Identification of novel recurrent CPSF6-RARG fusions in acute myeloid leukemia resembling acute promyelocytic leukemia

TO THE EDITOR:

Retinoic acid receptor γ (RARG) is a member of the nuclear receptor superfamily and shares high homology (90%) with retinoic acid receptor α (RARA) and retinoic acid receptor β (RARB).¹  So far, little is known about RARB or RARG fusion. Such et al reported the first case of RARG fusion in a male patient resembling classical acute promyelocytic leukemia (APL).²  The partner gene of RARG was identified as NUP98.²  Recently, Ha et al identified the PML gene as the second partner gene in a female APL patient.³  Here, we present the first recurrent RARG fusions in 2 acute myeloid leukemia (AML) patients mimicking APL.

Patient 1, a 48-year-old woman, was admitted because of dizziness, fatigue, and hypermenorrhea. Blood tests showed a hemoglobin level of 42 g/L, a platelet count of 92 × 10⁹/L, and a white blood cell count of 0.81 × 10⁹/L. Fibrinogen and d-dimer levels were 1.67 g/L (reference, 2.00-4.00 g/L) and 56.8 μg/mL (reference, 0.00-0.55 μg/mL), respectively. Prothrombin time and partial thromboplastin time were 13.5 seconds (reference, 10.5-13.0 seconds) and 25.3 seconds (reference, 23-35 seconds), respectively. Bone marrow (BM) smear showed hypercellularity, with 89% hypergranular promyelocytes with Auer rods (Figure 1A). The blasts were positive for CD13, CD33, and myeloperoxidase, partially positive for CD9 and CD64, but negative for HLA-DR, CD117, CD34, CD14 and CD11b by flow cytometry (supplemental Figure 1A, available on the Blood Web site). This patient was diagnosed with suspicion of APL. A BM sample obtained at diagnosis was processed after a short-term culture (24 hours) following standard RHG banding procedures. Fluorescence in situ hybridization (FISH) analysis was performed using a PML-RARA dual-color dual-fusion probe (Abbott Molecular, Des Plaines, IL) according to the manufacturer’s protocols (Figure 1B). Multiplex quantitative reverse transcription polymerase chain reaction (RT-PCR) was performed to detect 51 fusion transcripts, including PML-RARA, PLZF-RARA, NUMA1-RARA, STAT5B-RARA, PAKARIA-RARA, and NPM1-RARA. However, the t(15;17)(q24;q21) translocation was not detected by karyotyping; instead, a tetraploidy karyotype of 92, XXXX[2] was identified (Figure 1C). Both RT-PCR and FISH failed to detect the PML-RARA fusion transcript (Figure 1B). We performed targeted next-generation sequencing of the entire coding sequences of 382 known or putative mutational gene targets in hematologic malignancies and identified DNMT3A-G587fs mutation in this patient. She was initially treated with all-trans retinoic acid (ATRA) (30 mg/d, days 1-35) and arsenic trioxide (10 mg/d, days 1-15) combined with idarubicin (6 mg/m² per day, days 9, 13, and 14) and showed no response. She then received therapy with a course of idarubicin (6 mg/m² per day, days 1, 3, and 5), cytosine arabinoside (12 mg/m², hypodermic injection, every 12 hours, days 1-14), and granulocyte colony-stimulating factor, 250 μg, hypodermic injection, daily) combined with ATRA (30 mg/d, days 1-14) and arsenic trioxide (10 mg/d, days 1-14) and failed to achieved remission. Afterward, she received an induction therapy of decitabine (20 mg/m², days 1-5). Unfortunately, BM smear showed no response. She refused further chemotherapy and died of cerebral hemorrhage in July of 2017.

