Molecular Vision 2021; 27:107-116
Received 29 July 2020 | Accepted 16 March 2021 | Published 18 March 2021
The first three authors contributed equally to this study.
1Experimental Cardiology Unit, Department of Cardiovascular Medicine, University of Lausanne Medical School, Lausanne, Switzerland; 2Department of Computational Biology, University of Lausanne, Lausanne, Switzerland; 3Department of Ophthalmology, Hadassah Medical Center, Faculty of Medicine, The Hebrew University of Jerusalem, Jerusalem, Israel; 4Rappaport Faculty of Medicine, Technion- Israel Institute of Technology, Haifa, Israel; 5Institute of Molecular and Clinical Ophthalmology Basel (IOB), Basel, Switzerland; 6Department of Ophthalmology, University of Basel, Basel, Switzerland; 7Department of Genetics and Genome Biology, University of Leicester, Leicester, United Kingdom
Correspondence to: Carlo Rivolta, Institute of Molecular and Clinical Ophthalmology Basel (IOB), Mittlere Strasse 91, CH-4031 Basel, Switzerland; Phone: +41 61 568 7211; FAX: +41 61 265 8652; email: email@example.com
Purpose: Heterozygous mutations in the gene PRPF31, encoding a pre-mRNA splicing factor, cause autosomal dominant retinitis pigmentosa (adRP) with reduced penetrance. At the molecular level, pathogenicity results from haploinsufficiency, as the largest majority of such mutations trigger nonsense-mediated mRNA decay or involve large deletions of coding exons. We investigated genetically two families with a history of adRP, one of whom showed incomplete penetrance.
Methods: All patients underwent thorough ophthalmological examination, including electroretinography (ERG) and Goldmann perimetry. Array-based comparative genomic hybridization (aCGH) and multiplex ligation-dependent probe amplification (MLPA) were used to map heterozygous deletions, while real-time PCR on genomic DNA and long-range PCR allowed resolving the mutations at the base-pair level. PRPF31 transcripts were quantified with real-time PCR on patient-derived lymphoblastoid cell lines.
Results: We identified two independent deletions affecting the promoter and the 5′ untranslated region (UTR) of PRPF31 but leaving its coding sequence completely unaltered. Analysis of PRPF31 mRNA from lymphoblastoid cell lines from one of these families showed reduced levels of expression in patients versus controls, probably due to the heterozygous ablation of its promoter sequences.
Conclusions: In addition to reporting the identification of two novel noncoding deletions in PRPF31, this study provides strong additional evidence that mRNA-mediated haploinsufficiency is the primary cause of pathogenesis for PRPF31-linked adRP.
Retinitis pigmentosa (RP) is the most common form of inherited retinal dystrophy, a class of diseases characterized by progressive degeneration of photoreceptors, the light-sensing cells of the human retina . This condition affects approximately 1 in 3,500 individuals worldwide  and is usually inherited as a monogenic trait (dominant, recessive, or X-linked), despite being very heterogeneous at the genetic level .
PRPF31 is a ubiquitous pre-mRNA splicing factor, which is part of the U4/U6*U5 trimeric small nuclear ribonucleoprotein complex of the spliceosome, is essential for cell survival, and is highly conserved within all eukaryotes, from yeast to humans. In humans, PRPF31 is encoded by the PRPF31 (Gene ID: 26121; OMIM: 606419) gene, localized at chromosome 19q13.4 and constituted by 14 exons in its main mature isoform.
Despite the observation that complete PRPF31 deficiency is not compatible with life , heterozygous mutations of various types (nonsense, missense, and frameshift) are associated with nonsyndromic autosomal dominant retinitis pigmentosa (adRP) . The majority of the reported mutations result in premature stop codons leading to nonsense-mediated mRNA decay (NMD) of mutated transcripts, and therefore, to disease by haploinsufficiency . The link between defects in such a fundamental gene and a tissue-restricted phenotype such as RP has not been fully elucidated.
