Array-Based FMR1 Sequencing and Deletion Analysis in Patients with a Fragile X Syndrome–Like Phenotype

Background Fragile X syndrome (FXS) is caused by loss of function mutations in the FMR1 gene. Trinucleotide CGG-repeat expansions, resulting in FMR1 gene silencing, are the most common mutations observed at this locus. Even though the repeat expansion mutation is a functional null mutation, few conventional mutations have been identified at this locus, largely due to the clinical laboratory focus on the repeat tract. Methodology/Principal Findings To more thoroughly evaluate the frequency of conventional mutations in FXS-like patients, we used an array-based method to sequence FMR1 in 51 unrelated males exhibiting several features characteristic of FXS but with normal CGG-repeat tracts of FMR1. One patient was identified with a deletion in FMR1, but none of the patients were found to have other conventional mutations. Conclusions/Significance These data suggest that missense mutations in FMR1 are not a common cause of the FXS phenotype in patients who have normal-length CGG-repeat tracts. However, screening for small deletions of FMR1 may be of clinically utility.


Introduction
Fragile X syndrome (FXS) is an X-linked dominant disorder that is the most frequently encountered form of inherited intellectual disability. In 1991, the common causal mutation in FXS was identified to be a large CGG trinucleotide repeat expansion in the 59-untranslated region of the gene FMR1, the socalled full mutation [1]. Shortly thereafter, several groups identified FMR1 deletions in FXS patients, suggesting that multiple mutational mechanisms could give rise to the disorder [2,3,4,5]. The subsequent identification of an I304N FMR1 missense mutation in a severely affected FXS patient suggested that yet another class of FMR1 mutation was potentially a significant cause of disease [6]. However, while both trinucleotide repeat expansion [7] and FMR1 deletions [8] have proven to be the usual basis of FXS, no additional missense mutations have been identified in the subsequent 17 years.
Several groups have previously attempted to identify additional FMR1 missense mutations in patients without the full mutation but presenting with an FXS-like phenotype [9,10,11,12,13]. However, these previous studies were mutational screens and not designed to comprehensively evaluate the frequency of FMR1 missense mutations in FXS. Three of the studies surveyed fewer than ten FXS-like patients [9,10,12], while the other two studies used less proven detection methods to survey only a portion of the FMR1 coding sequence [11,13]. There is a lack of case reports and clinical studies detailing individuals with coding changes in FMR1 since FMR1 sequencing is rarely performed in the clinical setting. Thus, the frequency of such mutations responsible for a FXS clinical picture is not known.
In this study, we used array-based resequencing to search for missense mutations in FMR1 in a population of 51 unrelated FXSlike males. Despite achieving a high level of sequence coverage and accuracy, we did not identify any missense variants in FMR1, nor did we identify any novel noncoding variants likely to have a functional effect. Our method did, however, identify a pathogenic FMR1 deletion in a patient with FXS.

