A donor splice site mutation in CISD2 generates multiple truncated, non-functional isoforms in Wolfram syndrome type 2 patients

PBMCs were isolated from the whole venous blood of the healthy donator, patients and parents by density gradient centrifugation using Histopaque (Sigma-Aldrich, Milan, Italy), according to the manufacturer’s instructions. Written informed consent was obtained from all subjects enrolled in the study.

Total RNA was extracted using miRNeasy mini Kit (Qiagen, Milan Italy) and reverse transcribed with Superscript III reverse transcriptase (Invitrogen, Milan, Italy), according to manufacturer’s instructions.

Semi-quantitative PCR amplifications were performed using GoTaq DNA Polymerase (Promega, Milan, Italy). The PCR conditions were: denaturation at 95 °C for 2 min, followed by 24 (for actin B, ActB) and 35 (for CISD2) cycles at 95 °C for 30 s, 60 °C for 40 s, and 72 °C for 60 s.

Primer Sequences:

The SMARTer RACE cDNA Amplification kit (Clontech, California, USA) was used to extend the CISD2 mRNA in the PBMCs from the patients, according to the manufacturer’s instructions. The synthesis of the first strand cDNA was carried out with total RNA (1 μg) using a 5′ CDS primer A, SMART II A oligonucleotide and SMARTScribe Reverse Transcriptase. The 5′-RACE PCR reaction was performed with the high fidelity enzyme (SeqAmp DNA polymerase, Clontech) using a universal primer A mix (UPM, forward primer) and an antisense gene specific primer (GSP1) designed against the exon 3 of the CISD2. The nested 5′-RACE PCR was carried out using a specific reverse primers (GSP2) designed inner to the exon 3 of CISD2. The primer sequences are noted below (the bold sequences referred to the 15 nucleotides overlapped withpUC19 vector suitable for the in fusion cloning). The 5′-RACE products were separated by 2% agarose gel electrophoresis and the resulting bands were extracted from the gel, purified using NucleoSpin gel-PCR Clean Up Kit (Clontech), and cloned into pUC19. The positive clones that contained the inserts of the expected size were screened by colony PCR using M13F and M13R primers and sequenced by the Sanger method (BigDye Terminator v 3.1) and 3730XL DNA Analyzer (Applied Biosystems).

Primer Sequences:

The genomic DNA was isolated from buccal swab using QiAamp DNA Mini Kit (Qiagen, Milan, Italy). The genomic region flanking SNP was amplified with the high fidelity enzyme (SeqAmp DNA polymerase, Clontech) using primers noted below. The PCR products were sequenced by the Sanger method (BigDye Terminator v 3.1) and 3730XL DNA Analyzer (Applied Biosystems).

Primer sequences:

PBMCs were lysed in 50 mM Tris HCl (PH 7.6), 150 mM NaCl, 1% Nonidet P-40 containing protease and phosphatase inhibitor cocktail (Sigma). The protein concentration was determined by Bradford assay (Thermo Scientific, Milan, Italy). The total proteins were resolved on 14% SDS-Polyacrylamide gel, blotted onto PVDF membrane (GE Healthcare, Euroclone, Pero, Milan, Italy), and probed with the antibodies anti-CISD2 (1:1000, Proteintech, LaboSpace Srl, Milan, Italy) and β-tubulin (1:2000, Sigma). Proteins were detected with anti-rabbit secondary antibodies (1:100,000, GE Healthcare) and anti-mouse (1:10,000, GE Healthcare). Filters were developed with western Sure premium Chemiluminescent Substrate LI-COR using LI-COR instrument. β tubulin was used as normalizer.

Statistically significant differences between samples were determined in all the experiments by one-way ANOVA analyses; *, **, *** and **** indicated p < 0.05, p < 0.01 and p < 0.001, p < 0,0001, respectively. A p value < 0.05 was considered statistically significant.

