Published ahead of print on January 23, 2004, doi:10.1165/rcmb.2003-0330OC
© 2004 American Thoracic Society DOI: 10.1165/rcmb.2003-0330OC
The Splicing and Fate of ADAM33 Transcripts in Primary Human Airways FibroblastsBrooke Laboratories, Division of Infection, Inflammation and Repair, School of Medicine, Southampton General Hospital, Southampton, United Kingdom Address correspondence to: Robert M. Powell, The Brooke Laboratories, Level F, South Mailpoint 888, Southampton General Hospital, Southampton SO16 6YD, UK. E-mail: R.M.Powell{at}soton.ac.uk
The ADAM (A Disintegrin and Metalloprotease) family of Zn++-dependent metalloproteases are multidomain proteins involved in diverse cellular activities. Polymorphic variation in ADAM33 is strongly associated with asthma and bronchial hyperresponsiveness. Identification of those isoforms of ADAM33 that are expressed in airways is fundamental to dissecting the role of ADAM33 in asthma. Analysis of primary human airways fibroblasts has shown the presence of a number of alternatively spliced forms of ADAM33, including one encoding a putative secreted variant, and many transcripts lacking the metalloproteinase domain. The relative abundance of these transcripts has been quantified using reverse transcription real-time polymerase chain reaction, in both nuclear and cytoplasmic fractions of RNA. These results demonstrate that a number of splice variants of ADAM33 are transported into the cytoplasm. Ninety percent of ADAM33 mRNA is retained in the nucleus and the subtle differences in the composition of nuclear and cytoplasmic RNA suggest important events in the splicing and selection of ADAM33 transcripts. Western blot analysis confirmed that several protein isoforms of ADAM33 are expressed in primary airways fibroblasts. These findings demonstrate that ADAM33 exists in multiple isoforms, suggesting that it is a complex molecule that plays multiple roles within mesenchymal cells.
Abbreviations: A Disintegrin and Metalloprotease, ADAM • bronchial hyperresponsiveness, BHR • epidermal growth factor, EGF • human airways fibroblasts, HAF • metalloproteinase, MP • polymerase chain reaction, PCR • single nucleotide polymorphism, SNP
The ADAM gene family is a sub group of the zinc-dependent metalloproteinase (metzincin) superfamily (1, 2). ADAM proteins have a complex organization involving eight domains, the first six encoding signal sequence and pro, catalytic (metalloprotease [MP]), disintegrin-like, cysteine-rich, and epidermal growth factor (EGF) domains, which are anchored at the cell surface or Golgi apparatus by a transmembrane domain that is followed by a cytoplasmic domain with signaling specific sequences (3). This complex domain structure enables ADAM proteins to be involved in diverse functions including cell proliferation, differentiation, migration, and embryogenesis via ecto-domain shedding of growth factors, ligands and receptors, membrane fusion, and cell adhesion via interactions with integrins and extracellular matrix proteins (4–10). ADAM33 is the most recently described member of the ADAM family and is most closely related to human ADAM12, 15, 19 and Xenopus ADAM13, all of which possess proteolytic activity (11–13). ADAM33 is encoded on chromosome 20p13 and is expressed predominantly in mesenchymal cells and in the central nervous system, where it has been implicated in central nervous system development during embryogenesis (11, 14). ADAM33 RNA is found in abundance in airway fibroblasts and smooth muscle, but not in epithelial cells, T lymphocytes, or inflammatory cells that infiltrate the airway wall in asthma (15). ADAM33 has recently been identified as an asthma susceptibility gene as a result of a genome-wide scan in an outbred Caucasian population (15). Linkage analysis and association studies of single nucleotide polymorphisms (SNPs) and haplotypes, provided strong evidence for the involvement of several SNPs in ADAM33 in susceptibility to asthma and bronchial hyperresponsiveness (BHR), a cardinal feature of the disease (16, 17). Further support for a role for ADAM33 in regulating BHR comes from the identification of a quantitative trait locus for airways hyperresponsiveness (bhr1). This mapped to a region on mouse chromosome 2 (74cM) that is syntenic to 20p13 and very close to the location for the mouse ortholog of ADAM33 (73.9cM) (18). A number of ADAM proteins, including those most closely related to ADAM33, generate alternatively spliced transcripts. These variant forms have been shown to result in the generation of protein products with distinct functions, either by the deletion of functional domains, or by altering the cellular localization through the production of secreted or intracellular isoforms. Alternative splicing of ADAM12 produces a secreted variant (ADAM12-S), which promotes myogenesis when expressed in rhabdomyosarcoma cells grown as a xenograph (6) and obesity when expressed as a transgene in mice (12). An ADAM19 mini gene encoding only the domains downstream of the cysteine-rich region localizes to the cytosol and demonstrates novel biological effects following expression in neuronal cells (19). Other ADAMs that occur as secreted forms include ADAM 9 and 10, whereas isoforms of ADAMs 22, 28, 29, and 30 have variations within the cytoplasmic tail (20–24). Regarding ADAM33, several splice variants have previously been identified: the ß isoform lacking exon 17 was cloned in conjunction with the identification of mouse and human ADAM33, and the occurrence of a putative secreted form of ADAM33 (hsADAM33), resulting from a 37-base deletion in exon R, has also been predicted (25, 26). Several other variants within the pro-domain of ADAM33 have been reported in a lung library using a polymerase chain reaction (PCR)-based screening protocol using multiple primer pairs (600-bp amplicons with 100 bp overlap) (26). Because closely related ADAM members express a number of isoforms, we quantified specific splice variants of ADAM33 in primary human airways fibroblasts (HAF) and found that all novel splicing events that have previously been identified can be detected. Ninety percent of ADAM33 transcripts were constitutively retained within the nucleus, and the majority of cytoplasmic and nuclear transcripts lacked the exons which encode the MP domain. The existence of multiple protein isoforms of ADAM33 was confirmed by western blot analysis of fibroblast cell lysates.
