| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
INRS-Institut Armand-Frappier, Université du Québec, Pointe Claire, Québec, Canada H9R 1G6
ABSTRACT
Claudin 1 (CLDN1) is a tight junctional protein present in the epididymis. Limited information exists regarding the regulation of Cldn1 transcription. In the epididymis, the regulation of the 5' flanking region of genes coding for tight junctional proteins is unknown. The present objectives were to investigate the transcriptional regulation of the Cldn1 gene in the rat epididymis. A 1.8-kb sequence of the 5' flanking region of the rat Cldn1 gene was cloned. The transcriptional start site is an adenine located at the –198 position relative to the first codon, and 26 bp downstream of the putative TATA box. It is the only start site for the Cldn1 gene transcription in the rat epididymis. The Cldn1 promoter was inserted into a luciferase gene expression vector and transfected into a rat caput epididymal cell line (RCE-1). Sequential deletion analysis revealed that minimal promoter activity was achieved with the construct containing –61 to +164 bp of the promoter. This sequence contained a TATA box and two consensus SP1 binding sites. Electrophoretic mobility shift and supershift assays confirmed that SP1 and SP3 were present in RCE-1 cells and epididymal nuclear extracts, and that they bind to the 5' SP1 binding motif of the promoter. Site-directed mutagenesis of the 5' SP1 binding site resulted in a 4-fold decrease in transactivation of the minimal promoter sequence. These findings indicate that SP1 and SP3 bind to the Cldn1 promoter region, and that this interaction influences the expression of Cldn1 in the rat epididymis.
blood-epididymal barrier, epididymis, gene regulation, male reproductive tract,, RCE cells, SP1, tight junctions, transcription factors
Epididymal tight junctions play a critical role in the formation of the blood-epididymal barrier [1]. The formation and integrity of this barrier is believed to be essential for the maintenance of the luminal environment of the epididymis, which allows for the maturation of spermatozoa. Epididymal tight junctions were first identified by freeze-fracture electron microscopy [2]. These tight junctional complexes contain elaborate strands, which vary in their complexity along the different regions of the epididymis [3].
Tight junctions are comprised of a series of continuous, anastomosing, intramembranous particle strands [4–6]. These strands are comprised of integral transmembrane proteins, such as occludin and claudins (CLDNs; [7]), which are linked to membrane-associated proteins, such as junction-associated membrane proteins [8] and peripheral membrane proteins, including the membrane-associated guanylate kinase homologues (MAGUK; include tight junction protein 1 [TJP1], TJP2, and TJP3 [9–12]), as well as non-MAGUK proteins (e.g., symplekin, cingulin, and 7H6 antigen [13–15]). CLDNs are transmembrane proteins that are essential to tight junctions [16, 17, 8]. In addition to their contribution to the barrier function of tight junctions, CLDNs have also been associated with specific ion transport [8, 18, 19]. In the epididymis, occludin and CLDN1 have been shown to be present in epididymal tight junctions [20–22]. More recently, it has been reported that epididymal epithelial tight junctions are comprised of several CLDNs, including CLDNs 3, 4, 6–10, and 16 [23, 24].
There is limited information on the regulation of the blood-epididymal barrier and epididymal tight junctions. Suzuki and Nagano [25] reported that the number of strands was decreased in epididymal tight junctions of mice following orchidectomy. In the rat, Levy and Robaire [20] reported that, in aging brown Norway rats, epididymal tight junctions in the corpus region of the epididymis were compromised, suggesting that these were, in part, responsible for the loss of fertility associated with aging. Gregory et al. [22] showed that CLDN1 in the initial segment of the rat epididymis was regulated by testicular factors, including androgens. In other regions of the epididymis, however, testicular factors did not appear to influence CLDN1 expression.
The mechanisms regulating transcription of the Cldn1 gene are still emerging. Cldn1 gene expression is downregulated in breast cancer cells [26] and upregulated in colorectal cancer cells [27]. In MDCK epithelial cells, it has been reported that the expression of Cldn1 gene is inhibited by the Snail (SNAI1) and Slug (SNAI2) transcription factors that bind E-box elements of the human promoter, providing an explanation for the downregulation of the CLDN1 observed in breast cancer cells [28]. Recent studies have indicated that SNAI1 regulates Cldn1 at the level of translation [29]. The downregulation of CLDN1 in tumorigenic epithelial cells is thought to be dependent on additional factors, since SNAI1 and SNAI2 are also present in nontumorigenic cells. Transactivation of the human Cldn1 gene in colon cancer cells is mediated by two transcription factor (TCF) 4 binding motifs, which are bound by a catenin beta1-TCF/lymphocyte enhancer factor complex. There is no information on the factors that regulate Cldn1 transcription in the epididymis.
