|Home | About | Journals | Submit | Contact Us | Français|
Meckel syndrome (MKS) is an embryonic lethal, autosomal recessive disorder characterized by polycystic kidney disease, central nervous system defects, polydactyly and liver fibrosis. This disorder is thought to be associated with defects in primary cilia; therefore, it is classed as a ciliopathy. To date, six genes have been commonly associated with MKS (MKS1, TMEM67, TMEM216, CEP290, CC2D2A and RPGRIP1L). However, mutation screening of these genes revealed two mutated alleles in only just over half of our MKS cohort (46 families), suggesting an even greater level of genetic heterogeneity. To explore the full genetic complexity of MKS, we performed exon-enriched next-generation sequencing of 31 ciliopathy genes in 12 MKS pedigrees using RainDance microdroplet-PCR enrichment and IlluminaGAIIx next-generation sequencing. In family M456, we detected a splice-donor site change in a novel MKS gene, B9D1. The B9D1 protein is structurally similar to MKS1 and has been shown to be of importance for ciliogenesis in Caenorhabditis elegans. Reverse transcriptase–PCR analysis of fetal RNA revealed, hemizygously, a single smaller mRNA product with a frameshifting exclusion of B9D1 exon 4. ArrayCGH showed that the second mutation was a 1.713 Mb de novo deletion completely deleting the B9D1 allele. Immunofluorescence analysis highlighted a significantly lower level of ciliated patient cells compared to controls, confirming a role for B9D1 in ciliogenesis. The fetus inherited an additional likely pathogenic novel missense change to a second MKS gene, CEP290; p.R2210C, suggesting oligogenic inheritance in this disorder.
In recent years, functional and structural defects in primary cilia have been identified as the cause of an ever-increasing number of inherited diseases with a multi-organ phenotype, the ciliopathies (1,2). Primary cilia are evolutionarily conserved microtubule-based organelles that project from the cell surface and can play an important role in chemo-, mechano- and photosensation (1,2). Their assembly is tightly regulated and originates from the basal body (mother centriole) via protein transport mediated by the intraflagellar transport (IFT) machinery (3). During development, primary cilia have been shown to be crucial for multiple signal transduction pathways (4), including sonic hedgehog that when disrupted can trigger abnormal anterior–posterior patterning of the limb bud, and dorsoventral patterning of the neural tube (5), but also possibly Wnt signaling, including planar cell polarity (6). Due to its multiple roles, disruption of primary cilia results in an array of specific phenotypes that cause a multi-organ disease presentation (see below).
Meckel syndrome (MKS; also called Meckel-Gruber Syndrome [MIM249000,609884]) is a rare (1:10 000–1:140 000), embryonically lethal, pleiotropic, autosomal recessive disorder, which has been classed as a ciliopathy. Clinical features include bilateral renal cystic dysplasia or polycystic kidney disease (PKD), central nervous system defects (typically occipital encephalocele, but can include the Dandy-Walker malformation or hydrocephalus), postaxial polydactyly and biliary dysgenesis/congenital hepatic fibrosis (7,8). MKS is the most severe ciliopathy, but it has phenotypic and genetic overlap with other viable ciliopathies, such as Joubert syndrome (JBTS [MIM213300]), Bardet–Biedl syndrome (BBS [MIM209900]) and nephronophthisis (NPHP [MIM256100]) (8).
The first two genes associated with MKS were identified in 2006, MKS1 (17q22 [MIM249000]) and TMEM67 (MKS3; 8q22.1 [MIM609884]; 9,10). Further screening of MKS populations identified: CEP290 (12q21.3 [MIM610142]), previously associated with NPHP, Senior Loken syndrome [MIM270200], JBTS and Leber congenital amaurosis ([MIM204000]; 11,12); RPGRIP1L (16q12.2 [MIM610937]; 13–17), also associated with JBTS with cerebello-renal manifestations; and CC2D2A (4q15.33 [MIM612013]; 18), also associated with JBTS (17,19–21). However, mutation screening has suggested that these genes do not account for all MKS families (22). Recently, another MKS gene has been identified, TMEM216 (11q13.1 [MIM613277]), that also causes JBTS (23,24), while NPHP3 (3q22.1 [MIM608002]; 25) and specific BBS genes (BBS2, 16q21 and BBS4, 15q23 and MKKS, 20p12 [MIM606151, 600374, 236700]; 26) have been rarely associated with MKS-like phenotypes (without encephalocele). These further gene identifications illustrate the extreme genetic heterogeneity of MKS, but it seems certain that additional MKS genes are yet to be identified.
