Published ahead of print on July 18, 2003, doi:10.1165/rcmb.2003-0134OC
© 2004 American Thoracic Society DOI: 10.1165/rcmb.2003-0134OC Expression of Transient Receptor Potential C6 and Related Transient Receptor Potential Family Members in Human Airway Smooth Muscle and Lung TissueDivision of Therapeutics, University Hospital, Queens Medical Centre, Nottingham; and Novartis Respiratory Research Centre, Wimblehurst Road, Horsham, United Kingdom Address correspondence to: Prof. Ian P. Hall, Division of Therapeutics and Molecular Medicine, South Block, D Floor, University Hospital, Queens Medical Centre, Nottingham NG7 2UH, UK. E-mail: ian.hall{at}nottingham.ac.uk
Elevation of the intracellular free Ca2+ concentration regulates many functional responses in airway smooth muscle, including contraction, proliferation, adhesion, and cell survival. This increase in calcium can be achieved by a release from internal stores (sarcoplasmic reticulum) and/or entry across the cell membrane from the extracellular environment. The molecular identity of this calcium influx pathway in human airway smooth muscle (HASM) remains unclear. Functional studies using Fluo 4-loaded HASM suggest the presence of a histamine H1 receptor-activated Ca2+ entry pathway with characteristics similar to those seen with transient receptor potential (TRP) family homologs. Using a range of molecular and cell biological approaches we defined the expression pattern of transient receptor potential classics (TRPC) homologs in airway cells and tissue. Here we show that HASM and human bronchial epithelial cells both express TRPC1, -4, and -6, with HASM also expressing TRPC3 at the mRNA level. Identification of TRPC6 protein by western blot and confocal microscopy indicated that the protein is localized in specific cell types, suggesting that it plays an important role in regulating key functions in airway cells. These data demonstrate the expression of a range of TRPC homologs in the airway and the presence of a functional Ca2+ entry pathway with characteristics typical of TRPC family members. TRPC homologs may provide an important novel target for the treatment of airway disease.
Abbreviations: airway smooth muscle, ASM • Dulbecco's modified Eagle's medium, DMEM • fetal calf serum, FCS • human airway smooth muscle, HASM • human bronchial epithelial cells, HBEC • phosphate-buffered saline, PBS • polymerase chain reaction, PCR • reverse transcription, RT • Tris-buffered saline, TBS • transient receptor potential, TRP
Calcium homeostasis plays a key role in the regulation of many functional responses in airway cells. In airway smooth muscle (ASM), elevation of intracellular Ca2+ levels promotes contraction and plays a role in cell proliferation, adhesion, and survival (1). The source of Ca2+ for these responses has become clear over recent years. The initial contractile response of ASM is dependent upon the release of Ca2+ from intracellular stores as a consequence of receptor-mediated inositol 1,4,5 triphosphate (IP3) formation (2). However, the sustained elevation of intracellular free Ca2+ concentration ([Ca2+]i), which occurs in response to stimulation with agonists such as bradykinin and histamine, is due to influx of Ca2+ from the extracellular compartment (3). The molecular identity of the influx pathway has to date remained unclear. The influx pathway is relatively insensitive to L-type voltage-operated Ca2+ channel inhibition, but can be blocked by divalent and trivalent cations such as Ni2+ and La3+ (3). This influx pathway is an attractive candidate target for novel pharmacologic tools which might be effective ASM relaxants and hence could potentially act as bronchodilators. However, similar dihydropyridine-insensitive influx pathways have been described in other important airway cells including mast cells, bronchial epithelial cells, and inflammatory cells present within the airway wall (4–7). Therefore, to define the likely importance of receptor-mediated Ca2+ influx pathways as a novel target for new agents that may be of value in the management of airflow obstruction, it is important to define the molecular identity of the relevant influx pathways in the different airway cell types. One strong group of candidates for the Ca2+ influx pathway present in ASM is members of the transient receptor potential (TRP) family (8). Originally identified as a critical component of the Drosophila phototransduction pathway, TRP family members exhibit a number of features that are similar to those of the non–voltage-dependent Ca2+ influx pathway in ASM (9). There are seven known TRPC homologs in the human genome, encoding a group of structurally-related proteins able in transformed cell systems to form Ca2+ influx pathways. There has been considerable controversy regarding the characteristics of the different TRP homologs. Some reports have suggested that TRPC1, TRPC4, and TRPC5 demonstrate characteristics in keeping with store-dependent influx pathways, whereas TRPC3 and TRPC6 exhibit features in keeping with the receptor-operated influx pathway. However, others have suggested the characteristics of TRPC4 and TRPC5 to be more in keeping with receptor-operated pathways (10–14). In addition, recently, it has become clear that splice variants of TRPC4 and TRPC6 exist that show different characteristics, most notably in the case of a truncated TRPC4 splice variant that is able to directly activate the IP3 receptor present on the sarcoplasmic reticulum; this splice variant can also bind to calmodulin (15–17).