The objectives of this study were to clone the 5' flanking region of the rat Cldn1 gene, and to assess the regulation of the Cldn1 promoter in the rat epididymis, using an epididymal cell line as a model, in order to obtain a better understanding of the factors that regulate the expression of epididymal tight junctional proteins and the blood-epididymal barrier.
Isolation of the 5' Flanking Region of the Rat Cldn1 Gene
The Genome Walker kit (Clontech, Palo Alto, CA) was used to amplify a 1787-bp fragment. The first amplification reaction was done using Cldn1 REV3 (5' CGT AGA TGG CCT GAG CAG TCA CGA TGT 3') as the gene-specific primer 1 (+116 to +142 relative to the adenine of the methionine start codon) and adaptor primer 1. A second PCR was done using Cldn1 REV4 (5' CGC GTT GGC CAT GGC TCT TTT CT 3') as the gene-specific primer 2 (–11 to +12 relative to the adenine of the methionine start codon) and nested adaptor primer 2. These gene-specific primers were derived from an 893-bp rat cDNA previously cloned, corresponding to the complete coding sequence of Cldn1 plus 120 bp of untranslated sequence upstream of the ATG (GenBank accession no. AF195500). The longest PCR fragment was obtained from the SspI digested genomic DNA, gel purified (Qiagen, Mississauga, ON) and cloned into the PCRII-Topo vector (Invitrogen, Burlington, ON). Sequencing results (Sheldon Biotech, Montreal, PQ) allowed the design of an internal primer pair (Cldn1 Forw3 5' GGC CTT AGA TTC TCC TGT T 3' and Cldn1 REV5 5' AAA ACT GGA GGG AGA TAA AA 3') used for PCR amplification and complete sequence determination of the cloned DNA.
The promoter sequence was compared to the rat genomic sequence using the Basic Local Alignment Search Tool (BLASTN 2.2.12; available at http://www.ncbi.nlm.nih.gov/BLAST). Determination of putative transcription factor binding sites located upstream of the Cldn1 coding sequence was done using TRANSFAC (v4.0; Transcription Element Search System; available at http://www.cbil.upenn.edu/tess). The search was restricted to the rat species, and allowed no mismatch. The position of the putative TATA box was predicted using the Proscan software (v1.7; available at http://thr.cit.nih.gov/molbio/proscan).
Rapid Amplification of cDNA Ends (RACE)
Total RNA was extracted from the entire epididymis of adult Sprague-Dawley rats using the guanidine thiocyanate method [30]. The 5' portion of Cldn1 mRNA was amplified with the BD SMART RACE cDNA Amplification kit (Clontech) according to the manufacturer's instructions. Cldn1 REV3 and Cldn1 REV4 were used as gene-specific primers in two separate reactions, which were designed to amplify products with 130-bp size difference. The human transferrin gene that was supplied with the amplification kit was used as a positive control. Each PCR product was gel-purified and cloned into the PCRII-Topo vector. The resulting clones were analyzed with restriction enzymes, and two clones per PCR reaction were sent for automated sequencing (DNA Landmarks, Ste.-Julie, PQ). All procedures involving animals used in this study were approved by the university animal care committee.
Plasmid Construction for Gene Reporter Assays
Sequential deletion constructs of the Cldn1 promoter were produced by PCR amplification using one reverse primer and different forward primers, as listed in the Table 1. Restriction sites were inserted into the primers to insure that the cloning was unidirectional. The plasmid DNA containing the entire Cldn1 promoter was used as template. PCR products (100–200 µl) were ethanol-precipitated, gel-purified, and double-digested with NheI and KpnI (GE Healthcare, Baie D'Urfe, PQ). The resulting fragments spanned the promoter region from positions –1414, –613, –226, –173, –125, –61, and –7 to position +164 relative to the transcriptional start site. The digested products were loaded onto an agarose gel, separated by electrophoresis, and purified. These fragments were ligated into the NheI and KpnI cloning site of the pGL3-Basic vector upstream of the firefly luciferase gene (Promega, Madison, WI) using T4 DNA ligase (Invitrogen). Competent JM109 bacteria (kindly provided by Dr. M. Sylvestre, INRS) were chemically transformed with the seven different constructs. Colonies were analyzed by restriction enzyme digest, and positive clones were selected, grown, purified with a commercial kit (Plasmid Midi Kit; Qiagen), and stored at –20°C.
