Discovery and characterization of a novel CCND1/MRCK gene fusion in mantle cell lymphoma

The t(11;14) translocation resulting in constitutive cyclin D1 expression is an early event in mantle cell lymphoma (MCL) transformation. Patients with a highly proliferative phenotype produce cyclin D1 transcripts with truncated 3′UTRs that evade miRNA regulation. Here, we report the recurrence of a novel gene fusion in MCL cell lines and MCL patient isolates that consists of the full protein coding region of cyclin D1 (CCND1) and a 3′UTR consisting of sequences from both the CCND1 3′UTR and myotonic dystrophy kinase-related Cdc42-binding kinase's (MRCK) intron one. The resulting CCND1/MRCK mRNA is resistant to CCND1-targeted miRNA regulation, and targeting the MRCK region of the chimeric 3′UTR with siRNA results in decreased CCND1 levels. Electronic supplementary material The online version of this article (doi:10.1186/s13045-016-0260-7) contains supplementary material, which is available to authorized users.


Findings
Mantle cell lymphoma (MCL) is considered incurable upon relapse [1]. The hallmark of MCL is the t(11;14)(q13;q32) translocation [2] which results in the constitutive expression of cyclin D1 protein despite the short half-life (~30 min) of its cyclin D1 (CCND1) transcript [3]. The CCND1 mRNA has a long 3′UTR (~3.1 Kb) that contains numerous destabilizing elements [4,5]. MCL patients with highly proliferative tumors express CCND1 transcripts with truncated 3′UTRs correlating with reduced survival [6]. In some MCL patients, 3′UTR shortening is due to single nucleotide polymorphisms (SNPs) that result in the generation of an optimal proximal polyadenylation signal (pPAS) [6]. In other cases, no such mutations have been observed, suggesting that alterations in the activity of the cleavage and polyadenylation machinery is responsible. This process, termed alternative polyadenylation (APA) [7], has been the focus of much attention as cancer cells have shown a global tendency to utilize polyadenylation sites proximal to the stop codon to reduce the activity of microRNAs [8]. Aside from SNPs and APA, genomic deletions that result in shortened 3′UTRs have not been wellexplored and are the focus of this study [6].
We confirmed aberrantly high cyclin D1 protein expression in all three MCL cell lines compared to a control B lymphocyte cell line, RPMI-1788 (Fig. 1a).
To determine the state of the 3′UTR, we mapped the 3′end of CCND1 mRNA. We observed 3′rapid amplification of cDNA end (3′RACE) products in all the MCL lines that would indicate that 3′UTR shortening has occurred (Fig. 1b). Jeko-1 utilizes a non-consensus and unmutated pPAS (AATAAT) (Additional file 1). Interestingly, mutations in this same genomic region creating a consensus pPAS (AATAAA) have been observed in 3/15 MCL patients [6]. Hence, in Jeko-1, APA is the likely cause of CCND1 3′UTR shortening to allow for the use of a non-optimal pPAS.
Unexpectedly, in Granta-519 and SP-53, the 3′ RACE products were slightly larger than that observed from Jeko-1. Sequencing revealed that, in both cases, the 3′UTR of CCND1 is fused to the reverse complement of intronic sequences present in the myotonic dystrophy kinase-related Cdc42-binding kinase (MRCK) gene (Fig. 1c). Placement of this genomic region within the CCND1 3′UTR results in the use of a consensus polyadenylation signal (PAS) from MRCK, which triggers the subsequent addition of the poly(A) tail, creating a chimeric 3′UTR. The observed CCND1/MRCK fusion gene likely is formed by a second translocation event between chromosomes 11 and 14, which positions the full open reading frame of CCND1 and a truncated 3′UTR within intron one of MRCK (Fig. 1d).
The presence of the CCND1/MRCK chimeric mRNA was validated using chimera-specific primers (Additional file 1). Quantitative polymerase chain reaction (qRT-PCR) shows that MCL cell lines express at least twice as much CCND1 mRNA as other cancer cell lines (Fig. 2a). Surprisingly, we were able to detect the CCND1/MRCK fusion product in 8 out of 13 MCL patient DNA samples (Fig. 2b). These results suggest that this translocation event may be selected for in MCL and functions as a mechanism to shorten the CCND1 3′UTR.
StarBase analysis [9] identified 86 miRNAs that can potentially interact with the CCND1 3′UTR (Additional file 2). To determine the impact of the CCND1/MRCK fusion on miRNA regulation of CCND1, we generated a reporter plasmid with the chimeric 3′UTR placed downstream of luciferase. We also made constructs containing either the full-length (FL-CCND1) or the truncated CCND1 3′UTR from Jeko-1 (Tr-CCND1) and co-transfected them with three miRNA mimics known to repress CCND1. The FL-CCND1 reporter was downregulated in response to each mimic tested (Fig. 2c). However, all three mimics had no significant effect on either the Tr-CCND1 or CCND1/MRCK reporters. This observation parallels reports in glioblastoma where a recurrent FGFR3/TACC3 causes the truncation of the FGFR3 3′UTR, resulting in elimination of miRNA binding sites [10]. Intriguingly, transfection of siRNA-targeting of the MRCK component of the chimeric 3′UTR causes reduced cyclin D1 protein expression as effectively as siRNA targeting the CCND1 sequences (Fig. 2d).
These results provide an enhanced understanding of the strength of selection for CCND1 3′UTR shortening that occurs in MCL. In addition, the novel CCND1/MRCK gene fusion identified is a potential new diagnostic and/or therapeutic target for MCL.