To date, there is very little evidence linking mammalian TRPs to native function and no functional data in airway cells. However, a recent study has shown that murine TRPC6 expressed in HEK 293 cells produced almost the exact properties of the In view of the potential role of TRPC homologs as novel ASM targets for the therapy of airflow obstruction, the aims of the current study were to examine the expression pattern of TRP family homologs in ASM and airway epithelium, and to define the splice variants of TRPC4 and TRPC6 present in these cell types. In addition, because of data suggesting that TRPC6 has at least some characteristics similar to the molecular identity of the ASM Ca2+ influx pathway, we studied the distribution in the airway of TRPC6 at the protein level using two antibodies recognizing human TRPC6.
Materials Unless otherwise stated, all chemicals used in this study were purchased from the Sigma Chemical Co. (Poole, Dorset, UK).
Immunohistochemistry Three-micrometer paraffin sections were cut, floated on to water at 52°C, mounted onto Polysine slides (Surgipath Europe, Peterborough, UK), and dried overnight at 37°C. The sections were dewaxed in xylene and taken to 70% industrial methylated spirit. Endogenous peroxidase was blocked with hydrogen peroxide in methanol for 10 min. After washing in running water, the sections were incubated for 15 min at 37°C in 0.1% trypsin (ICN Biomedicals Ltd., Basingstoke, UK) in 0.1% aqueous calcium chloride which had the pH adjusted to 7.8 with 0.1M NaOH (20). Trypsin digestion was stopped by immersion in running water and the sections were washed in Tris-buffered saline, pH 7.6 (TBS) before beginning the PAP staining method. Unwanted background staining was reduced by incubation for 30 min in 0.1% bovine serum albumin in TBS. This was drained from the slides and the sections were incubated in the rabbit anti-TRPC6 diluted 1/1,000 in TBS overnight at 4°C. The sections were allowed to reach ambient before this was washed off with TBS. Swine anti-rabbit immunoglobulins (1/200) and Rabbit PAP (1/400) (DakoCytomation Ltd., Ely, UK) were applied sequentially (diluted in 0.2 M Tris-HCl pH 7.6), for 30 min at ambient. The sections were washed in TBS 3 x 5 min between the stages. Sites of reaction were visualized with 3,3'-diaminobenzidene using the method described by Graham and coworkers (21). The cell nuclei were counterstained with hematoxylin. The sections were dehydrated through graded industrial methylated spirit, cleared in xylene, and coverslipped using a Bayer coverslipper (Bayer plc, Newbury, UK). All data are representative of samples from at least three donor tissues. Peptide blocking was performed as follows. The TRPC6 peptide was dissolved in TBS at concentrations of 0.01–1 mg/ml. These solutions were used to prepare the 1/1,000 dilution of primary antibody. The mixtures of antibody and peptide were allowed to stand for 30 min and then centrifuged. The supernatants were applied to the sections in place of the TBS diluted primary and incubated overnight at 4°C. Staining was subsequently completed as already described.
