SEARCH

SEARCH BY CITATION

Keywords:

  • CFTR;
  • decision flowchart;
  • VUCS;
  • interpretation;
  • splicing

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods and Decision Criteria
  5. RESULTS
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Molecular diagnosis of cystic fibrosis and cystic fibrosis transmembrane regulator (CFTR)-related disorders led to the worldwide identification of nearly 1,900 sequence variations in the CFTR gene that consist mainly of private point mutations and small insertions/deletions. Establishing their effect on the function of the encoded protein and therefore their involvement in the disease is still challenging and directly impacts genetic counseling. In this context, we built a decision tree following the international guidelines for the classification of variants of unknown clinical significance (VUCS) in the CFTR gene specifically focused on their consequences on splicing. We applied general and specific criteria, including comprehensive review of literature and databases, familial genetics data, and thorough in silico studies. This model was tested on 15 intronic and exonic VUCS identified in our cohort. Six variants were classified as probably nonpathogenic considering their impact on splicing and eight as probably pathogenic, which include two apparent missense mutations. We assessed the validity of our method by performing minigenes studies and confirmed that 93% (14/15) were correctly classified. We provide in this study a high-performance method that can play a full role in interpreting the results of molecular diagnosis in emergency context, when functional studies are not achievable.


Introduction

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods and Decision Criteria
  5. RESULTS
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Cystic fibrosis (CF; MIM #219700), the most frequent severe autosomal recessive disease in the European population, is caused by mutations in the Cystic Fibrosis Transmembrane Regulator gene (CFTR or ABCC7; MIM #602421) [Kerem et al., 1989; Riordan et al., 1989; Rommens et al., 1989]. To date, more than 1,900 mutations, polymorphisms, or unclassified sequence variations have already been reported in the Cystic Fibrosis Mutation Database (CFMD), over 90% of which consisting of point mutations or small insertions/deletions (http://www.genet.sickkids.on.ca/). Besides the p.(Phe508del) (c.1521_1523delCTT) mutation responsible for approximately two thirds (66%) of all CF chromosomes, the remaining third of alleles is highly heterogeneous with only 20 mutations occurring at a frequency of more than 0.1% and a large fraction of private mutations [Bobadilla et al., 2002; Claustres et al., 2000]. As a reference laboratory of the French network for molecular diagnosis of CF, we apply strategies for in-depth analysis of the CFTR gene, which consist of exploring the entire coding sequence, its flanking regions and targeted intronic sequences associated with the search for large rearrangements [Dequeker et al., 2009]. Using this approach, we regularly solve complex cases not only in the context of CF or male infertility due to isolated congenital bilateral absence of the vas deferens (CBAVD; MIM #277180) [Claustres, 2005], but also in diverse “CFTR-related disorders” (CFTR-RDs) [Bienvenu et al., 2010; Whitcomb, 2010].

Some variants such as nonsense or frameshift mutations are readily classified as pathogenic mutations. However, the frequent identification of rare sequence alterations of unknown pathogenicity, also called variants of unknown clinical significance (VUCS) or unclassified variants (UVs), substantially complicates test interpretation and consequently genetic counseling of patients and families. According to data collected in the CFTR national collaborative knowledgebase [Bareil et al., 2010], such UVs would represent at least 20% of the sequence variations identified by all French collaborating laboratories (J Cyst Fibros. Vol. 11 Suppl. 1, Page S17; June 2012). In this context, the Clinical Molecular Genetics Society (CMGS; http://www.cmgs.org) published in 2007 practice guidelines for the Interpretation and Reporting of Unclassified Variants (UVs) in Clinical Molecular Genetics, based on epidemiologic, familial and clinical data associated with computational studies and in vitro functional assays. However, there is currently no integrated tool available for diagnostic laboratories, and the Bayesian approach developed for the interpretation of VUCS in BRCA1 and BRCA2 genes [Goldgar et al., 2004; Plon et al., 2008] is not fully suitable to rare autosomal recessive diseases.

We therefore decided to develop a decision tree based on the CMGS guidelines and particularly suitable for the interpretation of variants in the CFTR gene. We focused on a particular class of variants that do not affect the highly conserved (canonical) dinucleotides GT and AG at splice sites but still may result in splicing alterations. Indeed, if sequence changes at GT and AG dinucleotides are considered to result in splicing aberrations, prediction of the consequences of alterations at less-conserved positions of splice sites or within exons is more challenging. For this purpose, 15 variants (10 intronic and five exonic) were selected in our cohort of CF and CFTR-RD patients. We analyzed sequentially these variants using our flowchart that includes a set of general and specific criteria. The performance of our method was assessed by comparing this classification with the results of minigenes studies.

Methods and Decision Criteria

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods and Decision Criteria
  5. RESULTS
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

The complete decision tree is detailed in Figure 1. The main criteria and the standards used for their interpretation are listed below.

image

Figure 1. In-house decision flowchart for the classification of VUCS considering their potential impact on splicing. Frequency in general population, literature and LSDB (Locus Specific DataBases) data and cosegregation studies are considered as general criteria. All following steps assessing the impact of variants on splicing are specific criteria. Other situations “a”: (1) no genotype–phenotype correlation in the patient (2) no deleterious effect on protein for nonsynonymous variants in published functional studies (3) clinical case report of the variant with no functional study (4) new variant. Other situations “b”: (1) Variant in trans of a deleterious mutation (no other variant in cis) in patients (2) homozygous variant or in trans of another VUCS in patients (3) in cis of another VUCS in patients (no mutation found in trans) or in partner or relative in a familial context of cystic fibrosis and CFTR-RD (4) no other mutation found in patient or heterozygous in partner or relative in a familial context of cystic fibrosis and CFTR-RD.

Download figure to PowerPoint

General Criteria

Report of the variants in general population

The Short Genetic Variations database dbSNP (http://www.ncbi.nlm.nih.gov/snp/) and 1000 Genomes (http://www.1000genomes.org/) were used to collect the frequency of each variant in the general population or specific ethnic groups when available. The data of matched control populations is from studies performed in our laboratory on a panel of 100 individuals from the general population and from a European collaborative study of 191 normal individuals [Bombieri et al., 2000]. As mentioned in the CMGS guidelines, caution must be exerted for autosomal recessive conditions where heterozygous for pathogenic variants may be present at a high frequency in certain populations. In the particular case of the CFTR gene, only the mutation p.Phe508del is reported with a high frequency (0.8%) in control population then all newly identified variants with a frequency higher than 1% should be considered as probably neutral or of little clinical significance [Bombieri et al., 2000].

Description of the variants in literature and/or in databases

We retrieved information about the functional impact of the variants from the literature. Published functional studies were considered of major importance to classify variants either as probably neutral or probably pathogenic. Conversely, clinical descriptions and case reports were considered as decision criteria with insufficient information (Fig. 1: other situationsa), so that further exploration was required for the classification of the variant. In addition, we collected data from the CFMD (http://www.genet.sickkids.on.ca/cftr/Home.html) and we referred to the national collaborative knowledgebase [Bareil et al., 2010] that inventories all variations found in patients and healthy compound heterozygous subjects analyzed in nine specialized French laboratories.

