By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Centre de Recherche du Centre Hospitalier de l'Université Laval (CHUQ), 2705 boulevard Laurier, T1-49, Quebec City, Quebec, Canada G1V 4G2 (e-mail: firstname.lastname@example.org).
ABSTRACT: The epididymis is essential for the acquisition of sperm fertilizing ability and forward motility. After vasectomy, the flux and composition of the epididymal fluid are modified, causing possible sequelae to the occluded excurrent duct. Some of these sequelae may not be reversible following vasovasostomy, affecting sperm physiology and their fertilizing ability. We previously demonstrated that the epididymal expression in men of a major glycoprotein secreted by the epididymis, cysteine-rich secretory protein 1 (CRISP1), and its encoding mRNA are affected by vasectomy. In this study we showed that following vasectomy, the increased level of CRISP1 is not due to a secretory defect but to its accumulation in the intraluminal compartment of the cauda epididymidis. Western blot analyses were performed to determine the amount of CRISP1 associated with spermatozoa of men who had undergone surgical vasectomy reversal. Spermatozoa of vasovasostomized men are characterized by a significant increase (P < .05) in CRISP1 levels when compared with normal donors. There was no linear correlation between CRISP1 levels and the period of time elapsed between vasectomy and vasovasostomy. CRISP1 was also present in seminal plasma of normal and vasovasostomized men, but not in vasectomized individuals. The soluble concentration of CRISP1 was significantly higher (P < .05) in seminal plasma of vasovasostomized men when compared with normal men. Knowing that one of the proposed functions of CRISP1 is to modulate sperm capacitation, we evaluated the level of tyrosine protein phosphorylation of 2 AXAP proteins of the fibrous sheath, p81 and p105. Spermatozoa of vasovasostomized men were characterized by a 50% increase of protein tyrosine phosphorylation when compared to spermatozoa of normal men (P < .05). Our results are discussed with regard to the fertilizing ability of ejaculated spermatozoa of some vasovasostomized men.
In the United States alone, more than 525 000 vasectomies are performed yearly on men aged between 29 and 45 years of age, an incidence rate remaining stable at 10 per 1000 men (Barone et al, 2004, 2006). Worldwide, more than 100 million men rely on vasectomy for contraception purposes (Peterson and Curtis, 2005). Owing to changes in their personal life, an increasing number of men undergo surgical vasectomy reversal (vasovasostomy). In the United States alone, this surgical procedure is performed on 250 000–300 000 men yearly (Wood et al, 2003). Overall, vasectomized men constituted more than 15% of referrals to infertility clinics (Wood et al, 2003). Even though recently challenged (Silber and Grotjan, 2004), it is generally accepted that pregnancy rate following vasovasostomy is much lower than the surgical success of reanastomosis (Shannon, 1994; Kolettis et al, 2003; Busato, 2009). Many factors have been suggested to be responsible for the difficulty of fertility recovery following vas deferens reanastomosis (Nieschlag et al, 1997). Pregnancy rate decreases with increasing time elapsed between vasectomy and vasovasostomy (Abdelmassih et al, 2002), suggesting that sequelae to the male reproductive tract occurring during vasectomy may affect fertility recovery.
Leaving the testis, spermatozoa have to transit along the epididymis to acquire their full fertilizing potential (Cornwall, 2009). During this maturational process, the male gamete interacts with ions, lipids, and proteins within the intraluminal compartment. The transcriptome and the proteome of the epididymis show great variability along the epididymis, resulting in secreted protein–sperm interactions occurring in a sequential manner along the excurrent duct (Johnston et al, 2005; Dacheux et al, 2006; Dube et al, 2007; Thimon et al, 2007).
