Novel reciprocal fusion genes involving HNRNPC and RARG in acute promyelocytic leukemia lacking RARA rearrangement

Acute promyelocytic leukemia (APL) is a unique subtype of acute leukemia characterized by retinoic acid receptor alpha gene (RARA) rearrangements (RARA-rearranged APL), of which PML-RARa is the typical one. As an altered retinoic acid receptor, PML-RARA owns the ability to transmit oncogenic signals within the cell. PML-RARA competitively controls the DNA sequences of retinoic acid-responsive element (RARE) sites and has a negative effect on transcription, inhibits the activation of physiological ligands, and leads to the arrest of promyelocytic differentiation.1 However, there are some rare patients lacking RARA rearrangement although they share the same morphology and immunocytochemistry features with RARA -rearranged APL. The classification of such leukemias remains in chaos and they have been also termed as acute promyelocytic-like leukemias (APLL). For years the underlying molecular aberrations of APLL remained unrevealed, but in recent years, RARG and RARB rearrangenments have been found in APLL.62 RARA, RARB and RARG belong to the nuclear receptor superfamily, and their sequences and functions are highly similar. In the present study we identified a couple of novel reciprocal fusion genes involving the heterogeneous nuclear ribonucleoprotein C (HNRNPC) gene and the retinoic acid receptor gamma (RARG) gene in APLL. To our knowledges, the reciprocal fusions were identified in human disease for the first time.

A 43-year-old man was referred to our hospital with fever and a sore throat. Laboratory investigations revealed the following patient characteristics: (1) hemoglobin level, 125 g/L; (2) platelet count, 57×10/L; and (3) white blood cell count 12×10L (1% blasts and 86% abnormal promyelocytes). No signs of coagulopathy were evident. Morphologic analysis of the bone marrow aspirate showed 86.5% microgranular atypical promyelocytes (Figure 1A). Analysis from flow cytometry showed that the blasts were positive for CD33, CD13, CD45, and cMPO and negative for CD14, CD34, CD16, CD56, HLA-DR, B- or T-cell markers. Thus, the patient started all-trans retinoic acid (ATRA, 60 mg/day) and arsenious acid (10 mg/day) treatment immediately.

Figure 1.Morphology analysis of bone marrow smear. (A) Bone marrow aspirate at the first consultation. The abnormal promyelocytes are shown, featured with abundant cytoplasm, azurophilic microgranules, and concaved or irregular nucleinuclei. (B) Bone marrow aspirate evaluation of the initial treatment after three weeks. Response was not observed in the cell morphology. (Wright-Giemsa staining, 1,000× magnification).

Afterwards, chromosomal analysis revealed the majority of metaphases were der(14)ins(14; ¿) (q24; ¿)t(14;17¿) (q24;q11.2¿),der17t(14;17¿)(q24;q11.2¿). Fluorescence in situ hybridization did not identify the PML/RARA rearrangement. A series of reverse transcription PCR (RT-PCR) assays amplified the HOX11 gene and alternative transcripts of 31 fusion genes, including PML-RARA, NPM-RARA, PLZF-RARA, and BCR-ABL, which were all negative. Sanger sequencing detected no FLT3 mutation.

A combined chemotherapy (daunorubicin plus cytarabine) was administrated immediately following ATRA, then, because RARA rearrangment was lacking, arsenious acid (days 1-7) was suspended. After three weeks of ATRA therapy no response was observed in blood cell counts, and no evidence suggested differentiation syndrome. A second bone marrow biopsy (Figure 1B) was subsequently peformed and the blasts were almost unchanged (87.5%) which resulted in the withdrawal of ATRA. Next, the patient received a new course of induction chemotherapy alone (homoharringtonine, daunorubicin and cytarabine, HDA) and he achieved a complete response (CR) this time. Afterwards, he received five cycles of chemotherapy, of which three were consolidated therapies (HAD, HA, and mitoxantrone plus cytarabine) and two were cytarabine intensive treatments. Unfortunately, the leukemia relapsed 1 year later, and two cycles of re-induction chemotherapy, MAC (mitoxantrone, cytarabine and cyclophosphamide) followed by FLAG regiment, were sequentially administrated. Yet the blast counts in the bone marrow remained almost at the same level as at the disease onset (87.2%). Subsequently, ATRA in combination with arsenious acid therapy of sufficient dose and duration was resumed and continued for four weeks. However, all treatments failed to produce any effects. The patient eventually died from sepsis.

To identify molecular alterations, transcriptome sequencing (RNA-seq) analysis was performed using peripheral blood mononuclear cells which were obtained in the patient’ relapse phase. Total RNA (4.6 μg) was used to synthesize cDNA libraries, and rRNA were removed with the Ribo-Zero™ rRNA Removal Kit (Epicentre). The samples were prepared according to the Illumina protocol. The library was sequenced on an Illumina HiSeqTM 2000 platform. The SOAPfuse software was employed to discover the fusion transcripts.7 As a result, we identified a novel gene fusion event between HNRNPC and RARG, leading to the formation of a couple of reciprocal transcripts, HNRNPC-RARG and RARG-HNRNPC. In HNRN-PC-RARG, HNRNPC exon 3 was fused in-frame to RARG exon 4. Whereas in RARG-HNRNPC, RARG exon 9 was fused in-frame to HNRNPC exon 4 (Figure 2A-B).