Individual and family data components: position relative to another UV or to a deleterious mutation

Segregation studies of the parental alleles are essential to assign the variants in cis or in trans and will condition the following steps of the analysis. For this purpose, we collected all segregation data in parents and siblings to evaluate the cosegregation of the variants with the disease. We discriminated cases in favor of possible neutrality from situations requiring the evaluation of specific parameters (Fig. 1: other situationsb).

In Silico Analysis

Three different levels of sequence analysis were combined and final interpretation was based on concordant results of at least two of the three criteria.

First, the genomic sequence environment of the UVs was analyzed using two conventional splice prediction tools: Human Splicing Finder 2.4.1 (http://www.umd.be/HSF/HSF.html), which includes two different calculation algorithms (HSF and MaxEnt) [Desmet et al., 2009], and NNSplice (http://www.fruitfly.org/seq_tools/splice.html; Splice Site Prediction by Neural Network or SSPNN) [Reese et al., 1997]. We primarily considered the predictions concerning consensus splice sites or branch points and we used the previously defined decrease cutoff values of 10% (HSF and NNSplice) and 20% (MaxEnt) [Houdayer et al., 2008, 2012; Le Guedard-Mereuze et al., 2010]. The reliability of these cutoff values was further validated in our study (data not shown). When a de novo splice site was predicted, we considered that it could interfere with the natural spliceosome assembly when its score was higher than the wild-type (wt) version of the natural splice site.

Secondly, we took into account the interdependency between different positions of a given splicing motif [Carmel et al., 2004; Roca et al., 2008] using the SpliceAid 2 database (http://193.206.120.249/splicing_tissue.html) [Piva et al., 2012]. SpliceAid 2 is a Web resource collecting experimental RNA target motifs bound by splicing proteins in humans but also a large number of true 5′ and 3′ splice sites in human genes. This database aims to assess the representation of a given sequence as an efficient splice site in human and thus determine the impact of a nucleotide substitution on the whole motif. The results are expressed in “frequency percentiles” from the less to the most common splice sites in human (from 1 to 10). We considered as “noncritical” variants (1) leading to a splicing motif with a higher frequency score than the wt sequence or (2) creating a de novo splice site less represented than the natural splice site. Conversely, we considered that the nucleotide change could have a “critical” impact on splicing when the mutated splice site scored lower than the wt splice site or when the score of the de novo splice site was higher than the score of the natural splice site. When the score of the mutant splice site was equal to the wt, no conclusion could be drawn and this criterion was considered as “undefined.” As a third step, we evaluated the conservation of the nucleotide position affected by the substitution in a set of selected mammalian orthologs. Conservation scores of nucleotide sequence has been reported to be anticorrelated with SNPs rates, particularly from position −20 to +10 of exons, traducing the genetic constraints imposed by splicing [Castle, 2011]. We performed the alignment of the human genomic sequence of the CFTR gene (NG_016465.1) with 18 vertebrate orthologs on Ensembl Website (19 amniota vertebrates Pecan subset). The percentages of the wt and variant nucleotides were obtained by analyzing the multiple alignments with the Jalview software (http://www.jalview.org/). The wt nucleotide was considered as highly conserved when its frequency was higher than 90%, intermediately conserved when its frequency was between 50% and 90%, and poorly conserved when its frequency was lower than 50% and/or when the mutated nucleotide was found with a frequency of at least 10%.

In Vitro Functional Studies

Splicing reporter constructs

The impact on splicing of the 15 selected variants was tested using the pSPL3 exon-trapping vector [Burn et al., 1995], kindly provided by Dr I. Bottillo. For each sequence variation, the exon and intronic flanking sequences (about 200 bp) were amplified from the patient's genomic DNA using the High Fidelity Phusion® polymerase (Finnzymes, Espoo, Finland). Amplicons were inserted into the pSPL3 XhoI/NheI restriction sites, using the T4 DNA ligase High Concentration (Invitrogen, St Aubin, France) according to manufacturer's instructions. Control minigenes containing wt sequence were generated for each corresponding sequence variation. The sequence fidelity of minigene constructs was verified by direct sequencing.

Cell culture and transfection

Human bronchial cells BEAS-2B, known to exhibit a low level of endogenous CFTR expression, were cultured in DMEM supplemented with 5% fetal calf serum, 1% Ultroser® serum substitute (Pall Life Sciences BioPharmaceuticals, St Germain en Laye, France), 2 mmol/l glutamine, and 100 U/ml penicillin–0.1 mg/ml streptomycin. Cells were maintained at 37°C in 5% CO2 humidified atmosphere on 100 × 20 mm2 tissue culture dishes. Twenty-four hours after plating in six-well plates, about 80% confluent cells were transiently transfected with 1.5 µg of minigene constructs using the PolyFect® transfection reagent (Qiagen, Courtaboeuf, France) according to the manufacturer's instructions. Cells were harvested after 48 hr for transcripts analysis. Two independent transfections were done for each variant.

Transcripts analysis

Total RNA was extracted from BEAS-2B cells using RNeasy Plus kit (Qiagen, Courtaboeuf, France) according to the manufacturer's instructions. RNA quantity and quality were determined by the 260:280 nm absorbance ratios using a Biophotometer (Eppendorf, Le Pecq, France). RT-PCR was performed as previously described [Rene et al., 2011] using pSPL3-specific primers SD6 as forward primer and SA2 as reverse primer [Bottillo et al., 2007]. Five microliters of PCR products were resolved on 1.5% agarose gel. We quantified the rates (%) of aberrant transcripts for wt and mutant constructs using the Quantity One® (v. 4.6.9) software (Bio-Rad, Marnes-La-Coquette, France). The proportions of missplicing (∆missplicing) in mutants relative to wt were determined by calculating the differences in rates of aberrant transcripts between the wt and mutated constructs. ∆missplicing values lower than 50% were considered as nonsignificative and negative in the classification while ∆missplicing higher than 50% reflected a probable impact on splicing and were considered as positive. In addition, PCR products were sequenced using a Big Dye Terminators v1.1 Cycle Sequencing Kit (Applied Biosystems, Courtaboeuf, France).