Primary Cell Culture Bronchial biopsies were obtained by fiberoptic bronchoscopy from subjects with or without asthma (n = 4, average age = 21 [range 20–25], average FEV1 = 4.16 liters/min) or from subjects with moderate/severe asthma (n = 4, average age = 22 [range 20–25], average FEV1 = 2.66 liters/min, 3/4 were using inhaled corticosteroids), who had given their informed consent to participate in the study. All procedures were approved by the Southampton and South West Hampshire Ethics Committee. Primary human airway fibroblasts (HAF) were obtained by outgrowth from bronchial biopsies, as previously described (27). Briefly, biopsies were dissected into small pieces and incubated in a petri dish containing Dulbecco's Modified Eagle's Medium supplemented with 10% fetal bovine serum. Tissue samples were incubated in an humidified incubator at 37°C, 5% CO2 for  1 wk during which time fibroblasts migrated from the tissue and proliferated on the base of the culture dish. The cultures had fibroblastic morphology and were > 98% smooth muscle actin negative by immunofluorescent staining, indicating minimal contamination with smooth muscle or myofibroblasts (data not shown).
Probe and Primer Design
Quantification of ADAM33 and ADAM33 Splice Variants Using qPCR For analysis of ADAM33 expression, total RNA was extracted from cell lysates using TRIzol (Invitrogen, Paisley, UK) reagent and standard protocols before treatment with DNase1 (Ambion, Huntingdon, UK) to remove any possibility of mis-priming on contaminating genomic DNA. In a modification of this, the cytoplasmic and nuclear fractions were isolated by differential lysis of the plasma membrane with RNL buffer. Briefly, cells were resuspended in RNL buffer (Qiagen, Crawley, UK), incubated on ice for 30 s and the nuclei harvested by centrifugation at 4,000 x g for 3 min. The cytoplasmic RNA, now in the aqueous phase, was transferred to TRIzol reagent and the nuclei washed in RNL buffer to remove cytoplasmic contamination. The nuclei were then resuspended in TRIzol reagent, extracted and DNase1 treated as above. One microgram of total RNA was reverse transcribed using random hexamer primers and 100U of MMLV reverse transcriptase (Promega), all following the manufacturer's protocols.