Consequences of vasectomy on the epididymis have been studied with the use of different animal models. Effects of vasectomy on the epididymal histology vary from one species to the other, calling into question the relevance of animal models for vasectomy research (Bedford, 1976; McDonald, 2000). In animal models, vasectomy affects the epididymis in many ways. In contrast, effects of vasectomy on epididymal protein secretion, and consequently on sperm maturation, have been the object of few reports (Turner et al, 1999, 2000; Doiron et al, 2003; Saez et al, 2004; Lavers et al, 2006). In human epididymis, we showed that vasectomy affects the synthesis of specific proteins: P34H and its transcript are delocalized along the epididymis (Legare et al, 2001), and NPC2 (HE1) expression is down-regulated, whereas the expression of other epididymal-specific proteins, such as HE2 and HE5, is not affected (Legare et al, 2004). More globally, using microarray analysis, we showed that the human epididymal transcriptome is affected by vasectomy. One of the transcripts most affected by vasectomy is CRISP1, which is up-regulated in the corpus epididymidis under vasectomy (Thimon et al, 2008a). Knowing that cysteine-rich secretory protein 1 (CRISP1) is one of the major epididymal secreted proteins and that it plays important functions in sperm physiology (Roberts et al, 2006; Cohen et al, 2007), we further document effects of vasectomy on CRISP1 expression and the possible consequences of this deregulation on ejaculated spermatozoa following surgical vasectomy reversal. Our results suggest that sequelae to the human epididymis occurring during vasectomy may not be reversible following surgical reversal, affecting sperm physiology and probably fertilizing ability.
Materials and Methods
These studies were conducted with the approval of the ethical committee for research on human subjects of our institution.
Human epididymides were obtained from our local organ transplantation program as previously described (Legare et al, 1999). After obtaining family consent, testicles were removed while artificial circulation was maintained to preserve tissues assigned for transplantation. Donors were 26–50 years of age with no medical pathologies that could affect reproductive function, except vasectomy. Organs were brought to the lab on ice and processed within 3 hours after orchidectomy. Vasectomy was determined while dissecting the scrotal segment of the vas deferens. Epididymal tissues were dissected in caput, corpus, and cauda and minced in small tissue pieces. Tissues were immediately snap frozen in liquid nitrogen for subsequent RNA extraction or protein extract preparation. Pieces of tissue were fixed by immersion in 4% paraformaldehyde in phosphate-buffered saline (PBS), pH 7.4, dehydrated, and embedded in paraffin for immunohistological localization of CRISP1.
Other tissues of the reproductive tract were used for Western blot determination of CRISP1 expression. Testicular tissues were obtained when epididymis were dissected from organ donors. Segments of the vasa deferentia were obtained from healthy patients of proven fertility who were undergoing vasectomy. Prostate and seminal vesicle tissues were obtained from patients undergoing prostatectomy by laparoscopy under general anesthesia.
Semen samples were obtained by masturbation from healthy control donors or from vasovasostomized men undergoing postsurgical spermogram follow-up performed at the clinical andrology laboratory of our institution. The controls and vasovasostomized subjects were of the same reproductive age group. The time period elapsed between vasectomy and vasovasostomy varied from 3 to 9 years. In order to correlate the amount of CRISP1 associated with ejaculated spermatozoa with potential deregulation in epididymal expression of this secreted protein, semen samples from vasovasostomized men who didn't meet the World Health Organization's reference values of semen analysis (1999), including viability and motility, were not included in this study. After liquefaction, spermatozoa were pelleted and washed twice by centrifugation at 800 ×g in Dulbecco PBS. Seminal plasma and sperm pellets were frozen at −80°C until used.
In Situ Hybridization
In situ hybridization and tissue sections preparations were performed as previously described (Legare et al, 2004). Briefly, CRISP1 complementary DNA was generated by RT-PCR using polyA RNA from normal human epididymis. The oligonucleotides used as primers were 5′-GAA-GCC-TGC-CCA-AGT-AAC-TG-3′ and 5′-GGG-AGT-TAA-GGT-CTC-CAG-CA-3′ and the PCR product was subcloned into pGEM-T (Promega, Madison, Wisconsin). The plasmid was digested and the mRNA was transcribed using SP6 and T7 RNA polymerase (Roche, Laval, Canada) in the presence of digoxigenin-11-uridin-triphosphate (DIG)-UTP.
Epididymis cryosections were fixed with 4% paraformaldehyde for 5 minutes, incubated for 10 minutes in 95% ethanol/5% acetic acid at −20°C, and rehydrated. Target mRNAs were unmasked by enzymatic digestion with 10 μg/mL proteinase K in PBS for 10 minutes at 37°C. Sections were incubated in 0.2% glycine, postfixed for 5 minutes with 4% paraformaldehyde, acetylated with 0.25% acetic anhydride, 0.1 M triethanolamine, pH 8, for 10 minutes, and washed in PBS.