Generation of Human TRPC6 Monoclonal Antibody
Generation of Human TRPC6 Polyclonal Antibody in Rabbits
Isolation and Characterization of Tracheal Smooth Muscle Cells
The resulting cultures stained positive for
Cell Culture
Differentiation of Human Bronchial Epithelial Cells
Total RNA Isolation and Reverse Transcriptase–Polymerase Chain Reaction Polymerase chain reaction (PCR) amplification was performed using specific primer pairs for each TRPC channel. The PCR conditions were optimized for TRPCs 1–3 with an annealing temperature of 60°C and an annealing temperature of 55°C for the other TRPC homologs (TRPCs 4–6). Cycling was performed 35 times unless otherwise stated (see below) (95°C, then specific annealing temperature, then 72°C, all for 90 s) followed by 10 min at 72°C, and finally reactions were stored at 4°C. Different primer pairs were designed for each of the splice variants. Each of the truncated regions was amplified using a "Touchdown PCR" method (annealing temperature starts at 56°C for one cycle and then decreases by two degrees on subsequent cycles, until the annealing temperature of 46°C is reached, at which point this temperature is maintained for a further 25 cycles). Products were identified by size and confirmed by direct sequencing. As a control we used placental cDNA (reverse-transcribed from human placenta total RNA purchased from Clontech [Oxford, UK]), the tissue in which the variants were first described. For sequencing, 10 µl of the amplified products were separated by electrophoresis on 1% agarose/TAE (Tris, acetic acid, EDTA) gel, and the DNA bands visualized by ethidium bromide staining. These bands were then cut out and cleaned using Promega "Wizard" PCR purification system. Direct sequencing was performed on the resulting PCR products to confirm their identity. The smaller splice variant products were subcloned into the pGEM vector (Promega, Southampton, UK) before being sequenced.
Primer Design
TRPC1 (accession no. X89066), 5' GGA ACT AGC CCG GCA ATG 3' and 5' CTT TCA AGG GCT GGC CCC 3' (reverse). TRPC2 (accession no. X89067), 5' GCC TGC CAC TAT TTC ACC AT 3' and 5' TCA ATC ATC TTG CCA ATG GA 3' (reverse). TRPC3 (accession no. NM_003305), 5' ATG TAA CTA TGG TGG TCG 3' and 5' TGG ACT AGG AAC TAG ACT 3' (reverse). TRPC4 (accession no. AF063822), 5' CTG CCT TGT GTT GCT AGC 3' and 5' CTG CAT GGT CAG CAA TCA 3' (reverse). TRPC5 (accession no. NM_012471), 5' GGT CAA CTA CTC ACC GTA 3' and 5' CTT CCT TGC GTA TGG CAT 3' (reverse). TRPC6 (accession no. NM_004621), 5' GAC TCG TTT AGC CAC TCC 3' and 5' GCG ACC GTG ATC ACC ACT 3' (reverse). TRPC4ß (accession no. AF063823), 5' CCT CGA AGG AGT CTT CAA ATT CGG 3' and 5' CCA CCA CCT TCT CTG ACT TGA ATG G 3' (reverse). TRPC4
Western Blot Analysis
Confocal Microscopy
Measurement of [Ca2+]i The fluorescence was then continuously recorded at wavelengths of 485 nm excitation and 520 nm emission with programmed drug additions using a Flexstation (Molecular Devices, Wokingham, UK). A baseline was established for the first 15 s, after which 20 µl of the test inhibitor or vehicle was incubated with the cells for 3 min before stimulation with histamine (100 µM) to identify any store release. CaCl2 (2 mM final concentration) was added back to the buffer 4 min after histamine addition, to identify any influx from extracellular sources. Data are presented as change in fluorescence intensity (FI), measured in arbitrary fluorescence units, compared with baseline.
An initial series of experiments was performed to confirm the presence of a histamine-induced Ca2+ influx pathway in cultured HASM cells (Figure 1). Histamine induced release of Ca2+ from intracellular stores under low extracellular Ca2+ conditions, and also induced influx of Ca2+ across the membrane when Ca2+ (2 mM) was re-added to the extracellular environment (Figure 1A). Both of these responses were inhibited by the selective histamine H1 antagonist, mepyramine (1 µM) (Figure 1A). Neither the intracellular Ca2+ response to histamine in low extracellular Ca2+ conditions, nor the response to Ca2+ re-addition were sensitive to the L-type voltage gated Ca2+ channel antagonist nifedipine (Figure 1B). In contrast, SK&F 96365, an agent previously described to inhibit receptor-operated Ca2+ entry pathways mediated by TRPC3, TRPC6, and TRPC7 (24), concentration-dependently inhibited Ca2+ entry but not intracellular Ca2+ release (Figure 1C) at concentrations likely to be selective for these homologs.