The PCR probes were labeled with a 5'-reporter dye FAM (6-carboxy-fluorescein) and a 3'-quencher dye TAMRA (6-carboxy-N,N,N',N'-tetramethyl-rhodamine). Primers directed against 18S rRNA and a yakima yellow labeled probe were used as a normalizing control (Eurogentech, Seraing, Belgium). Before undertaking analysis of ADAM33 variant expression, each probe and primer set was validated for use with the
For each sample, measured in duplicate, the PCR reaction contained 25 ng of cDNA template, 250 nM fluorogenic probe, 900 nM of forward and reverse primers, 12.5 µl qPCR master mix (Eurogentech), in a final volume of 25 µl. Separate normalizing control samples were run which contained 1 µl of 18S rRNA primer and probe mix (Eurogentech). RT-negative samples were used to demonstrate that the signals obtained were RT-dependent and not due to genomic contamination, although this was irrelevant for all but two of the primer sets, since they do not bind to contiguous sequences on genomic DNA. The PCR protocol was as follows: 50°C, 2 min, 95°C, 10 min, followed by 40 cycles of denaturation 95°C, 15 s, and annealing/extension 60°C, 1 min. Thermocycling and real-time detection of PCR products were performed on an IcyclerIQ sequence detection system (Bio-Rad, Hercules, CA) and following completion of the PCR reaction, the thresholds for fluorescence emission baseline were set just above background levels on the FAM and yakima yellow dye layers. Expression levels in HAF for each amplicon were calculated using the
Standard PCR Detection of ADAM33 Splice Variants PCR for ADAM33 transcripts in total cell RNA was performed to detect possible spliced variants over larger regions of the transcripts using primers in the 5' and 3' UTR and internal forward and reverse primers in various domains (Table 2). Standard protocols were used throughout, with Amplitaq enzyme (Applied Biosystems, Warrington, UK), 20 pmol of forward and reverse primer, 5 mM MgCl2, and the following cycling program: 95°C for 5 min, 95°C for 15 s, 60°C 1 min, 72 2.5 min, and 35 cycles. PCR bands were sequenced directly using fluorescent dye terminator sequencing (BigDye version 3; Applied Biosystems) on an ABI 377 automated sequencer.
Immunodetection of ADAM33 in Airways Fibroblasts Fibroblasts, or COS7 cells transfected with ADAM33 for 24 h, were lysed into SDS sample buffer and separated by electrophoresis using 10% polyacrylamide gels. Western blot analysis was performed using an affinity purified rabbit anti-ADAM33 antibody (RP3; TriplePoint Biological, Forest Grove, OR) (working dilution: 0.2 µg/ml) raised against the C-terminal residues of the ADAM33 cytoplasmic domain. Detection was performed using HRP-conjugated anti-rabbit immunoglobulins (Jackson Immuno Research Laboratories Inc, West Grove, PA) at a dilution of 1:10,000. To confirm the specificity of the antibody, control blots were performed using the ADAM33 antibody after preadsorption with the immunizing peptide (2 µg/ml; TriplePoint Biological) (Figure 6). The specificity of the antibody for ADAM33 was confirmed using lysates of COS7 cells mock transfected, and transfected with full length ADAM33 + and – preincubation of the antibody with the immunizing peptide.
Identification of ADAM33 Splice Variants in HAF Cells The domain structure of ADAM33 and the previously identified splice variants are detailed in Figure 1. The D-Long exon contains a 3' extension of 39 bases; the resulting 13–amino acid insertion into the "PRO domain" includes an additional cysteine and has not been previously been identified in any other ADAM or matrix metalloproteinase PRO domain. The ß form of ADAM33 has been previously reported in mouse RNA, and was identified in this study using the exon M forward and exon V reverse primers, which amplify the 3' end of ADAM33 downstream of the disintegrin domain (26). This variant lacks exon Q in its entirety, resulting in a 26–amino acid deletion in-frame at the "cysteine rich/EGF" domain junction. The putative secreted isoform results from the deletion of 37 bases from the 5'end of exon R. This frame shift gives rise to a protein with no predicted transmembrane domain (25).
Validation of qPCR Amplicons
Semiquantitative Analysis of ADAM33 Amplicons in HAF Cells
Identification of Splice Variants that Lack a Functional MP Domain
Quantification of ADAM33 Transcripts that Contain the Contiguous Exons FGH and I The qPCR amplicon placed in exons G and H was originally designed to detect the presence of the MP domain. Because these exons were found to be present in ADAM33 transcripts that lacked all other MP encoding exons, it followed that the signal derived from this amplicon may overestimate the true level of expression of full-length or MP-expressing isoforms. To confirm this, the forward and reverse primers for this amplicon were moved into the adjacent exons (Figure 4A). As with other amplicons in this report, the specificity and amplification efficiency of these new amplicons was tested by performing a standard curve and resolving the PCR product on an agarose gel (data not shown). Consistent with the above findings, moving the forward or reverse primer outside of the GH "island" partially ablated the signal and detection between exons F and I was markedly lower in HAF (Figure 4B). Overall, these data show that the contiguous exons FGHI are present in < 5% of transcripts and a functional MP domain cannot be encoded in > 95% of ADAM33 transcripts. To confirm this finding, the FGHI amplicon was compared with a panel of other ADAM33 amplicons (Figure 4C). Overall, these data predict that 20% of transcripts contain this novel GH "island." The majority of ADAM33 transcripts ( 95%) lack the contiguous exons FGHI, which encode a region spanning the PRO and MP domains.
