Novel GUCA1A Mutations Suggesting Possible Mechanisms of Pathogenesis in Cone, Cone-Rod, and Macular Dystrophy Patients

Here, we report two novel GUCA1A (the gene for guanylate cyclase activating protein 1) mutations identified in unrelated Spanish families affected by autosomal dominant retinal degeneration (adRD) with cone and rod involvement. All patients from a three-generation adRD pedigree underwent detailed ophthalmic evaluation. Total genome scan using single-nucleotide polymorphisms and then the linkage analysis were undertaken on the pedigree. Haplotype analysis revealed a 55.37 Mb genomic interval cosegregating with the disease phenotype on chromosome 6p21.31-q15. Mutation screening of positional candidate genes found a heterozygous transition c.250C>T in exon 4 of GUCA1A, corresponding to a novel mutation p.L84F. A second missense mutation, c.320T>C (p.I107T), was detected by screening of the gene in a Spanish patients cohort. Using bioinformatics approach, we predicted that either haploinsufficiency or dominant-negative effect accompanied by creation of a novel function for the mutant protein is a possible mechanism of the disease due to c.250C>T and c.320T>C. Although additional functional studies are required, our data in relation to the c.250C>T mutation open the possibility that transacting factors binding to de novo created recognition site resulting in formation of aberrant splicing variant is a disease model which may be more widespread than previously recognized as a mechanism causing inherited RD.


Introduction
Autosomal dominant cone and cone-rod dystrophies (adCODs and adCORDs, resp.) are hereditary retinal disorders characterized by visual loss, abnormalities of color vision, visual field loss, and a variable degree of nystagmus and photophobia. CORDs are characterized by progressive loss of cone photoreceptor function followed by progressive loss of rod photoreceptor function and are often accompanied by retinal degeneration. In contrast, in inherited progressive CODs, only cone function is impaired, and retinal degeneration is often restricted to the central retina [1,2].
In recent years, the importance of aberrant splicing as a disease mechanism has changed from a situation in which this possibility was considered to be relatively rare and only relevant when one out of a few critical nucleotides in the splice sites consensus sequences at the exon/intron borders are changed to a situation where sequence variants in a gene may potentially disrupt splicing. Thus, in addition to the splice site sequences, other cis-regulatory elements which act as splicing enhancers (exonic and intronic, ESEs, and ISEs) or silencers (exonic and intronic, ESSs, and ISSs) and direct the splicing machinery to use the correct splice sites should be considered as important mutation spots [15]. The interactions between silencers, enhancers, and their cognate binding proteins play a critical role in the fidelity and regulation of pre-mRNA splicing [16]. To our knowledge, at least 10% of all mutations identified as causing humaninherited disease are known to alter consensus 5 -or 3splice sites, thereby inducing aberrant pre-mRNA splicing [17]. Although the mechanistic consequences of mutations on splice sites are fairly easy to interpret, evaluating precisely how inherited disease-causing mutations influence the loss or gain of ESE/ESS motifs remains challenging [18].
In this study, we report two Spanish families with an autosomal dominant retinal degeneration in which two novel GUCA1A mutations have been identified (Table 1). We investigated the relationship between the newly reported coding sequence mutation (c.250C>T) and splicing regulation using a set of bioinformatics techniques and predicted that ESE loss and ESS gain might be responsible for retinal dystrophy. Furthermore, we identified a range of retinal abnormalities within a single pedigree caused by novel GCAP1 p.L84F variation which is consistent with the broader phenotype associated with p.Y99C and p.P50L GUCA1A mutations [5,19].