Consistent with MKS being a ciliopathy, many of the MKS proteins have been localized to the centrosome, the pericentriolar region, or the cilium itself (1,23,27–29), and loss of the protein is often associated with disrupted ciliogenesis or centrosome defects (18,30–34). Recent advances in genomics allow mutation screening of enriched coding regions of the human genome in combination with next-generation sequencing to identify novel disease genes (35–37). In this study, we employed these methods to screen MKS families for mutations in known ciliopathy genes and others from defined ciliomes. We report the identification of a novel MKS gene, B9D1, which is mutated in a pedigree with a severe MKS phenotype, characterized fetal cells from this case and obtained data hinting at oligogenic inheritance.
In our MKS population of 46 families, mutation screening by Sanger sequencing of the most common MKS genes (MKS1, TMEM67, TMEM216, CEP290, CC2D2A and RPGRIP1L; 9,10,12,15,18) revealed two pathogenic variants in only 52% of the pedigrees (38; and manuscript in preparation). The remainder had a single likely pathogenic mutation (9 pedigrees) or no mutation detected (13 pedigrees). To identify novel loci, we enriched and next-generation sequenced the coding exons of 31 genes implicated in ciliopathies or ciliary-related signaling (Table 1) from 12 not fully genetically characterized MKS pedigrees using RainDance microdroplet-PCR (39,40) and the IlluminaGAIIx with 75 bp paired-end reads. Data were mined using the package NextGENe (SoftGenetics LCC). We employed Sanger-verified single nucleotide polymorphisms (SNPs) detected in the previous screening as positive controls for the optimization of data mining and for the detection of novel variants (see Materials and Methods for details). This analysis identified four novel mutations (see Table 2 and Discussion for details), most interestingly, one in a novel ciliopathy gene, B9D1, which has not previously been implicated in MKS or any other ciliopathy.
In pedigree M456, B9D1 (Chr. 17p11.2; 19.56 kb; 7 exons; CDS 612 bp) had a typical splicing mutation (c.505+2T>C; Fig. 1A–C). This gene was a strong candidate as it is one of three B9 domain-containing proteins, including the MKS protein MKS1, and B9D2 [MIM611951], that appear to have similar cilia-related functions (30,34; see Discussion for details). The typical splicing variant was detected with a sequence depth of 2722 reads and a mutant percentage of 48.79% in the father (R1946) of pedigree M456 (Fig. 1A and B). Sanger sequencing revealed likely hemizygosity of the c.505+2T>C mutation in the fetus since the father was heterozygous for the variant and the mother (R1945) did not carry the mutated allele (Fig. 1C).
RT–PCR amplification of RNA isolated from fetal fibroblasts supported hemizygosity in R1964. The PCR amplification, which spanned all seven exons of B9D1, resulted in a single smaller product (416 bp) compared with the normal-sized product of 513 bp (Fig. 1D). This size difference suggested variant-associated exon skipping, which was confirmed by Sanger sequencing. The sequence readout highlighted skipping of exon 4, resulting in a frameshift and a premature stop codon (p.W82CfsX44; Fig. 1E and F) after introduction of 44 non-conserved amino acids and a nearly complete disruption of the functional B9 domain (Supplementary Material, Fig. S1). Evaluation of the mutant protein by exogenous expression of a FLAG-epitope-tagged version of the R1964 transcript further supported the pathogenic nature of the splice mutation. The truncated protein associated with the c.505+2T>C change was not detected when exogenously expressed in U2OS cells, indicative of misfolding and immediate degradation (Supplementary Material, Fig. S2).