Correlation Study of the Classification Method and In Vitro Functional Studies

We assessed the sensitivity and the specificity of the method by comparing the predicted class of the 15 variants, based on the criteria included in the decision tree, and the results of in vitro functional studies [Le Guedard-Mereuze et al., 2010; Thery et al., 2011]. The sensitivity was the capacity of the method to correctly classify variants affecting splicing in vitro (TP: True Positive; FN: False Negative). The specificity was its ability to classify as probably neutral the variants that did not cause splicing alterations in vitro (TN: true negative; FP: false positive). Both parameters are given by the following formulas:

  • display math

RESULTS

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods and Decision Criteria
  5. RESULTS
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Based on the CMGS recommendations, we built a decision tree to guide the classification of VUCS in the CFTR gene using an interpretation model that associates general parameters and specific criteria for in silico analysis (Fig. 1). A set of 15 VUCS identified in CF or CFTR-RD patients or in cases of suspected CF on abnormal ultrasound signs during pregnancy was selected from our cohort of patients. Of the 10 intronic sequence variations, eight were located at noncanonical positions of splice sites (Fig. 2A). Three variants consisted of substitutions at position +5 (c.869+5G>A and c.1766+5G>A) and +3 (c.579+3A>G) of 5′ss and five variants in 3′ss affected position ranging from −15 to −3. Two variants were located outside the consensus splice sites (c.1209+18A>C and c.1766+73T>G). Five exonic sequence variations were also studied (Fig. 2B): c.454A>G (p.(Met152Val)) in exon 4, c.532G>A (p.(Gly178Arg)) in exon 5, c.1581A>G (p.( = )) in exon 10, c.2706C>G (p.(Ser902Arg)) in exon 15 and c.3014T>G (p.(Ile1005Arg)) in exon 17a.

image

Figure 2. Localization of the 15 variants selected for the correlation study according to their position relative to splice sites. A: Intronic sequence variations. B: Exonic sequence variations. Brackets: intron (IVS) or exon (E) number is indicated/given for each variant according to legacy numbering of CFTR exons.

Download figure to PowerPoint

Epidemiological Data

Data were available for only two out of the 15 variants (Table 1). The c.274-6T>C variant was reported in 0.3% of 382 genes from control population [Bombieri et al., 2000]. This was below the threshold for SNP (minor allele frequency >1%) and no frequency data could be retrieved from dbSNP. Consequently, this variant could not be classified as probably neutral. The variant c.1581A>G (p.( = )), not found in our control population studies, is reported in dbSNP (rs1800094:A>G) with a minor allele frequency of 0.7% and an increased frequency in the sub-Saharan African population (1.7% on 120 chromosomes analyzed). Interestingly the patient carrying this variant originated from the sub-Saharan region. This variant was therefore considered as neutral.

Table 1. Results Obtained for the 15 Variants Concerning the Criteria Included in Our in-House Decision Tree
    In silico studies 
Mutation namea HGVS [legacy] nomenclatureFrequency >1%Literature/ LSDB*Position/2d mutation or VUCSComputational predictionsbSpliceAid 2Ortholog conservationOverall interpretationc (impact on splicing)
  1. *Yes, reported with no additional data; CR, clinical Case Report (“+”: reported in patients with clinical and segregation data; “−”: reported as non pathogenic); FS, functional studies (“+”: alteration of transcripts -t- or protein -p-).

  2. **Sub-Saharan African population (120 chromosomes analyzed).

  3. a

    HGVS nomenclature: cDNA sequence with +1 corresponding to the A of the ATG translation initiation codon (NM_000492.3). Brackets: legacy nomenclature with +133 corresponding to the A of the ATG translation initiation codon.

  4. b

    Using Human Splicing Finder 2.4.1 (HSF and MaxEnt) and NNSplice (SSPNN): concordant results of at least two of the three criteria.

  5. c

    Interpretations based on a majority of concordant results for all criteria.

Sequence variations at 5′ (donor) splice sites
c.579+3A>G [711+3A>G]NoFS t+Trans severe mutationPositiveUndefinedHighProbably pathogenic
c.869+5G>A [1001+5G>A]NoCR +Trans severe mutationPositiveUndefinedHighProbably pathogenic
c.1209+18A>C [1341+18A>C]NoCR −Cis mild mutation; trans severe mutationNegativeNoncriticalPoorProbably neutral
c.1766+5G>A [1898+5G>A]NoCR +Cis VUCS; trans severe mutationPositiveCriticalIntermediateProbably pathogenic
c.1766+73T>G [1898+73T>G]NoCR +Unassigned VUCSNegativeNoncriticalPoorProbably neutral
Sequence variations at 3′ (acceptor) splice sites
c.274-6T>C [406-6T>C]NoCR −No other mutation/UVNegativeCriticalPoorProbably neutral
c.1585-8G>A [1717-8G>A]NoCR +Trans severe mutationPositiveCriticalIntermediateProbably pathogenic
c.1585-3T>C [1717-3T>C]NoCR +Trans severe mutation in asymptomatic subjectsNegativeCriticalPoorProbably neutral
c.2909-15T>G [3041-15T>G]NoCR +Trans severe mutationNegativeCriticalIntermediateProbably pathogenic
c.3964-3C>G [4096-3C>G]NoCR +HomozygousPositiveCriticalIntermediateProbably pathogenic
Exonic sequence variations
c.454A>G/p.(Met152Val) [M152V]NoCR +Trans severe mutationPositiveCriticalHighProbably pathogenic
c.532G>A/p.(Gly178Arg) [G178R]NoFS p+Trans severe mutationNegativeNoncriticalHighProbably neutral
c.1581A>G/p.( = ) [E527E]Yes**YesNo other mutation/UVNegativeNoncriticalPoorProbably neutral
c.2706C>G/p.(Ser902Arg) [S902R]NoCR +Trans severe mutationPositiveUndefinedIntermediateProbably pathogenic
c.3014T>G/p.(Ile1005Arg) [I1005R]NoCR +Trans severe mutationNegativeNoncriticalHighProbably neutral

International Literature and Locus-Specific Databases

All the information collected for the 15 variants are summarized in Table 1. According to the international literature, the c.3014T>G (p.(Ile1005Arg)) variant was reported in 0.4% of CF German chromosomes and in 1.7% of Estonian CF chromosomes [Bobadilla et al., 2002; Dork et al., 1994]. The c.869+5G>A variant was described in a Chinese CBAVD patient [Goh et al., 2007]. In vitro functional studies were available for two variants: c.579+3A>G (711+3A>G using legacy nomenclature) and c.532G>A (p.(Gly178Arg)). The first one was recently shown to induce aberrant splicing of exon 5, suggesting a probable involvement in CFTR-related disorders [Sheridan et al., 2011]. Although the c.532G>A (p.(Gly178Arg)) mutation is classically considered as severe, a previous functional study reported no loss of protein synthesis and maturation but a partial defect of the chloride channel activity [Seibert et al., 1997]. This result was rather in favor of moderate pathogenicity (class IV) according to the general classification of deleterious mutations in the CFTR gene [Welsh and Smith, 1993]. This discrepancy between the phenotype and the genotype made us suspect a potential effect on splicing.

