|Home | About | Journals | Submit | Contact Us | Français|
Birt-Hogg-Dubé syndrome(BHDS), caused by germline mutations in the folliculin (FLCN) gene, predisposes individuals to develop fibrofolliculomas, pulmonary cysts, spontaneous pneumothoraces and kidney cancer. The FLCN mutation detection rate by bidirectional DNA sequencing in the National Cancer Institute BHDS cohort was 88%. To determine if germline FLCN intragenic deletions/duplications were responsible for BHDS in families lacking FLCN sequence alterations, 23 individuals from 15 unrelated families with clinically-confirmed BHDS but no sequence variations were analyzed by real-time quantitative PCR (RQ-PCR) using primers for all 14 exons. Multiplex Ligation-Dependent Probe Amplification (MLPA) Assay and array-based comparative genomic hybridization (CGH) were utilized to confirm and fine map the rearrangements. Long Range PCR followed by DNA sequencing was used to define the breakpoints. We identified 6 unique intragenic deletions in 9 patients from 6 different BHDS families including four involving exon 1, one that spanned exons 2–5, and one that encompassed exons 7–14 of FLCN. Four of the six deletion breakpoints were mapped, revealing deletions ranging from 5688 to 9189bp. In addition, one 1341bp duplication, which included exons 10 and 11, was identified and mapped. This report confirms that large intragenic FLCN deletions can cause BHDS and documents the first large intragenic FLCN duplication in a BHDS patient. Additionally, we identified a deletion “hot spot” in the 5′-noncoding-exon 1 region that contains the putative FLCN promoter based on a luciferase reporter assay. RQ-PCR, MLPA and aCGH may be used for clinical molecular diagnosis of BHDS in patients who are FLCN mutation-negative by DNA sequencing.
Birt-Hogg-Dubé Syndrome (BHDS;MIM135150) is an autosomal dominant disorder that predisposes individuals to develop fibrofolliculomas, pulmonary cysts, pneumothoraces and kidney neoplasia(Birt et al., 1977; Toro et al., 1999). The BHDS locus was mapped to chromosome 17p11.2 (Schmidt et al., 2001) and germline mutations in a novel gene, FLCN, (MIM607273) were identified in affected members of BHDS families (Nickerson et al., 2002). The 14 exon gene encodes folliculin (FLCN), a novel protein that is highly conserved across species. The identification of somatic “second hit” FLCN mutations in the remaining wild-type allele in BHD-associated renal tumors suggests a tumor suppressor role, consistent with the Knudson “two-hit” hypothesis (Vocke et al., 2005). Additionally, FLCN mRNA levels were reduced in renal tumors from BHDS patients (Warren et al., 2004) and FLCN protein could not be detected in BHD renal tumors or Flcn heterozygous knockout mice (Hasumi et al., 2009).
A spectrum of germline FLCN mutations in patients with BHDS has been reported, including small insertion/deletions, splice-site, and nonsense mutations predicted to prematurely truncate the protein and, rarely, missense mutations (Toro et al., 2008; Schmidt et al., 2005). To date, 88% of 102 clinically-proven BHDS families in the NCI cohort have sequence-identified FLCN mutations (Toro et al., 2008).
Large intragenic rearrangements have been reported in the cancer predisposing genes von Hippel-Lindau (VHL[MIM608537]) (Franke et al., 2009; Hoebeeck et al., 2005; Hattori et al., 2006), succinate dehydrogenase subunit B (SDHB[MIM185470]) (McWhinney et al., 2004), fumarate hydratase (FH[MIM136850]) (Ahvenainen et al., 2008) and breast cancer 1 (BRCA1[MIM113705]) (Swensen et al., 1997). Recently Kunogi et al. (Kunogi et al., 2010) published the first report of large intragenic FLCN deletions in two unrelated BHDS patients. We speculated that large intragenic deletions and/or duplications in the FLCN gene might account for some of the 12% of families in the NIH BHDS cohort for whom FLCN sequence variants have not been identified.
Patients with clinically-confirmed Birt-Hogg-Dubé syndrome that were negative for a mutation by bidirectional DNA sequencing of the 14 FLCN exons were selected for this study. Most patients were seen at the Urologic Oncology Branch, National Cancer Institute, National Institutes of Health for clinical assessment and provided a peripheral blood sample for DNA extraction. Additional family members were evaluated during field trips. This study was approved by the Institutional Review Board of the National Cancer Institute. All patients provided written informed consent.