Tissue sections were prehybridized for 2 hours at 42°C with 250 μg/mL of salmon sperm DNA preheated in a hybridization buffer and incubated overnight at 42°C under coverslips with 25 μL of 5 μg/mL heat-denatured antisense of sense cRNA probed with DIG.
Nonspecific staining was blocked by incubation for 1 hour with 5% heat-inactivated sheep serum in Tris buffer. Hybridization reactions were detected with an alkaline phosphatase–conjugated anti-DIG antibody diluted 1:1000 in a blocking buffer. The hybridization signal was visualized after incubation with the phosphatase substrate, nitroblue tetrazolium chloride, and 5-bromo-4-chloro-3-indolylphosphate p-toluidine salt (GIBCO-BRL, Gaithersburg, Missouri). Epididymis sections from normal and vasectomized men were processed in parallel to allow comparison.
Sperm pellets (107) were resuspended in sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer (Laemmli, 1970) prepared without reducing reagent. After 10 minutes, spermatozoa were pelleted by centrifugation and the protein supernatant recovered and reduced by addition of 2% β-mercaptoethanol followed by 5 minutes of heating in boiling water bath.
Seminal plasma samples were centrifuged twice at 3000 ×g for 10 minutes to eliminate spermatozoa and other cellular constituents. Supernatants were precipitated with MEOH/CHCl3 and proteins were resuspended in sample buffer and submitted to SDS-PAGE (Laemmli, 1970). Protein concentrations were determined by amido black staining of dot blots (Chapdelaine et al, 2001).
Tissues were homogenized with a Polytron (Inter-Sciences, Markham, Canada) in a homogenization buffer (0.01 M Tris, 0.1 mM EDTA, 1% sodium dodecyl sulfate (SDS), and 1 mM phenylmethylsulfonyl fluoride). Homogenates were then centrifuged at 3000 ×g for 15 minutes at 4°C, supernatants were precipitated with MEOH/CHCl3, and proteins were resuspended in sample buffer and submitted to SDS-PAGE (Laemmli, 1970). Protein concentrations were determined by amido black staining of dot blots (Chapdelaine et al, 2001).
Goat polyclonal antibody against CRISP1 was purchased from Santa Cruz Biotechnologies Inc (Santa Cruz, California) and used at 1 μg/mL for Western blot analysis, 5 μg/mL for immunohistochemistry on epididymis, and 15 μg/mL on sperm smears. Normal goat serum was used as negative control in CRISP1 immunodetection protocols. Rabbit polyclonal antibody against prostatic acidic phosphatase (PAP) purified from human seminal plasma was used at 1/1000 (vol/vol). This antiserum was a generous gift of Dr R. R. Tremblay (Université Laval, Canada) and was used as a marker of prostatic protein secretions. Mouse monoclonal antibodies against α-tubulin (TUBA; Sigma, Oakville, Canada), β-actin (Sigma), and phosphotyrosine (clone 4G10; Upstate Biotechnology) were used at (vol/vol) dilutions of 1/50 000, 1/20 000, and 1/10 000, respectively. Rabbit anti-goat immunoglobulin G (IgG), goat anti-rabbit, and goat anti-mouse conjugated to horseradish peroxidase were purchased from BIO/CAN Scientific (Missisauga, Canada) and used at 1/5000 (vol/vol). Biotinylated rabbit-anti-goat secondary antibody was obtained from Dako Diagnostics (Mississauga, Canada) and used at 1/400 (vol/vol).