Regarding data from locus-specific databases (LSDB), four variants were reported in the CFMD in CF (c.1766+5G>A, c.1585-8G>A, c.454A>G) or CBAVD (c.2909-15T>G) patients. Conversely, c.1209+18A>C and c.274-6T>C were described as probably nonpathogenic. The variant c.1581A>G was reported [Cutting et al., 1992] with no additional information. In addition, 10 variants from our series were also reported by other laboratories in the French collaborative database. Five variants (c.1766+5G>A, c.1585-8G>A, c.3964-3C>G, c.532G>A, and c.3014T>G) were reported in CF patients, two variants (c.579+3A>G and c.2706C>G) were described both in CF and CFTR-RD patients, and three variants (c.1766+73T>G, c.274-6T>C, and c.1581A>G) were found in CFTR-RD patients.

Clinical and Segregation Data

Eleven variants were found in trans to a disease causing mutation in CF or CBAVD patients and one (c.3964-3C>G) was found in homozygous state in a CF patient from a consanguineous family (Table 1). The c.1209+18A>C variant was found in cis of p.(Gly1349Asp) in CF and CBAVD patients. The p.(Gly1349Asp) mutation was associated with mild clinical phenotypes and has been shown to induce gating dysfunctions of the CFTR channel [Bompadre et al., 2007]. The c.1209+18A>C variant could be considered as possibly neutral or could modulate the impact of p.(Gly1349Asp) in patients by an additive effect on splicing. The variant c.274-6T>C was identified in a pregnant patient referred because of abnormal ultrasound signs of fetal bowel. This is related to CF in 3%–7% of cases [Muller et al., 2002; Scotet et al., 2010]. Her partner was heterozygous for a severe CFTR mutation, in a context of secretory azoospermia. No other variant or disease-causing mutation could be found after the analysis of the complete coding sequence and intronic flanking regions in maternal DNA. The fetus was not analyzed. Finally, the variant c.1585-3T>C was described in trans to a severe mutation in a child diagnosed as possible CF through newborn screening. Sweat tests conducted later in this child were negative and the complete analysis of the family revealed two CF symptom-free brothers presenting the same genotype, but a possible male infertility has not been explored.

In Silico Analysis

Computational studies using the conventional splice prediction tools Human Splicing Finder 2.4.1 and NNSplice suggested that seven variants could affect splicing (Supp. Table S1). Four of them were predicted to lead to an exon skipping by decreasing the strength of the surrounding natural splice site and three would create a strong de novo splice site. Eight variants showed no predicted effect on consensus splice sites or branch points and were considered as negative (Supp. Table S1).

The analysis of exact splicing motives (wt and variants) using SpliceAid 2 database predicted seven variations as critical (all five variations within the consensus 3′ss, the c.1766+5G>A mutation at 5′ss and the c.454A>G missense mutation) and five as not critical. The potential impact of three variants remained undefined (Table 1).

The analysis of genomic sequence conservation in orthologs revealed that 10 variants out of 15, both intronic and exonic, were located at highly or intermediately conserved positions across species. The intronic positions +3 and +5 were highly or intermediately conserved in orthologs, when c.1209+18A>C and c.1766+73T>G affected poorly conserved nucleotides. In intron 3, the nucleotide T at position c.274-6 was less represented (26%) than the variant nucleotide C (68%) across species. The position c.1585-3 was also poorly conserved (T: 73% and C: 26%). The variants c.1585-8G>A, c.2909-15T>G and c.3964-3C>G affected intermediately conserved positions. We observed 89% conservation for positions c.2909-15T and c.3964-3C, close to the threshold of 90% considered for high degree of conservation. Nucleotides affected in exons 4, 5, and 17a were highly conserved, when the variant alleles c.2706C and c.1581G were found in 73% and 21% of species, respectively.

Finally, the overall interpretation of in silico results revealed a good correlation between nucleotide conservation in intronic sequences and predictions by Human Splicing Finder 2.4.1 and NNSplice (Table 1). Indeed, most variants affecting highly or intermediately conserved nucleotides were predicted to affect splicing (7/10; 70%), whereas all variants affecting poorly conserved positions obtained negative predictions (5/5; 100%). The results obtained using SpliceAid 2 were correlated with at least one of the other two in silico tools for 66% of the variants (10/15) and was decisive in the interpretation of the only variant with discordant results in computational predictions and ortholog conservation (c.2909-15T>G). By using SpliceAid 2, we reported no false negative predictions, two false positive predictions (c.274-6T>C and c.1585-3T>C) and no concluding predictions for the three remaining ones (c.579+3A>G, c.869+5G>A, and c.2706C>G/p.(Ser902Arg)).

The three levels of analysis (computational predictions, SpliceAid 2 and ortholog conservation) were concordant, in favor of a probable neutrality for the variants c.1209+18A>C, c.1766+73T>G, and c.1581A>G and in favor of a probable pathogenicity for the variants c.1766+5G>A, c.1585-8G>A, c.3964-3C>G, and c.454A>G.

In Vitro Functional Study

Results of the splicing patterns, for wt and mutated constructs, analyzed by gel electrophoresis are reported in Figure 3. Δmissplicing values (averages ± SEM) in mutants relative to wt are reported in Table 2. All results were confirmed by direct sequencing of RT-PCR products (data not shown).

image

Figure 3. Representative experiments of splicing patterns analysis for the 15 minigene constructs by agarose gel electrophoresis and percentages of aberrant splicing patterns measured in wt and mutant constructs (averages ± SEM of three independent experiments). A: Intronic variants in/close to 5′splice sites; B: intronic variants in 3′splice sites; C: exonic variants.

Download figure to PowerPoint

Table 2. Results of Ex Vivo Functional Studies of the 15 CFTR Variants Using Minigene Constructs and Correlation with the Overall Interpretation Using our Decision Tree
Mutation namea HGVS [legacy] nomenclatureResults in vitro studies (Δmissplicing percentages: average ± SEM)Interpretation using in-house algorithmConclusion
  1. Δmissplicing percentages represent the rate of aberrant splicing relative to the wt construct (see Fig. 3 for absolute missplicing values in wt and mutant constructs).

  2. a

    HGVS nomenclature: cDNA sequence with +1 corresponding to the A of the ATG translation initiation codon (NM_000492.3). Brackets: legacy nomenclature with +133 corresponding to the A of the ATG translation initiation codon.

