|
|
||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
APPETITE, OBESITY, DIGESTION, AND METABOLISM
Department of Biological Sciences, University of Alberta, Edmonton, Alberta, Canada
Submitted 8 November 2004 ; accepted in final form 1 March 2005
| ABSTRACT |
|---|
|
|
|---|
teleost fish; food intake; mRNA expression; bombesin
In mammals, it is well established that regulation of GH secretion from the anterior pituitary gland is under the reciprocal control of two hypothalamic hormones: a stimulatory growth hormone-releasing hormone (GHRH) found in the arcuate nucleus and an inhibitory somatostatin (SRIF) synthesized in the periventricular nucleus. Several lines of evidence suggest that, in addition to the intricate patterns of GHRH and SRIF release regulating GH directly at the pituitary level, SRIF modulates GH secretion indirectly through central regulation of GHRH-containing neurons (70). In recent years, ghrelin, a Ser3-octanoylated, 28-amino acid peptide that was recognized as the endogenous ligand of GH secretagogue receptor, has been included in the regulatory model of GH secretion (68, 77). Ghrelin stimulates GHRH outflow at the arcuate nucleus level and, through direct pituitary actions, serves as a functional antagonist of SRIF at the pituitary level (69). Ghrelin is mainly synthesized in the stomach and acts as an orexigen, thereby linking growth and energy balance (16, 38).
Growth in fish is regulated by the brain neuroendocrine-GH-insulin-like growth factor axis (58). The control of pituitary GH secretion in fish is multifactorial, with a balance of stimulatory and inhibitory neurohormones acting directly on the pituitary somatotrophs on a seasonal basis (57). SRIF is the major GH inhibitor in fish. Three cDNAs encoding for pre-pro-SRIF (PSS) have been cloned from goldfish brain (43): PSS-I, which encodes SRIF-14; PSS-II, which is potentially processed into gSRIF-28 that has [Glu1,Tyr7,Gly10]SRIF-14 at the COOH terminus; and PSS-III, which encodes [Pro2]SRIF-14 at its COOH terminus. All of these peptides were able to inhibit in vitro secretion of GH from goldfish pituitary cells (81). Moreover, in situ hybridization studies in goldfish also suggest that the three SRIF peptides could be involved in the control of pituitary function (8). On the other hand, GH release is stimulated by dopamine and a number of neuropeptides (57), including ghrelin (75). Notably, several of these neuropeptides such as CCK and bombesin (BBS)/gastrin-releasing peptide (GRP) and ghrelin have also been shown to regulate food intake in fish as well as in mammals (25, 59, 73, 74). In goldfish, a short-term relationship between circulating serum GH levels and feeding has been demonstrated (25). When fed a 2% wet body wt ration on a regular schedule, goldfish exhibit an acute elevation in serum GH at 30 min after feeding (25). It was speculated that CCK and/or BBS/GRP might integrate the regulation of satiation and the postprandial increase in circulating GH levels (45).
In goldfish, the effects of CCK and BBS on mRNA expression of SRIF isoforms have been recently studied (11). Both intraperitoneal and intracerebroventricular administration of CCK inhibits the expression of PSS-I but not PSS-II or PSS-III. On the other hand, BBS inhibits the expression of both PSS-I and PSS-II. These results suggest that, in addition to a possible direct effect at the pituitary level, both CCK and BBS can also act via SRIF peptides to control GH secretion (11). However, neither the involvement of SRIF peptides in the postprandial control of GH nor the effects of ghrelin on mRNA expression of SRIF peptides have been studied in fish. Furthermore, the possible interactions of ghrelin with BBS/GRP are also unknown in vertebrates.
In this report, we analyze the expression of mRNA for SRIF variants in the forebrain and ghrelin in the gut. The expression is studied in relation to feeding, postprandial GH secretion, and intraperitoneal injections of BBS and/or ghrelin peptides.
| MATERIALS AND METHODS |
|---|
|
|
|---|
Goldfish (Carassius auratus) of the common or comet varieties, with body weights ranging from 25 to 40 g, were purchased from Mount Parnell Fisheries (Mercersburg, PA) and maintained in 300-liter flow-through aquaria at 17°C under the simulated natural photoperiod of Edmonton, Alberta, Canada, for at least 2 wk before experiments were started. The fish were fed with commercially prepared Unifeed NU-Way trout ration (United Feeds, Calgary, AB, Canada). Sexually regressed or early recrudescent fish were used for the experiments. Goldfish were anesthetized with 0.02% tricaine methanesulfonate (MS-222, Syndel Laboratories, Vancouver, BC, Canada) for 2 min before treatments and tissue collection. The experimental procedures, approved by the University of Alberta, were performed according to standards of the Canada Council on Animal Care.
