Chimeric NUP98–NSD1 transcripts from the cryptic t(5;11)(q35.2;p15.4) in adult de novo acute myeloid leukemia

Abstract

The t(5;11)(q35;p15.4) is a clinically significant marker of poor prognosis in acute myeloid leukemia (AML), which is difficult to detect due to sub-telomeric localization of the breakpoints. To facilitate the detection of this rearrangement, we studied NUP98–NSD1 transcript variants in patients with the t(5;11) using paired-end RNA sequencing and standard molecular biology techniques. We discovered three NUP98–NSD1 transcripts with two fusion junctions (NUP98 exon 11-12/NSD1 exon 6), alternative 5′ donor site in NUP98 exon 7, and NSD1 exon 7 skipping. Two of the transcripts were in-frame and occurred in all t(5;11) samples (N = 5). The exonic splicing events were present in all samples (N = 23) regardless of the NUP98–NSD1 suggesting that these novel splice events are unassociated with t(5;11). In conclusion, we provide evidence of two different NUP98–NSD1 fusion transcripts in adult AML, which result in functional proteins and represent suitable molecular entities for monitoring t(5;11) AML patients.

Keywords:

• NUP98
• NSD1
• 11p15.5 translocation
• acute myeloid leukemia
• alternative splicing

Introduction

In de novo and therapy-related lymphoid and myeloid malignancies, 11p15.5 rearrangements with the nucleoporin-98-kDa (NUP98) gene involve more than 30 different partner genes [1–3]. Within the diverse spectrum of NUP98 rearrangements, the nuclear receptor-binding SET-domain protein 1 (NSD1) gene, located at 5q35, is the most commonly observed NUP98 fusion partner in acute myeloid leukemia (AML) [4,5]. The relative frequency of the NUP98–NSD1 fusion in cytogenetically normal pediatric and adult AML (CN-AML) was previously shown to be 16% and 2%, respectively. Beyond AML, NUP98–NSD1 has been reported in rare cases of myelodysplastic syndromes (MDS), and acute biphenotypic leukemia [6–8]. Several studies have observed that NUP98–NSD1 characterizes a clinically significant group of AML patients [4,8–11]. Furthermore, over 80% of NUP98–NSD1 patients co-express internal tandem duplications in the FLT3 gene (FLT3-ITD), and co-occurrence of these two aberrations has been shown to result in especially high induction failure and poor survival in AML [8,10].

At the chromosomal level, the t(5;11) juxtaposes the N-terminus of NUP98 to the C-terminus of NSD1 [12] (5′-NUP98–NSD1-3′), and occasionally the N-terminus of NSD1 to the C-terminus of NUP98 (5′-NSD1–NUP98-3′) [8]. The rearrangement leads to haploinsufficiency of the NUP98 and NSD1 genes, which participate in the nucleocytoplasmic transport of small molecules [13] and in chromatin-mediated transcriptional regulation [14], respectively. NSD1 is required for normal growth and development, thus haploinsufficiency of NSD1 results in Sotos syndrome [15] that is characterized by abnormal excessive growth [15]. Haploinsufficiency of NUP98 is associated with premature separation of sister chromatids, severe aneuploidy and ill-timed degradation of securin [16]. The NUP98–NSD1 fusion protein is known to accumulate in the nucleus [8,17] and transform hematopoietic precursors through epigenetic changes that prevent myeloid cell differentiation [18]. Thus far, two chimeric NUP98–NSD1 transcripts have been described in MDS, joining NUP98 exon 12 to NSD1 exon 6 and NUP98 exon 11 to NSD1 exon 6 [6]. While the NUP98 exon 12/NSD1 6 transcript is well-known aberration in AML, the sole expression of NUP98 exon 11/NSD1 exon 6 fusion transcript has only been described in one pediatric AML patient with a NUP98–NSD1-like gene expression signature [11]. To date, co-expression of these two NUP98–NSD1 transcripts in AML has not yet been reported.