Patient samples and sample preparation
All the samples from MCL patients were obtained as deidentified DNA from the Indiana Biobank, a biorepository run by Indiana University. These patients had all been diagnosed with MCL with ICD-9 codes from 200.40 to 200.48. All specimens were collected by Indiana Biobank from patients after informed consent, as approved by the Indiana University Institutional Review Board. PCR was done using Phusion DNA Polymerase (New England Biolabs) per the manufacturer's protocol. HF Phusion buffer was used together with MgCl2. Primers were designed to provide the largest DNA PCR product possible, and were validated to work with genomic DNA. The primer sequences used were CCND1 forward 5´TCCG

mRNA extraction, 3'RACE, PCR and qRT-PCR
Total mRNA was extracted using TRIzol reagent (Life Technologies). After DNase (Promega) treatment, cDNA synthesis was carried out using M-MLV Reverse Transcriptase (Life Technologies). For 3´RACE, oligo(dT25)T7 primer was used in cDNA synthesis, and the first round of PCR was performed using the CCND1 primer 5´TGGTGAACAAGCTCAAGTGG and oligo(dT25)T7. The second round of PCR was performed using a nested CCND1 primer 5´TGGCATTTTGGA-GAGGAAGTG and a T7 primer. All PCR was performed using pfu polymerase. The resulting PCR product was cloned using Zero Blunt TOPO PCR Cloning Kit for Sequencing (Life Technologies) and sequenced (Lonestar Labs, TX). The qRT-PCR protocol used, as well as the amplicons used to measure 7SK and CCND1, were previously described [8]. The primers used to detect CCND1-MRCK were the CCND1 forward primer 5´GAGGAGGAAGAGGAGGAGGAGGAGGT and the MRCK reverse primer 5´TCCAATTCTGCTA GACCTTTGTGATA. Primers used to detect the control 18S rRNA were 5´CAGCCACCCGAGATTGAGCA and 5´TAGTAGCGACGGGCGGTGTG.

RNAi and luciferase assays
Either control siRNA [8], CCND1-specific siRNA (SASI Hs 01-00213909 from Sigma), or MRCK target-specific siRNA, (5´GAUCCGGAACGAACUA GAGTT) were electroporated into MCL cells using the Nucleofactor Kit R (Amaxa), following the manufacturer's protocol (program A-023 on Nucleofactor II device). Warm cell culture media (500ul) was added after electroporation, and the cells were replated and harvested for Western blot 72 hours later. To conduct luciferase assays, HeLa cells were plated (4.5x104 per well) in a 24 well plate. After 24 hours, cells were transfected with 100ng of microRNA mimics using Lipofectamine 2000 (Life technologies). The MISSION microRNA mimics used were: miR-15a, miR-19a, miR-202, and the miRNA mimic negative control #2 (Sigma). After 24 hours, the cells were then transfected with 50ng of psi-check-2 dual luciferase plasmid hours. The cells were lysed one day after plasmid transfection, and luciferase assays were performed using the Dual-Luciferase Reporter Assay System (Promega).

Western blot
Cells were lysed using RIPA buffer and 20ug of each protein sample was resolved on a 12% SDS-PAGE gel. After transfer to a PVDF membrane, blocking was done for 1hr in 5% non-fat milk resuspended in PBS (+0.001 % Tween 20). The membranes were probed with the monoclonal cyclin D1 antibody (DCS6 from Cell Signaling) and polyclonal alpha tubulin antibody (Abcam). After probing with HRP-conjugated secondary antibody, proteins were detected by luminol.

miRNA identification
We used predicted CCND1-miRNA interactions using starBase software, which consolidates TargetScan, Pic-Tar, PITA, Miranda and RNA22 miRNA prediction softwares, and overlaps the data with CLIP-Sequencing (CLIP-Seq.) data. We used a high stringency cutoff, where only miRNAs supported by >/=3 CLIP-Seq. experiments were selected, to reduce false positives [9].