Paraffin sections were prepared from fixed epididymal tissues. Ejaculated spermatozoa from normal and vasovasostomized men were washed, smeared on microscopic slides, fixed with cold ethanol, and processed for CRISP1 localization. Endogenous peroxidase activity was quenched with 3% H2O2 (vol/vol) in PBS for 30 minutes. Nonspecific binding sites were then blocked with 10% goat serum in PBS for 1 hour. Tissue sections and sperm smears were incubated overnight at 4°C with anti-CRISP1-specific antibodies diluted in PBS. Control sections were processed in parallel by replacing the primary antibodies by goat nonspecific IgGs. Sections were subsequently incubated with biotinylated rabbit anti-goat antibody for 30 minutes followed by an incubation with ABC reagent for 30 minutes. Immunostaining was revealed using 3-amino-9-ethylcarbazole. Harris hematoxylin-counterstained tissue sections and sperm smears were mounted under a coverslip with an aqueous mounting medium (Sigma). Slides were observed under a Zeiss Axioskop2 Plus microscope (Toronto, Canada) linked to a digital camera from Diagnostics Instruments (Sterling Heights, Michigan). Images were captured with the spot software (Diagnostics Instruments).
Proteins were separated on a 12% SDS-polyacrylamide gel and transferred to nitrocellulose membranes (Towbin et al, 1979). After saturation with 5% milk in PBS-0.1% Tween-20, membranes were incubated with anti-CRISP1, anti-PAP, antitubulin or antiphosphotyrosine antisera. Rabbit anti-goat, goat anti-rabbit or goat anti-mouse IgG conjugated to horseradish peroxidase were used for chemiluminescent detection of proteins (ECL reagent; Amersham, Baie d'Urfée, Canada). CRISP1, PAP, TUBA, and phosphotyrosine-containing protein signals were quantitated by densitometry in the linear range of film exposure and expressed as arbitrary units.
Statistical analyses were performed by analysis of variance using Super ANOVA software (ABACUS Concepts, Berkeley, California). Results were compared by Student-Newman-Keuls test. Differences were considered significant at P < .05.
Cellular Localization of CRISP1 Along the Epididymis From Normal and Vasectomized Men
We previously showed that CRISP1 is up-regulated in the corpus segment of the epididymis after vasectomy, resulting in an accumulation of CRISP1 in the cauda segment (Thimon et al, 2008a). Herein, immunohistochemistry confirmed that CRISP1 was undetectable in the caput epididymidis, and an increasing amount of this protein was revealed in the corpus and cauda segments of normal tissues. Our results showed that CRISP1 was accumulating in the intraluminal compartment of the cauda epididymidis of vasectomized men. Very small amounts of CRISP1 were immunodetectable in the epididymal epithelium, suggesting that vasectomy did not affect the synthesis, secretion, or reabsorption of this secreted protein. In both normal and vasectomized tissues, CRISP1 was undetectable in the interstitial tissues of the epididymis (Figure 1). In situ hybridization localization of CRISP1 mRNA was in agreement with the immunohistological localization of the translational product as well as with our previous observations (Thimon et al, 2008a). In both normal and vasectomized tissues, CRISP1 mRNA was expressed by the epithelium in the corpus and cauda segments and was undetectable in the caput (Figure 2).
Effect of Vasectomy on CRISP1 Associated With Spermatozoa and in Seminal Plasma
CRISP1 was associated with the acrosomal region of ejaculated spermatozoa from both normal and vasovasostomized men. The immune detection signal varied significantly from one sperm cell to the other in both normal and vasovasostomy samples (Figure 3). The goat IgG did not stain the acrosomal region, but the equatorial segment was nonspecifically stained for unknown reasons.
Western blot analyses were performed on protein extracted from a constant number of spermatozoa from different men. As expected, CRISP1 was detected as a single band of ∼32 kDa. The commercially available antibody was specific to CRISP1, as shown by the absence of immunodetectable band revealed when Western blots were probed with a control serum (data not shown). When expressed as a ratio of a constitutive protein TUBA, the quantity of CRISP1 associated with a constant number of ejaculated spermatozoa was significantly higher in vasovasostomized men when compared with normal donors (Figure 4) with P < .05. There was no correlation between the amount of CRISP1 associated with a constant number of spermatozoa of vasovasostomized men and the time elapsed between vasectomy procedure and the vasovasostomy, at least in the time interval of 3–9 years of vasectomy (Figure 5). The percentage of motile spermatozoa in the semen samples was not correlated with the amount of CRISP1 associated with washed ejaculated spermatozoa (data not shown).
CRISP1 was also found as a soluble form in the seminal plasma. When Western blots were performed on 20 μg of seminal plasma proteins, CRISP1 was detectable in semen samples of both normal and vasovasostomized men. As found for CRISP1 associated with spermatozoa, CRISP1 was significantly (P < .05) more concentrated in seminal plasma of vasovasostomized men when compared with samples from normal men. Interestingly, CRISP1 was undetectable in seminal plasma of vasectomized men (Figure 6). This correlated with the fact that CRISP1 was undetectable on Western blots of 20 μg of protein extracted from human seminal vesicles and prostate tissues (Figure 7).
Effect of Vasectomy on Human Sperm Protein Phosphotyrosine Content
One of the proposed functions of CRISP1 is its involvement in the control membrane signaling events leading to sperm protein tyrosine phosphorylation (Gibbs et al, 2008). Furthermore, protein tyrosine phosphorylation levels were significantly lower when spermatozoa of crisp1−/− knockout mice were incubated in capacitation medium (Da Ros et al, 2008). Using an anti-phosphotyrosine antiserum, we thus compared the basal protein tyrosine phosphorylation level in a constant number of ejaculated spermatozoa from vasovasostomized and normal men that are characterized by high and low levels of CRISP1 respectively. More precisely, tyrosine phosphorylation of 2 AXAP proteins, p81 and p105, was quantitated by densitometry. In washed ejaculated spermatozoa, the level of tyrosine phosphorylation of p81 and p105 was significantly increased (P < .05) in spermatozoa of vasovasostomized men when compared to ejaculated spermatozoa from normal control individuals (Figure 8). The level of phosphorylation of p81 and p105 increased in parallel in sperm of both normal and vasovasostomized men during a 4-hour incubation period in a capacitating medium. Thus, the difference in the level of tyrosine phosphorylation between the 2 populations of spermatozoa at time zero remained the same during 4 hours of capacitation (data not shown).
Vasectomy has major consequences on epididymis histology in men. The thickness of the epithelium is modified all along the epididymis and the area of the intraluminal compartment is greatly enlarged, especially in the distal segments (McDonald, 1996, 2000; Legare et al, 2001). These modifications can have consequences on the function of the excurrent duct and its protein secretion activity. Studies using laboratory rodents suggest that overall protein synthesis is decreased in the epididymis under vasectomy, at least in the caput segment (Turner et al, 1999). Obviously, these studies cannot be extrapolated to human. During vasectomy, the human epididymal transcriptome is greatly affected all along the epididymis, suggesting that protein composition of the intraluminal compartment is modified when compared with fertile men (Thimon et al, 2008b). If not reversible following surgical vasectomy reversal, these modifications can have consequences on sperm maturation occurring as a result of epididymal protein secretion (Cuasnicu et al, 1984a,b).
CRISP1 expression is greatly affected by vasectomy, being up-regulated in the corpus segment (Thimon et al, 2008a). Vas deferens obstruction in the rat results in a decrease of the overall protein synthesis in the caput epididymidis, CRISP1 being the most affected protein (Turner et al, 1999). In this species, CRISP1 remains low after vasovasostomy (Turner et al, 2000). In rodents, crisp1 is expressed in the caput epididymis whereas the human CRISP1 transcript is detectable in the corpus and cauda segments (Thimon et al, 2007). In the human, the corpus epididymidis is the more active segment in protein synthesis, whereas in the rodent, the initial segment and the caput are known to be the more active in translational activity (Cornwall et al, 2002). This discrepancy of CRISP1 expression pattern between species may be the reason why the expression of this gene is differently affected by vasectomy when comparing humans and rodents. Interestingly, CRISP1 is up-regulated by more than 10-fold in the caput of nonobstructive azoospermic men when compared with normal individuals (Dube et al, 2008). Even though these authors have not investigated the effect of azoospermia on CRISP1 expression all along the epididymis, it appears that nonsecretory azoospermia and vasectomy affect the epididymal transcriptome in different ways. Regulation of epididymal gene expression thus appears to be complex and under the control of different factors.