It was previously predicted that a homozygous CISD2 mutation in the donor splice site of intron 1 (NM_001008388.4:c.103 + 1G > A) caused exon 1 to be skipped in two Italian sisters affected by WFS2 [3]. To analyse the effect of the c.103 + 1G > A mutation on the mRNA, a reverse transcription polymerase chain reaction (RT-PCR) assay covering almost the entire cDNA was performed on RNA extracted from PMBCs derived from the patients (1 and 2) and the heterozygous unaffected parents (3 and 4), respectively. The PBMCs derived from healthy donor were also included as a control (CTR). Overlapping primers designed against each of the three exons of the CISD2 gene were coupled with two primers designed against the 3′-UTR and exon 3, allowing for the dissection of four fragments representing progressive 5′ deletions (Fig. 1a and b). The primers designed against exons 2 and 3 amplified the expected PCR products of 991 and 790 nt in all samples, although very faint levels were detected in the patients (Fig. 1c). As predicted, no amplification was observed in patients using the primers to amplify the CISD2 region from exon 1 to exon 3, while PCR products of the expected size (1031 and 227 nt) were obtained in the remaining samples (Fig. 1c). The quantitative PCR analysis performed with the primer set designed against exons 2 and 3, confirmed the reduced amounts of CISD2 mRNA amplified in the patients compared with those obtained in either the parents or the healthy control (less than 99%) (Fig. 1d), suggesting the strong instability of the mutant mRNA. A decrease in mRNA levels of approximately 64% was also observed in the heterozygous samples compared to the mRNA levels of the control (Fig. 1d). Finally, the primer set designed to amplify exon 1 validated a decrease of CISD2 mRNA of approximately 65% in the parents and undetectable signals in the homozygous samples (Fig. 1d).

Together, the results indicate a potential skipping of exon 1 in the transcripts of the patients, which severely affects the mRNA stability.

To validate the putative exon 1 absence demonstrated by RT-PCR, a 5′-RACE experiment was performed on total RNA derived from PBMCs of the patients and the healthy control. A gene-specific primer (GSP1) designed against exon 3 and a forward universal primer (UPM) were used for the PCR step that followed the cDNA synthesis in the 5′-RACE process. The expected PCR product of approximately 500 nt was extended in the healthy control, while smeared signals were detected in both homozygous samples, reinforcing the very low amount of mutant mRNA (data not shown). To increase specificity and sensitivity, a nested PCR with an inner GSP2 primer was performed on the primary PCR products which resulted in the extension of two fragments in both patients: one of approximately 400 nt in size, which was the more represented fragment, and one of 800–900 nt, which was less enriched (Fig. 2a).

The PCR products were cloned, and several independent clones were screened based on their molecular length. A direct sequencing analysis performed on ten different clones for each patient showed that the 400-nt product contained three different transcripts resulting from the whole or partial absence of exon 1 from the mature transcript. One product consisted of exon 2 perfectly spliced with exon 3 joined to a short stretch of genomic DNA sequence (86 nucleotides in size) located 57,022 nt upstream of the beginning of exon 2 (Fig. 2b, a1). Computer analysis indicated that this small segment of genomic DNA (Chr4:102,828,100–102,828,185, coordinates referred to GRCh38/hg38 of the human genome assembly, www.​genome.​ucsc.​edu) was located within an CpG island (CpG:141, chr4:102,826,475 102,828,235) and partially overlapped with the pugo gene (Chr4:102,827,193–102,829,052) (Additional file 1: Figure S1). There is no NCBI RefSeq model for the pugo gene, but Homo sapiens cDNA sequences in GenBank that were aligned on the genome and clustered in a minimal non-redundant way by the manually supervised AceView program (www.​ncbi.​nlm.​nih.​gov/​ieb/​research/​acembly), support two unspliced and three alternatively spliced variants. Among these variants, the unspliced pugo variant a, which is the only variant that putatively encodes a protein (DA090506 and DB032576), matched 86 nt of its non-coding region with the CISD2 variant a1.