ArrayCGH analysis of the fetus and the mother highlighted that the apparent loss of heterozygosity was due to a de novo deletion of the B9D1 locus in the fetus. The deletion spans 1.713 Mb at chromosome 17p11.2, including the complete B9D1 locus (Fig. 1G). Additionally, 18 other genes were deleted, including the disease locus for the autosomal recessive Sjögren-Larsson syndrome ([MIM 270220]; ALDH3A2 [MIM609523]), which does not phenotypically overlap with MKS, six snoRNAs and one microRNA of unknown function (Supplementary Material, Table S1).
The affected fetus (R1964) was detected by ultrasound at 13 weeks of gestation with posterior encephalocele and abnormal posterior fossa, bilaterally enlarged multicystic dysplastic kidneys and no bladder. Polydactyly, that is typical in MKS1 (MIM249000) (9), was not noted but the fetus had bilateral club feet and shortened limbs. By 16 weeks, the cystic kidneys were grossly enlarged, distorting the abdomen and compressing the diaphragm. The pregnancy continued until delivery at 35 weeks; the baby survived 1.75 h. Examination at birth indicated an enlarged abdomen, an irregular-shaped posterior encephalocele (2 × 2 × 4 cm), bilateral clubbed feet and ambiguous genitalia. No autopsy was performed. Chromosome analysis of a 400-band karyotype was grossly normal.
To determine whether B9-containing proteins are more generally associated with MKS, we analyzed 24 not fully genetically resolved MKS pedigrees for B9D1 and B9D2 variants. However, we found only common intronic polymorphisms, one synonymous change in B9D1 (c.594A > C) and two polymorphic missense changes: B9D1, p.R61W and B9D2, p.I11M, all recorded in dbSNP.
In R1964, we also identified a novel change in a second MKS gene, CEP290. Sanger sequencing showed that the heterozygous variant, p.R2210C, was inherited from the mother (Figs. 1A and and2A).2A). The amino acid substitution is located at a well-conserved site in multiple CEP290 orthologs (Fig. 2B) and utilizing various bioinformatics tools the change scored as a highly likely pathogenic change (Table 2), suggesting it is of structural/functional importance. Sanger sequencing and arrayCGH analysis of R1964 did not identify a second CEP290 mutation. These data highlight possible tri-allelic inheritance that may be playing a role in disease causation or, at least, in modifying the disease presentation.
Employing R1964 chorionic villi (CV)-derived fibroblasts, we investigated ciliogenesis compared with karyotypically normal CV fibroblasts. Following induction of cilia biogenesis by serum starvation, mutant cells showed a significantly lower level of ciliation (2.9 vs. 45.3% P< 0.0001) compared with the normal cells (Fig. 3A–C). In addition, in the small number of R1964 cells that formed primary cilia, it appears that cilia biogenesis may be impaired as the cilia appeared shorter and not as well defined as in the control cells (Fig. 3B, bottom panel). Centrosome duplication defects, which have previously been associated with MKS1 mutations (31), were also observed in the R1964 cells at a higher percentage (24.3 vs. 8.5% P= 0.0025) compared with the normal cells (Fig. 3C).
Given the additional CEP290 mutation (p.R2210C) found in this family, we analyzed the localization of this protein in the R1964 fibroblasts. Previously, CEP290 was found to co-localize with the centriolar satellite protein PCM-1, scattered around the centrosome, while CEP290 depletion was shown to disrupt PCM-1 protein complex formation (41). Consistent with these data, we observed aggregation of CEP290 around the centrosomes and centrioles in three representative patterns of variable diffusion (Fig. 3D). However, there were no notable localization differences observed between the mutant R1964 cells and a human placental fibroblast cell line. Depending on the stability of the mutant protein, this localization may represent protein derived from both alleles or from just the normal protein.