|
The 5' SP1 binding site in pGL3 constructs containing the –61 to +164 fragment of the Cldn1 promoter was mutated using the QuickChange II Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA). A four-base mutation (GGCG to TTAC) was introduced according to the manufacturer's instructions using a sense mutagenic primer 5' GGT ACC GCT GCA TGC CGG GAG GTT ACG AGC TGC TTT AAA TCG 3' and its complementary strand as the antisense primer (Operon Biotechnologies, Huntsville, AL). The resulting selected clones were sequenced (DNA Landmarks) and grown. Plasmids were purified (Plasmid Midi Kit) and stored at –20°C.
Cell Culture and Transfections
RCE-1 cells were cultured in Dulbecco modified Eagle medium/Ham nutrient mixture F12 (Sigma), containing the appropriate supplements, in a humidified incubator at 32°C with 5% CO2, as previously described [31]. Cells were seeded at a density of 100 000 cells per well in collagen IV-coated 24-well plates (BD Biosciences, San Jose, CA). Cells were allowed to adhere overnight. The next day, cells were washed with PBS, and the medium was replaced with 500 µl per well of medium without antibiotics. The cells were transfected using 2 µl of Lipofectamine 2000 (Invitrogen) and 1 µg of each of the pGL3 constructs combined with 100 ng of phRL-TK vector (Promega), in a total volume of 100 µl. The phRL-TK vector expresses the Renilla luciferase from the herpes simplex virus thymidine kinase promoter and was used as an internal control for transfection efficiency. A promotorless pGL3-Basic empty vector was used as a negative control. A plasmid containing the luciferase gene under the control of the Rous sarcoma virus long terminal repeat (pRSV-L) was used as a positive control (kindly provided by Dr. R.S. Viger). All experiments were done in triplicate. The cells were placed in the incubator for 24 h; medium was subsequently removed, and the cells were washed with PBS. Cells were lysed directly in the plate using 125 µl of Passive Lysis Buffer (Promega). Firefly and Renilla luciferase activities of a 20-µl aliquot were determined using the Dual-Luciferase Assay kit (Promega) and an MLX microtiter plate luminometer (Dynex Technologies, Chatilly, VA). Protein concentration for each sample was determined using a commercial Bradford reagent (Bio-Rad Laboratories, Mississauga, ON). Relative luciferase activities were expressed as the ratio of firefly to Renilla luciferase activity, normalized to the protein concentration.
Electrophoretic Mobility Shift Assays (EMSA)
Nuclear extracts were prepared from epididymides (initial segment or caput epididymidis) of adult Sprague-Dawley rats and from RCE cells with a commercial kit (Nuclear Extraction kit; Active Motif, Carlsbad, CA). Protein concentrations were measured using Bradford reagent. Three oligonucleotides spanning the region of interest (positions –9 to –56 relative to the transcription start site; Table 2) were synthesized, as were their respective complementary strands (Invitrogen). A 10-pmol aliquot of the sense oligonucleotide was end-labeled with 50 µCi of [
-32P] ATP using 5 U of T4 polynucleotide kinase and reaction buffer (GE Healthcare) in a total volume of 50 µl. Following a 30-min incubation at 37°C, the labeled oligonucleotide was heated at 65°C for 10 min and incubated for 30 min at room temperature with a 2.5 molar excess of its antisense counterpart preheated in the same manner. The double-strand oligonucleotide was purified on a Sephadex G-50 column (Nick column; GE Healthcare) and the radioactive content of a 2-µl aliquot of the resulting fraction was determined in order to ensure a specific activity of at least 5000 cpm/fmol. Nuclear protein extract preparations (10 µg; a pool of three separate individuals) were mixed with 1x binding buffer (10 mM Tris-HCl, pH 7.5, 4% glycerol, 1 mM MgCl2, 0.5 mM EDTA, 50 mM NaCl, 0.5 mM DTT) and poly-dI-dC (50 µg/ml), and preincubated, when appropriate, with unlabeled oligonucleotide or antibodies (0.5–2 µg; Santa Cruz Biotechnologies, Santa Cruz, CA) for 30 min at room temperature. Labeled oligognucleotides (30 fmol) were then added to the mixture for a final volume of 15 µl, and incubated for an additional 30 min at room temperature. The binding reaction products were then mixed with 3 µl of 6x DNA loading buffer and separated on a 6% polyacrylamide gel. Gels were dried and exposed overnight with a phosphorus screen and scanned with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
|
Using a commercial kit and specific primers derived from a previously cloned cDNA [22], we isolated and cloned a fragment of the Cldn1 gene, containing 1775 bp of the untranslated sequence, 5' upstream of the translational start codon. This sequence showed 96% identity with the Rattus norvegicus chromosome 11 genomic contig (NW_047356.1) when analyzed with BLASTN. Figure 1 represents part of the sequence, including a 433 bp upstream fragment of the translation start codon. This sequence part shows a 98% homology with the genomic contig as analyzed with BLASTN. The TATA box was predicted by Proscan analysis to be located between positions –27 to –33. A large number (52) of transcription factor binding sites identified in the rat species were found using TRANSFAC software.