Given the functional data suggesting that the agonist-induced Ca2+ entry pathway in HASM has characteristics suggestive of the presence of TRPC channels, we next defined the TRPC isoforms (including known splice variants of TRPC4 and 6 [Figure 2]) present in airway smooth muscle and epithelial cells. We used sequence specific primers designed to selectively amplify individual TRP family members and an RT-PCR approach to identify which TRPC family members are expressed in relevant airway cell types. Where possible, primers were designed to amplify products spanning intron/exon boundaries to ensure expression profiles were not misinterpreted by genomic DNA contamination. The cell types studied were primary cultured human ASM, cultured undifferentiated human bronchial epithelial cells (HBEC), differentiated HBECs following culture at an air–liquid interface, and the transformed epithelial cell line BEAS 2B. We found TRPC1 to be ubiquitously expressed. In contrast, TRPC3 was not expressed in HBECs (both undifferentiated and differentiated) despite being expressed in the BEAS 2B cell line, and both TRPC4 and TRPC6 were absent from BEAS 2B cells despite being expressed in the other cell types studied (Figure 3, Table 1). Because of the additional products in some experiments, the identity of all TRPC homologs seen by RT-PCR was confirmed by sequencing of the relevant band excised from gels to confirm identity of the products obtained.
Having demonstrated the presence of both TRPC4 and TRPC6 in ASM and HBEC, we next defined the splice variants of these TRPC family members present in these cell types. Both the TRPC4ß splice variant and the  316–431 deletion TRPC6 (accession no. AJ271067) variant were expressed in ASM and HBEC (Figure 4, Table 2). The molecular identity of all the splice variants observed was confirmed by direct sequencing, either of the PCR product excised from the gel (TRPC4ß) or following subcloning of the PCR product into pGEM (316–431 TRPC6).
The characteristics of the ASM Ca2+ influx pathway, which is activated in response to agonists such as histamine (see above), share a number of features exhibited by full-length TRPC6 expressed in other cell types (18). In particular, Ca2+ influx appears to be dependent upon presence of agonist and was inhibited at appropriate concentrations by SK&F 96,365. Therefore, we extended our molecular studies to expression of TRPC6 at the protein level by using both a polyclonal and a monoclonal antibody recognizing TRPC6. To confirm the specificity of the monoclonal antibody we performed Western blot analyses using lysates from ASM and HBEC: lysates from BEAS 2B cells, as expected, produced no specific bands (data not shown). Both ASM and undifferentiated HBEC lysates produced a 97-kD band corresponding with the expected size of full-length TRPC6. In both cases a band of lower molecular weight was also observed which may represent the TRPC6 splice variant identified in these cells (Figure 5). A third band was also seen with a higher molecular weight, the identity of which is unclear, although it is possible that it may represent a glycosylated form of TRPC6.
We next examined the distribution of TRPC6 within ASM and HBEC by confocal microscopy. HBEC were differentiated at an air–liquid interface as previously described (21) giving rise to an organized epithelium similar to that seen in vivo. Both mucus-secreting and ciliated cells were present in the apical layer (Figure 6A), but absent from the basal cell layer, as previously described (23). In differentiated HBEC populations, TRPC6 was localized to the apical surface, being present in ciliated epithelial cells and mucus-secreting cells but absent from basal cells (Figure 6B). The apparent apical distribution of TRPC6 in differentiated epithelial cultures is not due to lack of penetration by the antibody because specific staining of desmoplakin can be observed at all levels in the epithelial culture (Figure 6C). In ASM, TRPC6 is found both with a wide cytoplasmic distribution and generally showed evidence of subcellular localization (seen as punctate staining in some regions of the cytoplasm) (Figure 6D).
To extend these studies to examine the distribution of TRPC6 in tissue ex vivo, we used the polyclonal antibody against TRPC6 in tissue slices from human lung sections. Specific staining for TRPC6 was seen throughout the ASM layer, on both apical and basal epithelial cells and on mucosal gland cells within the lung (Figure 7). Although the heaviest staining was observed in epithelial cells, all ASM examined demonstrated specific staining, indicating that TRPC6 is widely expressed in ASM cells.