PRPF31-associated RP displays sometimes incomplete penetrance, which is, in fact, a clinical hallmark of the involvement of this gene in the pathology. That is, carriers of heterozygous mutations in PRPF31 may or may not develop the disease throughout their life, and therefore, either be affected by progressive blindness or else be completely asymptomatic (while, however, still being able to pass pathogenic mutations to their offspring) [7,8]. This differential phenotype seems to depend, in turn, on the differential expression of the PRPF31 allele that does not bear the mutation. High expressors compensate for the protein deficiency induced by NMD on the mutated allele and have normal vision, whereas low expressors suffer from retinal degeneration due to an inadequate amount of functional PRPF31 protein [9,10]. The 5′ untranslated region (UTR) of the gene TCF3 fusion partner (TFPT; Gene ID: 29844; OMIM 609519) overlaps with that of PRPF31; the two genes are arranged in a “head-to-head” configuration and are transcribed from opposite DNA strands. TFPT has not been fully characterized, but its protein product has been identified as a molecular partner of transcription factor 3 (TCF3 or E2A) in childhood leukemia, and seems to function as a proapoptotic factor, putatively involved in cerebral apoptosis . It has also been hypothesized that PRPF31 and TFPT share the same bidirectional promoter, and that expression of the two genes is coregulated [12,13]. Furthermore, the genomic region comprising TFPT and PRPF31 is characterized by the presence of multiple repetitive Alu elements in intergenic and intragenic sequences, putatively promoting recombination events, such as deletions, duplications, and copy number variations in general .
In this article, we report the identification of two noncoding deletions spanning PRPF31 and TFPT core promoters as well as PRPF31’s 5′ UTR, in two families with adRP (one non-penetrant). This work shows that, in addition to mutations affecting PRPF31’s open reading frame, disease can be triggered by promoter-induced hypoexpression of the non-mutated PRPF31 coding sequence, presumably according to the same mechanisms of haploinsufficiency that are typical of all NMD-inducing mutations in PRPF31 detected thus far.
We studied two unrelated families of Ashkenazi Jewish descent (MOL0931 and TB228) with a history of adRP. All participating individuals signed a consent form, in agreement with the Declaration of Helsinki and the ARVO statement on human subjects. The study was approved by the institutional review boards of our respective institutions. Ocular evaluation included measurement of best-corrected visual acuity (BCVA), fundus examination after pupillary dilatation, fundus photography, spectral-domain optical coherence tomography (SD-OCT), full-field electroretinogram (ffERG), and visual field testing.
Peripheral blood mononuclear cells (PBMCs) were collected from the five members of the family MOL0931, and lymphocytes were immortalized from peripheral blood using the Epstein-Barr virus, to produce lymphoblastoid cell lines (LCLs). They were grown in Roswell Park Memorial Institute (RPMI) 1640 (Gibco, Waltham, MA) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin, as described in the literature .
Total genomic DNA was purified from peripheral blood leukocytes using the illustra Nucleon BACC genomic DNA extraction kit (GE Lifescience, Chicago, IL). RNA was extracted from the LCLs using the RNeasy mini kit (Qiagen, Hilden, Germany), according to the manufacturer’s instructions. Total RNA was then retrotranscribed using oligo dT and random hexamers in combination with the GoScript™ Reverse Transcription System (Promega, Madison, WI), as reported in the manufacturer’s protocol. Control RNA was extracted and retrotranscribed to cDNA from 13 LCL samples derived to the CEPH reference panel , according to the same conditions.
Genomic DNA samples of MOL0931–1 and MOL0931–2 were tested with array-based comparative genomic hybridization (aCGH) targeted for PRPF31 at GeneDx (Gaithersburg, MD). The array contains DNA oligonucleotide probes in or flanking PRPF31 exons. The array is designed to detect most single-exon deletions and duplications. Data analysis was performed with the Agilent Genomic Workbench software (Santa Clara, CA), with gene-specific filtering with the Cartagenia BENCH software (Agilent). Fragment analysis multiplex ligation-dependent probe amplification (MLPA) of the PRPF31 gene in individual TB228-1 was performed using the SALSA MLPA Probemix P235-B3 (MRC-Holland, Amsterdam, the Netherlands), according to the manufacturer’s instructions.
Real-time PCR was performed using the SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA). We used three identical replicates for each real-time experiment. GAPDH (Gene ID: 2597 ; OMIM: 138400) and human beta actin (ACTB; Gene ID: 60; OMIM: 102630) cDNA derived from mRNA was used to normalize PRPF31 mRNA expression during quantification.
To confirm the absence of genomic DNA in cDNA preparations, control reactions lacking the reverse transcriptase enzyme (−RT) were performed. These preparations were run in parallel and used as negative controls for each RT–PCR.