|
RACE experiments were done using the same specific primers in two separate reactions, and two products with a difference in size of 130 bp (369 bp and 239 bp) were gel-purified and cloned successfully (Fig. 2A; lanes 1 and 2, respectively). Two positive clones for each reaction were sequenced, and the transcription start site was determined to be an adenine at position –198 relative to the ATG. Sequencing results of the four clones (Fig. 2B) gave an additional guanine located at the junction between the genomic sequence and the BD SMART II A oligonucleotide.
|
Luciferase activity experiments were conducted with RCE-1 cells to identify the regions of the promoter that were important for the transcriptional regulation of the Cldn1 gene. Each of the seven different constructs used expanded to the –35 position relative to the ATG start codon. Reporter vectors containing the full-length Cldn1 promoter (–1414) as well as the three successive deletion constructs (–613, –226, and –173) showed similar transactivation potential in the RCE-1 cell line (Fig. 3). A decrease in the luciferase activity was observed when the cells were transfected with the plasmid containing the fragment starting at the position –125 of the Cldn1 promoter. No predicted transcription factor could be associated with this specific region of the promoter (between the Cldn1-310pb and Cldn1-248pb primers; Fig. 1). Conversely, luciferase activity was maintained at levels similar to those obtained with the longest constructs in the transfection experiments, using a reporter vector containing the fragment spanning the –61 to +124 positions of the promoter (Fig. 3). This region contains the putative TATA box and two potential SP1 transcription factor-binding sites (Fig. 1). The –7 construct excludes these three recognition sequences, and resulted in a loss of luciferase activity when this vector was transfected into the cells (Fig. 3).
|
To determine if epididymal nuclear proteins that bind specifically to the Cldn1 promoter could be detected, we performed EMSA on the region responsible for minimal promoter activity (construct –61). Two oligonucleotides, A and B, span the promoter sequence specific to the –61 construct, which is associated with elevated luciferase activity or the minimal promoter. Oligonucleotide A (positions –9 to –36) contains the putative TATA box and the 3' SP1 binding site. Nuclear proteins from RCE-1 cells, as well as from the initial segment and caput epididymidis of adult rat, could specifically bind to this region of the promoter (Fig. 4; Oligo A, Table 2). A weak binding complex of high molecular weight was observed with the nuclear extract prepared from the rat epididymis, and was compared to those prepared from the RCE-1 cell line. Nuclear proteins from the cell line also exhibited a lower molecular weight binding complex, which could be out-competed using excess cold oligonucleotide (Fig. 4). Since this oligonucleotide (Oligo A) contained an SP binding site, supershift assays were performed using SP1 and SP3 antibodies. However, the binding complex was not supershifted in preparations with nuclear proteins from RCE-1 cells or from rat epididymal tissues (initial segment or caput epididymidis), suggesting that the transcription factors are not SP1 or SP3 (data not shown).