Clinical Assessment and Sample
Collection. This study adhered to the tenets of the Declaration of Helsinki for research involving human subjects. A three-generation Spanish family with an autosomal dominant retinal dystrophy was ascertained (Figure 1(a)). Five affected and 9 unaffected individuals were recruited by the Hospital Fundación Jimenez Diaz, Madrid, Spain. After informed consent was obtained, each patient was assessed through a funduscopy, an optical coherence tomography (OCT) using macular cube scan (512 × 128 protocol), an electrooculogram (EOG), a full-field electroretinogram (ERG), and a multifocal ERG (mfERG) using standard electrophysiologic methods. Peripheral blood was obtained by venipuncture, and genomic DNA was extracted from lymphocytes according to standard protocols [20].
Family 141     A cohort of 61 unrelated Spanish patients clinically diagnosed with COD, CORD, or MD and family history consistent with an autosomal dominant mode of inheritance were additionally screened for mutations in GUCA1A gene.
In addition, 200 randomly selected DNA samples (400 chromosomes) from healthy control individuals obtained from the European Collection of Cell Cultures (Human Random Control DNA Panels, ECACC, Salisbury, UK) were analyzed to assess the frequency of sequence changes in the normal population.

Genotyping and Linkage Analysis.
Genotyping of 10 members of the Spanish pedigree (Family 141, Figure 1(a)) was performed with 262,264 single nucleotide polymorphisms (SNPs) from the Affymetrix Gene Chip Mapping 250K Nsp Array (AROS Applied Biotechnologies, Aarhus, Denmark). LOD scores were calculated for the markers by multipoint linkage analysis using Genehunter program in the easyLINKAGE Plus package v5.08 [21]. The call rate, defined as the percentage of successful genotype calls among subjects, was used as a measure of data quality. The results reported in this study are based on 95% call rate. Uninformative SNPs were removed from the data by Merlin [22]. We modeled the disease as an autosomal dominant trait with 100% penetrance.
All regions with suggestive LOD scores (>1.5) were additionally saturated with microsatellite (STR) markers designed according to the information obtained from Marshfield, GDB Human Genome Database (http://research .marshfieldclinic.org/) and Ensembl genome data resources (http://www.ensembl.org/). Multiplex microsatellite genotyping of all available family members (Figure 1(a)) was performed with 3-5 markers per reaction using QPCR mix (ABgene, UK). Data collection and allele identification were performed using GeneScan and GeneMapper v4.0 software (Applied Biosystems Incorporated, Foster City, CA, USA).

Mutation
Screening. Sequence analysis of the coding exons and intron-exon junctions of candidate genes within the critical interval, IMPG1, ELOVL4, RDS, GUCA1A, and RIMS1, was performed on ABI3730xl genetic analyser (Applied Biosystems Incorporated, Foster City, CA, USA) according to the manufacturer's protocol. Novel nucleotide sequences were compared to the published cDNA sequence according to the human genome build 37 (hg19, 2009).

Restriction Enzyme Analysis.
To confirm the presence of c.250C>T variant, exon 4 of the human GUCA1A gene was directly amplified from genomic DNA. A sample of genomic DNA (100 ng) from patients and healthy individuals was used in a 20 L reaction mixture containing 0.5 M of forward (5 GTCTAGGAAGACAGATAGGTC 3 ) and reverse (5 CAAGGAAGGGAAGGAACGTG 3 ) primers, 20 M of each dNTP, 1× PCR buffer, and 1 U of MyTaq DNA polymerase (Bioline, London, UK). After an initial denaturation of 95 ∘ C for 5 min, 30 cycles were performed at 95 ∘ C for 30 s, 62 ∘ C for 20 s, and 72 ∘ C for 50 s, with a final extension step of 72 ∘ C for 5 min. The PCR products (517 base pairs length) were digested with SmlI according to the manufacturer's instructions (New England BioLabs, Beverly, MA, USA) and separated by electrophoresis in a 3% agarose gel.