Considering the sub-telomeric location of the breakpoints, detection of t(5;11) by traditional chromosome banding can be challenging. In addition, the potential for alternative spliced versions of the NUP98–NSD1 transcripts can complicate minimal residual disease monitoring. Application of newer technologies such as next-generation genome and RNA sequencing (RNA-seq) could overcome these challenges. By applying paired-end RNA-seq to samples from de novo AML patients and analyzing the data with a bioinformatic pipeline designed to detect fusion transcripts, we were able to efficiently detect NUP98–NSD1 transcripts in t(5;11) positive samples that included alternatively spliced versions with novel splice junctions. In this study, we show that whole transcriptome sequencing is an informative method for identifying clinically significant fusion transcripts and novel splice junctions present in t(5;11) AML patients that can be followed by targeted methods for molecular detection.

Materials and methods

Study patients

Bone marrow (BM) aspirates were collected after written informed consent from three adult de novo AML patients with NUP98–NSD1 fusion (NUP98–NSD1⁺), ten patients without NUP98–NSD1 (NUP98–NSD1⁻) and two healthy donors. The karyotypes of all study patients’ BM metaphase cells were obtained by standard G-banding. The present study was approved by the Local Institutional Review Board at the Helsinki University Central Hospital (permit numbers 239/13/03/00/2010, 303/13/03/01/2011, Helsinki University Hospital Ethics Committee) and conducted in accordance with the Declaration of Helsinki. The clinical and demographic characteristics of the NUP98–NSD1⁺ study patients are shown in Table 1.

Table 1. Clinical and demographic characteristics of the NUP98–NSD1⁺ patients.

Patient	Gender	Age	Disease	FAB subtype	Survival from Dg (days)	BM blast count at Dg (%)	WBC count at Dg (×E9/L)	B-neut at Dg (×E9/L)	Karyotype
600	Male	54	Acute monocytic leukemia	M5b	244	20	223.0	22.0	46 XY, t(5;11)(q35;p15.4) del (3q), t(12;15)
3600	Female	39	Acute myeloblastic leukemia with maturation	M2	1175^a	40	13.1	3.3	46 XX, t(5;11)(q35;p15.4)
3660	Male	58	Acute monocytic leukemia	M5	545	81	32.0	10.0	46 XY, t(5;11)(q35;p15.4) del (9)(q21–22)

Dg: diagnosis; FAB: French–American–British morphological classification of acute leukemias; WBC: white blood cell; B-neut: peripheral blood neutrophils.

^a The patient remains in complete remission.

Sample preparation

Bone marrow mononuclear cells (MNCs) were isolated by gradient centrifugation (Ficoll-Paque; GE Healthcare, Little Chalfont, UK). Total RNA was extracted from the BM MNCs with the RNeasy Mini Kit (Qiagen, Hilden, Germany) or Total RNA Purification Kit (Norgen Biotek, Thorold, Canada). Genomic DNA was isolated from BM MNCs using the DNeasy Blood & Tissue Kit or QIAamp DNA Mini Kit (Qiagen, Hilden, Germany). The DNA and RNA samples were quantified with the Qubit^® Fluorometer (Thermo Fisher Scientific, Carlsbad, CA, USA) and the quality of the RNA assessed using 2100 Bioanalyzer instrument and RNA Nano chips (Agilent Technologies, Santa Clara, CA, USA).

Array-based comparative genomic hybridization

Array-CGH was used to explore the presence of small genomic deletions that could indicate sites for DNA breakpoints associated with genomic rearrangements. For this, three micrograms of digested genomic DNA was labeled and processed as previously described [19]. Sample processing and hybridization to 244K Human CGH Microarrays (Agilent Technologies, Santa Clara, CA, USA) containing up to 1 × 10⁶ highly sensitive 60-mer oligonucleotide probes was performed according to a standard protocol from the manufacturer. The array-CGH data was analyzed with Genomic Workbench Standard Edition 5.0 (Agilent Technologies, Santa Clara, CA, USA).