Analysis of Nuclear and Cytoplasmic Factions of RNA To further explore the relevance of these novel splice variants, cytoplasmic mRNA was separated from the nuclear fraction and the levels of ADAM33 isoforms and other genes measured in both compartments (Figure 5A). The ratios of mRNA in these compartments vary greatly between genes, but ADAM33 was atypical in retaining the highest proportion of mRNA within the nucleus ( 90%). The occurrence of the previously identified splice variants was also quantified in nuclear and cytoplasmic fractions (Figure 5B). The proportion of cytoplasmic transcripts that contain exons FGH and I increased in the cytoplasmic fraction, indicating these transcripts are preferentially exported, presumably as a consequence of nuclear degradation of mis-spliced transcripts. mRNA encoding the putative secreted form of ADAM33 was also detected in the cytoplasm. Surprisingly, all other amplicons had the same proportions relative to the 3'UTR in each compartment. This indicates that nuclear mRNA composition is broadly reflective of that seen in the cytoplasm and that the nuclear mRNA is 100% processed across the splice junctions that are spanned by these amplicons. An exception to this was the exon-D long amplicon, which showed a dramatic reduction in the cytoplasmic fraction. Because exon D-long was completely undetectable in some cytoplasmic extracts and only very low in others, we believe it to be a nuclear splicing intermediate with a relatively long half-life. Because the D-long amplicon spans intron 3 (Figure 1B), splicing at the 5' end of ADAM33 is necessary for the detection of exon D-long with this assay, and must also proceed rapidly after or during transcription.
Western Blot Analysis of ADAM33 Expression in Airways Fibroblasts To determine whether alternatively spliced forms of ADAM33 mRNA are translated into protein, immunoblot analysis was performed on lysates of human airways fibroblasts using an affinity-purified antibody raised against a peptide derived from the cytoplasmic tail of ADAM33. To demonstrate the specificity of this antibody for ADAM33, it was tested using lysates of COS7 cells transfected with full-length ADAM33. This showed that the antibody readily detected a protein band at 120 kD and that its recognition was completely abolished when the antibody was preadsorbed with an excess of the immunizing peptide (Figure 6A); based on a previous report (28), this is likely to represent unprocessed ADAM33. As in the previous report, we also observed a higher molecular band whose properties have not been characterized. When applied to lysates of primary airways fibroblasts, the ADAM33 antibody detected several protein bands whose immunoreactivity could also be blocked using the immunizing peptide, suggesting that these are isoforms of ADAM33. We observed a faint protein band at 120 kD that is consistent with unprocessed ADAM33, and a band at 100 kD that would be equivalent to processed ADAM33. However, we also detected several other smaller proteins, whose detection was partially or completely ablated by preadsorption with the immunizing peptide. The bands between 50 and 60 kD are consistent with clones 3, 4, 5, and 6 (Figure 3), which lack the MP domain but express the domains downstream of the disintegrin domain in conjunction with a variable N-terminal region. The smallest band, at 37 kD, is the approximate size of clone 1, expressing half of the cysteine-rich domain and the downstream C-terminal portion of ADAM33.
Previous studies have provided strong evidence that polymorphic variation in ADAM33 is a genetic determinant for the development of BHR (15). Because a number of significant SNPs have been mapped to introns within ADAM33, we hypothesize that they may regulate the splicing of ADAM33 transcripts. We have used PCR to identify the expression of a number of novel ADAM33 transcripts and have designed TaqMan qPCR assays to measure the relative prevalence of these species in cultured human airways fibroblasts. We have employed the TaqMan technology to target the predicted domains of ADAM33 and have also designed primers that detect novel splice junctions (Figures 1B and 1C). The threshold for detection during amplification will depend on the starting template concentration and the priming efficiency of individual primers sets. Because these efficiencies have been measured and found to be equivalent (97 ± 5%, Table 1), the contrasting levels of detection indicate different concentrations of starting template and different relative expression levels of these ADAM33 exons or splice variants. The overall abundance of domains such as the MP domain has been inferred as well as more detailed quantification of specific splice variants (Figure 1). The qPCR data in this report is supported by conventional PCR, cloning, and sequencing—TaqMan qPCR providing accurate quantification across relatively small regions of the transcript, whereas standard PCR across multiple exons has provided a wider context in which novel variants should be viewed. The agreement between the two approaches is excellent: regions predicted to be absent by qPCR, such as the deletions within MP domain and of exon E (Figure 2), are not found in clones obtained by standard PCR (Figure 3B). Conversely the finding that exons G and H can be detected as an "island," using standard PCR (Figure 3B), has been validated by performing detailed quantification of the exons in this region (Figure 4B).
