Published ahead of print on July 19, 2007, doi:10.1164/rccm.200701-164OC
© 2007 American Thoracic Society doi: 10.1164/rccm.200701-164OC
Characterization of the BMPR2 5'-Untranslated Region and a Novel Mutation in Pulmonary Hypertension1 Division of Medical Genetics, Departments of Genetics and Cardiovascular Sciences, University of Leicester, Leicester, United Kingdom; 2 Genomic Medicine Institute, Lerner Research Institute, and Taussig Cancer Center, Cleveland Clinic, Cleveland, Ohio; 3 Department of Genetics, Case Western Reserve University School of Medicine, Cleveland, Ohio; 4 Division of Genetics and Molecular Medicine, Department of Medical and Molecular Genetics, Kings College London, London, United Kingdom; and 5 Division of Respiratory Medicine, Department of Medicine, University of Cambridge School of Clinical Medicine, Addenbrookes and Papworth Hospitals, Cambridge, United Kingdom Correspondence and requests for reprints should be addressed to R. C. Trembath, M.B. B.S., F.R.C.P., F.Med.Sci., Division of Genetics and Molecular Medicine, Kings College London, 7th Floor, Guy's Tower, Guy's Hospital, London SE1 9RT, UK. E-mail: richard.trembath{at}genetics.kcl.ac.uk
Rationale: Familial pulmonary arterial hypertension results from heterozygous inactivating mutations of the BMPR2 gene. Traditional mutation analysis identifies pathogenic mutations in some 70% of linked families. We hypothesized that the apparent shortfall is due to mutations located in the promoter region of the gene, resulting in abnormal gene regulation. Objectives: To identify mutations in untranslated sequence regulating BMPR2 transcription. Methods: DNA upstream of the coding region was analyzed by direct sequencing in 16 families. Reverse transcription-polymerase chain reaction analysis and rapid amplification of cDNA ends of normal human lung RNA were used to investigate transcription of this region. Transcript levels were assessed by allele-specific expression analysis and inhibition of nonsense-mediated decay in lymphoblastoid cell lines. Measurements and Main Results: The wild-type transcriptional start site of BMPR2 was defined, 1,148 bp upstream of the ATG. Within this region, we identified a double-substitution mutation, predicted to form a cryptic translational start site, in one family. The mutant transcript contains a premature stop codon predicted to trigger nonsense-mediated decay. Expression analysis in the patient's cell line indeed showed reduced expression of the mutant transcript that could be restored to normal by inhibiting nonsense-mediated decay. Conclusions: Activation of a cryptic translation initiation site is a novel mutational mechanism in this disorder. These results demonstrate that the 5'-untranslated region of BMPR2 is considerably longer than previously thought, emphasizing the need to fully characterize the BMPR2 promoter and the importance of analyzing noncoding regions in patients with pulmonary arterial hypertension who are negative for mutations within the coding region and intron–exon junctions.
Key Words: pulmonary hypertension • BMPR2 • 5'-untranslated region mutation • Kozak sequence • cryptic translation initiation site
Pulmonary arterial hypertension (PAH) is a dramatic vascular disorder characterized by the occlusion of small pulmonary arteries leading to a sustained elevation of mean pulmonary arterial pressure and progressive right heart failure (1). In the absence of treatment, the condition is typically fatal within a 3-year time frame. The majority of patients develop disease as a sporadic event and are described by the term "idiopathic" (IPAH), whereas others have a recognized familial record of disease (FPAH). Common histopathological features include intimal thickening and medial hypertrophy, due to proliferation of endothelial and smooth muscle cells, together with matrix deposition in the adventitia (2).
Heterozygous germline mutations of the gene BMPR2, encoding a type II bone morphogenetic protein receptor of the transforming growth factor (TGF)- Screening of the coding sequence and intron/exon boundaries of BMPR2 together with systematic analysis for gross gene rearrangements has revealed mutations in at least 70% of subjects with familial PAH (6–8). With the exception of loss-of-function mutations of the activin receptor–like kinase 1 (encoded by ACVRL1), identified in a subgroup of subjects with hereditary hemorrhagic telangiectasia and PAH, no locus other than BMPR2 has been implicated in familial disease (9). Thus, one possible explanation of this apparent shortfall is that missing mutations reside in regions of the BMPR2 gene not routinely screened by present mutation detection strategies, namely upstream regulatory elements, and intronic and 3'-untranslated sequences. Because the genomic structure of BMPR2 spans approximately 190 kb, direct analysis of all intronic and untranslated sequences in subjects with unknown mutations remains impractical. As a first step, we therefore focused on analyzing part of the 5'-untranslated region and promoter of the gene.