Primers for quantitative amplification of genomic DNA in family MOL0931 are listed in Supplementary Table 1. The primers were designed to match unique genomic regions using Primer 3 plus, and the in silico amplification product was tested using the UCSC in-Silico PCR tool. Assessment of hemizygous genotypes in the carriers of mutations in PRPF31 and in men versus women, for the XIST (Gene ID: 7503; OMIM: 314670) gene, was performed as described previously . In short, the readouts of these amplifications were standardized by measuring the signals produced by a portion of XIST and then normalized to signals from a bona fide diploid portion of PRPF31, located on exon 8 (primers 5′-CCT GAC CAA CCT CTC CAA GA-3′, 5′-GTA GAC GAG AAG CCC GAC AG-3′). The ratio between the threshold cycle (Ct) values of regions to be investigated for the copy number and the Ct value of exon 8 gave us an indication of the mono- or biallelic genetic content.
The long-range PCR encompassing exon 3 of TFPT and the intronic proximities of exon 2 of PRPF31 was achieved with use of the high-fidelity Phusion DNA polymerase (TaKaRa, Shiga, Japan), primers 5′-GGA TGT GGA AGT GGG AGA GA-3′ (forward) and 5′-GGC GAT GGT CTT GAC TGA AT-3′ (reverse), and a “touchdown” cycling profile. In essence, following an initial denaturation step at 98 °C for 3 min, the samples were subjected to the following conditions: 15 cycles at 98 °C for 10 s + 66 °C (decreasing by 0.7 °C per cycle) for 30 s+ 72 °C for 9 min, followed by 25 cycles at 98 °C for 10 s + 55 °C for 30 s + 72 °C for 9 min. The amplification buffer (20 µl) was composed of Phusion Buffer, 4 µl; 10 mM deoxynucleotide triphosphate (dNTPs), 0.4 µl; forward primer 10 µM, 1 µl; reverse primer 10 µM, 1 µl; 10 ng/µl DNA, 10 µl; dimethyl sulfoxide (DMSO), 0.6 µl; Phusion polymerase, 0.2 µl; and H2O, 2.8 µl. All the amplification products from patients and controls were sequenced with the Sanger method by using 16 different primers spanning the entire fragment (Supplementary Table 2). Standard PCRs to confirm the deletion breakpoints were performed with the use of the GoTaq polymerase (Promega), a standard amplification buffer, the cycling conditions 94 °C for 3 min, [94 °C for 30 s + 60 °C for 30 s+ 72 °C for 30 s] for 30 cycles, and 72 °C for 3 min, and the primers listed in Supplementary Table 3.
Family MOL0931 of Ashkenazi Jewish origin included three individuals affected with RP: the index case (MOL0931–1), his sister (MOL0931–2), and his paternal grandmother (MOL0931–5). The proband’s father had no symptoms, suggesting dominant inheritance with incomplete penetrance (Figure 1). The index case was noted to have impaired night vision at age 11, but only at age 17 was the diagnosis of RP established, based on funduscopic findings and non-detectable electroretinogram (ERG) responses under all stimulus conditions. His VA was 20/25 in the right eye and 20/30 in the left eye. Intraocular pressure (IOP) was within normal limits, and anterior segments were quiet and clear with no evidence of cataracts. There were strands and floaters in the vitreous as well as +1 cells. Funduscopic findings were those characteristic of RP, including pallor of the optic discs, attenuated retinal vessels, and widespread grayish spots of atrophy, as well as bone spicule-like pigmentation throughout the retina, with somewhat better preservation of retinal color in the macular area. There were cystoid macular changes and mild epiretinal membranes. One year later, at age 18, VA was 20/40 in both eyes, and the cystoid changes had worsened considerably. Oral administration of acetazolamide 125 mg three times daily yielded significant reduction in the degree of macular edema (Figure 2A–D). Goldmann visual fields at age 18 were constricted to 10–15 degrees from fixation using the V4e target, and there were preserved islands of vision in the far temporal periphery of both eyes (Figure 2E,F). His sister, at age 15, had much better fields, with mild to moderate superior and nasal constriction but well-preserved temporal fields with the V4e target (Figure 2G,H). OCT performed at the age of 22 showed preservation of the outer nuclear layer (ONL) and the ellipsoid zone (EZ) in the foveal region, but marked thinning of the ONL and loss of the EZ outside this area (Figure 2I,J). The father of the two siblings (MOL0931–3) was asymptomatic. His VA was 20/20 in both eyes, and his clinical ophthalmological exam was within normal limits without funduscopic signs of retinal degeneration. ffERG testing performed at the age of 51 years showed normal dark-adapted rod b-wave responses and normal light-adapted cone responses, including normal cone flicker implicit times. The dark-adapted mixed rod-cone response to an International Society for Clinical Electrophysiology of Vision (ISCEV) standard flash showed normal amplitude of the a-wave, while the b-wave was reduced to approximately 70% of the lower limit of normal (Table 1).