Subjects and Samples
This study was approved by the Emory University Institutional Review Board (IRB ID: 1317-2004). All patients and/or legal guardians gave written informed consent to participate in this study. We recruited 51 unrelated intellectually disabled males who previously tested negative for the FMR1 full mutation (.200 CGG repeats) and exhibited at least two of the FXS-like features listed in Table 1. Forty-seven of the patients were of European descent and four were of African descent. A focused clinical history and either a blood or saliva specimen were obtained from each patient. DNA was extracted from the obtained specimens using standard methods as were isolation of lymphoblastoid cells from whole blood. The following PCR conditions were used for LR-PCR-A: initialization at 95uC for 4 minutes; 37 cycles of denaturation at 95uC for 30 seconds and annealing/elongation at 60uC for 4 minutes; and a final elongation step of 72uC for 9 minutes. Each LR-PCR-B, -C, and -D reaction contained 50 ng of genomic DNA, 100 ng of each primer, 4 ml of dNTPs (Takara), 2.5 ml of Ex Taq Buffer (Takara), and 0.4 ml of Ex Taq (Takara), in a total of 25 ml. The following PCR conditions were used for LR-PCR-B: initialization at 94uC for 4 minutes; 30 cycles of denaturation at 94uC for 20 seconds and annealing/elongation at 64uC for 8 minutes; and a final elongation step of 68uC for 13 minutes. The same conditions were used for LR-PCR-C, but 35 cycles of denaturation and annealing/elongation were used instead of 30. The same conditions used for LR-PCR-C were used for LR-PCR-D, but the annealing/elongation at 64uC was continued for 9 minutes instead of 8 minutes. The expected sizes of the LR-PCR amplicons were confirmed by gel electrophoresis.
Sequencing-by-hybridization. FMR1 sequencing was performed on Custom Resequencing Arrays (Affymetrix, Santa Clara, CA), designed to provide coverage of all 17 FMR1 exons and the FMR1 promoter, plus 200 bp of flanking intronic sequence ( Figure 1). Patient sample amplicons were processed for sequencing-byhybridization according to the Affymetrix CustomSeq Resequencing Array protocol, Version 2.1, with the following exceptions. The four LR-PCR amplicons per patient were pooled in equimolar fashion to a total of 4 mg and digested with 0.2 units of DNase I (Promega, Madison, WI) at 37uC for 3 minutes, yielding digestion products between 100-600 bp. Labeling, hybridization, and array processing were performed as per the protocol.
Variant Detection and Confirmation. Base-calling was performed with the ABACUS statistical method [14] using the POPGEN genotyping software [15]. Putative variants were confirmed by traditional Sanger sequencing of fresh LR-PCR amplicons. Both POPGEN data and DNA chromatograms were inspected manually with the SeqScape software (Applied Biosystems, Foster City, CA).

Western Blotting
Immunoblotting was performed using standard methods. Briefly, patient and control lymphoblastoid cells were lysed with a standard Triton X-100-based lysis buffer. The lysate protein concentrations were measured with the Bradford assay. Proteins were denatured by heating at 95uC for 3 minutes and separated by polyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane. To assess protein loading and transfer, the membrane was reversibly stained with Ponceau S. The membrane was blocked for one hour in blocking buffer (10 g dry milk, 200 ml Tween-20, and 100 ml PBS), probed with primary antibody (anti-FMRP 1a or anti-eIF4e) overnight, and probed for one hour with horseradish-peroxidase conjugated anti-mouse secondary antibodies. Proteins were detected by chemiluminescence (ECL, GE Healthcare, Piscataway, NJ).

Sequence Accuracy
Across the 51 FXS-like patients sequenced by array hybridization, 99.6% of bases were called with high reliability, as determined by a quality score of 30 or greater. The high level of sequence accuracy is further demonstrated by the identification of known polymorphisms. As seen in Table 2, we detected all seven SNPs catalogued in dbSNP (build 130) for which the population frequency has been measured in HapMap samples. For the sake of comparison, we weighted the HapMap frequency data by the racial distribution of our patient population. None of the SNPs were found to be at a statistically different frequency in the FXS-

Novel FMR1 Sequence Variants
Notably, no novel variants were detected in the FMR1 coding sequence in the population of 51 FXS-like males. However, two novel intronic variants, c.52-47A.G and c.105-179G.T, were identified in FMR1 ( Table 3). As an assessment of possible functional relevance, we examined the mammalian conservation of these nucleotide positions and their genomic regions using phyloP and phastCons scores, respectively [16]. Because both variants are located in poorly conserved genomic regions (phastCons of 0.01), it is likely that they represent rare neutral variants that lack functional significance.

Array-Based Deletion Detection
In addition to detecting point mutations, resequencing arrays allow the detection of deletions. In one FXS-like patient, we identified a 355 bp deletion extending from 220 bp upstream of the CGG repeat through the second codon of the FMR1 coding sequence (i.e. hg18, chr.X: 146801041-146801395). After confirming this deletion with Sanger sequencing, we assessed its effects on FMRP translation. As shown in Figure 2, immunoblot analysis of patient lymphoblastoid cell line lysates revealed an absence of FMRP expression.