Figure 2.Detection of the reciprocal HNRNPC/RARG and RARG/HNRNPC fusion. (A) and (B) SOAPfuse obtained Span reads and junction reads of HNRNPC/RARG and RARG/HNRNPC respectively from RNA-sequencing data. (C) and (E) Gel picture of the amplified fragment. M: DNA marker; lane 1: PCR amplicons; lane 2: blank controls. (D) and (F) Sanger sequencing showing the junction of the HNRNPC/RARG and RARG/HNRNPC genes respectively. From the nucleotide sequence surrounding the fusion regions, amino acid sequences may be in frame. (G) Deduced schematic overview of the structures of reciprocal rearrangements, and the construction of the corresponding chimeric proteins is illustrated in figure (H). DBD: DNA binding domain; LBD: ligand binding domain; RRM: RNA recognition motif.

Total RNA was also extracted from the same blood mononuclear cells and reverse transcribed into cDNA. PCR was performed, and the following primers were used to amplify HNRNPC-RARG mRNA: 5′-TGTTCATACGGGCTTTGCCTTCGTTCAG-3′ for HNRNPC and 5′-CATGGCTTGTAGACCAGAGGAGGCGGAG-3′ for RARG. The following primers were used to amplify RARG-HNRNPC mRNA: 5′-ACAAGCTGTAGGAGCCACTGCTGGAAGC-3′ for RARG and 5′-CAAAAGAGGACCCGTACATCTCCGCTGC-3′ for HNRNPC. After Sanger sequencing, the fusion transcript amplicons were compared with GenBank sequences using the BLAST program. The amplicon size of HNRNPC-RARG fusion was 186-bp (Figure 2C), as predicted. Sanger sequencing demonstrated that the amplicon sequence can be completely aligned to that of RNA-seq (Figures 2D). The HNRNPC 5′-region encodes an RNA recognition motif (RRM), and the segment from RARG encodes a DNA binding domain (DBD, Figure 2G). We also found a reciprocal chimeric transcript. A 213-bp RARG-HNRNPC fusion product was specifically amplified from the patient’s cDNA (Figure 2E). Also Sanger sequencing con-firmed the 100% sequence identity (Figure 2F). The RARG 5’-region encoding the ligand-binding domain was fused to the HNRNPC 3’-region, where a cluster of phosphorylation sites is located (Figure 2H). PCR with primers flanking the HNRNPC breakpoint was also performed, but an amplicon of chimera was not obtained.

HnRNP C is a ubiquitously expressed RNA-binding protein (RBP). RBP are believed to influence pre-mRNA metabolism such as splicing, polyadenylation, stability, transport, and translation mediated by internal ribosome entry site.98 HnRNP C also plays an essential role in cell progression and the regulation of several DNA repair proteins.1210 RARA, RARB, and RARG are three RAR subtypes which share highly similar sequences and functions. A study on RAR knockouts observed that RARG overexpression enhanced multipotent hematopoietic stem cell (HSC) self-renewal and the loss of RARG promoted HSC differentiation. Thus, RARG seems to act as a major regulator maintaining the balance between HSC self-renewal and differentiation.13

Acute myeloid leukemias mimicking acute promyelocytic leukemia, or acute promyelocytic-like leukemias (APLL), share the same morphology and immunocytochemistry features with typical APL, and little is known about the molecular mechanisms of APLL. In this study, we identified a novel fusion gene HNRNPC-RARG and its reciprocal in APLL. The sequences and function of the RARG and RARA are highly alike, and therefore can logically explain the similarity of the biological characteristics between the two entities. Three other fusion genes harboring RARG have been found in APLL. Such et al. reported a case of APLL carrying a NUP98-RARG fusion transcript. The patient had CR after standard chemotherapy and then received consolidation therapy and intensification by autologous stem-cell transplantation. He relapsed 2 years later, and re-induction therapy introduced ATRA besides chemotherapy. Although the patient achieved once again CR, ATRA treatment effect could not be evaluated in vivo. In fact, the follow-up studies have demonstrated the patient’s leukemic cells were not sensitive to ATRA in vitro.143 A novel PML-RARG fusion was identified in a patient of APLL thereafter, and two kinds of fusion transcripts were detected. While the administration of ATRA continued for nine days, response could not be evaluated.4 Interestingly, cells carrying the artificial PML-RARG fusion protein responded well to retinoic acid in vitro. Recently, three cases with recurrent CPSF6-RARG fusions which were composed of three isoforms according to RARG break points were reported, and the main functional domains of RARG were retained. Unfortunately, they showed resistance to treatment with ATRA or ATRA plus arsenic. Two patients died after failing induction therapy and only one patient achieved CR benefiting from combined chemotherapy.65 Our patient also showed a resistance to ATRA at onset and at relapse. The patient received two cycles of combined chemotherapy before CR, and re-induction failed at relapse, all of which indicate drug resistance. Moreover, poor prognosis was observed likewise. Thus, it may be speculated that in comparison with RARA-rearranged APL, functions of RARG chimeric transcripts might be different. Combined chemotherapy should be administrated during induction therapy unhesitatingly rather than the alternate of ATRA or arsenic. Moreover, all the evidence above confirms that RARG rearrangements are not random but recurrent genetic abnormalities.

In conclusion, we present a novel HNRNPC-RARG fusion gene and its reciprocal in APLL, and suggest that at least a portion of APLL have RARG gene rearrangements. RARG-rearranged APLL may be a novel candidate subtype of APL.