The second family (TB228) is an Ashkenazi Jewish family, segregating adRP with no indication for reduced penetrance (Figure 1). The index patient (TB228–1) was diagnosed with RP at the age of 47 years. At this age, his scotopic and photopic ffERG recordings were severely reduced. He reported that his father and his brother had RP, but no clinical data were available. He has two sons. One (TB228–3) had normal vision. The second son (TB228–2) had night blindness since childhood and was diagnosed with RP at the age of 8 years. At the age of 11 years, his scotopic and photopic ERG recordings were severely reduced, and by the age of 17 years, his ffERG was nonrecordable. At the age of 38 years, he underwent cataract surgery in his left eye. A year later, OCT imaging demonstrated thinning of the retina with flattening of the foveal depression in both eyes and an epiretinal membrane in the left eye. At the age of 43 years, he had severe constriction of the visual field. At this time, his BCVA was 20/20 (right eye) and 20/100 (left eye). Funduscopy revealed extensive retinal pigment deposits and optic disc pallor on both eyes. Summary clinical findings are reported in Table 1.
For family MOL0931, genetic analysis for the two affected siblings including screening of previously reported mutations using the Asper mutation detection array and Sanger sequencing of candidate exons was negative. Whole exome sequencing (WES) analysis was subsequently performed on DNA samples of the three affected individuals but did not reveal any potential pathogenic variants. Array CGH analysis of the two affected siblings allowed identification of a putative heterozygous deletion involving exon 1 of PRPF31, which is noncoding, and possibly other regions of the same gene.
For family TB228, TB228–1 and TB228–2 underwent whole exome sequencing, with negative results. Fragment analysis MLPA of the PRPF31 gene in individual TB228–1 indicated a putative heterozygous deletion involving exon 1 of PRPF31.
Prompted by the array CGH (family MOL0931) and MLPA (family TB228) data, we hypothesized and investigated the presence of a large heterozygous deletion, which is a common cause of pathogenesis in PRPF31-associated RP . To this end, we designed a series of very-short range (maximum 143 bp) real-time PCRs by using as a template the genomic DNA from the three patients and the asymptomatic obligated carrier from family MOL0931, as well as from four control subjects.
Of the 12 PCRs performed, four PCRs were compatible with the hemizygous presence of template DNA (Figure 3). Interestingly, these amplifications were all contiguous and flanked on both sides by PCRs outputting results from diploid DNA, in agreement with the presence of a putative heterozygous deletion within the PRPF31 gene. Specifically, the last positive diploid amplifications were in intron 2 of TFPT and intron 1 of PRPF31, from the centromeric and telomeric ends, respectively, indicating that such deletion could include exons 1 and 2 of TFPT and exon 1 of PRPF31. Therefore, we set up a long-range PCR (8.3 kb) that could span the breakpoint of the deletion, by anchoring primers on exon 3 of TFPT and a portion of intron 1 of PRPF31 neighboring exon 2, that is, the closest regions devoid of repetitive DNA elements (Figure 4A). Sequencing of this amplification product identified the breakpoint of a 3,567 bp deletion (hg19; chr19: g.54113356-54116922del), originating in intron 2 of TFPT and ending in intron 1 of PRPF31, likely as a consequence of an Alu-mediated recombination (Figure 3, Figure 5). This event was further validated with a short-range “diagnostic” PCR performed directly on the genome of patients and controls, with primers located immediately across the breakpoint (Supplementary Table 3, Figure 4B) and with Sanger sequencing of the PCR product. In these tests, only DNAs from patients result in a signal, as the distance between the forward and reverse primers in the absence of the deletion would be too large to yield any PCR product in standard conditions (Figure 4A,B).