We used a combination of two novel technologies (PCR-based exon-enrichment and next-generation sequencing) to explore the genetic complexity of 12 genetically unresolved MKS pedigrees. This led to the identification of four novel variants including the typical splicing change (c.505 + 2T > C) in a novel MKS/ciliopathy gene, B9D1 (Fig. 1 and Supplementary Material, Fig. S1), and subsequent analyses revealed a de novo deletion of the normal allele. Further, we highlighted potential oligogenic inheritance or modifying effects in the pedigree since the fetus inherited a single likely pathogenic missense change in a second MKS gene, CEP290 (Fig. 2).
B9D1 is evolutionarily conserved in ciliated organisms and localizes to the basal body and primary cilia in mammalian cells, and to the transition zone in Caenorhabditis elegans (30,34). The protein has also been shown to interact with the other B9 domain proteins, B9D2 and MKS1 by yeast two-hybrid analyses (42), while in C. elegans these proteins are thought to complex at the base of the cilium (34). Additionally, depletion of B9D1 by shRNA has been shown to decrease cilia number in mIMCD3 cells (30). Consistent with this, the M456 fetal cells (that have likely completely lost functional B9D1; Supplementary Material, Figs. S1 and S2) had a very low level of ciliated cells compared with the control.
B9-containing proteins are also implicated in cilia biogenesis/function in whole organisms where knockdown or conditional deletion of B9D2 results in defects in cilia stability/formation in Paramecium tetrauelia (43) and in impaired ciliogenesis, cystic kidney disease and hydrocephalus in the stumpy mouse mutant (44). The phenotype observed in R1964 is similar to stumpy or to MKS1 (9,33), with the notable exception of bilateral club feet rather than the polydactyly typically seen in MKS1. It seems unlikely that such a severe reduction in ciliated cells as observed in the R1964 mutant fibroblasts can occur in every tissue as this would be unlikely compatible with the degree of fetal development seen here. For example, a knockout mouse model of the IFT protein IFT88 [MIM600595] (45) displays early to mild-gestation lethality due to complete loss of cilia biogenesis. Instead, it is likely that tissue-specific reductions in ciliation occurred in the fetus, as has been reported in ciliopathy knockout mouse models of MKS1 and RPGRIP1L (28,46).
Oligogenic inheritance has been suggested in BBS, a milder ciliopathy. There, both triallelism and digenic inheritance (47), as well as alleles from a second gene significantly modifying the phenotype (48), have been reported. Recently, single possible mutant alleles of two genes, WDPCP [MIM613580] and TTC21B [MIM612014], implicated in other ciliopathies, have also been described in MKS patients (49,50). In two other patients in our screen, single strongly scoring variants in NPHP3 or the autosomal recessive PKD (ARPKD [MIM2632001]) gene, PKHD1 [MIM606702] (this variant was previously described as an ARPKD mutation; 51) were detected in combination with one or two CC2D2A mutations (Table 2), further emphasizing genetic complexity in MKS.
The significance of the CEP290 mutant allele in addition to the loss of B9D1 in M456 is unclear. The importance of CEP290 as a MKS disease gene is well documented (11,12)—in our population 13% (6/46) of pedigrees have two CEP290 mutations, making it one of the most frequent MKS genes (manuscript in preparation). Its role as a key player in oligogenic inheritance or as a modifier is further supported in our population in which three additional families had one likely CEP290 mutation—one with a single pathogenic mutation, one with a CEP290 nonsense mutation and a strongly scoring RPGRIP1L missense variant and the third with two strongly scoring missense changes to CEP290 and CC2D2A (manuscript in preparation).
In the M456 family, the additional CEP290 mutation could possibly be associated with the severe limb phenotype. Also, the CEP290 variant may explain why B9D1 is only rarely a cause of MKS. The combination of an additional variant in a B9D1 mutation background as reported here is especially interesting since additional loss of the NPHP genes nphp-1 (MIM607100) or nphp-4 (MIM606966) is required to disrupt ciliation in C. elegans B9 protein mutants (30,34). On the other hand, recessive mutations at MKS1 alone generate a severe MKS phenotype (9), and conditional loss of B9D2 in the mouse generates an MKS-like phenotype (44). Also, B9D1 is a small gene (small mutational target) and MKS1 is only a common cause of MKS because of the Finnish ancestral mutation (9).