RQ-PCR was conducted on an ABI Prism 7000 Sequence Detection System (Applied Biosystems) using SYBR Green Master Mix (Applied Biosystems, Foster City, CA). All data analysis was performed with the 2-ΔΔCt method in Microsoft Excel version 2007. Primer pairs were designed to amplify the 14 exons of the FLCN gene (GenBank accession AF517523)(Kunogi et al., 2010) and ~ 8kb of 5′-genomic sequence (Supporting Information Table 1). All primers were evaluated using the National Center for Biotechnology Information BLAST program.
PCR reaction components included 250nM of each primer, 10ng template DNA and SYBR Green Mix in a final volume of 20ul. RQ-PCR conditions were as follows: 10min at 95°C, 40 cycles of 95°C for 15sec and 60°C for 60sec. Zinc finger protein gene (ZNF80 [MIM194553]) and human serum albumin gene (ALB [MIM103600]) were selected as endogenous controls (Hattori et al., 2006). Genomic DNA from a BHDS patient with a germline missense mutation in the FLCN gene was used as a negative control. All samples were run in triplicate. Threshold values were set at 0.8–1.3 for normals, 0.45–0.74 for deletions and 1.6–1.8 for duplication. Experiments were replicated at least twice if a deletion/duplication was suspected.
Deletion/duplication analysis of the FLCN gene was performed by MLPA according to manufacturer’s instructions with a commercially available probe mix (P256, FLCN probe mix, MRC-Holland, Amsterdam, The Netherlands). In brief, 200ng of genomic DNA in a final volume of 5ul was heated for 5min at 98°C. After cooling to room temperature, 1.5ul of probe mix and 1.5ul SALSA hybridization buffer (MRC-Holland) were added, followed by heat denaturation (1min at 95°C) and hybridization (16h at 60°C). Ligation was performed by adding 32ul of ligation mix at 54°C for 15min, and stopped by incubation for 5min at 98°C. PCR amplification was carried out for 35 cycles in a final volume of 40ul. PCR products were separated by capillary electrophoresis using an ABI 3130Xl (Applied Biosystems) with an internal size standard (LIZ 600; Applied Biosystems). Data analysis was performed using GeneMarker v1.6 (SoftGenetics, State College, PA). Threshold values were set at 0.75–1.3 for normals, 0.4–0.65 for deletions and 1.4–1.6 for duplications. Patient data was normalized to data from two deletion-negative controls.
An Agilent Custom High–Definition (HD) (Agilent, Santa Clara, CA) CGH array was designed to investigate the FLCN gene copy number using probes from the Agilent HD-CGH database. Seventy-three HD probes within a 25kb genomic region containing the FLCN gene were computationally pre-selected with an average probe density of ~3 probes/kb. In the 50kb flanking regions 5′ and 3′ to FLCN, a fade-out design achieved an average density of ~1 probe/kb diminishing to an average of ~1 probe/40kb over the entire genome. The array was printed on an Agilent 4x44K Customer array and processed according to manufacturer’s protocol. Briefly, 0.5ug of patient genomic DNA and normal human reference DNA (Promega, Madison, WI) were fragmented by AluI/RsaI digestion, labeled with Cy3/Cy5 fluorescent dyes and hybridized at 65°C for 24hrs. After hybridization and washing, arrays were scanned using Agilent Microarray Scanner. Data were extracted with Agilent Feature Extraction Software (v10.7.1.1) and analyzed with Agilent DNA Analytics 4.0 software (v4.0.85).
Expand Long Range dNTPack (Roche, Indianapolis, IN) was used according to the manufacturer’s instructions using RQ-PCR primers adjacent to deleted regions and additional nested primers (Supporting Information Table 2). DNA fragments were gel purified using DNA Gel Extraction Kit (Qiagen, Valencia, CA).
All purified DNA products were sequenced bidirectionally using the Big Dye Terminator v.1.1 Cycle Sequencing Kit (Applied Biosystems) according to manufacturer’s specifications and run on an ABI 3130xl Genetic Analyzer. RQ-PCR and LR-PCR nested primer pairs were used for sequencing.