Ethics approval was granted by the Leicestershire local research ethics committee. Patient ascertainment and DNA extraction were performed as previously described (5). Three primer pairs were designed to encompass 2.2 kb of DNA upstream of the translation initiation (ATG) site. Samples from probands from each of 16 FPAH families and 11 independent subjects with IPAH were sequenced. Each had previously been comprehensively screened for coding sequence, exon–intron boundary mutations, and gross rearrangement of BMPR2 (5, 8). Sequencing was performed with a BigDye terminator version 2.1 kit (Applied Biosystems, Foster, City, CA) and an Applied Biosystems fluorescence-based sequencer. Peripheral blood samples for RNA isolation were collected into PAXgene tubes (Qiagen, Valencia, CA) and processed according to the manufacturer's protocol. Lymphoblastoid cell lines were established through the Human Genetic Cell Bank Service (European Collection of Cell Culture, Porton Down, UK) and cultured in RPMI with GlutaMAX I (Invitrogen, Carlsbad, CA) and 10% fetal calf serum, at 37°C with 5% CO2. To inhibit nonsense-mediated decay, the medium was supplemented with puromycin (250 µg/ml; Sigma-Aldrich, St. Louis, MO) for 10 hours. Total RNA from puromycin-treated and untreated cells was isolated with an RNeasy kit (Qiagen). All RNA samples were treated with DNase I (Invitrogen) and then reverse transcribed using random hexamer priming and SuperScript II (Invitrogen) as directed. A parallel negative control sample, omitting the reverse transcriptase enzyme, was analyzed in each case. Allele-specific quantitation of single-nucleotide polymorphism rs6435156, located in the 3'-untranslated region of BMPR2 (see dbSNP: http://www.ncbi.nlm.nih.gov/SNP/index.html), and the 5'-untranslated mutation was performed by primer extension reaction, using a SNaPshot kit (Applied Biosystems) according to the manufacturer's protocol (primers available on request). Two replicates were performed on two cDNA samples from the same source for each assay, a total of four measurements per variant per individual. The mean peak height ratio obtained from replicates for each cDNA sample was normalized against the mean ratio obtained from seven to nine replicates of genomic DNA samples heterozygous for the same variant, on the basis that heterozygous DNA samples contain a 1:1 allele ratio. Significance values were determined by two-tailed t test for each sample against the DNA control. Rapid amplification of cDNA ends (RACE) was performed on DNase-I treated total RNA from normal adult human lung (Stratagene, La Jolla, CA), using a GeneRacer kit (Invitrogen) 5'-RACE protocol. All steps were conducted according to the manufacturer's protocol. Aliquots of ligated RNA were reverse transcribed with either random hexamers or a gene-specific primer, using SuperScript III reverse transcriptase (Invitrogen) as recommended. Nested polymerase chain reaction (PCR) amplifications were performed with Taq DNA polymerase with PCR buffer IV (ABgene, Epsom, UK) and 2.5% dimethyl sulfoxide. Forward primers complementary to the ligated RNA oligonucleotide from the GeneRacer kit were used in combination with gene-specific reverse primers (Table 1) at the recommended concentrations. The first round of PCR was performed for 40 cycles with annealing at 60°C and extension at 68°C for 2 minutes, and nested PCR was performed for 25 cycles. Products were cloned with a TOPO TA cloning kit (Invitrogen). Colonies were screened by PCR using M13 forward and reverse primers to check for the presence of inserts and then expanded in liquid culture. Purified plasmid DNA was sequenced in both directions with M13 primers.