RNA sequencing

One to three micrograms of DNase-treated (RNase-Free DNase Set; Qiagen, Hilden, Germany) total RNA (RIN >8) from each patient was sequenced. The RNA samples were depleted of ribosomal RNAs (Ribo-Zero rRNA Removal Kit; Epicentre, Madison, WI, USA), purified (RNeasy^® MinElute^® Cleanup Kit; Qiagen, Hilden, Germany), and reverse transcribed to double stranded cDNA using SuperScript^® Double-Stranded cDNA Synthesis Kit (Thermo Fisher Scientific, Carlsbad, CA, USA) with random hexamers (New England BioLabs, Ipswich, MA, USA). The RNA-seq libraries were prepared using Illumina-compatible Nextera technology (Epicentre, Madison, WI, USA), bar-coded, and enriched by ligation PCR according to the manufacturer’s instructions (Illumina, San Diego, CA, USA). PCR products were column purified and sequenced to an average length of 317 (104–530) nucleotides per mate using Illumina HiSeq™ 2000 instrument. Sequenced reads were extracted from TopHat v2.0.3 [20] alignment using in-house Perl scripts and executed in FusionCatcher [21]. Fusion gene supporting reads were aligned against a reference sequence and visually inspected in integrative genomics viewer (IGV) version 2.3.59 [22]. Initially, the sequenced FASTQ reads were aligned against human reference genome (HG18), but to improve the visualization and quantification of reads, we created FASTA format fusion constructs containing all exons of both partner genes. The 5′-NUP98–NSD1-3′ and 5′-NSD1–NUP98-3′ constructs were generated comparably by extracting NUP98 (NM 139131.3) and NSD1 (NM 022455.4) sequences from PubMed (NCBI nucleotide database), and by annotating them using a general feature format (GFF) annotation file. The GFF file was created using annotation files from GenBank (NCBI) and Perl scripts. The start and end exon positions for NUP98–NSD1 and the reciprocal NSD1–NUP98 were annotated accordingly. The GFF-file was needed for TopHat2 alignment between FASTQ reads and the fusion. After alignment, all reads (single and paired) aligned to the fusion were extracted and processed. PCR duplicates, non-primary reads, and reads with low mapping quality (>10) were filtered out before acquiring the read counts with SAMtools [23].

Cloning and Sanger sequencing

One microgram of total RNA was reverse transcribed to cDNA using SuperScript™ III First-Strand Synthesis System for RT-PCR (Thermo Fisher Scientific, Carlsbad, CA, USA) with Oligo (dT) primers. NUP98-NSD1 was amplified from the cDNA template using Phusion™ high-fidelity (HF) DNA polymerase (Thermo Fisher Scientific, Carlsbad, CA, USA), 5 × Phusion HF buffer, dNTP mix (10 mM each), and oligonucleotide primers: NUP98 5′-ATGTTTAACAAATCATTTGG-3′ and NSD1 5′-CTACTTCTGTTCTGATTCTG-3′. The cycling conditions for PCR were as follows: an initial denaturation at 95 °C for 2 min, 35 cycles of denaturation for 20 s at 95 °C, annealing for 10 s at 54 °C, and elongation for 100 s at 72 °C with final extension time of 5 min. PCR fragments were separated on 1% agarose gel in the presence of SYBR^® Safe DNA Gel Stain (Thermo Fisher Scientific, Carlsbad, CA, USA) and gel-purified using NucleoSpin^® Gel and PCR Clean-up Kit (Macherey-Nagel, Duren, Germany). The purified fragments were cloned into the pCR^® 2.1-TOPO^® vector and introduced into One Shot™ Stbl3™ Chemically Competent Escherichia coli cells according to TOPO^® TA-cloning protocol (Thermo Fisher Scientific, Carlsbad, CA, USA). NUP98–NSD1⁺ colonies were identified with colony-PCR, and propagated overnight in Luria broth supplemented with ampicillin (100 μg/ml). The plasmids were extracted with the NucleoSpin^® Plasmid Easy Pure Kit (Macherey-Nagel) and sequenced. Bidirectional Sanger sequencing using oligonucleotide primers (Table S2) spanning across the full-length NUP98–NSD1 was performed in ABI3730xl DNA Analyzer using BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, South San Francisco, CA, USA). Base calling was done using Sequencing Analysis v5.2 software, while the generated trace files were inspected with Sequencher™ 5.2 (Gene Codes Corporation, Ann Arbor, MI, USA) and BLAST algorithm (National Center for Biotechnology Information [NCBI] nucleotide database). The NUP98 (NM 139131.3) and NSD1 (NM 022455.4) sequences were used as reference for the alignment.