Recently, next-generation sequencing was employed to identify mutations in conventionally PCR-amplified exons of 18 NPHP ciliopathy genes, identifying mutations in 30/120 NPHP families (52). Here we have extended this approach to employ exon-enrichment and next-generation sequencing to identify a novel MKS gene and reveal genetic complexity using micro-droplet PCR amplified exons. The high level of sensitivity (see Materials and Methods) that we obtain in our experiment is illustrated by the identification of a TMEM67 variant (G250R; M459) that was previously missed by Sanger sequencing in a patient with a second TMEM67 change (Table 2). Since this technology has the potential to be scaled to screen ~2300 genes (39,40) in a single reaction, exceeding the number of genes currently annotated in the human ciliome (~2024 genes) (53); we believe this may be a suitable method for screening a large, defined number of candidates that are justified biologically. Consequently, this approach should be considered a viable option compared with whole-exome analysis for mutation screening ciliopathy pedigrees while allowing pooling of indexed samples to reduce sequencing costs.
In conclusion, we provide strong evidence that B9D1 is a novel MKS gene. The initial mutation was identified using the novel approach of RainDance exon-enrichment and next-generation sequencing, highlighting its suitability to characterize complex human diseases, such as ciliopathies. That we detected B9D1 mutations in only one pedigree emphasizes the genetic heterogeneity of MKS. The cellular phenotype we observed in mutant fibroblasts highlighted the importance of B9D1/CEP290 in ciliation, and stresses the fact that B9D1 should be considered in MKS and other ciliopathy diagnostic testing. Additionally, our data emphasize that CEP290 alleles may act as disease modifiers in MKS.
This study was approved by the relevant Institutional Review Boards, and all family members gave informed consent. Family M456 was a non-consanguineous family referred for research testing from within the USA. All clinical records were reviewed by an experienced clinician. Blood samples were collected from parents, and CV cells and fetal tissue were available from the fetus, for DNA extraction and phenotypic analysis. DNA was isolated utilizing the Qiagen Puregene kit.
Initial mutation screening was performed by conventional Sanger sequencing after PCR amplification of all coding exons (±20 bp). PCR amplification was performed using the PlatinumTaq protocol (Invitrogen) with an annealing temperature range of 58–62°C and an elongation time of 1 min/kb. M13-tailed sequencing primers were used for amplification allowing a standardized sequencing reaction that was outsourced to Beckman Coulter Genomics.
The 31 genes (Table 1) were selected from the Cilia Proteome v3.0 Database (http://v3.ciliaproteome.org/cgi-bin/index.php) and prioritized for screening based on their direct implication as a ciliopathy gene or in vivo data supporting their cilia-dependent signaling role. The primer design was performed by RainDance using the Primer3 software. The primers had a standard annealing temperature of 57–59°C amplifying a product between 250 and 600 bp. We excluded all amplicons that amplified non-coding exons. RainDance synthesized the primers and performed droplet PCR exon enrichment utilizing the RDT-1000.
After exon enrichment, the samples had a concentration between 30 and 72 ng/μl with a total volume of 10 μl. Successful enrichment was tested by RainDance utilizing the Agilent Bioanalyzer DNA 12000 kit. All samples had the greatest size frequency between 300 and 700 bp. Prior to sequencing, the samples were blunt-ended and 5′ phosphorylated using the Quick Blunting Kit (New England Biolabs; NEB) and concatenated with the T4 Quick Ligase Kit (NEB) to enable successful random fragmentation during the next-generation sequencing library preparation. For optimal ligation, 5 μl of T4 ligase was added three times, first after 4h then twice after 8h. The ligase was inactivated by adding ethylenediaminetetraacetic acid (EDTA) (30 mm). Successful concatenation was tested using the Agilent Bioanalyzer DNA High Sensitivity kit. The maximum fragment size of each sample ranged between 2500 and 10 000 bp, while only a minute fraction (less than 10%) remained unconcatenated.