TRP homologs are an attractive target for the development of novel therapeutic agents for the treatment of airflow obstruction. Sustained contraction of ASM is dependent upon entry of Ca2+ across the cell membrane (4) and the characteristics of the agonist-induced influx pathway present in ASM demonstrated in the studies described in this paper show a number of features consistent with those of TRPC family members in transformed cell systems (3, 12, 13). In particular, the prolonged influx of Ca2+ following agonist stimulation is insensitive to L-type voltage-dependent Ca2+ channel antagonists such as nifedipine, but sensitive to inhibition by SK&F 96365 at appropriate concentrations (  10 µM). In addition, in previous studies the sustained entry of Ca2+ following stimulation with histamine was sensitive to inhibition by the addition of the H1 selective inhibitor mepyramine, indicating that Ca2+ influx requires the continued presence of agonist, again in keeping with receptor-mediated stimulation of the influx pathway (3). However, TRPC family members have been identified in many other cell types likely to be represented in the lung, including epithelial cells, inflammatory cells, and nerves (5, 24–27). Therefore, we set out to determine the TRPC family members present in a number of commonly used cell systems, which have been established to examine airway pharmacology. To ensure that RT-PCR approaches were specific for the cell types of interest, we used primary cell cultures of relevant cell types in which contamination by small numbers of other cells would be avoided. The models chosen were human ASM cells, which display many characteristics of ASM ex vivo (28), both undifferentiated and differentiated HBEC cultures (23), and the frequently used bronchial epithelial cell line BEAS 2B. We observed differential expression of TRPC family members in these different cell types, most notably with TRPC3 being absent from HBEC cultures and with TRPC6 being absent from BEAS 2B cells. The lack of expression of TRPC6 in BEAS 2B cells, a widely used epithelial cell line, is presumably because of partial dedifferentiation in vitro, and suggests these cells may not be a good model for studying calcium handling in airway epithelial cells. Because the functional characteristics of TRPC6 make it an attractive target in ASM (13), we extended our studies to examine TRPC6 expression at the protein level by generating two antibodies recognizing human TRPC6. In studies using a monoclonal TRPC6 antibody, we found TRPC6 expression to be limited to the apical layer of HBEC cultures following differentiation at an air–liquid interface. In studies on tissue slices of human lung we observed in addition positive staining in the basolateral layer as well as the apical surface. The explanation for the lack of staining in the basal cell layer in the in vitro differentiated airway epithelial system is unclear, although we believe it is not due to the inability of the antibody to access the cells in the deeper layer of the epithelium as we observed positive staining with a control (desmoplakin antibody). Further studies examining protein expression of other TRP family members will depend upon the availability of suitable antibodies. The molecular identity of the Ca2+ entry pathway in ASM remains an important issue to resolve. At present there are no effective small molecule antagonists available for the study of TRPC homologs, and hence, although the TRPC family remains an attractive candidate for this pathway, studies are limited to expression profiling approaches. In the current study, we observed expression of TRPC1, TRPC3, TRPC4, and TRPC6 in ASM. TRPC2 was not expressed in ASM, although as this is believed to be a pseudogene it is unlikely to be functionally relevant. TRPC1 and TRPC4 demonstrate characteristics in at least some recombinant cell systems more in keeping with a store-operated (capacitative entry) channel (11), and although this may be important in ASM for replenishing intracellular stores, it is perhaps less likely to be important for agonist-induced Ca2+ influx. In keeping with an additional role for store-operated channels in these cells is the observation that thapsigargin (2 µM) induces a relatively small, slow rise in intracellular Ca2+ with a further rise on readdition of Ca2+ (data not shown). We believe, therefore, that TRPC3 and TRPC6 of the currently identified TRPC family members are the best candidates for the molecular identity of the Ca2+ pathway present in ASM, although additional homologs are undoubtedly present at the mRNA level and ASM also expresses an entry pathway with store operated characteristics. In recombinant cell systems, TRPC6 demonstrates a number of features, suggesting that it most closely resembles the features of the agonist-induced ASM Ca2+ influx pathway (3, 18). In