Sequencing of the genomic region upstream of BMPR2 identified several known polymorphisms, and one novel sequence variant (case 7858), a GC > AT double-substitution mutation 944 bp upstream of the translation start site, designated c.*–944/5GC > AT (Figure 1A). Analysis of other family members demonstrated segregation with the disease phenotype (Figure 1B). This variant was not found amongst 400 ethnically matched normal control chromosomes.
Although we were looking for possible mutations affecting transcription factor–binding sites or other regulatory sequences, inspection of the sequence flanking the mutation revealed that it generated a novel ATG (Figure 1A) and thus might serve as an alternative translational start site, if expressed. From publicly available sequence data, the 5'-untranslated region of BMPR2 was not known to extend this far upstream. Reverse transcriptase–PCR analysis, using the same primers as for genomic DNA sequencing but on DNase I–treated leukocyte RNA from the patient, confirmed that this region is indeed transcribed (data not shown). To completely exclude the possibility that this result was due to residual DNA contamination and to confirm expression in lung, we then amplified human lung cDNA, using a forward primer upstream of the mutation site and a reverse primer in exon 2. Intron 1 of BMPR2 is 87.25 kb in length and thus a PCR product could be generated only from correctly spliced RNA. A 1.3-kb product was amplified as predicted (Figure 2), confirming that the region containing the mutation is expressed in both leukocytes and lung and that the 5'-untranslated region of BMPR2 is therefore considerably longer than previously believed.
We then performed 5'-RACE on normal human lung RNA to establish the location of the transcriptional start site. Initially, we used primers within the PCR amplicon in which the mutation was contained. A single RACE product was obtained, with a transcriptional start site 204 bp upstream of the mutation at position –1148 relative to the wild-type ATG. To determine whether this was the only cap site, RACE was then repeated with primers located in exon 2. The first round of this 5'-RACE PCR yielded four bands, the largest of which was approximately 1.4 kb in size. A second round of PCR was then performed with nested primers to provide specificity, as recommended in the protocol. Genuine products should be 122 bp smaller after the secondary PCR compared with the primary amplification. The 1.4-kb fragment did indeed show the correct size change, whereas the smaller bands did not reamplify at the predicted size and were suspected to be artifacts. The nested PCR products were subjected to TOPO TA cloning and 22 colonies were amplified, yielding inserts of four discrete sizes. At least two clones of each size were sequenced in both directions. Sequences of the two longest clones corresponded to contiguous BMPR2 cDNA commencing at position –1147. Smaller clones were confirmed to be PCR artifacts, not representing genuine RACE products. The base immediately preceding position –1147 is a guanine residue, indicating that it represents a full-length mRNA transcript with the methyl-guanine cap at position –1148. This corresponds exactly to the cap site identified in the first RACE experiment primed from only 200 bp downstream, suggesting that the –1148 cap site marks the predominant BMPR2 transcript in adult lung. The possibility of alternative transcriptional start sites in other tissues has not been investigated. ATG sequences can serve as translation initiation start sites only when embedded within a consensus sequence known as the Kozak sequence (10). To obtain an independent score of the consensus fit for this variant, we used the ATGpr program (see http://www.hri.co.jp/atgpr/) (11). As shown in Table 2, the variant creates an ATG with a better match to the Kozak consensus than the accepted translation start site for BMPR2, because in the latter the base immediately following the ATG is an adenine residue rather than the strongly preferred guanine. This suggests that the variant creates a cryptic translational start site that would be used in preference to the normal site. However, the open reading frame from this cryptic ATG is only 330 bases long and ends in a premature stop codon upstream of the wild-type ATG. The variant transcript would likely therefore be recognized and targeted by the nonsense-mediated mRNA decay mechanism.
To test this hypothesis, we undertook allele-specific quantitation of the relative abundance of mutant versus wild-type transcripts in leukocyte cDNA obtained from the patient. This showed that the mutant allele was significantly reduced compared with the wild-type allele (Figure 3A). Both patient 7858 and an unaffected sibling, who did not carry the mutation, were heterozygous for the single-nucleotide polymorphism (SNP) rs6435156, located in the 3'-untranslated region of BMPR2. Allele-specific analysis of this SNP confirmed significant imbalance in expression between the two alleles in the patient, whereas the unaffected sibling showed equal expression (Figure 3A). We also established a lymphoblastoid cell line from patient 7858 and treated it with puromycin, a protein synthesis inhibitor that blocks nonsense-mediated decay. Allele-specific quantitation of the mutant allele was performed on cDNA derived from treated and untreated cells. As a control, cells from patient 8066 (with the known splice site mutation that leads to generation of a premature stop codon) were similarly analyzed. In both cases, puromycin treatment significantly increased expression of the mutant allele relative to the wild-type allele (Figure 3B). Finally, to confirm that the recovered transcript did not harbor another, previously undetected, nonsense mutation, the puromycin-stabilized cDNA was re-sequenced in its entirety and no other mutation was identified.