Sample library preparation and next-generation sequencing were performed at the Mayo Clinic Advanced Genomics Technology Center. The concatenated samples were sonicated using the Covaris E210 (duty cycle: 10%, intensity: 5, and burst: 200) and library preparation was performed using the Illumina Next Reagents (NEB). Successful library preparation was checked using the Agilent Bioanalyzer DNA 7500 and sequenced on the IlluminaGAIIx using 75 bp paired-end indexed reads. Initial cluster formation was performed on 7 pM of enriched/concatenated/sonicated DNA using the Illumina Cluster Station. A final cluster density with an average of 217 594 clusters/lane was sequenced, followed by individual read base calling using the Pipeline v2.0 software (Illumina).
The next-generation sequencing reads were analyzed using the variant discovery software NextGENe (SoftGenetics LLC). The reads were aligned against the GenBank file of the 31 genes and their isoforms. For reliable Indel discovery, we utilized one cycle of consolidation when analyzing the data. Consolidation aligns the paired-end reads, merges overlapping sequences and creates a consensus sequence by elongating the original read while maintaining the information of that read. Consequently, reads with larger Indels are less likely disregarded as poorly resolved sequence. For a variant to be called, we required the read containing the variant to match 80% to the aligned position, the variant to be covered by at least 30 reads and the variant to be present in 10% of all reads aligned to the given position. To filter all detected variants meeting the given criteria, we used Sanger-detected SNPs of each pedigree as a baseline. First, we assured 100% sensitivity of all prior Sanger-sequenced SNPs using these criteria. Next, we determined the SNP with the lowest coverage of reads and the SNP with the lowest mutant percentage. Both parameters were then used to detect novel variants by filtering the initial report for variants that had at least one-half of the lowest coverage and one-half of the lowest mutant percentage of any known SNP. Overall, we obtained a specificity of 99.19% covering 613/618 exons with a mean coverage greater than 100-fold. Multiplexing of three indexes in one lane (one index/sample) obtained on average 24 105 total reads of which 76.49 ± 1.82% mapped to the target region. All positive controls had a mean coverage of 3224 ± 2827 fold if multiplexing three indexed samples per lane.
All identified novel variants were analyzed for their potential pathogenic character utilizing various bioinformatics tools. First, they were compared against the dbSNP 131 database. Variants recorded as known SNPs were excluded from further analysis and scored as likely polymorphic. Subsequently, all remaining missense changes were scored for likely pathogenicity utilizing AlignGVGD (http://agvgd.iarc.fr/agvgd_input.php), SIFT (http://sift.jcvi.org/) and PolyPhen (http://genetics.bwh.harvard.edu/pph/index.html). Multi-sequence alignments contained at least eight orthologs, including mammals, birds and fish for the AlignGVGD and SIFT analysis. Variants were designated as highly likely (B) or likely (C) pathogenic as previously described (54). Variants that scored as likely pathogenic with two of the bioinformatics tools were checked for segregation within the family employing Sanger sequencing. Segregation as well as bioinformatics analysis of novel variants is summarized in Table 2.
The segregation of the two detected variants in M456, CEP290: p.R2210C and B9D1: c.505 + 2T > C, was analyzed by Sanger sequencing. To re-sequence the NextGENe-detected variants, the exon containing the change was PCR amplified using primers located in the flanking introns.
The PCR products were purified (ExoSAP-IT, Affymetrix) and sequenced on the ABI PRISM 3730xl DNA analyzer. Sequence analysis was performed with the Sequencher v4.8 software (Gene Code Cooperation). Similar segregation analysis was done for other detected variants (Table 2).
The arrayCGH analysis was performed at the Mayo Clinic Cytogenetics Laboratory. We provided 1 μg of fetal DNA (isolated from fetal tissue) and blood DNA from the mother to run the array. In order to cover the region surrounding B9D1 on chromosome 17 and CEP290 in high density, we ran a one million-feature chip on both samples (Agilent SurePrint G3 1X1M). We analyzed the array data using the Genomic Workbench v5.0 software (Agilent).