Clinical Evaluation of Patients.
Clinical evaluation including fundus photography and electrophysiology was performed in a three-generation Spanish pedigree with autosomal dominant retinal degeneration with cone and rod involvement (Family 141, Figure 1(a)). All affected individuals complained of very early loss of central vision within the first two decades (between age of 2 and 17 years), with subsequent gradual deterioration of the color vision and mild photophobia. There was no evidence of nystagmus. Clinical characteristics of patients are summarized in Tables 2 and 3. Funduscopy revealed a range of macular phenotypes, from normal or mild changes to severe macular atrophy (Figures 2(a)-2(j)). This variation was not related to the age of the subjects. Despite the normal fundus appearance in patient III:1, OCT scans showed that the macular region was abnormally thin. However, retinal layers at the macular area were still well conserved in this subject (Figure 3(a)). Instead, the affected mother (subject II:1) clearly presented disorganization of the retinal layers at the macular region, and reduction in the retinal thickness also appeared (Figure 3(b)). In the presence of common genotype, electrophysiologic examination of the patients detected intrafamilial variability of the phenotype involved either the cones only or both cone and rod systems with photopic amplitudes more markedly reduced than the scotopic (Figure 4). The oldest subject (individual I:1) had bilaterally undetectable cone response and subnormal rod ERGs (Figure 4(b)), indicating a cone-rod phenotype with severe loss of macular function ( Figure 5(b)). Patients II:1, III:1, and III:4 all presented from moderate (III:4) to significantly reduced (II:1 and III:1) cone ERGs and normal rod responses on full-field ERG (Figures 4(c), 4(e), and 4(f)), suggestive of cone dysfunction. Additional evidence for macular dysfunction is presented as detected by mfERG ( Figure 5(d)) and OCT imaging (Figure 3). In patient II:3, the full-field cone and rod ERGs were normal in both eyes (Figure 4(d)), but the mfERG showed decreased central amplitudes reflecting the macular dysfunction ( Figure 5(c)); findings are consistent with isolated severe bilateral macular degeneration and resulted in a loss of the central vision. No abnormalities were observed in the EOGs in any patient ( Table 2). All affected individuals studied in detail were later found to carry the p.L84F GUCA1A mutation.

Genotyping and Linkage
Analysis. Five affected and nine unaffected subjects of the Spanish family (Figure 1(a)) were genotyped with Affymetrix 250K SNP microarray, and genome-wide linkage analysis was used to search for cosegregation of the markers with the disease phenotype. Parametric     except the interval located on chromosome 6 which could not be further refined. Affected individuals shared a common haplotype tested with microsatellite markers spanning the linked region and generating maximum LOD score of 2.10 at = 0.0 (Figure 1(a)).

Mutation Analysis.
According to Ensembl human genome database, the 55.37 Mb critical interval on chromosome 6p21.31-q15 included 295 known genes among which were previously described retinal degeneration genes. Positional candidate genes were IMPG1 and ELOVL4, mutations in which cause adMD [34,35]; as well as RDS, GUCA1A, and RIMS1, implicated in autosomal dominant COD and CORD [9,36,37]. By direct sequencing of the coding exons and intron-exon junctions of the candidate genes, we found only known polymorphisms except a novel heterozygous substitution c.250C>T in GUCA1A, replacing leucine with phenylalanine at amino acid residue 84 (Figure 1(b)). Segregation analysis showed that the novel mutation (p.L84F) co-segregated with the disease in the pedigree (Figure 1(a)), but it was not detected in 200 unrelated normal controls. This variant was not reported in public sequence repositories (including the 1000 Genomes Project database, http://www.1000genomes.org/). A restriction fragment length analysis confirmed the c.250C>T mutation showing that the transition from C (wt) to T (mutation) results in loss of SmlI restriction site. Wildtype samples produced two fragments of 276 bp and 241 bp, while the restriction target site (5 -CTCAAG-3 ) in exon 4 of GUCA1A was destroyed by the mutation (Figure 1(c)).
We hypothesized that the c.250C>T mutation could function by disrupting a splicing enhancer and/or by the creation of a splicing silencer because recognition of splice sites can either be assisted by splicing enhancers or suppressed by splicing silencers. Therefore, we scanned the wild-type and c.250C>T mutant GUCA1A exon 4 sequence for ESEs and ESSs using different web-based programs. The results generated by ESEfinder 3.0 program indicated that the 250T nucleotide abolishes SF2/ASF binding motif (CTCAAGG) (Figure 1(g)). An overlapping SC35 binding motif (GGTCCTCA) is also disrupted, while a second SC35 binding site (GTCCTCAA>GTCTTCAA) is not affected by the mutation, but its score has been decreased from 3.572 to 3.341 (Table 4). Although Human Splicing Finder calculates the score according to a different algorithm [26], the two analyses gave similar results.
The ASD and ESRsearch programs identified that both ESE (which promotes exon inclusion) and ESS (which removes identified exon from the final product) elements present on GUCA1A exon 4. These programs confirmed the loss of SF2/ASF ESE site (CTCAAGG) caused by the mutant allele. In addition, the c.250C>T mutation creates a new ESS site for binding of SRp20 (GTCTTCAAG) and strengthens the binding motif of the transacting splicing repressor hnRNP I (CTTGGTCCTC>CTTGGTCTTC). ESRsearch program predicted the mutant TCTTC motif as a gain of novel hnRNP I binding ESS (Figure 1(g)).