Statistical analysis

Statistical analysis was performed with Prism software version 6.0 (GraphPad Software, La Jolla, CA, USA). The novel splice events were compared between NUP98–NSD1⁺ versus NUP98–NSD1⁻ samples using unpaired nonparametric Mann–Whitney rank comparison test. The t-tests between groups were two-tailed, and exact p values below .05 were considered statistically significant.

Results

Detection of NUP98–NSD1 transcripts and exonic rearrangements

Array-CGH showed that the index patient carried narrow cryptic deletions within NUP98 and NSD1 genes at positions 11p15.4 and 5q35.2, respectively (Figure S1). The 0.11 Mb (chr11: 3,617,261–3,723,126) and 0.05 Mb (chr5: 176,546,244–176,595,458) microdeletions indicated unbalanced t(5;11)(q35.2;p15.4)/NUP98–NSD1 translocation. The chromosomal breakpoints putatively occurred within one oligonucleotide probe length (60-mer) from positions chr11: 3,723,126 and chr5: 176,595,458 (NCBI36/HG18). To verify the NUP98–NSD1 fusion sequence, we cloned it from the index patient’s leukemic cells and identified two in-frame (NUP98–NSD1_v1-v2) and one out-of-frame fusion (NUP98–NSD1_v3) transcripts with mRNA lengths of 5,676, 5,562, and 5,432 base pairs (GenBank^® accession numbers: KU695532, KU695533, KU695534). The Sanger sequenced fusion transcripts revealed fusion junctions between NUP98 exon 12/NSD1 exon 6 and NUP98 exon 11/NSD1 exon 6, alternative 5′ donor site in NUP98 exon 7 (nucleotides 757-783), and NSD1 exon 7 skipping (nucleotides 3922-4192) (Figure 1(A–C)). The nucleotide positions were counted from the first nucleotide of ATG start codon in the corresponding NCBI reference sequence (NM 139131.3 and NM 022455.4).

Figure 1. NUP98–NSD1 transcripts detected from the index patient. (A) Exon structures of NUP98 (NM 139131.3) and NSD1 (NM 022455.4), plotted with GenomeGraphs version 3.1 [26]. Bar thickness represents the exon size in base pairs and line lengths the distance between exons. (B) The schematic illustrates three NUP98–NSD1 transcripts detected by sequence analysis of plasmids with full-length NUP98–NSD1. (C) The fluorescent peak trace chromatograms revealed two fusion junctions and alternatively spliced exonic regions. (D) The cDNA sequences were translated using EMBOSS Transeq and analyzed with Simple Modular Architecture Research Tool (SMART) [27]. The structural domains and repeats, including FG- (red), FGFG- (green), GLFG- (blue), and GLFGFG-repeats (orange) were drawn to scale using MyDomains-Image creator. The domain structure abbreviations are: Drosophila Su(var)3-9 and ‘Enhancer of zeste’ proteins (SET), post-SET domain (PS), associated with SET domain (AWS), coiled-coil domain (CC), plant homeodomain (PHD), and Gle-2 binding sequence (GLBS).

