4 cells ( Figure S1A).
Our data likely indicate that ATO but not ATRA has the ability to reactivate LMO2 expression in APL cells.
Collectively, our findings identify LMO2, as a downstream target of PML-RARα, whose dysregulated expression is associated with the failure of erythropoiesis in APL. Our data not only reveal a molecular mechanism of PML-RARα-mediated erythropoiesis inhibition but also provides evidence that PML-RARα has broad impacts on multiple lineages of blood cells rather than myeloid lineage only.

ACKNOWLEDG EMENTS
The work was supported in part by National Natural Science Foun-

CONFLI CT OF INTEREST
The authors declare no conflict of interest.