BAC-cloned human DNA (CTD2 2504A7;Invitrogen, Carlsbad, CA) was double-digested with KpnI and SacI (New England Biolabs, Waltham, MA) to release a fragment containing 4248bp of 5′ flanking sequence, FLCN exon 1 (228bp) and 673bp of intron 1 that was gel purified using a QIAquick Gel Extraction kit (Qiagen). The fragment was ligated to linearized phosphatase-treated pGL3-Basic vector (Promega) using DNA Ligation Kit (Takara Bio Inc., Madison, WI) according to manufacturer’s protocols. The ligation mixture was directly used for transformation into One Shot® Stbl3™ Chemically Competent E. coli cells (Invitrogen). Plasmid DNA was extracted using Nucleospin Plasmid kit (Machery-Nagel, Deer Park, NY). The insert (2038bp) of the FLCN promoter/exon 1 construct was confirmed as correct by restriction mapping.
To generate the promoter/exon 1 deletion mutant construct, the wild-type FLCN promoter/exon 1 construct was digested with SacII (New England Biolabs, Ipswich, MA) to release a 1396bp DNA segment containing 835bp of the CpG island that included all of exon 1. A self-ligation of the resulting ~8.6kb exon 1 deletion construct was performed with the DNA ligation kit (Takara Bio Inc.) and used directly for transformation as above. Plasmid DNA was prepared using Nucleospin Plasmid extraction kit (Macherey-Nagel). The deletion construct was confirmed by restriction mapping as above.
HEK293A cells were transfected with pGL3-Basic empty vector, FLCN wild-type and promoter/exon 1 deletion mutant vectors using Lipofectamine 2000 according to the manufacturer’s instructions and incubated overnight at 37°C in a CO2 incubator. Reporter assays using the Dual-Luciferase Reporter 1000 Assay System (Promega) were performed 24hours later according to the manufacturer’s instructions. The samples were read in a 96-well plate using MicroLumatPlus LB96V (Berthold Technologies, Oak Ridge, TN). Data analysis was performed in Microsoft Excel version 2007.
Patient 11 tumor and normal kidney tissue from a sporadic kidney cancer patient were collected at surgery at the Urologic Oncology Branch, NCI, flash frozen and maintained in liquid N2 until use. RNA was extracted from tumor tissues using Trizol following homogenization with gentleMACS™ Dissociator (Miltenyi Biotec, Bergisch Gladbach, Germany). cDNA was generated using SuperScript III First-Strand Synthesis SuperMix for qRT-PCR (Invitrogen). To eliminate the contaminating wild-type sequence, the mutant cDNA band from patient 11 tumor RNA was cloned into a TOPO TA cloning vector (Invitrogen) prior to sequencing. A PCR product from patient 11 genomic DNA containing the breakpoint between exon 11 and duplicated exon 10 (exon 10′) was cloned into a TOPO TA cloning vector (Invitrogen).
DNA from 23 patients in 15 unrelated families with clinically-proven BHDS was analyzed for FLCN deletions and duplications. We detected six unique intragenic germline deletions [40% (6/15) of families] by RQ-PCR, which were confirmed by MLPA and aCGH. The results summarized in Table 1 include one deletion spanning exons 2–5, one deletion encompassing exons 7–14, and four deletions that include exon 1.
Patient 1 of Family A had a FLCN deletion that spanned exons 2–5, which was identified by RQ-PCR and confirmed by MLPA (Figs. 1A and 1B). LR-PCR was used to amplify the deleted region and produced a ~3kb mutant product compared to the ~12.3kb wild type genomic sequence, predicting a deletion of ~9kb (Fig. 1C). Bidirectional sequencing of the mutant PCR product, confirmed a 9189bp deletion (Fig. 1D; Supporting Information Table 3) that eliminated exons 2–5, including the translation start codon within exon 4, preventing normal translation. The deletion boundaries involving the repeat elements AluSq in intron 1 and AluSx in intron 5 generated the breakpoint sequence: AluSq’-AluSx’ 5′-GCCATTGCAC-TCCAGCCTGG-3′.