Nucleotide to amino acid conversion (Figure 1(D)) indicated that the phenylalanine (F), glycine (G), and leucine (L) possessing nontandem repeats and Gle2p-binding (GLEBS)-like motif at the NH₃ terminus of NUP98 were consistent and functional in all three NUP98–NSD1 transcripts. Each fusion transcript had eight GLFG-repeats, FGFG-repeat, GLFGFG-repeat, and non-tandem FG-repeats. The FG-repeat counts in NUP98–NSD1_v1, NUP98–NSD1_v2, and NUP98–NSD1_v3 were 38, 36 and 39, respectively. The C-terminal functional domains of NSD1 within the hybrid proteins remained unchanged in all except in NUP98–NSD1_v3, which was truncated due to exon skipping related frame shift.

We re-analyzed the RNA-seq data from patient 600 using a modified alignment strategy and confirmed that the novel splice events identified through cloning and Sanger sequencing were neither PCR-introduced artifacts nor RNA-seq library preparation errors (Figure 2). In sample 600_2, 93% (75/81) of the NUP98–NSD1 supporting reads supported the fusion between NUP98 exon 12 and NSD1 exon 6 (NUP98–NSD1_v1), and 7% (6/81) the NUP98 exon 11/NSD1 exon 6 fusion (NUP98–NSD1_v2). The percentages were similar (91% and 9%) in the relapse sample 600_3 collected 2 months later. The reciprocal 5′-NSD1–NUP98-3′ was undetectable in both samples.

Figure 2. IGV views of two NUP98–NSD1 fusion junctions and alternatively spliced regions in the index patient detected by paired-end RNA sequencing. The figure shows supporting FASTQ-reads for fusion junctions between NUP98 exon 11/NSD1 exon 6, and between NUP98 exon 12/NSD1 exon 6 (A), alternative 5’ donor site of NUP98 exon 7 (B), and exon skipping of NSD1 exon 7 (C). The color bars on top represent different nucleotides: A (green), C (blue), T (red), and G (orange). The middle panel (grey) show RNA-sequencing data coverage. NUP98 (NM 139131) is transcribed from the – strand, and NSD1 (NM 022455) from the + strand.

Validation of fused exons and exonic rearrangements with paired-end RNA seq

We performed RNA-seq on additional samples with (N = 4) and without (N = 16) t(5;11) to acquire more evidence for chimeric NUP98–NSD1 fusion transcripts and exonic rearrangements. As shown in Figure 3(A), all positive samples co-expressed the two in-frame NUP98–NSD1 transcripts. In these samples, NUP98–NSD1_v1 was the predominant transcript, however, 11% (95% confidence interval 6.0–16.6%) of the NUP98–NSD1 supporting FASTQ reads supported the NUP98–NSD1_v2 fusion. Its relative frequency increased during disease progression from 9.4% (9/96) to 18.1% (13/72) in patient 3660 and from 7% (6/81) to 9% (7/78) in patient 600. Interestingly, NSD1 exon 7 skipping and alternative 5′ donor site of NUP98 exon 7 were detected from all study samples regardless of NUP98–NSD1 (Figure 3(B,C)). In the NUP98-NSD1⁺ group, 8% (95% CI 5.3–9.9%) of the reads supported an alternative 5′ donor site of NUP98 exon 7, while it was 6% (95% CI 5.3–7.4%) in the negative group (OR 1.183). In both groups, 6% (95% CIs 0–13.1% and 3.4–9.4%) of the reads spanning across NSD1 exon 7 supported its skipping. No statistically significant differences in NSD1 exon 7 skipping or alternative 5′ donor site was found between the NUP98–NSD1⁺ and NUP98–NSD1⁻ groups (p-values .33 and .38) suggesting these are previously unknown normal splicing events of the wild type NUP98 and NSD1 genes.

Figure 3. RNA-seq supporting read counts for the NUP98–NSD1 fusion junctions and splicing events. (A) The stacked bars show total number of reads supporting the two NUP98–NSD1 fusion junctions in five t(5;11)(q35.2;p15.4) positive samples from three adult de novo AML patients. The scatter plots visualize the percentage of RNA-seq reads supporting the alternative 5′ donor site of NUP98 exon 7 (B) and NSD1 exon 7 skipping (C) in NUP98–NSD1⁺ (N = 5) and NUP98–NSD1^- patient samples (N = 16). The error bars illustrate the mean percentage of reads with 95% confidence interval.