Peptides
BBS was obtained from American Peptide (Sunnyvale, CA). Goldfish ghrelin(119) with an octanoyl modification in the Ser3 residue was synthesized in the laboratory of Dr. Jean E. Rivier (Clayton Foundation Laboratories for Peptide Biology, The Salk Institute of Biological Sciences, San Diego, CA). Stock solutions were made in fish physiological saline (0.65% NaCl) or Milli-Q water, aliquoted, and stored at 20°C. Aliquots were subsequently thawed and diluted in fish physiological saline before use.
Intraperitoneal Injections
Fish were anesthetized, and 12 µl/g body wt of either fish saline (0.65% NaCl) or peptide dissolved in fish saline were administered into the peritoneal cavity just caudal to the pelvic fins, using a 25-gauge, 5/8-in. needle attached to a 100-µl Hamilton syringe. After injection, fish were returned to their respective aquaria and allowed to recover.
Experimental Design
Periprandial changes in serum GH levels and the forebrain expression of PSS genes. Goldfish were acclimatized to a scheduled feeding time for 2 mo. The fish were fed once daily with 2% body wt ration. On the day of the experiment, one group of fish was sampled at different times with respect to the scheduled feeding time. As a control, a group of fish did not receive food at the scheduled feeding time (unfed group). Under deep anesthesia, blood was collected and the animals were euthanized by spinal transection, and then forebrain tissue was collected. Forebrain tissue was maintained at 70°C until RNA extraction. Blood was allowed to clot overnight at 4°C, centrifuged and serum collected, and stored at 20°C until GH determination.
Time course of the effects of ghrelin on serum GH levels and the forebrain expression of PSS genes. Ghrelin (100 ng/g body wt) was administered intraperitoneally; at different times postinjection, the animals were anesthetized, and blood and forebrain samples were collected and stored as described previously.
Effects of increasing dosages of ghrelin on serum GH levels and the forebrain expression of PSS genes. Ghrelin (0100 ng/g body wt) was administered intraperitoneally. After 45 min, the fish were anesthetized, and forebrain samples were collected and processed as described above.
Interaction between ghrelin and BBS. Ghrelin (50 ng/g body wt) and/or BBS (100 ng/g body wt) were administered intraperitoneally. After 45 min, the animals were anesthetized, and forebrain samples were collected and processed as described above.
Effects of BBS on serum GH levels and the forebrain expression of PSS genes, before and after the feeding time, in fed and unfed conditions. Goldfish were randomly selected for placement into three groups and fed once daily with a 2% body wt ration. The fish were acclimatized to a scheduled feeding time for 2 mo. On the day of the experiment, 3 h before the scheduled feeding time, one group of fish received saline or BBS (100 ng/g body wt ip) injections and were sampled 45 min after injection. Of the remaining groups, only one group received food at the scheduled feeding time (fed group). Three hours after the scheduled feeding time, fish received saline or BBS (100 ng/g body wt ip) injections. Fish were sampled 45 min after injection under deep anesthesia. Forebrain tissue was stored at 70°C until RNA extraction. Blood was allowed to clot overnight at 4°C, centrifuged and serum collected, and stored at 20°C until GH and glucose determination. From the preprandial- and postprandial-fed groups, the most rostral portion of the gut, the "J" region (62), was also collected, cleaned in fish saline, and kept at 70°C until RNA extraction to determine ghrelin expression levels as described earlier (73).
GH Radioimmunoassay
Serum GH concentrations were determined using a double antibody radioimmunoassay, as described previously (47).
Preparation of Total RNA and Slot-Blot Quantification
We extracted total RNA from goldfish forebrain using Trizol reagent (Life Technologies, Gaithersburg, MD), based on the guanidinium thiocyanate-phenol-chloroform extraction method (13). The integrity of the RNA was verified in a denaturing agarose gel, which was stained with ethidium bromide. Slot-blot analysis was used to quantitate mRNA levels for the three PSS genes. Total RNA was blotted onto a Hybond-N nylon membrane (Amersham Pharmacia Biotech, Buckinghamshire, UK) using a Bio-Dot SF blotting apparatus (Bio-Rad Laboratories, Hercules, CA). Specific cDNA probes for PSS-I, PSS-II, PSS-III, ghrelin, and goldfish
-actin were amplified by PCR according to Canosa et al. (10) and Unniappan et al. (73) and labeled with [
-32P]dCTP using the Rediprime II Kit (Amersham Pharmacia Biotech) based on the random primer method (18). The labeled probes were purified through a silica cartridge using the QIAquick nucleotide removal kit (Qiagen, Santa Clarita, CA). The specificity of the probes and the quantitative relationship between amount of total RNA and the measured signal were previously demonstrated (10, 73). Hybridization was performed as previously described (10). Briefly, the membranes were prehybridized in hybridization solution (0.5 M Na2HPO4, 7% SDS, 1 mM EDTA, 1% BSA) for 3 h. The hybridization solution was then replaced with fresh hybridization solution, and 106 counts·min1·ml1 of labeled probe were added. After overnight hybridization at 65°C, the membranes were washed three times for 30 min with wash solution (40 mM Na2HPO4, 1 mM EDTA, 1% SDS) and exposed to a PhosphorImager screen for 72 h for PSSs or 10 days for ghrelin probes. As an internal control, the membranes were stripped and reprobed with [
-32P]dCTP-labeled cDNA probe for goldfish
-actin. The radioactive signal was scanned using a PhosphorImager 445 SI (Molecular Dynamics, Sunnyvale, CA) and quantified by Image Quant Software (Molecular Dynamics).