RQ-PCR for patients 2 and 3 in Family B identified a deletion spanning exons 7–14 (Supporting Information Figure 1). By MLPA, a deletion of at least exons 8–14 was detected (Supporting Information Figure 2). The discrepancy in results generated by the RQ-PCR and MLPA methods could be due to differences in the locations of the probes and primers for these two techniques since the MLPA probe for exon 7 is located just 5′ of the RQ-PCR primers for this exon. Our methods could not map the extent of this deletion precisely enough for standard long-range PCR amplification to be successful. Other methods such as whole genome sequencing of this region using targeted capture could be applied to fine map the deletion in Family B. Probes designed for aCGH for both patients defined the deletion to include the termination codon (Supporting Information Table 4), but could not clearly determine the status of exon 7, nor the exact extent of the deletion 3′ of exon 14.
Four of the six deletions (67%) identified in this study (Families C-F) encompassed exon 1 of FLCN. A representative example, Family D (Fig. 2A), is shown in the panels of Fig. 2. Both RQ-PCR (Fig. 2B) and MLPA (Fig. 2C) confirmed a deletion of exon 1 in patients 5 and 6 in Family D, and in Families C, E and F (Supporting Information Figures 1 and 2 and data not shown), but neither method gave information about the deletion boundaries. To determine the 5′-boundary of the deleted sequences in these BHDS families, RQ-PCR with an expanded panel of primers within the ~ 8kb genomic sequence 5′ of exon 1 was utilized. RQ-PCR localized all of the exon 1 deletions in Families C-F to a 4kb region consisting of 5′ flanking sequences, exon 1 and a portion of intron 1 (Fig. 3A; Supporting Information Figure 1). Due to the density of aCGH probes in the exon 1 region, we were able to more finely map these deletions using aCGH (Supporting Information Table 4). Primer pairs for LR-PCR were subsequently designed to amplify the mutant fragments in each family. The deletion breakpoints for patient 4 in Family C were identified by bidirectional sequencing of the mutant PCR products (Supporting Information Table 3), and this DNA sample was subsequently used as a positive control for Families D-F. Representative aCGH, LR-PCR, and deletion breakpoint sequencing results for patients 5 and 6 of Family D are shown in Figs. 2D, 2E, and 2F, respectively. Three of the breakpoints identified in the four exon 1 deletion families were characterized using these methods (Table 1). All of the deletions were different, ranging from 5688bp to 6645bp in size, and none included COPS3, the gene located approximately 44kb 5′ of exon 1 in FLCN. Efforts to map the breakpoints of Family F were unsuccessful.
The genomic sequences 5′ of exon 1 and within intron 1 have an unusually high number of Alu repeat elements identified by RepeatMasker software (38.4%) compared to the rest of the FLCN gene (26%) that could be responsible, by homologous recombination, for some of the deletions involving exon 1 (Fig. 3A). The deletions in Families C and E are flanked by Alu elements, but the Family D deletion was not flanked by any identifiable repetitive or homologous sequence (Supporting Information Table 3). The mechanism that generated this FLCN deletion is therefore unknown. Family E probably underwent a complex, partially Alu-mediated rearrangement, resulting in deletion of two sequences, which flanked a short sequence that became inverted. Most probably, an initial inversion event brought a 4006bp sequence and a 2764bp sequence adjacent to one another, permitting a subsequent 6645bp deletion but retaining a 125bp sequence in the reverse orientation (Fig. 3B; Supporting Information Table 3). The AluY and AluSg elements that flanked the 2764bp sequence may have mediated one, but not both, of these events.
To rule out the possibility that deletions of the non-coding exon 1 sequences were common polymorphisms, MLPA was performed on 52 unrelated patients not affected with BHDS and no exon 1 deletions were found (data not shown). Furthermore, identification of exon 1 deletions in an additional affected family member from each of Families D and E with the same breakpoints as the probands supported their pathogenicity in BHDS.