vascular smooth muscle cells from a variety of tissue sources, e.g., portal vein and cerebral arteries, TRPC6 has already been shown to be involved in receptor-stimulated Ca2+ influx in smooth muscle cells (18, 19). These data provide further evidence to suggest that TRPC6 may have a potentially important role in the control of the vascular smooth muscle contractile response. The data presented here suggest that TRPC6 may also fulfill a similar role in airway smooth muscle function. In view of this, we concentrated on using TRPC6 antibodies to examine the tissue distribution of TRPC6 at the protein level. We found expression of TRPC6 both in cultured human ASM cells (the same cells that we also found to express a Ca2+ entry pathway with characteristics compatible with TRPC6) and also ASM in vivo. However, it must be recognized that there may be redundancy in this influx pathway with both TRPC3 and -6 (and indeed other homologs) contributing to the pathway. A further level of complexity may be added because of the potential for TRPC family members to form multimeric units: indeed, the most likely conformation of the Ca2+ influx pathway would consist of a complex of four TRPC units (29–31). Although these may be composed of single TRPC family members, the possibility exists that the channel may in fact be composed of different TRPC family members yet staying within the confines of the subfamilies (TRPC4/5 or TRPC3/6/7) (22, 31), and with further added complexity being created by the existence of splice variants, which could also form part of the channel (16). With this in mind, it is important to note that splice variants of TRPC4 and TRPC6 are expressed in ASM. At present, there are no good tools to examine the complexes formed at the cell membrane, and further studies will be dependent upon the generation of suitable antibodies and ideally specific small molecule antagonists. In summary, therefore, our studies have demonstrated that ASM does not contain a single TRPC family homolog that can form a Ca2+ entry pathway uniquely expressed in ASM. However, this does not rule out TRP family members such as TRPC6 as an attractive target. There are many examples of other useful targets which are widely expressed but for which valuable therapeutic agents exist, including the L-type voltage-dependent Ca2+ entry pathway in vascular smooth muscle (32, 33). Further studies are therefore required to generate the necessary tools to start to address the importance of TRPC family members as therapeutic targets in ASM with the aim of generating a novel class of ASM relaxants.
This work was funded by the National Asthma Campaign (UK). R.C. is a recipient of a Ph.D. studentship sponsored by the Novartis Respiratory Research Centre, Horsham, UK. The authors are indebted to Dr. Adrian Rogers, Department of Biology and Biochemistry, University of Bath, UK, who generated and characterized the hybridoma cell line secreting the human TRPC6 monoclonal antibody used in this study and characterized the human TRPC6 monoclonal antibody itself. We gratefully acknowledge Henry Danahay, Hazel Atherton, and Gareth Jones for their assistance with the human bronchial epithelial cell (HBEC) studies and thank Alex Aitken for support with the splice variant PCR work. Received in original form April 11, 2003 Received in final form June 6, 2003
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Abstract
Introduction
1-adrenoceptor–activated Ca2+ permeable cation channel identified in portal vein smooth muscle (
(accession no. AF063824), 5' GCG AGC TGC TGA TAA CTT GAG AAG 3' and 5' CCA CCA CCT TCT CTG ACT TGA ATG G 3' (reverse). TRPC4
(accession no. AF063825), CGT TAT TTG AGA CAC TGC AGT CCC and 5'CAT TGC AGC AAC GTA TCG CTT CACC (reverse). TRPC6 splice site, 5' TGA CTC GTT TAG CCA CTC and 5' TGA TGC TGC GAA AAT TGC (reverse). 
97 kD) in HASM and human undifferentiated bronchial epithelial cell lysates was performed using Western blot analysis. Confluent 75-cm2 flasks of cells were lysed in 100 µl of buffer (1% Triton-X-100, 10 mM Tris-HCl pH8, 150 mM NaCl, 0.05% Tween 20) and centrifuged at 10,000 x g for 10 min, and the supernatant saved. Protein concentration of the samples was estimated using a protein assay kit (BioRad, Hertfordshire, UK) which allowed 20 µg of sample to be loaded per well. The samples were separated using 10% Bis-Tris precast gels (Invitrogen, Paisley, UK) under reducing conditions. The protein was then transferred using a semi-dry method on to PVDF membranes (Millipore) and probed with an anti-TRPC6 monoclonal antibody (Novartis). Amplification of the signal was achieved using a biotinylated anti mouse secondary antibody followed by an avidin/biotin complex conjugated to horseradish peroxidase (Vectorstain ABC kit; Vector Labs, Peterborough, UK) and detected by chemiluminescence (ECL reagent; Amersham). 