|
Oligonucleotide B, the sequence of which corresponds to the Cldn1 promoter sequence from position –34 to –56 (Oligo B; Table 2), contains a 5' Sp1 consensus sequence. When compared to the 3' SP1 putative binding site, this site showed two additional bases, which could account for the consensus rat sequence. Two protein-DNA complexes were observed in EMSA when Oligo B was incubated with nuclear extracts prepared from either RCE-1 cells, initial segment, or caput epididymidis (Fig. 5A). These two bands were specific and could be out-competed by an excess of unlabeled Oligo B. An excess of unlabeled Oligo B mutated in its SP1 binding consensus sequence could not compete with the binding complex, suggesting that binding proteins were associated with the SP binding site. As shown in Figure 5A, supershifted complexes were obtained when nuclear extracts were preincubated with SP1 antibody (0.5 µg). These supershifts led to the disappearance primarily of the higher molecular weight complex. The use of more than 0.5 µg of antibody slightly inhibited the binding reaction (data not shown). Supershift experiments using an SP3 antibody resulted in a supershift of the lower molecular weight band in either RCE-1 cells (Fig. 5B), initial segment, and caput epididymidis (data not shown). These results suggest that SP3 can also bind to the SP binding site on Oligo B, and both transcription factors may regulate Cldn1 minimal promoter activity.
|
To elucidate if the 5' SP1 binding site was implicated in the transactivation of the Cldn1 gene observed in vitro, we mutated the reporter construct containing the –61 to +164 minimal promoter sequence. This mutation was the same as that introduced into the mutated Oligo B for the EMSA experiments. The insertion of the mutation was confirmed by sequencing. Transfection of the mutated –61 construct into the RCE-1 cells resulted in a 4-fold loss in luciferase activity as compared to wild-type constructs (Fig. 6), thus supporting the notion that this SP binding site is necessary for Cldn1 minimal promoter activity.
|
The regulation of the blood-epididymal barrier remains poorly understood. Studies have shown that occludin and the family of CLDN proteins are localized to the tight junctions between epithelial principal cells of the epididymis [21–24]. We have previously reported that CLDN1 is expressed throughout the epididymis [22]. In all regions of the epididymis, it is localized both in the area of the tight junction and along the lateral plasma membrane of adjacent principal cells and between basal cells. CLDN1 is expressed by RCE-1 cells, and is localized along the plasma membrane surrounding the cells [31].
The mouse and human Cldn1 promoter regions contain a region of high homology (89% identity) that comprise an SP1 binding site upstream of a TA-containing element thought to operate as an atypical TATA box [32]. These motifs are also conserved in the rat (see Fig. 1). A second SP1 binding site located at the 3' proximity of the TA-element is present in the mouse and rat genes, but not in the human ortholog. On a computational basis, Krämer et al. [32] predicted the transcriptional start site to be located at positions –237 and –212 upstream of the ATG in the human and mouse species, respectively. These positions would correspond to the T nucleotide at position –200 in the rat sequence. Our experiments using 5'-RACE confirmed the actual transcriptional start site as an adenine located at position –198 relative to the ATG translational start site, 27 bp downstream of the putative TATA box (Fig. 1). This nucleotide is contained in a context similar to an Inr consensus sequence (Py Py A(+1) N T/A Py Py; reviewed in [33]), except for the two last pyrimidines. The human Cldn1 promoter sequence matches precisely the Inr consensus motif, in contrast to the rat and mouse sequences.
Sequencing results of the clones obtained in the RACE experiments suggest that we isolated the full-length mRNAs, at least their 5' ends, and reverse transcribed the methylated guanine forming the CAP structure, yielding an additional G nucleotide not present in the genomic sequence (Fig. 2B). Our RACE results indicate that, even though Cldn1 has a noncanonical TATA box, the transcription of Cldn1 results in the synthesis of a single transcript with a common 5' end. Previous studies on the 4-ene-steroid-5
-reductase 1 gene in the epididymis, which also has a noncanonical TATA box, suggested that multiple transcripts were synthesized as a result of the noncanonical TATA box [34]; this is not the case for Cldn1 in the epididymis.
Transfection experiments revealed that the entire Cldn1 promoter was sufficient to drive the expression of a reporter gene in the RCE-1 rat epididymal cell line. Different levels of transactivation were obtained when the promoter was analyzed by sequential deletion, suggesting that a number of regulatory factors may be associated with Cldn1 expression in the epididymis. Regions with full transactivation activity (–1414 to –125 and –61 to –7) and a region with repressor activity (–125 to –61) were also identified.