  3. IF, in frame; OF, out-of-frame; TP, true positive; TN, true negative; FN, false negative.

Sequence variations at 5′ (donor) splice sites
c.579+3A>G [711+3A>G]Exon 5 skipped IF (82.58 ± 4.94)Probably pathogenicTP
c.869+5G>A [1001+5G>A]Exon 6b skipped IF (81.45 ± 4.56)Probably pathogenicTP
c.1209+18A>C [1341+18A>C]No splicing defect (5.64 ± 2.42)Probably neutralTN
c.1766+5G>A [1898+5G>A]Exon 12 skipped IF (77.49 ± 3.96)Probably pathogenicTP
c.1766+73T>G [1898+73T>G]No splicing defect, alternative transcripts (4.60 ± 4.56)Probably neutralTN
Sequence variations at 3′ (acceptor) splice sites
c.274-6T>C [406-6T>C]No splicing defect (1.22 ± 2.30)Probably neutralTN
c.1585-8G>A [1717-8G>A]Insertion (IF) 6pb (TAATAG) of intron 10 (91.31 ± 2.38)Probably pathogenicTP
c.1585-3T>C [1717-3T>C]No splicing defect (1.68 ±1.83)Probably neutralTN
c.2909-15T>G [3041-15T>G]Exon 16 skipped OF (81.23 ± 3.07)Probably pathogenicTP
c.3964-3C>G [4096-3C>G]Exon 22 skipped OF (85.81 ± 3.64)Probably pathogenicTP
Exonic sequence variations
c.454A>G/p.(Met152Val) [M152V]Deletion (IF) last 36 bp exon 4 (90.66 ± 1.79)Probably pathogenicTP
c.532G>A/p.(Gly178Arg) [G178R]Exon 5 skipped IF (78.76 ± 4.74)Probably neutralFN
c.1581A>G/p.( = ) [E527E]No splicing defect (3.81 ± 2.03)Probably neutralTN
c.2706C>G/p.(Ser902Arg) [S902R]Deletion (OF) last 203 bp exon 15 (88.92 ± 3.30)Probably pathogenicTP
c.3014T>G/p.(Ile1005Arg) [I1005R]No splicing defect (2.20 ± 1.04)Probably neutralTN
Splicing outcome of intronic sequence variations

As predicted by computational studies, neither variant c.1209+18A>C (intron 8) nor c.1766+73T>G (intron 12) was found to alter splicing when compared with wt constructs (respectively 5.64 ±2.42 and 4.60 ±4.56 percent Δmissplicing) (Fig. 3A). In our series, we found alternative transcripts in exon 12 wt construct, concordant with previous data [Haque et al., 2010]. On the other hand, the three variations located in the donor splice sites induced exon skipping and led to an in-frame deletion of exon 5 (c.579+3A>G: 82.58% ±4.94), exon 6b (c.869+5G>A: 81.45% ±4.56), and exon 12 (c.1766+5G>A: 77.49% ±3.96). Concerning the five variants at 3′ss (Fig. 3B), constructs carrying the c.274-6T>C or c.1585-3T>C sequence variations revealed normal splicing patterns compared with wt constructs. Alterations c.2909-15T>G and c.3964-3C>G were found to lead to aberrant splicing in 81.23% (±3.07) and 85.81% (±3.64) of transcripts respectively, both consisting of the out-of-frame skipping of exon 16 or 22 of the CFTR gene. Finally, the size of c.1585-8G>A transcripts appeared to be slightly greater than the wt one and direct sequencing of RT-PCR products revealed the presence of the six last nucleotides of intron 10 (TAATAG) at the 5′end of exon 11, leading to the in-frame inclusion of two consecutive premature termination codons TAA and TAG (data not shown).

Splicing outcome of exonic sequence variations

The variants c.1581A>G and c.3014T>G did not induce aberrant splicing of exon 10 and exon 17a, respectively, compared with the wt construct (Fig. 3C). Surprisingly and contrary to the predictions of the three algorithms used for the in silico analysis, the c.532G>A substitution was found to be deleterious for splicing and to cause the loss of the entire exon 5 in 78.76% (±4.74) of transcripts with a significant proportion of residual normal transcripts (21.24%). The analysis of exonic enhancer and silencer cis-regulatory sequences with the Human Splicing Finder tool revealed that the mutation created a potential consensus-binding site for the splicing repressor hnRNPA1 and for the SRp40 protein (Supp. Fig. S1). Finally, consistent with the in silico predictions, the variants c.454A>G and c.2706C>G created de novo donor splice sites used in place of the natural splice sites and induced the in-frame loss of the last 36 nucleotides of exon 4 in 90.66% (±1.79) of transcripts and the out-of-frame loss of the last 203 nucleotides of exon 15 in 88.92% (±3.30) of transcripts, respectively.

Overall Interpretation of VUCS and Performance of the Method

By using the multicriteria decision steps included in our flowchart, seven variants were classified as probably neutral regarding their impact on splicing; these include two missense mutations, c.532G>A (p.(Gly178Arg)) and c.3014T>G (p.(Ile1005Arg)), which could lead instead to a CFTR protein structure and/or function alteration (Table 1). This conclusion is in line with a previous study showing that the missense variant p.(Gly178Arg) alters CFTR activity [Seibert et al., 1997]. The direct consequences of p.(Ile1005Arg) on CFTR protein should be further evaluated using specific bioinformatics and functional studies, to confirm or exclude its potential pathogenicity. The eight other variants in our study, c.579+3A>G, c.869+5G>A, c.1766+5G>A, c.1585-8G>A, c.2909-15T>G, c.3964-3C>G, c.454A>G, and c.2706C>G, were considered as probably pathogenic on splicing.

The overall correlation between this classification and the in vitro functional studies was high. Fourteen variants out of the fifteen (93.33%) were correctly classified regarding experimental results (Table 2). The specificity of our method was 100%, as we observed six true negative and no false positive interpretation. Its sensitivity was high (88.89%) with eight true positive correctly predicted and only one false negative variant, the exonic c.532G>A variant, classified as probably neutral on splicing (Table 1) but leading in vitro to a high proportion of transcripts lacking CFTR exon 5.

Discussion

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods and Decision Criteria
  5. RESULTS
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

The decision tree developed in this study is based on the general approach for predicting the pathogenicity of VUCS. International guidelines recommend a combination of general (epidemiological and familial data) and specific criteria depending on the type of the variant. A bias can be introduced if analysis only takes into account the position of the VUCS in the coding versus intronic region. Indeed, the effect of an exonic nonsynonymous variant will be evaluated only at protein level, ignoring the possible consequences of the nucleotide substitution on splicing. We built our classification method avoiding this preselection by evaluating, as a first step, the impact of all types of VUCS on pre-mRNA splicing. Then only nonsynonymous variants predicted to have no impact on splicing would secondarily be evaluated for their impact on the protein (Fig. 1). We performed in depth study of 15 representative CFTR variants to evaluate performance of this method compared to in vitro functional studies, and finally validated their classification as probably neutral or probably pathogenic on splicing.

The interest of our flowchart is to perform a sequential analysis from general parameters to accurate specific criteria, where each step acts as a filter to guide the classification of variants (Fig. 4). We considered epidemiological data as the first step of the analysis as all variants reported with a frequency higher than 1% in the matched population can be classified as probably neutral (e.g., c.1581A>G). The following step is the review of the international literature and LSDB. In our series, the variant c.579+3A>G could correctly be classified as probably pathogenic, since it was reported four times in CF and CFTR-RD patients in trans of severe mutations in the national collaborative knowledgebase [Bareil et al., 2010]. Moreover, it has recently been shown to affect splicing [Sheridan et al., 2011]. This database also provided valuable information for evaluating the cosegregation of variants with the disease in several families. For instance, the variant c.1585-3T>C found in two symptom-free brothers who carried the same genotype could be classified as possibly neutral.

image

Figure 4. Sequential classification of the 15 VUCS using our decision tree. *The remaining unclassified variant (c.3014T>G/p.(Ile1005Arg)) revealed no impact on splicing and has to be tested at protein level.