Broadly speaking, these results indicate a complex variety of splicing pathways including the deletion of exon Q (ß form, 25% prevalence), the partial removal of exon R (hsADAM33, Comparisons between nuclear and cytoplasmic RNA gave surprising results. Only the 3'UTR primer set can detect all ADAM33 transcripts, whereas the other assays require a specific splicing event(s) to generate the contiguous primer binding sites (Figure 1). Because we anticipated the presence of unspliced and partially spliced intermediates in the nucleus, we predicted lower signals relative to the 3'UTR within this fraction. That this was not the case indicates that although 90% of ADAM33 mRNA is constitutively found in the nucleus, introns are spliced out to completion at all but one of the locations we have examined. Where nuclear transcripts do persist in a partially spliced form is between exons 4 and 5 such that the D-long variant is abundant in the nucleus (40% prevalence) and very rare in the cytoplasmic fraction. Because exon D-long was completely undetectable in some cytoplasmic extracts we believe it to be a splicing intermediate and not an expressed isoform; the absence of the D-long variant in cytoplasmic RNA also provides evidence that it was not contaminated with nuclear RNA and that the other cytoplasmic ADAM33 mRNA species are not an artifact, but are transcripts available for translation. The presence of splicing intermediates with a relatively long half-life that are not exported from the nucleus highlights the importance of performing quantification of transcripts in the cytoplasmic fraction when studying numerous splice forms. As with CTGF and ADAM17, the greater proportion of ADAM33 mRNA was retained in the nucleus (Figure 5A). Quantification of total cell RNA for these genes will overestimate the level of fully processed transcript made available for translation. However, for those splice variants of ADAM33 detected in the cytoplasm, levels of expression determined using total cell RNA will broadly reflect the proportions of differently spliced forms (Figure 5B). Exon D-long was identified only in transcripts that contain no other intronic sequences (Figure 3B), indicating that this is a late splicing event. Overall the data here present a picture of rapid splicing followed first by large deletions across the MP domain and then by the trimming of exon 4 to the predicted the wild-type splice site. The regulation of this late splicing event may be a key step in permitting the export of nascent ADAM33 transcripts, because theoretically, the conjoining of splicing to mRNA export is a sensible way to regulate the availability of transcript (29). Because transcription and splicing are already known to occur concurrently, this model would also be consistent with current thinking on RNA processing within the nucleus. This large pool of nuclear mRNA (90% of total) indicates that either a reservoir of ADAM33 transcript is available to respond rapidly to stimulation, or that 90% of primary transcripts are fated for destruction within the nucleus. The equal proportions of the ß form and GH island between the nucleus and cytoplasm suggests that most detectable nuclear transcripts are of similar composition to those in the cytoplasm, and therefore not necessarily committed to terminal degradative pathways. However, the increased proportion of cytoplasmic transcripts that contain exons FGH and I (2% nuclear compared with 6% cytoplasmic) suggests a role for the full-length protein and shows that splicing-dependent nuclear selection also regulates the fate of ADAM33 transcripts. A similar case can be made for the hsADAM33 transcript, which can also be detected in the cytoplasmic RNA fraction (Figure 5B). A role for the MP domain is supported by alignments that demonstrate the presence a consensus zinc binding site within the MP domain and by a recent study showing proteolytic cleavage of a model peptide by recombinant ADAM33 in vitro (28).
Of the cytoplasmic transcripts that lack the MP encoding exons,
The cysteine-rich and EGF domains of ADAM proteins have been linked to cell adhesion and membrane fusion events, respectively (6, 30) The deletion of exon Q gives rise to the ß form, which we show to be expressed in Although the functional domains of ADAM33 are easily inferred from the sequence, the biological role and domains of ADAM33 that maybe relevant to the pathogenesis of asthma are unknown. Without exception, all studies of ADAM33 proteins have focused on the function of the full-length molecule, or on the MP domain in isolation. The present study suggests that isoform(s) that lack the MP domain will be abundant in human airways fibroblasts and will require further detailed study. Lastly, we have shown that regulation of splicing, nuclear RNA turnover, and the selection of transcripts for export to the cytoplasm, all act to determine the fate of primary ADAM33 transcripts. It may be within these processes that polymorphisms within the ADAM33 gene exert their effect(s).
This work was supported by the Wellcome Trust (UK) and by the Asthma, Allergy and Inflammation Research Charity (AAIR, UK). J.W.H. is an MRC Research Training Fellow. The authors are grateful to Professor Gill Murphy for her critical review of this manuscript. Received in original form September 9, 2003 Received in final form December 22, 2003
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Abstract
Introduction
or 3'UTR amplicon which was assigned the value 1 (
2% of transcripts. The primers located in the 3'UTR gave a signal equal to that seen with the EGF-