In this study, we focused on the minimal required promoter region of the Cldn1 gene located between nucleotides –61 to +124 from the transcriptional start. The minimal promoter contains conserved SP1 binding sites on each side of a putative TATA box. Even if other transcription factor binding sites were predicted, these sites were thought to be actual binding motifs in previous analyses of the mouse and human promoters [32]. SP1 and SP3 are ubiquitously expressed transcription factors that bind to GC-box elements of promoters and regulate the transcription of a variety of genes [35]. SP1 and SP3 have been shown to be expressed in the rat epididymis, and were localized predominantly to the nuclei of principal cells in the initial segment and caput epididymidis [34]. Results from this study show that both SP1 and SP3 are present in nuclear extracts prepared from the cell line and bind to the 5' SP1 binding sequence. The DNA-protein complexes were comparable to those observed with nuclear extracts prepared from the initial segment and caput epididymidis of the rat (see Fig. 5A). This interaction appeared to be necessary to the transactivation observed in the epididymal cell line, since the luciferase activity of the minimal promoter construct was lost when the SP binding site was mutated. SP1 has previously been implicated in the transcriptional regulation of two other epididymal genes that are highly expressed in the rat and mouse epididymis: 4-ene-steroid-5-reductase type 1 and glutamyl transpeptidase [36, 34]. Although these genes have similar GC-box and TATA cis-regulatory elements present in their respective promoters, these cis-sites appear to be selectively utilized in the structural context of each of these promoter, facilitating differential expression of these genes in the epididymis. The fact that SP1 and SP3 are localized primarily to the initial and caput regions of the epididymis may, in part, explain why CLDN1 expression is regulated by testicular factors only in the initial segment of the rat epididymis [22].
SP1 and SP3 transcription factors could not be detected in the protein-DNA complexes formed with the 3' SP1 binding sequence. This binding site is not conserved in the human sequence, and does not appear to be strongly implicated in the transcriptional regulation of Cldn1. It is possible that other factors of the SP family bind to this motif, or that the specific bands obtained in the EMSA experiment arose from the interaction of proteins with the putative TATA box. Further experiments will be needed to confirm this.
SP1 and SP3 transcription factors are implicated in the transactivation of other Cldn genes. Honda et al. [37] reported that the human Cldn4 promoter contains two SP1 binding sites and that both are necessary for the transcription of Cldn4 in ovarian cancer cells. Although SP3 could also bind to these sites, its knockdown did not affect the CLDN4 protein levels. Epigenetic modifications to the SP1 binding region of the Cldn4 promoter were implicated in the regulation of transcription. In kidney cell lines, Luk et al. [38] showed that an SP1 binding site was necessary for regulating the activity of the mouse Cldn19 promoter. This promoter contains no TATA box and has an Inr element with little homology with the consensus sequence. It has been suggested that other factors are involved in Cldn19 gene transcription, since the expression level of SP1 is relatively low in mouse kidney cells, even though CLDN19 is highly expressed in this tissue. Interestingly, Cldn1 transcripts are also present in high levels in the mouse kidney compared with other tissues [39].
It is clear that, even if transcription factors, such as SP1 and SP3, are necessary for the expression of the Cldn1 gene in the epididymis, other factors play a role in regulating the transcription of Cldn1. Results from our sequential deletion analysis suggest that inhibitory factors interact with the –125 to –61 region of the Cldn1 promoter. Further experiments will be needed to identify the factors implicated in this regulation. In conclusion, this is the first report of Cldn1 gene regulation in the epididymis, and indicates that SP1 and SP3 transcription factors are involved in the regulation of epididymal Cldn1. This information is the basis for understanding not only the regulation of CLDN1, but also the blood-epididymal barrier.
ACKNOWLEDGMENTS
M. Gregory is thanked for her helpful suggestions.
FOOTNOTES
1Supported by a NSERC-CIHR collaborative health grant awarded to D.G.C. 
Correspondence: 2Daniel Cyr, INRS-Institut Armand Frappier, Université du Québec, 245 Hymus Boulevard, Pointe Claire, PQ, Canada H9R 1G6. FAX: 514 630 8850; e-mail: daniel.cyr{at}iaf.inrs.ca
Received: 14 September 2006.
First decision: 20 October 2006.
Accepted: 17 January 2007.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