Download figure to PowerPoint

Classically, several dedicated splice prediction tools can be used for the assessment of the impact on splicing of VUCS. They were shown to act as efficient filters [Houdayer et al., 2012; Vreeswijk et al., 2009], however their performance could be improved as they generate some false positive or false negative predictions [Le Guedard-Mereuze et al., 2010; Thery et al., 2011]. In the present work, performance of those tools was concordant with previous studies. Then, to further improve in silico predictions, we included two additional levels of analysis (SpliceAid 2 and ortholog conservation) specifically suited to the assessment of splicing. Overall, we observed a good correlation between nucleotide conservation among orthologs and in vitro studies. All intronic and three exonic variants affecting highly and intermediately conserved nucleotides led to aberrant splicing in vitro. Conversely, all variants affecting poorly conserved nucleotides revealed no impact on splicing. The second tool, SpliceAid 2, was initially developed to reference experimental RNA target motifs bound by splicing proteins in humans. However, true 5′ and 3′ splice sites, polypyrimidine tracts, and branch point sequences were added [Piva et al., 2012]. We used this feature complementary to ortholog sequence conservation. Ten variants gave concordant results with in vitro functional studies (five negatives and five positives). Two negative variants after functional studies gave false positive results (predicted as “critical”) and three variants considered as “undefined” had an impact on splicing in vitro. This could be due to the inflexibility of the examined parameters in SpliceAid2, considering exact splicing motives found in human genes. There is certain plasticity within splice sites, that can explain the occurrence of tolerate variations (SNPs): a nucleotide change at a given position can be compensated by changing another position of the sequence, thus maintaining the base pairing with splicing factors above a minimal number [Le Guedard-Mereuze et al., 2009]. However, SpliceAid 2 gave results consistent with at least one of the two other criteria used for the in silico analysis in 10 of the 15 variants analyzed. Furthermore, it contributed to classify one additional variant (c.2909-15T>G in intron 15) as probably pathogenic upon splicing compared with conventional splice prediction tools, suggesting that our strategy is likely to improve the performance of the classification of VUCS.

After in silico analysis, two variants remained unclassified: c.532G>A and c.3014T>G. The c.3014T>G was correctly classified as probably neutral on splicing and will require an additional evaluation to determine its potential impact on the CFTR protein. The only false negative interpretation in our approach concerned the exonic mutation c.532G>A (p.(Gly178Arg)), classified as probably neutral before functional studies. In minigene experiments, the c.532G>A (p.(Gly178Arg)) led to skipping of exon 5 in 78% of the transcripts. We hypothesize that this mutation may modify auxiliary splicing regulatory elements. According to Human Splicing Finder predictions, it could create a high-affinity binding site for hnRNPA1 which functions as a splicing repressor and is commonly involved in alternative splicing [Matlin et al., 2005]. Further functional studies are necessary to confirm in silico analyzes [Spurdle et al., 2008], especially for the identification of trans regulating factors binding to the wt and mutated sequences [Disset et al., 2006; Luo and Reed, 2003). Thus, the c.532G>A (p.(Gly178Arg)) led to only partial exon skipping, so that its pathogenicity could rely on a cumulative effect of incomplete aberrant splicing and the partial defect of the chloride channel activity [Seibert et al., 1997] due to the amino acid change in the residual full-length transcripts. These observations point out that exonic mutations affecting cis regulatory elements are still difficult to apprehend without experimental analysis, that should be performed when possible [Spurdle et al., 2008]. Because testing every nonsynonymous variant is not feasible, priority should be given to alterations showing inconclusive predicted effect on the protein and/or discrepancy between genotype and phenotype.

Finally, our method showed a high performance level with 100% specificity and about 89% sensitivity. The 15 variants studied here reflect the panel of situations encountered in CF molecular diagnosis as they are both intronic (3′ splice sites from −20 to −3; 5′splice sites and other intronic positions) and exonic (missense and synonymous) and affect constitutively and alternatively spliced exons of the CFTR gene. Thus, its performance is expected to be reproducible whatever the set of variants tested. Furthermore, the use of the decision tree would have alleviated in vitro functional studies in 87% cases (13 variants/15). It is also worth mentioning that three exonic variants were reclassified as probably pathogenic toward splicing, as might be the case for nearly 20% of disease alleles originally classified as missense mutations [Lim et al., 2011]. This demonstrates the need to consider splicing assessment in first-line of a classification model to prevent misclassification of spliceogenic variants that could have no or negligible direct effect on protein function [Robinson et al., 2012].

We assume that our decision tree may be useful to orientate further studies that are necessary to characterize the functional impact of splicing alterations on the CFTR protein and refine genotype/phenotype correlation. Although functional assays are contributive in many cases, the clinical significance of some detected aberrant splicing events is still challenging, particularly for exon deletions that do not disrupt the coding frame. This is also the case for mutations leading to partial exon skipping, as the threshold of pathogenic/tolerated aberrant splicing is difficult to establish. The issue becomes more problematic when aberrant splicing occurs in an alternatively spliced exon. Indeed, the proportion of alternative splicing for a given exon is an important parameter to consider [Spurdle et al., 2008], and can differ depending on cell type and tissue due to the differential expression of splicing factors [de la Grange et al., 2010]. The use of nasal epithelial cells of patients, the only validated model accessible to noninvasive collection, could provide a definite evaluation of the variant [Carvalho-Oliveira et al., 2004; Costa et al., 2011]. In the same way, a supplementary blood sample for transcripts analysis is required for the classification of VUCS in BRCA1 or BRCA2 gene [Houdayer et al., 2012; Spurdle et al., 2008]. Thus, the identification of abnormal splicing patterns, both qualitative and quantitative, associated with the evaluation of their consequences on the CFTR protein will determine the functional impact of the variant according to the classification proposed by in 1993 [Welsh and Smith, 1993] and validate genotype–phenotype correlation.

In conclusion, if in vitro studies are efficient to assess the pathogenicity of rare variants in the CFTR gene, VUCS are often identified in an emergency context (CF suspicion in fetuses, partner of a CF patient or heterozygous subject when a pregnancy is underway) in which these studies may not be implemented. We demonstrate that our classification method plays here a full role in interpreting the results of molecular diagnosis, and it will now have to be definitely validated on a larger series. It will help in the harmonization of practices for the interpretation of VUCS following the requirements for Quality Assurance and Certification in molecular diagnosis. Finally, after refining the interpretation criteria to ensure the relevance of the predictions, this method could be adapted to other autosomal recessive diseases, for which no model is currently available.