Discussion
In this study, we identified a three-generation Spanish pedigree representing intrafamilial heterogeneity of retinal degeneration with cone and rod involvement. The clinical observations revealed a range of retinal abnormalities, including isolated macular dysfunction, cone dystrophy and cone-rod dystrophy-findings very similar to previous descriptions of the mixed phenotype associated with p.P50L and p.Y99C GCAP1 mutation [5,19]. The very early onset of visual loss and the fast progression of the retinal dystrophy in the presented family contribute to the wide variability of phenotypes associated with GCAP1 mutations. Combined with previous studies, our findings support the hypothesis of additional modifying factors (genetic and environmental) involvement, which are responsible for the modulation of the phenotype in patients harboring the same GUCA1A mutation [5].
Linkage to the GUCA1A locus was identified, and two novel missense mutations, p.L84F and p.I107T, predicting different mechanisms of pathogenesis were found in exon 4 of the GCAP1 coding gene (Figures 1(a)-1(e)). The Ile107 residue is located in a phylogenetically conserved motif within the functional EF3 hand domain of the guanylate cyclase-activating protein 1 (Figures 1(f) and 1(h)). The result of PSIPRED analysis suggested that the novel p.I107T mutation led to secondary structure changes affecting EF3, EF4, and two C-terminal -helices which may interfere with the correct folding of the two helix-loop-helix Ca 2+binding hands. A number of missense mutations associated with either loss or gain of function and producing COD, CORD, and MD dominant phenotypes have been already discovered. By using PSIPRED, we predicted that each of these mutations (p.P50L, p.E89K, p.Y99C, p.D100E, p.N104K, p.T114I, p.I143NT, p.L151F, p.E155G, and p.G159V) specifically alter the conformation of the protein which is essential for the Ca 2+ binding ( Table 1).
The p.I107T mutation represents the second naturally occurring mutation in the 12-amino-acid-loop of the third GCAP1 EF hand. Previously, a mutation (p.N104K) affecting the Ca 2+ -binding loop of EF3 has been found to be causative for adCOD [7]. It has been also reported that a mutation of a residue flanking EF3 (p.Y99C) disrupts the N-terminal helix of the helix-loop-helix conformation of EF3, severely affecting Ca 2+ binding at this site [6]. Another pathogenic mutation of a flanking hydrophobic residue (p.I143NT) was observed in EF4, emphasizing the importance of an intact Nterminal helix for Ca 2+ binding [8]. Other mutations linked to cone dystrophy (p.E155G and p.L151F) affect exclusively Ca 2+ coordination in EF4 [9][10][11]. Common biochemical effect of the mutations affecting one of the functional EF hands consists of the inability of the mutant GCAPs to inhibit photoreceptor GC in the dark when Ca 2+ is elevated [38]. The result is that cGMP levels are elevated in mutant photoreceptors and a larger number of cation channels remain open, which eventually leads to elevated Ca 2+ concentration and to photoreceptor cell death [39]. Taken together, all this data and the fact that the mutation is of the missense type suggest similar functional consequence of p.I107T, and the mechanism of disease causation should be considered as either loss or gain of function that affects calcium homeostasis within the photoreceptor outer segment. The mutation p.L84F is located in another phylogenetically conserved motif flanking the functional EF2 hand (Figures 1(f) and 1(h)). Although PolyPhen-2 predicted probably damaging effect for this mutation, no conformational changes of the mutant protein were shown by PSIPRED program (Table 1). Therefore, we used a set of computational approaches to determine if this substitution alters specific cisacting elements that are important for correct splicing, such as ESEs and ESSs, and acts at an earlier RNA level.