Discussion

In this study, our objective was to facilitate the molecular detection of NUP98–NSD1 by investigating NUP98–NSD1 fusion transcripts in adult de novo AML. We acquired experimental evidence of three chimeric NUP98–NSD1 transcripts and demonstrated that the two in-frame transcripts previously reported in MDS [6] are also co-expressed in adult AML. The co-expression was found in serial samples from two different patients, indicating that both NUP98–NSD1 transcripts are clonally stable during disease progression. The clonal stability and clinical relevance of NUP98–NSD1 is well established [8,9]. The absence of reciprocal NSD1–NUP98 in our study patients is in agreement with previous studies, which have suggested NUP98–NSD1 as the sole initiating oncoprotein [6,24]. Our results showed that NUP98–NSD1_v1, joining NUP98 exon 12 to NSD1 exon 6, is the predominant transcript. However, an alternative transcript NUP98–NSD1_v2 with NUP98 exon 11 joined to NSD1 exon 6 was also detected in all t(5;11) samples. Nevertheless, both fusion transcripts result in in-frame fusion proteins that retain the same functional domains of NUP98 and NSD1. Additional studies are required to assess the functional activity of the different transcript variants.

Increased NUP98–NSD1 expression has previously been shown to correlate with hematological relapse [8,9]. In the largest AML cohort screened for NUP98-NSD1 thus far [10], most patients remained positive for the fusion at the end of cycle 1 of induction chemotherapy (EOI-1), which strongly indicates that NUP98–NSD1 is an important tool for evaluating treatment response and MRD, either after chemotherapy or stem cell transplantation. As recommended previously by Grimwade and colleagues [25], sequential RT-qPCR monitoring coupled with preemptive therapy may eventually lead to individualized management of patients with rare gene fusions such as NUP98–NSD1. Due to emerging clinical significance of NUP98–NSD1, especially among pediatric AML patients [8], it is advisable that all newly diagnosed AML patients without favorable genetic abnormalities should be screened upfront for t(5;11) by fluorescence in situ hybridization (FISH) and followed by monitoring of NUP98–NSD1 transcripts by RT-PCR. Here, we show that RNA-seq is a sensitive method to detect NUP98–NSD1 fusion transcripts as well as the different fusion splice variants. Although data from additional t(5;11) patients should be analyzed, our initial findings suggest that the fusion joining exon 11 of NUP98 to exon 6 of NSD1 should be included to the molecular diagnostics panel in AML. This is supported by a previous study, which indeed detected the sole expression of NUP98 exon 11/NSD1 exon 6 fusion in a pediatric AML patient with NUP98–NSD1-like gene expression signature [11]. In addition, as next generation sequencing applications are incorporated into clinical laboratory diagnostic practice, cryptic translocations such as NUP98–NSD1 should be more readily detected. To address the therapeutic needs of NUP98–NSD1⁺ myeloid malignancies, further intensive studies of patients with t(5;11) are warranted.

Potential conflict of interest

Disclosure forms provided by the authors are available with the full text of this article online at https://doi.org/10.1080/10428194.2017.1357174.

Acknowledgements

The authors would like to thank the patients and healthy donors who participated in the study. We also thank personnel of the FIMM Sequencing Lab and Technology Center, including Pekka Ellonen, Aino Palva, Pirkko Mattila, Sari Hannula, Aino Vaittinen, and Henrikki Almusa for their technical assistance and expertise. We express our appreciation to technicians Minna Suvela and Siv Knaappila for their help with sample processing.

Additional information

Funding

The study was funded by Tekes: Finnish Funding Agency for Innovation [40336/09], Syöpäsäätiö, and the European Regional Development Fund [A31859]. Personal grant support was received from the Väre Foundation for Pediatric Cancer Research and Emil Aaltonen Foundation (JK).