Acknowledgments

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods and Decision Criteria
  5. RESULTS
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We acknowledge all members of the technical team for CF molecular diagnosis: Jean-Pierre Altieri, Caroline Guitard, Carine Templin, and Fanny Verneau. We thank Julie Miro and Victoria Viart for their technical support in functional studies. We are also grateful to all the collaborating members of the French molecular database and to the clinicians who initially referred the patients. Finally, we thank the French Association against CF (Vaincre la Mucoviscidose) for its constant support.

Disclosure statement: The authors declare no conflict of interest

References

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods and Decision Criteria
  5. RESULTS
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information
  • Bareil C, Theze C, Beroud C, Hamroun D, Guittard C, Rene C, Paulet D, Georges M, Claustres M. 2010. UMD-CFTR: a database dedicated to CF and CFTR-related disorders. Hum Mutat 31(9):10111019.
  • Bienvenu T, Sermet-Gaudelus I, Burgel PR, Hubert D, Crestani B, Bassinet L, Dusser D, Fajac I. 2010. Cystic fibrosis transmembrane conductance regulator channel dysfunction in non-cystic fibrosis bronchiectasis. Am J Respir Crit Care Med 181(10):10781084.
  • Bobadilla JL, Macek M, Jr., Fine JP, Farrell PM. 2002. Cystic fibrosis: a worldwide analysis of CFTR mutations–correlation with incidence data and application to screening. Hum Mutat 19(6):575606.
  • Bombieri C, Giorgi S, Carles S, de Cid R, Belpinati F, Tandoi C, Pallares-Ruiz N, Lazaro C, Ciminelli BM, Romey MC, Casals T, Pompei F, et al. 2000. A new approach for identifying non-pathogenic mutations. An analysis of the cystic fibrosis transmembrane regulator gene in normal individuals. Human Genet 106(2):172178.
  • Bompadre SG, Sohma Y, Li M, Hwang TC. 2007. G551D and G1349D, two CF-associated mutations in the signature sequences of CFTR, exhibit distinct gating defects. J Gen Physiol 129(4):285298.
  • Bottillo I, De Luca A, Schirinzi A, Guida V, Torrente I, Calvieri S, Gervasini C, Larizza L, Pizzuti A, Dallapiccola B. 2007. Functional analysis of splicing mutations in exon 7 of NF1 gene. BMC Med Genet 8:4.
  • Burn TC, Connors TD, Klinger KW, Landes GM. 1995. Increased exon-trapping efficiency through modifications to the pSPL3 splicing vector. Gene 161(2):183187.
  • Carmel I, Tal S, Vig I, Ast G. 2004. Comparative analysis detects dependencies among the 5′ splice-site positions. RNA 10(5):828840.
  • Carvalho-Oliveira I, Efthymiadou A, Malho R, Nogueira P, Tzetis M, Kanavakis E, Amaral MD, Penque D. 2004. CFTR localization in native airway cells and cell lines expressing wild-type or F508del-CFTR by a panel of different antibodies. J Histochem Cytochem: Off J Histochem Soc 52(2):193203.
  • Castle JC. 2011. SNPs occur in regions with less genomic sequence conservation. PloS ONE 6(6):e20660.
  • Claustres M. 2005. Molecular pathology of the CFTR locus in male infertility. Reprod Biomed Online 10(1):1441.
  • Claustres M, Guittard C, Bozon D, Chevalier F, Verlingue C, Ferec C, Girodon E, Cazeneuve C, Bienvenu T, Lalau G, Dumur V, Feldmann D, et al. 2000. Spectrum of CFTR mutations in cystic fibrosis and in congenital absence of the vas deferens in France. Hum Mutat 16(2):143156.
  • Costa C, Pruliere-Escabasse V, de Becdelievre A, Gameiro C, Golmard L, Guittard C, Bassinet L, Bienvenu T, Georges MD, Epaud R, Bieth E, Giurgea I, et al. 2011. A recurrent deep-intronic splicing CF mutation emphasizes the importance of mRNA studies in clinical practice. J Cystic Fibrosis: Off J Eur Cystic Fibrosis Soc 10(6):479482.
  • Cutting GR, Curristin SM, Nash E, Rosenstein BJ, Lerer I, Abeliovich D, Hill A, Graham C. 1992. Analysis of four diverse population groups indicates that a subset of cystic fibrosis mutations occur in common among Caucasians. Am J Hum Genet 50(6):11851194.
  • Dequeker E, Stuhrmann M, Morris MA, Casals T, Castellani C, Claustres M, Cuppens H, des Georges M, Ferec C, Macek M, Pignatti PF, Scheffer H, et al. 2009. Best practice guidelines for molecular genetic diagnosis of cystic fibrosis and CFTR-related disorders–updated European recommendations. Eur J Hum Genet 17(1):5165.
  • Desmet FO, Hamroun D, Lalande M, Collod-Beroud G, Claustres M, Beroud C. 2009. Human Splicing Finder: an online bioinformatics tool to predict splicing signals. Nucleic Acids Res 37(9):e67.
  • Disset A, Bourgeois CF, Benmalek N, Claustres M, Stevenin J, Tuffery-Giraud S. 2006. An exon skipping-associated nonsense mutation in the dystrophin gene uncovers a complex interplay between multiple antagonistic splicing elements. Hum Mol Genet 15(6):9991013.
  • Dork T, Mekus F, Schmidt K, Bosshammer J, Fislage R, Heuer T, Dziadek V, Neumann T, Kalin N, Wulbrand U, Wulf B, von der Hardt H, et al. 1994. Detection of more than 50 different CFTR mutations in a large group of German cystic fibrosis patients. Hum Genet 94(5):533542.
  • Goh DL, Zhou Y, Chong SS, Ngiam NS, Goh DY. 2007. Novel CFTR gene mutation in a patient with CBAVD. J Cystic Fibrosis: Off J Eur Cystic Fibrosis Soc 6(6):423425.
  • Goldgar DE, Easton DF, Deffenbaugh AM, Monteiro AN, Tavtigian SV, Couch FJ. 2004. Integrated evaluation of DNA sequence variants of unknown clinical significance: application to BRCA1 and BRCA2. Am J Hum Genet 75(4):535544.
  • Grange P, Gratadou L, Delord M, Dutertre M, Auboeuf D. 2010. Splicing factor and exon profiling across human tissues. Nucleic Acids Res 38(9):28252838.
  • Le Guedard-Mereuze S, Vache C, Molinari N, Vaudaine J, Claustres M, Roux AF, Tuffery-Giraud S. 2009. Sequence contexts that determine the pathogenicity of base substitutions at position +3 of donor splice-sites. Hum Mutat 30(9):13291339.
  • Haque A, Buratti E, Baralle FE. 2010. Functional properties and evolutionary splicing constraints on a composite exonic regulatory element of splicing in CFTR exon 12. Nucleic Acids Res 38(2):647659.