In semen, CRISP1 is found in the soluble fraction and in association with spermatozoa. Both forms are in higher concentrations in vasovasostomized men with successful reanastomosis of the vas deferens. CRISP1 remains localized to the acrosomal region of spermatozoa in both normal and vasovasostomized men. Thus, higher expression of CRISP1 in vasovasostomized men does not affect the interaction of this secreted protein with spermatozoa. As described in all studied mammalian species (Gibbs et al, 2008), human CRISP1 gene has been reported to be expressed only in the epididymis (Hayashi et al, 1996; Kratzschmar et al, 1996). Some authors, however, have shown expression of human CRISP1 gene also in the vas deferens and seminal vesicles (Nolan et al, 2006). Our results show that human CRISP1 is not detectable in the soluble fraction of semen of vasectomized men, suggesting that its secretion is in fact restricted to the epididymis. The presence of CRISP1 can thus be considered as a good marker of excurrent duct permeability and can be used in the clinic to discriminate between secretory and obstructive azoospermia. Although many functions have been attributed to sperm-bound CRISP1, the role of soluble CRISP1 in semen remains to be established.
Studies in rat and mouse showed that CRISP1 has decapacitation properties and that it mediates both sperm–zona pellucida binding and sperm–egg plasma membrane interactions (Roberts et al, 2007; Cohen et al, 2008). CRISP1 knockout mice have recently been generated, showing that male fertility was not affected. When incubated for 60 to 90 minutes in capacitated medium, protein tyrosine phosphorylation was greatly decreased in CRISP1−/− sperm when compared to CRISP1+/+ and CRISP1+/− male gametes (Da Ros et al, 2008). This observation is in contradiction with the proposed role of CRISP1 as a decapacitation factor. It is interesting to note that spermatozoa of vasovasostomized men characterized by a higher level of CRISP1 show a higher level of protein tyrosine phosphorylation before capacitation. The function of CRISP1 in the cascade events leading to capacitation remains puzzling. Whether or not the increase of sperm-associated CRISP1 in some vasovasostomized men affects their ability to interact with the egg's zona pellucida and/or plasma membrane remains unknown.
CRISP1 is not the only gene affected by vasectomy in men (Legare et al, 2001, 2004). We showed that vasectomy affects the expression of 2 other epididymal secreted proteins, NPC2/HE1 and P34H, affecting biochemical parameters of ejaculated spermatozoa in 40% to 50% of vasovasostomized men investigated (Legare et al, 2006). Relocalization of P34H expression along the epididymis jeopardizes the acquisition of P34H by maturing spermatozoa, resulting in the inability of ejaculated spermatozoa from vasovasostomized men to bind to the zona pellucida (Boue and Sullivan, 1996; Guillemette et al, 1999). NPC2/HE1 is another epididymal protein that is down-regulated under vasectomy. NPC2/HE1, having a cholesterol-binding pocket (Okamura et al, 1999; Ko et al, 2003; Vanier and Millat, 2004; Legare et al, 2006), has been hypothesized to be involved in cholesterol efflux during sperm maturation (Okamura et al, 1999; Legare et al, 2006). Down-regulation of NPC2/HE1 thus results in an increase of sperm membrane cholesterol, affecting the ability to properly undergo capacitation and the acrosome reaction in some vasovasostomized men (Legare et al, 2006). As for NPC2 and P34H, an abnormal quantity of CRISP1 associated with ejaculated spermatozoa is detectable in about 50% of non-azoospermic vasovasostomized men. Sperm abnormalities associated with P34H, NPC2/HE1, and CRISP1 expression are not found in the same vasovasostomized individuals (data not shown). It thus seems that sequelae to the epididymis occurring under vasectomy vary from one man to the other or that the recovery from these abnormalities occurs differently in a population of vasovasostomized men. The time during vasectomy or the time elapsed between sperm analysis and surgical vasectomy reversal does not seem to influence these recovery processes.
In conclusion, deregulation of CRISP1 expression is an example of sequelae to the epididymis under vasectomy that can have consequences on sperm physiology in vasovasostomized men. This can explain why fertility recovery is jeopardized in some vasovasostomized men for whom vas deferens recanalization is surgically successful. Furthermore, soluble CRISP1 in seminal plasma can be used as a marker of excurrent duct permeability in men.
The collaboration of the Québec Transplant nurses and coordinators as well as associated surgeons is acknowledged. The authors also acknowledge the technicians of the andrology laboratory of our institution.
Supported by Canadian Institutes for Health Research grant MOP-89863 (R.S.).