PCR analysis and Sanger sequencing on DNA from family TB228 revealed a shorter deletion (g.54113892-54116404del; 2,513 bp) at the same locus in affected individuals TB228–1 and TB228–2. This event, which removed the same exons that were ablated in patients from family MOL0931, was not present in unaffected individual TB228–3 (Figure 4B). The deletion’s boundaries included intron 2 of TFPT and intron 1 of PRPF31, within the vicinity of two short interspersed nuclear elements (SINE) repetitive elements (Figure 3, Figure 5).
Molecular pathogenesis of mutations in PRPF31 is almost invariably associated with DNA variants that lead to reduced expression of the allele that carries the mutation, via a mechanism of haploinsufficiency [9,10]. In most cases, this is the consequence of mutations leading to premature termination triplets in the coding sequence and therefore, to nonsense-mediated mRNA decay of the transcript . However, in at least one instance reported in the literature the same pathological effect is produced by a single-base mutation in the promoter, leading to hypoexpression of wild-type PRPF31 mRNA . To test whether the deletions identified in this study, involving noncoding parts of the gene and encompassing its promoter, could induce pathogenesis by the same mechanism, we assessed PRPF31 expression in LCLs from the four carriers of mutations from family MOL0931 and compared it to that of LCLs from 13 controls, according to well-established protocols for this specific analysis [9,10,17]. Fresh biological material from family TB228 was unavailable, and therefore, this test could not be performed on members from this pedigree.
In agreement with previous literature testing coding mutations in PRPF31, individuals carrying the noncoding chr19: g.54113356-54116922del mutation had markedly reduced levels of PRPF31 transcripts compared to controls, whereas MOL0931–3 had intermediate amounts of mRNA, slightly higher than those of patients (Figure 6).
In adRP associated with mutations in PRPF31, pathogenicity is prevalently triggered by hemizygosity of its transcripts (haploinsufficiency). This, in turn, is due to three main causes: single-base or small mutations leading to premature termination codons and therefore, to NMD (most cases) , large deletions involving one or several coding exons [14,18,19], or heterozygous mutations in regulatory regions. To the best of our knowledge, there is only one example of the latter mechanism, involving a single base deletion in the 5′ UTR of PRPF31 .
In this work, we report the identification of two deletions spanning exon 1 (noncoding) of PRPF31, as well as its core promoter. In contrast to the largest majority of mutations detected thus far, both mutations reported here leave PRPF31’s open reading frame untouched. However, similar to all other mutations investigated thus far, these deletions also result in a reduction in PRPF31 transcripts, and therefore, presumably in a pathogenic mechanism related to haploinsufficiency of the PRPF31 protein. In the present case, however, the reduction in PRPF31 expression should not be triggered by NMD, but by ablation of the gene’s promoter sequences, as defined by Rose et al. . The levels of PRPF31 mRNA expression displayed by cell lines from individual MOL0931–3, having an intermediate value between patients and controls, are also typical, although they are not really predictive of his present or future clinical status [9,10].
It is also remarkable that two mutational events within the same genomic interval, of similar nature, and resulting in the same phenotype were of independent origin. This is even more striking if we consider that the two families in which the deletions occurred were from the same ethnic group, and therefore, we would have expected the presence of an ancestral mutation, common to both pedigrees. A likely explanation for these findings is that the presence of multiple repetitive DNA elements in this region makes it particularly unstable and prone to events of uneven recombination, resulting in turn in deletions or duplications. In summary, in addition to reporting the identification of two novel noncoding mutations in PRPF31, these results provide an additional element of support for the notion that retinal pathogenesis is due to the gene’s hypoexpression, rather than to the mere presence of mutations in the gene’s coding sequence.
Appendix 1. Supplementary table 1.
Appendix 2. Supplementary table 2.
Appendix 3. Supplementary table 3.
This work was supported by a grant from the Swiss National Science Foundation (grant 31003A_176097, to CR) and a grant from the Foundation Fighting Blindness USA (BR-GE-0214–0639-TECH to TB, DS, and EB). The authors thank Devora Marks-Ohana for technical assistance and Sitta Föhr for her help with the copyediting of the manuscript. We warmly thank all patients and their family members for participating in this study.