  • Houdayer C, Caux-Moncoutier V, Krieger S, Barrois M, Bonnet F, Bourdon V, Bronner M, Buisson M, Coulet F, Gaildrat P, Lefol C, Leone M, et al. 2012. Guidelines for splicing analysis in molecular diagnosis derived from a set of 327 combined in silico/in vitro studies on BRCA1 and BRCA2 variants. Hum Mutat 33(8):12281238.
  • Houdayer C, Dehainault C, Mattler C, Michaux D, Caux-Moncoutier V, Pages-Berhouet S, d'Enghien CD, Lauge A, Castera L, Gauthier-Villars M, Stoppa-Lyonnet D. 2008. Evaluation of in silico splice tools for decision-making in molecular diagnosis. Hum Mutat 29(7):975982.
  • Kerem B, Rommens JM, Buchanan JA, Markiewicz D, Cox TK, Chakravarti A, Buchwald M, Tsui LC. 1989. Identification of the cystic fibrosis gene: genetic analysis. Science 245(4922):10731080.
  • Lim KH, Ferraris L, Filloux ME, Raphael BJ, Fairbrother WG. 2011. Using positional distribution to identify splicing elements and predict pre-mRNA processing defects in human genes. Proc Natl Acad Sci USA 108(27):1109311098.
  • Luo MJ, Reed R. 2003. Identification of RNA binding proteins by UV cross-linking. Curr Prot Mol Biol. Edited by Ausubel et al. Chapter 27:Unit 27 2.
  • Matlin AJ, Clark F, Smith CW. 2005. Understanding alternative splicing: towards a cellular code. Nature reviews. Mol Cell Biol 6(5):386398.
  • Le Guedard-Mereuze S, Vache C, Baux D, Faugere V, Larrieu L, Abadie C, Janecke A, Claustres M, Roux AF, Tuffery-Giraud S. 2010. Ex vivo splicing assays of mutations at noncanonical positions of splice sites in USHER genes. Hum Mutat 31(3):347355.
  • Muller F, Simon-Bouy B, Girodon E, Monnier N, Malinge MC, Serre JL. 2002. Predicting the risk of cystic fibrosis with abnormal ultrasound signs of fetal bowel: results of a French molecular collaborative study based on 641 prospective cases. Am J Med Genet 110(2):109115.
  • Piva F, Giulietti M, Burini AB, Principato G. 2012. SpliceAid 2: a database of human splicing factors expression data and RNA target motifs. Hum Mutat 33(1):8185.
  • Plon SE, Eccles DM, Easton D, Foulkes WD, Genuardi M, Greenblatt MS, Hogervorst FB, Hoogerbrugge N, Spurdle AB, Tavtigian SV. 2008. Sequence variant classification and reporting: recommendations for improving the interpretation of cancer susceptibility genetic test results. Hum Mutat 29(11):12821291.
  • Reese MG, Eeckman FH, Kulp D, Haussler D. 1997. Improved splice site detection in Genie. J Comput Biol: A J Computat Mol Cell Biol 4(3):311323.
  • Rene C, Paulet D, Girodon E, Costa C, Lalau G, Leclerc J, Cabet-Bey F, Bienvenu T, Blayau M, Iron A, Mittre H, Feldmann D, et al. 2011. p.Ser1235Arg should no longer be considered as a cystic fibrosis mutation: results from a large collaborative study. Eur J Hum Genet 19(1):3642.
  • Riordan JR, Rommens JM, Kerem B, Alon N, Rozmahel R, Grzelczak Z, Zielenski J, Lok S, Plavsic N, Chou JL, et al. 1989. Identification of the cystic fibrosis gene: cloning and characterization of complementary DNA. Science 245(4922):10661073.
  • Robinson DO, Lin F, Lyon M, Raponi M, Cross E, White HE, Cox H, Clayton-Smith J, Baralle D. 2012. Systematic screening of FBN1 gene unclassified missense variants for splice abnormalities. Clin Genet 82(3):223231.
  • Roca X, Olson AJ, Rao AR, Enerly E, Kristensen VN, Borresen-Dale AL, Andresen BS, Krainer AR, Sachidanandam R. 2008. Features of 5′-splice-site efficiency derived from disease-causing mutations and comparative genomics. Genome Res 18(1):7787.
  • Rommens JM, Iannuzzi MC, Kerem B, Drumm ML, Melmer G, Dean M, Rozmahel R, Cole JL, Kennedy D, Hidaka N, et al. 1989. Identification of the cystic fibrosis gene: chromosome walking and jumping. Science 245(4922):10591065.
  • Scotet V, Dugueperoux I, Audrezet MP, Audebert-Bellanger S, Muller M, Blayau M, Ferec C. 2010. Focus on cystic fibrosis and other disorders evidenced in fetuses with sonographic finding of echogenic bowel: 16-year report from Brittany, France. Am J Obstet Gynecol 203(6):592 e1592 e6.
  • Seibert FS, Jia Y, Mathews CJ, Hanrahan JW, Riordan JR, Loo TW, Clarke DM. 1997. Disease-associated mutations in cytoplasmic loops 1 and 2 of cystic fibrosis transmembrane conductance regulator impede processing or opening of the channel. Biochemistry 36(39):1196611974.
  • Sheridan MB, Hefferon TW, Wang N, Merlo C, Milla C, Borowitz D, Green ED, Mogayzel PJ, Jr., Cutting GR. 2011. CFTR transcription defects in pancreatic sufficient cystic fibrosis patients with only one mutation in the coding region of CFTR. J Med Gen 48(4):235241.
  • Spurdle AB, Couch FJ, Hogervorst FB, Radice P, Sinilnikova OM. 2008. Prediction and assessment of splicing alterations: implications for clinical testing. Hum Mutat 29(11):13041313.
  • Thery JC, Krieger S, Gaildrat P, Revillion F, Buisine MP, Killian A, Duponchel C, Rousselin A, Vaur D, Peyrat JP, Berthet P, Frebourg T, et al. 2011. Contribution of bioinformatics predictions and functional splicing assays to the interpretation of unclassified variants of the BRCA genes. Eur J Hum Genet 19(10):10521058.
  • Vreeswijk MP, Kraan JN, van der Klift HM, Vink GR, Cornelisse CJ, Wijnen JT, Bakker E, van Asperen CJ, Devilee P. 2009. Intronic variants in BRCA1 and BRCA2 that affect RNA splicing can be reliably selected by splice-site prediction programs. Hum Mutat 30(1):107114.
  • Welsh MJ, Smith AE. 1993. Molecular mechanisms of CFTR chloride channel dysfunction in cystic fibrosis. Cell 73(7):12511254.
  • Whitcomb DC. 2010. Genetic aspects of pancreatitis. Annu Rev Med 61:413424.

Supporting Information

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods and Decision Criteria
  5. RESULTS
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Disclaimer: Supplementary materials have been peer-reviewed but not copyedited.

FilenameFormatSizeDescription
humu22291-sup-0001-si.pdf284KSupplementary Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.