Taken together, we consider these data as strongly suggesting that the c.*–944/5GC > AT variant is pathogenic and leads to nonsense-mediated mRNA decay. We recognize, however, that this variant could be in linkage disequilibrium with another as yet unidentified disease-causing mutation, potentially lying deep within untranscribed regions of the gene.
Activation of a cryptic translation initiation site appears to be an uncommon or poorly described mutational mechanism, having previously been reported in only a few conditions, for example,  -thalassemia, hemochromatosis, and the apolipoprotein-(a) gene (12–14). Bioinformatic analysis of the BMPR2 c.*–944/5GC > AT mutation described here is remarkable, in that the novel ATG motif created by this mutation is a considerably better match to the consensus Kozak translation initiation sequence than the accepted wild-type initiation site for BMPR2 translation. Furthermore, we have demonstrated that the mutation site is immediately downstream of the only mRNA cap site identified in human lung and thus the novel ATG would be the first potential initiation site encountered by the ribosomal machinery. The overall exon structure of the gene is maintained and thus the mutant transcript would be read as containing a nonsense mutation within exon 1. The distance between the translation start site and nonsense mutation is a critical determinant of nonsense-mediated mRNA decay; a distance of 330 bp in this mutant transcript is predicted to confer full sensitivity to nonsense-mediated mRNA decay (15). Indeed, we have documented an in vivo reduction in expression of the mutant allele relative to wild-type that can be recovered by blocking nonsense-mediated mRNA decay, strongly supporting the hypothesis that this mutation generates a transcript that is targeted for degradation. We cannot exclude the possibility that protein translation might reinitiate at the wild-type ATG. However, the instability of the mutant transcript and the relative weakness of the wild-type Kozak sequence suggest that, if any reinitiation occurs, it would likely be insufficient to restore protein levels to normal. Combined with segregation of the mutation with the disease phenotype and its absence in 400 control chromosomes, all available data thus support the hypothesis that this is a pathogenic variant. The identification of this mutation in the 5'-untranslated region of BMPR2 underlines the importance of sequencing the noncoding regions of this gene in all cases of familial PAH that are negative for coding or splice site mutations and structural gene rearrangements. We analyzed all 16 such families available to us and identified this one mutation, suggesting that about 6% of mutation-negative PAH families may have mutations in the 5'-untranslated region. Our discovery that the transcription initiation site is more than 1 kb upstream of the wild-type translation initiation site was surprising, but the high level of conservation of this region between human, mouse, and rat (Figure 4) suggests that this also holds true in other species. This underlines the importance of fully characterizing the promoter of this gene, both for understanding the regulation of BMPR2 and identifying additional sites that might be mutated in PAH.
The authors are indebted to the families for their continued participation in these studies. The authors thank Rachel Harrison for sample collection, Phyllis Harbor for technical support, and the Genomics Core Facility of the Lerner Research Institute for sequencing services.
Supported by a program grant from the British Heart Foundation (BHF) to R.C.T. and N.W.M. (RG/2000012). V.J. is supported by a BHF Ph.D. studentship (FS/02/082/14741).
* These authors contributed equally to this manuscript. Originally Published in Press as DOI: 10.1164/rccm.200701-164OC on July 19, 2007 Conflict of Interest Statement: M.A.A. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. R.D.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. V.J. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. N.W.M. received research grant funding from Novartis plc in 2004 for £260,000 over 3 years. He has acted as a consultant to Novartis plc and has given educational talks sponsored by Actelion and GlaxoSmithKline. R.C.T. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form January 31, 2007; accepted in final form July 18, 2007
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENATRY
