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.
What's known on the subject? and What does the study add?
Caveolae are specialised regions of bladder smooth muscle (BSM) cell membranes where specific signalling pathways are regulated. Caveolin proteins are involved in caveolar biogenesis and function as signal transduction regulators. Expression of caveolin-1, -2, and -3 has been previously identified in the bladder; however, the distribution and relative expression of these proteins have not been defined.
The present data show significant differences in the spatial distribution of caveolin proteins throughout the bladder wall. Region dependent variations in the co-localisation of caveolin subtypes in detrusor SM were also detected. These findings support the premise that the unique spatial pattern of caveolin proteins associated with BSM cells may enable regionally distinct functional responses to common stimuli.
• To determine the regional expression profile of caveolin isoforms (integral membrane proteins abundant in caveolae), the spatial relationships among caveolin proteins within specific smooth muscle (SM) regions and the extent of their molecular interactions in bladder SM (BSM).
MATERIALS AND METHODS
• Regional differences in the expression of caveolin family members were determined by quantitative reverse transcriptase-polymerase chain reaction and Western blot of RNA and protein extracted from the base, body and dome of rat bladders.
• To evaluate the distribution of caveolin-1 (Cav-1), Cav-2 and Cav-3 within the bladder, longitudinal tissue sections from the base to dome were processed for confocal microscopy and quantified for intensity of immunoreactivity (IR) and extent of co-localisation.
• Interactions among Cav-1, Cav-2 and Cav-3 were determined by co-immunoprecipitation.
• Differential expression of Cav-1 and Cav-3 was detected among bladder regions, with lowest expression in the bladder base relative to the dome.
• Cav-1 was highly expressed in all regions, although an increase in IR from submucosa to serosa was detected in each region.
• The distribution of Cav-2 IR generally paralleled Cav-1, but progressively decreased from submucosa to serosa in each region.
• Cav-3 expression predominated in the medial region of BSM increasing progressively from base to dome, but was poorly expressed in the outer SM layer particularly in the dome.
• Cav-1 co-precipitated extensively with both Cav-2 and Cav-3. Co-precipitation between Cav-3 and Cav-2 was also detected.
• The isoform-specific spatial distribution and distinct molecular interactions among caveolins in BSM may contribute to the contractile heterogeneity of BSM cells and facilitate differential modulation of responses to local stimuli.
• As BSM caveolae regulate key signalling processes involved in contraction, altered expression of caveolin proteins may generate a regional imbalance in contraction/relaxation responses, thus leading to bladder dysfunction.
The plasma membrane of bladder smooth muscle (BSM) is decorated with abundant caveolae [1,2], small flask-shaped invaginations of the sarcolemma arranged in longitudinal arrays between intervening dense bands. Although relatively unexplored in the bladder, caveolae have been shown to subserve a multitude of essential cellular processes, as evidenced by the growing list of pathological conditions and diseases that have been attributed to the loss of these organelles . In other tissues, these specialised structures are considered to be critical players in the organisation, integration and modulation of various signal transduction processes, given that many receptors, signalling intermediates and ion channels involved in cell signalling are enriched in these microdomains . Consequently, caveolae create a favourable microenvironment for modulating cell-dependent signal transduction in which relevant participating molecules are segregated or sequestered to facilitate or dampen downstream signalling events  .
Caveolae are stabilised by caveolins, integral membrane proteins that serve as scaffolds for interacting proteins and function as signal transduction regulators. The caveolin protein family consists of three caveolin subtypes that exhibit a cell and organ-specific expression pattern . Caveolin-1 (Cav-1) and Cav-2 are similarly distributed and abundantly expressed in endothelial cells, adipocytes and fibroblasts, while Cav-3 expression is limited to skeletal muscle and some cardiac muscle and SM. Cav-1 and Cav-3, but not Cav-2 can drive the formation of caveolae. However, Cav-2 may facilitate caveolin-1-mediated biogenesis of caveolae or influence caveolar morphology . Thus the specific combination of caveolin isoforms that are expressed in particular cell types may contribute to the complexity and diversity of caveolae-dependent signalling.
The heterogeneity of caveolin subtype expression is particularly evident in muscle cells. In vascular SM, Cav-1 and Cav-3 are co-expressed in arteries and arterioles; in contrast, the presence of caveolin-1 in SM of veins is not coincident with Cav-3 expression . Moreover, Cav-1 is absent from skeletal muscle in which Cav-3 is abundantly expressed. Airway SM expresses only Cav-1 and Cav-2 . Cav-1 and Cav-3, but not Cav-2, are expressed in atrial myocytes, while Cav-3 and Cav-2, but not Cav-1 are co-expressed in ventricular myocytes . In the jejunum, Cav-1 and Cav-2 are expressed in all SM layers, but Cav-3 was detected only in the longitudinal and outer circular SM layer, not in the inner circular layer . Uterine longitudinal SM expresses each caveolin isoform . These findings collectively show the diversity of caveolin expression that is not only cell-type dependent, but particularly for SM cells, is also organ and region specific.
Previous studies have shown that all three caveolins [13,14], are expressed in BSM. However, the spatial distribution and relative expression of these proteins are unclear. Given the importance of caveolins in regulating cell signalling, an adequate characterisation of their expression patterns and tissue localization in the bladder is essential for isolating their site of action and determining their contribution to regional specificity in BSM function. Therefore, the purpose of the present study was to establish the level of expression of these caveolae-related proteins, determine their tissue distribution and identify the extent of their co-localisation in BSM.
MATERIALS AND METHODS
Male adult (aged 8–10 weeks) Sprague–Dawley rats were housed with free access to water and food. Euthanasia was induced by CO2 asphyxiation according to an approved Institutional Animal Care and Use Committee protocol.
After removing the mucosa using a stereomicroscope, bladders were cut into thirds corresponding to regions of the base, body and dome. The base consisted of tissue between the bladder neck and the level of the ureteric orifices. Tissue above the ureteric orifices was cut in half transversely and then both margins were trimmed by 2 mm to essentially isolate the region of the body from the dome. Bladder samples were placed in RNA-later stabilisation reagent and RNA was extracted by Total RNA Mini Kit (Qiagen). Isolated RNA was used for two step real-time RT-PCR. Complementary DNA (cDNA) was generated using a high capacity cDNA Archive Kit (Applied Biosystems). Quantitative real-time RT-PCR was performed using Taq-Man Universal Master Mix, Taq-Man probes with forward and reverse primers for Cav-1, Cav-2 or Cav-3 (1.25 µL) and 50 ng RNA in a total volume of 25 µL in optical tubes (ABI Prism 7700 laser sequence detector, Applied Biosystems). The PCR cycle number that generated a fluorescence signal above threshold (threshold cycle, CT) was determined. For each sample, we used 18s rRNA as an internal control for template load and to correct for amplification efficiency. Gene expression levels from samples corresponding to bladder body and dome were compared to expression in the bladder base. Results were expressed in folds of gene expression using relative quantitation of ΔΔCT.
CONFOCAL LASER-SCANNING MICROSCOPY
After euthanasia, urinary bladders were quickly removed from adult male rats through a lower abdominal incision. Bladders were opened on the anterior aspect, and longitudinal segments of tissue that extended from the base to the dome were isolated between the ureteric orifices. Bladder tissue was then placed in OCT, quickly frozen and stored at −80 °C until sectioning. Full-thickness sagittal sections were cut on a microtome (8–10 µm) to include regions from the base to the dome and from the mucosa to serosa. Sections were fixed with cold acetone (10 min.), washed in PBS, and then blocked with 5% BSA in PBS with 0.03% triton X-100 for 1 h. Sections were incubated overnight with primary antibody (rabbit anti-Cav-1, goat anti-Cav-3, Santa Cruz; mouse anti-Cav-2, BD Bioscience). After extensive washing in PBS, sections were incubated with AlexaFluor 488, AlexaFluor 594 or AlexaFluor 647-conjugated secondary antibodies for 2 h at room temperature. Bladder sections were mounted using Fluoromount-G (Southern Bioteck). Specificity of immunoreactivity was confirmed by the absence of fluorescence in sections in which primary antibody was either omitted or pre-incubated with excess blocking peptide. Tissue sections from different rats were placed on the same slide to ensure processing under identical conditions and reduce staining variation.
Tissue sections were examined using a confocal laser-scanning microscope (Zeiss LSM 710) equipped with 40X-water immersion and 63X-oil immersion objectives (1.4 Numerical Aperture; NA). Optical slices were acquired with pinhole diameter set to 1 Airy Unit and averaged over two scans at a pixel size of 0.085 µm. When more than one fluorophore was used, sections were scanned sequentially at each excitation wavelength to minimise emission crosstalk. Laser gain and power settings were adjusted to optimise the dynamic range of the detector, without generating signal in control sections. All settings remained constant for each sample.
From full-length sagittal cryosections consisting of the dome, body and base of the bladder, images were acquired from muscle bundles located within each third of the muscularis propria in the transverse direction (from submucosa to serosa) designated as inner, medial and outer layers. At least two images were obtained from each of these layers, without reference to bundle orientation. From each image, 10 regions of interest were identified and the mean intensity of fluorescence was determined after thresholding each channel to remove background. This analysis was repeated in three separate bladders processed under identical conditions. For double-labelled samples, the overlapcoefficient (,where S1 = channel 1 signal intensity and S2 = channel 2 signal intensity)  was determined in high resolution images. This coefficient represents the ratio of colocalising pixels to total pixels above threshold. Quantitative analysis of images was accomplished using Zen software (Carl Zeiss Microimaging).
WESTERN BLOT ANALYSIS
After removing the mucosa, bladders were divided into three pieces, representing the base, body and dome of the organ. To compare the expression of caveolin isoforms in each bladder region, tissue lysates were obtained using RIPA lysis buffer (Santa Cruz) supplemented with phenylmethylsulfonyl fluoride (2 mm), sodium orthovanadate (1 mm) and a protease inhibitor cocktail [E-64 (28 µm), aprotinin (1.6 µm), bestatin (80 µm), pepstatin (30 µm), AEBSF (2 mm) and leupeptin (40 µm)]. Protein concentrations were determined using the bicinchoninic acid (BCA) protein assay by measuring the absorbance at 280 nm with a biophotometer (Eppendorf). Equal amounts of protein lysates (8 µg) were loaded onto a 4–12% SDS polyacrylamide gel and separated by electrophoresis. Proteins were then transferred to a 0.2-µm pore nitrocellulose membrane (Invitrogen). Non-specific binding was inhibited by incubating membranes in Tris-buffered saline (TBS) with 5% dry milk for 1 h. Membranes were incubated overnight at 4 °C with each specific primary antibody. β-actin (Santa Cruz) was used as an internal control. Unbound antibody was removed by extensive washing with TBS containing Tween 20 (0.05%). Membranes were incubated for 1 h at room temperature with horseradish peroxidase-conjugated secondary antibody (Santa Cruz). After washing, the membrane was incubated with a chemiluminescence substrate (Western Lightning plus-ECL, Perkin Elmer), and immunoreactive bands were visualised by exposure of membrane to audioradiographic film (Kodak Biomax). The absolute intensity of each protein band determined by densitometry was normalised by the intensity of the internal control β-actin and differences in relative intensity in each region were determined.
IMMUNOPRECIPITATION (IP) OF CAVEOLIN ISOFORMS
The molecular interaction between caveolin isoforms was determined by IP experiments. The mucosa was removed from each bladder by microdissection. The remaining SM layer was lysed in IP extraction buffer (Dynabeads Co-immnoprecipitation kit, Invitrogen) supplemented with NaCl (100 mm), MgCl2 (2 mm), DTT (1 mm), and protease inhibitors [E-64 (28 µm), aprotinin (1.6 µm), bestatin (80 µm), pepstatin (30 µm), AEBSF (2 mm) and leupeptin (40 µm)]. In parallel, specific primary antibodies directed against Cav-1, Cav-2 and Cav-3, as well as equal concentrations of mouse, goat or rabbit isotype control IgG, were covalently immobilised onto the surface of M-270 epoxy magnetic beads (1.5 mg, Dynabeads® Antibody Coupling Kit, Invitrogen) according to the manufacturer's protocol. The antibody-coupled beads were added to diluted bladder lysates (1 mg/mL) and incubated under agitation for 1 h at 4 °C. After incubation, the beads with immobilised proteins were collected by placing the tubes in a magnetic field (DynaMagTM-2, Invitrogen). IP complexes were eluted in buffer (Invitrogen) and separated from the antibody-coupled beads by magnet. Finally, purified protein complexes were re-suspended in sample loading buffer, separated by SDS-PAGE, and transferred to nitrocellulose membrane. Interaction among caveolin isoforms was evaluated by immunoblotting Cav-1, Cav-2 and Cav-3 IPs with each caveolin antibody. Immunoreactive bands were detected by incubating with chemiluminescence substrate followed by exposure to film.
For each caveolin isoform, Western blotting or fold change in gene expression was calculated from each region of the bladder, and differences between base, body and dome were analysed using a one-way anova followed by Student-Newman-Keuls test or Tukey test for data that was not normally distributed. For image analysis, the mean intensity of immunoreactivity, overlap coefficients and co-localisation coefficients were analysed by two-way anova of bladder regions and muscle layers. Data are presented as the mean (sem) with P < 0.05 considered to indicate statistical significance.
DISTRIBUTION OF CAVEOLIN GENE EXPRESSION
Gene expression for all the three caveolin isoforms was detected in rat bladder tissue by real-time RT-PCR. Relative to their respective mRNA levels in the bladder base, expression of both Cav-1 and Cav-3 was significantly up-regulated in the dome of the bladder (almost two- and four-fold increase, respectively). In contrast, Cav-2 gene expression appeared to be uniformly expressed in the base, body and dome of the bladder (Fig. 1A).
DISTRIBUTION OF CAVEOLIN PROTEIN EXPRESSION
All caveolin proteins were highly expressed in BSM tissue. Antibody against Cav-1 detected two distinct bands corresponding to Cav-1α and Cav-1β isoforms. Using a Cav-2 antibody that can detect three isoforms, bands corresponding to Cav-2α and Cav-2β were clearly identified, but Cav-2γ immunoreactivity was weak. The regional expression pattern largely concurred with the corresponding gene expression. Expression of caveolin proteins was highest in the body of the bladder (Fig. 1B,C). Compared with the base, the relative expression of Cav-1 and Cav-2 were not different in the dome, but expression of Cav-3 was greater in the dome than in the base.
LOCALISATION OF CAVEOLIN PROTEINS
Immunofluorescence analysis confirmed the protein expression of all caveolin isoforms in BSM. There was a distinct punctate distribution on the periphery of BSM cells, consistent with membrane labelling. Intramural blood vessels were intensely immunoreactive to caveolin proteins, particularly Cav-2. Immunoreactivity was most intense for Cav-1, which was uniformly distributed in SM cells throughout the bladder with greatest expression in the outer layer of the base and body and the medial layer of the dome (Fig. 2A). The mean intensity of immunoreactivity in the inner layer increased from base to dome. In the medial layer, Cav-1 immunoreactivity was significantly lower in the body than in the base or dome, while in the outer layer, the mean intensity was greatest in the base of the bladder. Cav-2 immunoreactivity in muscle bundles progressively decreased from base to dome, and from inner to outer SM regions (Fig. 2B). The intensity of immunoreactivity for Cav-3 appeared the weakest among the isoforms. Cav-3 was predominately localised in the medial muscle layer, increasing from base to dome, but was poorly expressed in the outer layers of the base, body and dome (Fig. 2C). In the inner layer, Cav-3 expression was significantly lower in the base than in the body or dome.
MOLECULAR INTERACTION OF CAVEOLIN ISOFORMS
In bladder lysate IP with Cav-1, immunoblots corresponding to Cav-2 and Cav-3 proteins were detected (Fig. 3). These physical interactions were confirmed by the reciprocal detection of intense Cav-1 immunoreactive bands in lysates IP with either Cav-2 or Cav-3. Cav-2 immunoreactivity was also detected in Cav-3 IP samples and immunoreactivity corresponding to Cav-3 was found in Cav-2 IP samples. No immunoreactivity for either Cav-1, Cav-2 or Cav-3 was detected in control lanes in which IPs were performed with non-specific rabbit, mouse or goat IgG.
DISTRIBUTION OF CO-LOCALISED CAVEOLIN ISOFORMS
In double-labelled immunofluorescence images, Cav-1 was co-localised with both Cav-2 and Cav-3. Quantitative analysis of high power images (63X) showed that co-localisation of Cav-1 and Cav-2 was similar among regions, except in the medial layers of the body and dome in which the overlap coefficients were significantly lower than the inner and outer layers (Fig. 4). Cav-1 was more co-localised with Cav-2 in the inner layers of all regions whereas Cav-2 co-localisation with Cav-1 was greater in the outer layer of all regions.
The overlap coefficient for Cav-1 and Cav-3 was lowest in the outer layer of all bladder regions and was significantly higher in the inner layer of the dome compared with the inner layer of the base (Fig. 5). Cav-3 immunoreactivity that was co-localised with Cav-1 was relatively evenly distributed among layers and regions. In contrast, Cav-1 was poorly co-localised with Cav-3 in the outer layers of all regions of the bladder, consistent with the weak immunoreactivity of Cav-3 in this layer as shown in Fig. 2C.
In all regions of the bladder, Cav-2 and Cav-3 co-localisation was greatest in the medial layer (Fig. 6), particularly in the dome. The proportion of Cav-3 that was co-localised with Cav-2 was significantly lower in the outer layers of all regions, consistent with the poor immunoreactivity for both Cav-2 and Cav-3 in the outer layers. The differential regional distribution as well as the isoform specific co-localisation patterns among caveolin proteins can be appreciated in merged images from triple labelling procedures (Fig. 7). In the outer SM layer of the bladder where Cav-3 immunoreactivity is relatively weak, Cav-1 and Cav-2 co-localisation predominates, whereas in the inner layer where Cav-2 immunoreactivity is relatively weak, co-localisation of Cav-1 and Cav-3 predominates.
As dynamic signal transduction microenvironments, caveolae regulate and facilitate signalling events by promoting the spatial segregation or sequestration of various signalling molecules . Caveolin proteins that stabilise caveolar morphology serve as essential signalling mediators by interacting with molecular components of specific signalling cascades and assembling protein complexes. Cav-1 and Cav-3 share the ability to induce caveolae biogenesis and interact with signalling intermediaries in cells in which the other isoform is absent. Therefore, the role of these proteins in cell signalling has generally been studied only in cell types in which either Cav-1 or Cav-3 predominate. We and others have previously shown that urinary BSM, unlike non-SM cells and certain other SM systems, expresses all three caveolin isoforms [13,14,17]. As the functional attributes of caveolae are probably determined by the complex relationship among the different combinations of caveolin proteins represented in caveolae, we comprehensively examined the caveolin isoform distribution and interaction to begin to unravel the role of caveolae and caveolins in modulating BSM cell contraction.
The present data show an isoform-specific and region-dependent distribution of caveolin gene and protein expression in BSM. Consistent with previous observations [14,18], widespread Cav-1 expression was detected throughout the bladder wall. Interestingly, regional differences in Cav-2 immunoreactivity did not parallel that of Cav-1, but showed graded attenuation of intensity in BSM across layers from submucosa to serosa. Although Cav-1 is generally considered the major regulatory caveolin subtype in SM, substantial Cav-3 expression was detected in BSM tissue. The evident region-dependent differences in Cav-3 immunoreactivity were distinct from the distribution patterns of Cav-1 and Cav-2, suggesting that caveolin isoform expression is independently regulated.
Although all three caveolin isoforms have been detected in most types of SM, Cav-3 appears to be far less expressed than Cav-1 or Cav-2 in arterial SM and is even absent in venous SM . These regional variations in Cav-3 expression in a particular cell type is consistent with previous studies in intestinal SM in which expression in outer circular and longitudinal muscles contrasted with the absence of Cav-3 in the inner circular layer . Similarly, in the ureter, Cav-3 is expressed in the outer circular muscle but not the inner longitudinal muscle , and in the stomach, Cav-3 is expressed in the inner circular but not outer longitudinal layer. Consistent with these location-specific findings, the present data indicated a heterogeneous Cav-3 distribution with minimal expression in the outer layer of all regions of BSM. This spatial diversity of caveolin expression has not been previously described in the bladder, but may determine the local functional properties of BSM cells.