Patient 2, a 51-year-old woman, was admitted because of fever, chest pain, and paraphasia. Blood tests showed a hemoglobin level of 65 g/L, a platelet count of 45 × 10⁹/L, and a white blood cell count of 20.15 × 10⁹/L. Fibrinogen, fibrin degradation products, and d-dimer levels were 1.66 g/L (reference, 2.00-4.00 g/L), 341.2 μg/mL (reference, 0.00-5.00 μg/mL), and 189.4 μg/mL (reference, 0.00-0.55μg/mL), respectively. Prothrombin time and partial thromboplastin time were 13 seconds (reference, 10.5-13.0 seconds) and 35.2 seconds (reference, 23-35 seconds) respectively. BM smear showed hypercellularity with 87.5% hypergranular promyelocytes (Figure 1D). The blasts were positive for CD13, CD33, cytoplasmic myeloperoxidase, and CD9, partially positive for CD34, but negative for HLA-DR, CD2, CD7, CD10, CD11c, CD14, and CD38 by flow cytometry (supplemental Figure 1B). Both quantitative RT-PCR and FISH failed to detect the PML-RARA fusion transcript from the BM sample (Figure 1E). The t(15;17)(q24;q21) was not detected by karyotyping; instead, a del(12)(p12)[2]/46,XX[18] was detected (Figure 1F). Targeted next-generation sequencing identified WT1 and K-RAS mutations in this patient. She was initially treated with ATRA (25 mg/m²/ d1-d28, 15 mg/m², d29-d42) and daunorubicin (60 mg/m², days 1-3). Although the coagulation function returned to normal, BM aspiration at days 14 and 42 showed no response. She then received a course of daunorubicin and Ara-C chemotherapy (daunorubicin 60 mg/m², days 1-3, and Ara-C 100 mg/m², days 1-7) and achieved morphologic remission. The patient received 2 courses of high-dose cytarabine consolidation therapy followed by 2 courses of standard 7+3 chemotherapy. She remained complete remission until the last follow-up in November of 2017.

Cytogenetic, FISH, and RT-PCR analysis demonstrated the absence of t(15;17)(q24;q21) and PML/RARA in both patients. In order to characterize the molecular aberrations, we performed RNA sequencing on the total RNA of BM samples and found a recurrent CPSF6-RARG fusion in both patients (supplemental Figure 2). Whole-genome sequencing analysis results revealed that the breakpoint in 12q15 were located at the intron 4 of CPSF6 in both patients (Figure 1G). There are 2 breakpoints in the intron 3 or 5′ untranslated region and telomeric of exon 9 of RARG (Figure 1G). The 3′ region of RARG (from exon 1 or exon 4 to exon 9) was reversed and fused in-frame with the 5′ region of CPSF6 gene (from exon 1 to exon 4) (Figure 1G). RT-PCR and Sanger sequencing analysis confirmed CPSF6-RARG in-frame fusion in both patients (Figure 1H-J). The longer transcription, named CPSF6-RARG-L, harbored a point mutation from 805G to C in patient 2, resulting in a change from glycine to arginine at 269 (Figure 1J). The CPSF6-RARG fusion protein in both patients combines the RNA recognition motif domain of CPSF6 and the main RARG domains of DBD and LBD (Figure 1J). Compared with NUP98-RARG¹  and PML-RARG,²  the breakpoint of RARG in patient 1 is consistent with NUP98-RARG, and the breakpoint of RARG in patient 2 is the same as the PML-RARG transcript.

To compare the cellular localization of CPSF6-RARG fusion protein with wild-type RARG and CPSF6, myc-tagged versions of these proteins were expressed in HeLa cells. Immunofluorescence analysis showed that myc-RARG and myc-CPSF6 have a diffused distribution in nucleus, whereas CPSF6-RARG-L and CPSF6-RARG-S exhibit a similar intranuclear distribution (Figure 2A). Furthermore, we examined the transcriptional properties of CPSF6-RARG-L and CPSF6-RARG-S. RARE luciferase reporter experiments were performed. Compared with RARG, CPSF6-RARG-L repressed the expression of luciferase reporter to a level comparable to PML-RARA, whereas CPSF6-RARG-S showed a comparable transcriptional activity with RARA or RARG (Figure 2B). In the presence of ATRA, both CPSF6-RARG fusions and RARG showed weak luciferase induction, which was in marked contrast with the significant luciferase induction by RARA (Figure 2C). These results indicated that both CPSF6-RARG fusions may exert similar transcriptional effects on RARG downstream targets.

It was reported that an artificial PML-RARG fusion showed an oncogenic potential comparable to that of PML-RARA.^4,5  Therefore, CPSF6-RARG might be assumed to have similar oncogenic functions. The partner gene also plays a crucial role in the biological properties of fusion proteins. CPSF6 is one subunit of a cleavage factor required for 3′ RNA cleavage and polyadenylation processing. CPSF6 and CPSF5 form a protein complex binding to RNA substrates that promotes RNA looping.⁶ CPSF6 was reported to be fused with PDGFRB in a patient with myeloproliferative neoplasm with eosinophilia⁷  and FGFR1 in a patient with 8p11 myeloproliferative syndrome.⁸ 

In summary, we identified novel CPSF6-RARG fusions in 2 patients with AML resembling APL. This is the first report of a recurrent fusion transcript involving the RARG gene. It will be necessary to conduct further studies to determine the prevalence and leukemogenic mechanisms of CPSF6-RARG fusion in AML mimicking APL.