Discussion
We have sequenced the promoter, exons, and splice junctions of FMR1 in 51 unrelated patients with several classic features of FXS but without the full mutation utilizing resequencing arrays. Two novel intronic variants were identified which likely have no functional effect. Notably no missense or promoter mutations were found. As the largest sequencing analysis of FXS-like patients to date, these data suggest that FMR1 sequence variants are not a significant cause of the FXS phenotype.
At the present time, two missense changes in FMR1 have been identified, the benign and polymorphic p.A145S variant (rs29281) and the p.I304N mutation previously detected in a severely affected FXS-like patient [6]. It is surprising that these are the only missense changes that have been found in FMR1. In comparison, over 100 distinct point mutations in the nearby gene MECP2 have been shown to cause Rett syndrome, despite the fact that the gene is smaller and more recently identified than FMR1 [17]. Furthermore, because a functional absence of the FMR1 gene product is compatible with life, albeit associated with the symptoms of FXS, missense changes in FMR1, which in many cases would be less damaging than a loss-of-function, should not lead to embryonic lethality.
Since there is no reason to assume the FMR1 gene is less mutable than any other gene, why are conventional mutations uncommon among patients presenting with FXS-like features but without the full mutation? There are several possible explanation for absence of missense mutations. First, unlike Rett syndrome or many other Mendelian syndromes, the phenotype of FXS is subtle and variable. This makes a firm clinical diagnosis often difficult, even for an experienced clinician. Second, many syndromic aspects of FXS individually are not unusual in a developmentally delayed male population (i.e. language delay) and our criteria of only two features (Table 1) for study inclusion may have been too lenient. Third, it is possible that the phenotypic consequence of missense mutations might be distinct from classic FXS, leading to   a more subtle isolated developmental/behavioral phenotype, such nonspecific intellectual disability, or even autism, learning disability, anxiety disorder or attention-deficit/hyperactivity disorder, without overall intellectual disability. Similarly, a FMR1 missense mutation could selectively alter the function of only one domain of FMRP, thereby causing a specific FXS-like symptom, such as connective tissue defects or macro-orchidism, in the absence of an overall FXS-like phenotype. Given the already high level of genetic heterogeneity among patients with developmental disability [12,18,19], this heterogeneity may be further com-pounded by any of these possibilities. Perhaps accepting the unavoidable heterogeneity and sampling a much larger cohort with minimal clinical criteria (i.e. diagnostic laboratory samples submitted to ''rule out FXS'') would be profitable. While much more costly, recent advances in sequencing-by-synthesis may allow such studies. The current study confirms the known importance of occasional FMR1 deletions responsible for FXS. The deletion we identified extends from 220 bp upstream of the CGG repeat through the second codon of the FMR1 coding sequence, and results in the absence of FMRP expression in patient tissues. While the exact breakpoints are unique, this deletion belongs to a well-characterized class of deletions that result from the instability of the CGG trinucleotide repeat region [8,20]. This deletion, as a null mutation, would be expected to present with a FXS phenotype as the FMR1 full mutation is also a functional null mutation. Since FMR1 deletions are not specifically screened for clinically and are usually found secondary to CGG-repeat screening, many small deletions and perhaps duplications may be missed in routine testing of patients with a FXS presentation. Therefore screening for small FMR1 copy number variation might be clinically useful and could be accomplished by targeting FMR1 for high density coverage in clinical arrays screened by comparative genome hybridization Western blot of lymphoblastoid cell lysate from a healthy control, a fragile X patient, and a patient harboring a novel deletion in the 59UTR of FMR1. The protein eIF4e was assessed as a loading control. doi:10.1371/journal.pone.0009476.g002