The second transcript, where exon 2 correctly spliced with exon 3, was joined to a short stretch of genomic DNA sequence (83 nucleotides in size) located at 24004 nucleotides upstream the beginning of exon 2 (Fig. 2b, a2). Computer analysis performed on this small segment of genomic DNA (Chr4:102,861,131–102,861,212) did not identify the existence of any genes. In both variants, in silico analysis (NetStart1, http://​www.​cbs.​dtu.​dk) predicted a translational start site in the beginning of exon 2 which resulted in a frame shift, introducing amino acids changes and a premature stop codon (Fig. 2c, a1 and a2). Finally, the third 5′-RACE product resulted in the exclusion of the last 70 bases of exon 1 from the mature transcript (Chr4:102,869,118–102,869,187; NG_008636.2:5141–5210), which shifted the open reading frame and introduced downstream amino acid changes in addition to a premature stop (Fig. 2b, a3 and c, a3). In this isoform, the transcriptional start site was mapped at 111 and 43 nt upstream of the ATG for patients 1 and 2 respectively (Fig. 2c, a3). The frequency of the mutant isoforms in each patient is reported in Fig. 2d. The coexistence of multiple mutant CISD2 variants that diverged from one another by only the N-terminal region is also demonstrated by the sequence chromatograms from the 5′-RACE products that were not subcloned: the region spanning from exon 2 to exon 3 is represented by a single peak pattern, while the remaining N-terminal region is made up of multiple peak patterns (Additional file 2: Figure S2). The risk of false positive splice variant identification is very low since the three isoforms were detected in both siblings (two independent samples) although at different frequency.

The direct sequencing analysis performed on five clones for each patient showed that the upper 5′-RACE products may result in an inefficiently spliced pre-mRNAs that retained a large segment of intron 1 in the mRNA. In patient 1, the putative pre-mRNA retained the first 268 nt of intron 1 (Chr4:102,869,299–102,869,455; NG_008636.2: 5322–5478) in the final transcript, and the transcriptional start site was mapped to the same position identified in the smaller variant a3 (Fig. 2b, b1 and Additional file 3: Figure S3, b1). In patient 2, 646 nt region of intron 1 (Chr4102874083–102,874,671 and NG_008636.2: 10,204–10,694) was joined to exon 2 that was perfectly spliced with exon 3 (Fig. 2b, b2 and Additional file 3: Figure S3, b2). This product might be a prematurely terminated 5′-RACE fragment that did not reach the 5′ end of the template.

Together, the data demonstrated that the c.103 + 1G > A mutation functionally impaired mRNA splicing, producing multiple splice variants devoid of all or part of exon 1 that encoded for truncated non-functional isoforms.

To further investigate the consequences of the c.103 + 1G > A mutation on the protein product, Western blot analysis was performed on PBMCs using a polyclonal antibody raised against the CISD2 C-terminus which encompasses the CDGSH iron sulfur domain absent in the mutant isoforms. The Western blot analysis did not reveal the presence of the protein in the patients, whereas a decrease of approximately 50% was observed in the heterozygous parents compared with that observed in the healthy control (Fig. 3).

Together, these data indicate that the c.103 + 1G > A mutation caused a complete loss of CISD2 protein expression in the WFS2 patients and a decrease of approximately 50% in the unaffected parents.

The direct sequencing performed on isoform a3 and putative pre-mRNA b1 showed that patient 1 was a homozygous carrier of single nucleotide polymorphism (SNP) rs223332 (Fig. 2c, a3; Additional file 2: Figure S2, b1). This SNP, which is a substitution of a guanine to thymine 56 nt upstream of the CISD2 ATG (NG_008636.2:g.5052G > T), is annotated in the dbSNP database (http://​www.​ncbi.​nlm.​nih.​gov/​SNP/​) with a minor allele frequency (G) of 40%. The SNP genotyping performed on the genomic DNA isolated from buccal swabs from all family members and a healthy control indicated that the parents were heterozygous for this SNP, while both patients and the healthy control were homozygous carriers (Fig. 4). Considering the high frequency in the population, this SNP alone might be inert, but it may exert a functional effect when associated with CISD2 mutations or other simple and/or complex traits.