Since all of the exon 1 deletions mapped in Families C-E and predicted by aCGH in Family F included additional sequence 5′ to exon 1, we speculated that the FLCN promoter was also deleted, thereby preventing transcription of the mutant copy of FLCN. Bioinformatics resources (UCSC human Genome Browser and Proscan Version 1.7) were used to predict the FLCN promoter through identification of CpG islands, open chromatin regions, histone and transcription binding regions. Exon 1 was found to encompass part of the putative promoter. To demonstrate that the common region lost in the exon 1 deletions was functionally important in FLCN regulation, we performed luciferase reporter assays. Based on the predictions of the putative promoter region, we designed the wild-type FLCN DNA insert for the luciferase reporter vector to include a 5149bp region that contained 4248bp 5′ of exon 1, exon 1 (228bp) and 673bp of intron 1. The FLCN promoter/exon 1 deletion mutant vector lacked exon 1, a common region deleted in BHDS Families C-F, and retained only ~130bp at the 3′ end of the predicted CpG island (Fig. 4A). The mutant vector demonstrated a 31-fold decrease in activity (p<0.001, student t test) when normalized to the activity of the wild type vector (Fig. 4B). These data confirm that the exon 1 deletions found in the germline of BHDS Families C-F are functionally important, and provide the “first hit” leading to loss of FLCN tumor suppression and tumorigenesis in BHDS.
The MLPA results for patients 10 and 11 in BHDS Family G suggested a duplication of exons 10 and 11 (Supporting Information Figure 2), which was confirmed by RQ-PCR (Fig. 5B). cDNA from patient 11 kidney tumor RNA was used to amplify exons 7–12. A 750bp PCR product generated from both patient 11 and control samples corresponded to the wild type cDNA sequence (749bp) (Fig. 5C). An additional ~1kb PCR product, amplified from patient 11 but not the control, was the predicted size (987bp) of the mutant FLCN allele containing an exon 10–11 duplication. Sequence analysis demonstrated that exons 10 and 11 were duplicated in tandem in the mature transcript (Supporting Information Table 3; Fig. 5A). The source of a smaller band present in the control but not patient sample was unknown and not analyzed further.
To map the breakpoint between exons 11 and 10′ (duplicated exon 10), genomic DNA from patients 10 and 11 was amplified with exon 11 RQ-PCR forward primer and exon 10 RQ-PCR reverse primer producing a mutant product measuring ~750bp that was absent in the control sample (Fig. 5D). This region could not be amplified from wild-type FLCN since the primers are facing opposite directions. Bidirectional sequencing of the 778bp fragment using the same PCR primers revealed a sequence containing the 3′ end of exon 11, 410bp of the 5′-end of intron 11 joined to 154bp of the 3′-end of intron 9, followed by the 5′-end of exon 10 (Fig. 5E; Supporting Information Table 3). The breakpoint was not flanked by Alu repeat elements and only a small percentage of repetitive sequences was found by RepeatMasker in the vicinity of the duplication (11.51% from intron 9 to intron 11, of which 8% were Alus).
The exon 10–11 duplication in BHDS Family G (c.1063-154_1300+410dup; p.Glu434GlyfsX35) is predicted to produce a frameshift and premature termination codon. Interestingly, a somatic mutation in the wild-type copy of the FLCN gene was identified in the patient 11 kidney tumor that generated a stop codon in exon 12 (data not shown) providing the second “hit” to inactivate the tumor suppressor FLCN in the BHD renal tumor.
Chart reviews were performed on 9 patients from 6 unrelated families excluding patients 2 and 3, from Family B, for whom phenotypic data were incomplete. The phenotypic data from deletion and duplication patients were pooled to analyze genotype-phenotype correlations and compared to patients with FLCN point mutations. Point mutations include small insertions/deletions, nonsense, missense, and splice-site mutations (Toro et al., 2008) (Table 1; Supporting Information Figure 3).
The average age at diagnosis for BHDS for the FLCN deletion/duplication families was 47 (range 29–60) compared to 48 (range 31–71) in BHDS patients with point mutations. Age of onset for skin and lung manifestations was not reported (Schmidt et al., 2005). Ninety-one percent (10/11) of patients and 86% (6/7) of families had fibrofolliculomas, excluding patient 4 of Family C who had perifollicular fibromas. Sixty-four percent (7/11) of patients and 71% (5/7) of families had lung cysts. Of those, only patient 9 of Family F developed pneumothoraces (n=3). Twenty-seven percent (3/11) of patients and 29% (2/7) of families had kidney tumors. Among the promoter/exon1 deletion families, only patient 4 of Family C developed bilateral multifocal kidney tumors. Four patients (36%) from four unrelated families (Families A, D, F and G) had thyroid findings including 3 with hypothyroidism, one of whom had become thyroid-dependent following a total thyroidectomy and radiation for papillary thyroid cancer. Patient 5 of Family D had a benign thyroid nodule. With the exception of the thyroid findings, the frequency of symptoms reported in patients with FLCN deletions/duplications, or point mutations was similar (Supporting Information Figure 3).