A previous study has proposed that, according to Burnstock's functional classification of SM as multiunit, intermediate and unitary models (based on the extent of innervation and electrical coupling), Cav-3 expression broadly correlates with an intermediate type of SM . These authors also linked Cav-3 with preferential expression in highly contractile cells, particularly those oriented circumferentially. These classifications apply adequately for the expression of Cav-3 in the inner circular muscle of the stomach, pupillary sphincter muscle and certain resistance arteries . In rat bladder, SM bundles form a meshwork arrangement with variable orientation rather than assembling in distinct layers . Although orientation-specific contractions have not been measured in BSM bundles, differential contractile responses have been shown in bladder tissue incised from longitudinal and circular directions [21,22]. Thus, the present finding of restricted expression of Cav-3 to the inner SM layers of the bladder may provide a molecular basis for regional variability in SM contractility. The role of Cav-3 in defining SM phenotype may have important implications under disease conditions that cause bladder remodelling in which the orientation and function of BSM cells can be markedly altered. For example, previous studies have shown that spinal cord injury causes a shift from a relative predominance of longitudinal over circumferential BSM orientation to an equally bidirectional orientation .
The present IP results confirmed specific molecular interactions among all caveolin family members in rat bladder tissue, consistent with a previous report in mouse bladders . While Cav-1 interaction with Cav-2 as stable hetero-oligomeric complexes is well described in many cell types, the formation of Cav-3 complexes with Cav-1 or Cav-2 is muscle-specific. In a myoblast cell line that co-expresses Cav-1 and Cav-2 but not Cav-3, recombinant expression of Cav-3 allows Cav-1 and Cav-2 to form complexes with Cav-3; in contrast, overexpression of Cav-3 in fibroblasts with endogenous Cav-1 and Cav-2 does not establish interaction of Cav-3 with either Cav-1 or Cav-2 . Notably, in atrial myocytes and bronchial SM, cells that endogenously express Cav-1 and Cav-3, molecular interaction between these isoforms has recently been described [10,25]. We identified a similar reciprocal Cav-1–Cav-3 association in BSM, suggesting that caveolin family members can physically interact within the same protein complex. These distinct molecular interactions may provide a mechanism for regional specificity in caveolae-mediated signalling mechanisms that regulate bladder contraction.
We used quantitative co-localisation analysis to objectively show differences in the distribution of caveolin proteins in BSM and compared the extent of overlap among isoforms using confocal microscopy images. There was co-localisation among all caveolin family members to variable extents in a region-dependent manner. Although >45% of Cav-1 and Cav-3 immunoreactivity was co-localised, pronounced diversity among regions of the bladder and within SM bundles was apparent. Moreover, within the same cell, co-localisation of Cav-1 and Cav-3 coincided with independent localisation of these isoforms, consistent with a pattern previously described in stomach and bladder . However, unlike cardiac myocytes , in which Cav-1 and Cav-3 co-localise in the atrium but Cav-3 is expressed without Cav-1 in the ventricle, individual BSM cells expressing Cav-3 appeared to be consistently immunoreactive for Cav-1, whether spatially separated and/or co-localised. Despite poor Cav-3 expression in the outer muscle layer of all bladder regions, it co-localised with Cav-2, albeit weakly, in regions with evident immunoreactivity. Thus the present results suggest that two or more caveolin isoforms can be co-expressed in the same membrane microdomain in BSM and that the particular combination of isoforms present in BSM caveolae is spatially regulated.
Morphologically and biologically distinct SM cell populations within the arterial wall have been previously described . In the bladder, phenotypic heterogeneity contributes to regional differences in potassium channel activity and contractile properties of SM cells . As key regulators of cell signalling, the spatial distribution of caveolin isoforms as well as their isoform-specific association with membrane receptors and signalling proteins in caveolae probably contributes to a broad range of contractile responses among genetically similar cells . In the bladder, the unique relative co-expression profile of these isoforms and molecular interactions found here may underpin the functional diversity of BSM cells, allowing differential responses to local stimuli and impacting the regulation of detrusor contraction. Moreover, altered expression or interaction of these proteins with subsequent dissociation from caveolae-related signalling molecules may generate a regional imbalance in contraction/relaxation responses, and thus contribute to bladder dysfunction.
Supported by Medical Research service, Department of Veteran's Affairs, Washington, DC.