RNA was isolated from the fetal-derived cell line and from a normal 20-week gestation placental cell line (CRL-7548-ATCC) using the RNAeasy kit (Qiagen). cDNA was synthesized by RT–PCR utilizing the Superscript III kit (Invitrogen). To assess whether the B9D1 variant altered splicing, we PCR amplified the cDNA using a forward and reverse primer within the first and last coding exon of the gene.
|B9D1 c.505 + 2T > C||Forw-ex1||5′-TTCTACTCATGGTCAACGGGCAGG-3′|
The amplicons of the normal and the fetal (R1964) cDNA were assessed by agarose gel electrophoresis (1.5% agarose gel) and by Sanger sequencing.
Full-length WT B9D1 was amplified from a HEK293T cDNA preparation using a forward primer linked to a BamH I site followed by the start codon of B9D1 and a reverse primer linked to a NotI site after the stop codon of B9D1 (restriction sites: italics, Start/Stop codon: bold).
The WT B9D1 amplicon was cloned into pcDNA3.1(+) FLAG-BamH I. The c.505 + 2T > C mutant allele (with a frameshifting deletion of exon 4) was cloned from a cDNA preparation of R1964 primary CV cells (see cDNA analysis above). Using unique internal restriction sites (NcoI—start codon; DraIII—exon 7), the mutant cDNA was cloned into the 3X-FLAG-WT B9D1 pcDNA3.1(+) construct.
The N-terminal FLAG-tagged cDNA constructs were transfected into U2OS cells using electroporation (Nucleofector II, Amaxa). The cells were selected using 300 μg/ml of geneticin (G418, Invitrogen) for 72 h. Whole cell lysates were isolated using RIPA buffer (Upstate) and total RNA extractions performed as described above. Protein expression analysis was performed by western blotting with 20 μg of protein loaded on a 4–12% Bis–Tris gel (Invitrogen) and transferred onto a 0.4 μm polyvinylidene fluoride membrane. The exogenous B9D1 product was detected using the α-FLAG antibody (F7425; Sigma) and the α–γ-tubulin antibody (T6557; Sigma) was used as a loading control.
Cells from fetus R1964 were collected by chorionic villus sampling (CVS) and compared with normal primary CV fibroblasts (obtained from a phenotypically and karyotypically normal fetus). Fibroblasts were cultured in advanced DMEM/F-12 (Invitrogen) supplemented with 30% fetal bovine serum (FBS), glutamine, penicillin/streptomycin (Gibco) and plasmocin (InvivoGen). Prior to analysis, cells were grown to confluence, and serum starved (0% FBS) to induce ciliation and grown for an additional 3 days. In cells prepared for CEP290 analyses, ciliation was not induced and a normal placental fibroblast cell line (ATCC, CRL-7548) was employed as a control (cultured in same media supplemented with 5% FBS).
Monoclonal antibodies T6793 and T6667 (Sigma) to acetylated α-tubulin and γ-tubulin were employed to label cilia and centrosomes, respectively. For CEP290 localization studies, a polyclonal antibody against CEP290 (NB100-86991; Novus Biologicals; recognizing residues 2429–2479) and a monoclonal antibody against centrin (20H5, Dr Jeffrey Salisbury, Mayo Clinic, Rochester) were also employed.
All cells were fixed in methanol and prepared for analyses as described previously (31). Slides were visualized using a Zeiss AxioObserver (Carl Zeiss) microscope at a magnification of ×100.
Statistical significance of cell phenotypes were determined as previously described (31).
This work was supported by a grant from the Bernard E. and Edith B. Waterman Charitable Foundation through the Center of Individualized Medicine, Mayo Clinic; the National Institute of Diabetes and Digestive and Kidney diseases (grant number DK059597); and Mayo Graduate School (K.H.).
We wish to thank Drs Eric Wieben and Franklyn Prendergast (Mayo Center of Individualized Medicine) for support and encouragement; Bruce Eckloff and Yean Kit Lee for technical assistance; Vickie Kubly for coordinating sample collection; Dr Daniel Van Dyke for making available the normal CVS cells; and Dr Jeffrey Salisbury for kindly providing the centrin antibody. MKS families and referring physicians and counselors are thanked for their participation.
Conflict of Interest statement. None declared.