We have identified six intragenic deletions and one duplication in 47% (7/15) of unrelated BHDS families and characterized 5 of 7 breakpoints. Our data confirm that, in addition to protein-altering FLCN frameshift, missense, nonsense and splice-site mutations, BHDS can be caused by large intragenic deletions as first described by Kunogi et al. (Kunogi et al., 2010). Significantly, we have identified a “hot spot” in the exon 1/promoter region for FLCN deletions and present here the first reported case of a large intragenic FLCN duplication.
Alu repeats are associated with and may explain the majority of FLCN deletions characterized in our cohort. Sixty percent (3/5) of the deletions identified (Families A, C and E) were flanked by Alu repeats (SINEs), reported to be involved in Non-Allelic Homologous Recombination (NAHR) (Deininger et al., 1999) in Alu-mediated deletions implicated in other human cancers including VHL, breast cancer, Ewing’s sarcoma and HNPCC (Lehrman et al., 1986; Mauillon et al., 1996).
The FLCN intragenic deletions in Families E and D involved more complex structural rearrangements. The inversion and subsequent intragenic deletion in Family E was the result of a recombination involving AluSg- and AluY-flanked sequences, and another recombination event of unknown mechanism. The Family E deletion retained AluY sequences at the 5′-deletion boundary but no definable repetitive elements at the 3′-deletion boundary; an AluJb is located 982bp away. A similar mechanism has been described for the bleeding disorder Glanzmann Thrombasthenia (GT), where an Alu-mediated inversion occurred followed by an Alu-mediated deletion(Li et al., 1993). The intragenic FLCN deletion in Family D had no repetitive elements or homology to another DNA sequence on either side of the deletion boundaries. The closest repetitive elements were THE1B located 436bp from the 5′-boundary, and AluJ, located 1106bp from the 3′-deletion boundary. As has been suggested, (Stankiewicz et al., 2003; Lee et al., 2006) it is possible that repetitive elements near the gene could facilitate the deletion.
Low copy repeats (LCR) are DNA elements ranging from 1–200kb in size with >90% homology, which have been implicated in many chromosome 17 genetic rearrangements (Shchelochkov et al., 2010). The density of these elements in the human genome ranges from 5–10%. However, their frequency in the proximal region of chromosome 17 is 23%, possibly accounting for the high rate of genetic rearrangements (Shchelochkov et al., 2010). Stankiewicz et al. investigated the breakpoints of 18 genetic rearrangements in chromosome 17 and identified 9 patients (50% of their cohort) with LCRs at one breakpoint and non-LCR DNA sequences at the other boundary as identified in Family E (Stankiewicz et al., 2003). Inoue et al. (2002) also described 2 Pelizaeus-Merzbacher (PLP1) families with deletions but no identifiable homologous sequences at the breakpoint boundaries. Possible mechanisms for the generation of these deletions include NAHR using only small segments of homology or Non-Homologous DNA End-Joining (NHEJ) where no homology is necessary (Lieber et al., 2003). Another possibility is that novel repetitive elements could be involved in these rearrangements. These systems may not be mutually exclusive of each other.
We have identified the first intragenic duplication in the FLCN gene. Although Alu-mediated intragenic duplications have been well-described, (Schichman et al., 1994; Yap et al., 2006) no Alu repeats or other repetitive elements were found at the breakpoint junction in BHDS Family G. We speculate that nonhomologous recombination or small homologous mechanisms are involved in the generation of the tandem duplication. Several examples in the literature, including two Duchenne muscular dystrophy families (Hu et al., 1991) and a split hand–split foot malformation 3 (SHFM3) family (de Mollerat et al., 2003) in which duplication events occurred at breakpoints without homologous sequences, further demonstrate that nonhomologous duplication events occur, although the mechanism remains poorly understood.