Two of the major factors in establishing exon identity are serine-and arginine-rich proteins (SR proteins), among which are SF2/ASF and SC35, and the heterogeneous nuclear ribonucleoproteins (hnRNPs) [15]. SR proteins promote the initial stages of spliceosome assembly by binding to ESEs and recruiting basal splicing factors to adjacent splice sites or by antagonizing the effects of ESS elements [40]. In contrast, hnRNPs member among which is hnRNP I (also known as polypyrimidine tract-binding protein, PTB), mediate the repressive effects of silencers and can alter recruitment of the core splicing machinery [41]. Surprisingly, since SR proteins are typically thought to promote exon inclusion, it has been demonstrated that interactions between SRp20, PTB, and other hnRNPs create an exon silencing complex that promotes exon skipping and cause inherited disease [42].
Our in silico analyses of exon 4-wild type and mutant sequence indicated that naturally occurring SF2/ASF (splicing factor 2/alternative splicing factor) ESE motif (CTCAAGG) was abolished (SF2/ASF: 250C score = 2.216, threshold = 1.956) suggesting that this heptamer might be essential to distinguish a functional exon and its disruption results in loss of activity. In addition, it was shown that the c.250C>T mutation disrupted (SC35: 250C score = 3.059, threshold = 2.383) a binding ESE motif (GGTCCTCA) and weakened (SC35: 250C score = 3.572, 250T score = 3.341, and threshold = 2.383) a second recognition site for the RNA-binding protein SC35 (GTCCTCAA>GTCTTCAA) ( Table 4). A search for potential splicing silencer regulatory elements within the GUCA1A exon 4 showed that the 250T allele created a new ESS site (GTCTTCAAG) for binding of SRp20 which is not present with 250C and strengthened another binding motif (TCCT>TCTT) for hnRNP I/PTB (Figure 1(g)), both designated as exon silencer partners by Sterne-Weiler and coauthors [42]. It has been shown that the loss of an ESE consensus motif for SF2/ASF can cause human-inherited disease [43]. There are many other examples for disease-associated mutations that alter splicing enhancers or silencers and promote exon skipping, which support our findings [42,[44][45][46][47][48]. Similarly, disease-causing mutations in introns that cause missplicing by inducing the inclusion of intronic sequences as exons (pseudoexon inclusion) and function by strengthening of preexisting weak pseudosplice sites or by creating new splice sites have been discovered [49][50][51].
Based on this data, we concluded that the c.250C>T mutation may function by disruption of SF2/ASF and SC35 binding ESEs and activation of cryptic ESS sites (for binding of hnRNP I and SRp20) which promotes in-frame skipping of exon 4 from the spliced mRNA. Thus, the expected consequence of this mutation is production of aberrant transcript that, if translated, would encode a truncated (50 amino acids shorter) protein copy suggesting either gain of function or haploinsufficiency with cone cells predominantly affected because of the higher concentrations of RetGC1 and GCAP1 in cone photoreceptors.

Conclusions
In conclusion, we suggest that mutation associated with disruption of ESE and activation of ESS cis-elements can be applied to photoreceptor degeneration. Importantly, this means that similar molecular mechanism may cause retinal degeneration much more frequently than reported for aberrant spliced mRNA so far.
However, because of their predictive nature, these findings need to be validated with experimental evidences in order to see the real impact of the reported missense mutations on the structure and conformation of GCAP1.