Our results demonstrate the importance of utilizing RQ-PCR, MLPA and/or aCGH as diagnostic methods in BHDS patients who are FLCN mutation-negative by DNA sequencing, especially for the FLCN promoter/exon1 region, where 67% of the deletions identified were located, and no point mutations had been reported (Toro et al., 2008; Schmidt et al., 2005). However, deletions and duplications involving exons did not account for all cases of BHDS in patients who were mutation-negative by sequencing (8/15 families, 53% of this cohort). Our RQ-PCR method will detect deletions and duplications within exonic regions of FLCN, but will miss smaller deletions, duplications and other genetic rearrangements within introns (especially large introns 1, 3, 8 and 9) as well as epigenetic alterations and variations in the distal end of the 3′UTR.
The deletions and duplication we characterized most likely affect either the amount or function of the FLCN protein. Based on the marginal activity of the exon1/promoter deletion in the luciferase reporter assay, the exon 1 deletions in BHDS Families C-F would be predicted to dramatically reduce FLCN expression from the mutant allele suggesting that the commonly deleted region in these exon 1 deletion families contains the FLCN minimal promoter region.
The exon 2–5 deletion in BHDS Family A includes the initiation codon in exon 4, preventing normal translation. The duplication of exons 10 and 11 in BHDS Family G alters the reading frame of the transcribed mRNA resulting in a premature termination codon. The exon 7–14 deletion in BHDS Family B removes much of the coding region and the termination codon. If any protein were made from the encoded mRNA, its function would most likely be disrupted since FNIP1 and FNIP2, the FLCN-interacting proteins that also interact with AMPK, bind to the C-terminus of the FLCN protein (Baba et al., 2006; Hasumi et al., 2008; Takagi et al., 2008). However, in most cases, the generation of a premature termination codon would result in degradation of the FLCN mRNA by nonsense mediated decay (NMD) (Chang et al., 2007).
Phenotypic findings in the BHDS patients with FLCN deletions or duplication are very similar to those in the point mutation-positive patients (Schmidt et al., 2005). Although major conclusions cannot be drawn from the small number of deletion and duplication-positive families (n=7), a few observations can be made. The most prominent findings are fibrofolliculomas, followed by lung cysts and kidney neoplasms. One patient had perifollicular fibromas (PFF), which have been described as part of the BHDS phenotype (Toro et al., 2008). Notably, only one of 6 patients in the deletion-positive families developed kidney tumors. No significant difference was noted in the frequency of observed/reported phenotypic features between point mutation-positive patients and patients with a deletion or duplication. Additionally we report 6 of 15 patients with thyroid findings. A few BHDS cases have been described with thyroid adenomas or multinodular goiter, (De La Torre et al., 1999; Drummond et al., 2002) but the question of whether FLCN plays a role in thyroid pathology, or BHD-associated thyroid findings reflect the high prevalence of thyroid disease in the general population (Rallison et al., 1991), will await larger studies.
In conclusion this study confirms that large intragenic deletions in FLCN, in addition to sequence-altering germline mutations, are causative for BHDS, and reports the first large FLCN duplication in a BHDS patient. Large intragenic deletions and duplications of FLCN may account for at least 5% of cases of BHDS. Consequently RQ-PCR, MLPA and/or aCGH should be employed for clinical molecular diagnosis of BHDS in patients who are FLCN mutation-negative by DNA sequencing.
We thank the National Institutes of Health Undergraduate Scholarship Program for their funding, James Peterson and Lindsay Middelton for providing pedigree and phenotypic data, and Dr. Makiko Kunogi for helpful discussions regarding RQ-PCR. This research was supported in part by the Intramural Research Program of the National Institute of Health, National Cancer Institute, Center for Cancer Research and funded in part with federal funds from the National Cancer Institute, National Institutes of Health, under contract HHSN261200800001E. The content of this publication does not necessarily reflect the views or policies of the Department of Heath and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.
The URLs for data presented are as follows:
National Center for Biotechnology Information BLAST program:
Detection of Alu repeat sequences with RepeatMasker: http://www.repeatmasker.org
UCSC human genome browser: http://genome.ucsc.edu/
Proscan Version 1.7: http://www-bimas.cit.nih.gov/