Data Analyses
The mRNA levels for each gene were expressed as a ratio of specific gene mRNA to
-actin and then normalized as a percentage of the control (saline or time 0) group. The normalized data were subjected to ANOVA followed by Tukey's (assuming equal variance) or Dunnett's (equal variance not assumed) multiple comparisons tests. Differences were considered significant when P < 0.05.
| RESULTS |
|---|
|
|
|---|
Similar to previous findings in our laboratory (28), serum GH levels were significantly increased at 20 min after feeding (Fig. 1A). Serum GH levels returned to baseline 60 min after feeding. On the other hand, in the unfed group, the serum GH levels remained unchanged (Fig. 1A).
|
|
To identify the peripheral signal that triggers the changes in forebrain expression of PSS genes, we evaluated the effects of intraperitoneal administration of ghrelin on the expression levels of PSS in the goldfish forebrain. In this experiment, a dosage of 100 ng/g body wt of ghrelin(119) induced a transient rise of serum GH at 15 min after intraperitoneal injection (Fig. 2A), returning to the baseline at 45 min postinjection. The administration of ghrelin(119) had different effects on PSS gene expression (Fig. 2B). Ghrelin(119) treatment significantly increased the PSS-I expression levels, whereas it significantly decreased PSS-II gene expression at 45 min postinjection. No significant effects on PSS-III were found (Fig. 2B). Ghrelin(119) significantly increased PSS-I gene expression at only the maximum dosage (100 ng/g body wt) assayed, whereas it significantly decreased PSS-II expression at all dosages tested (Fig. 3). No significant effects were found on PSS-III expression (Fig. 3).
|
As previously shown, CCK and BBS act on SRIF peptide gene expression (11). It is believed that these peptides act postprandially, inducing satiety, and are also involved in GH regulation (28, 45). Table 1 summarizes the effects of CCK, BBS, and ghrelin on the forebrain expression of PSS-I and PSS-II. Although CCK had no effects, both BBS and ghrelin reduced PSS-II gene expression. Furthermore, the effects of ghrelin and BBS differed from one another in the regulation of PSS-I gene expression. Although BBS significantly decreased PSS-I expression, ghrelin showed a nonsignificant tendency to increase it (Table 1); a higher dosage of ghrelin(119) (100 ng/g body wt) increased PSS-I expression (Figs. 2B and 3). Thus we hypothesize that BBS and ghrelin could interact in the postprandial control of GH. In an attempt to verify this hypothesis, the effects of coadministration of these peptides on the expression levels of SRIF precursors were investigated. The results, shown in Fig. 4, confirm that intraperitoneal administration of 50 ng/g body wt of ghrelin(119) has no significant effect on PSS-I expression levels (Fig. 4A), whereas it inhibits the expression of PSS-II (Fig. 4B). On the other hand, 100 ng/g body wt of BBS reduced both PSS-I and PSS-II expression levels (Fig. 4, A and B). Neither ghrelin nor BBS had effects on PSS-III gene expression (Fig. 4C). Coadministration of ghrelin blocked the effects of BBS on the expression of PSS-I (Fig. 4A) but not on PSS-II (Fig. 4B).
|
|
It is already known that serum ghrelin levels are higher before meals and in fasted animals, whereas serum ghrelin levels are significantly decreased after a meal (73). To verify the interaction between ghrelin and BBS in a more physiological approach, the serum ghrelin levels were manipulated by the feeding status. Intraperitoneal administration of BBS 3 h before feeding increased serum GH levels (Fig. 5A, left). Under these conditions, BBS had no effects on PSS-I expression levels, whereas it decreased PSS-II gene expression levels (Fig. 5A, right). In unfed animals, 3 h after the scheduled feeding time, BBS administration had no effects on serum GH levels or PSS-I gene expression levels but still significantly reduced the expression of PSS-II (Fig. 5B). On the other hand, when the fish were fed and the effect of BBS administration was assayed 3 h after feeding, BBS significantly increased serum GH levels, whereas it reduced both PSS-I and PSS-II gene expression (Fig. 5C).
|
BBS mimics the actions of endogenous GRP, which controls gastric function shortly after meals, reducing the gastric acid secretion and increasing the gastrin release (49). In this experiment, we tested the hypothesis that ghrelin gene expression in the gut could be under the control of GRP. We determined ghrelin mRNA expression levels in the goldfish gut 3 h before (preprandial) or 3 h after (postprandial) the scheduled feeding time in fish injected with saline or BBS (100 ng/g body wt). As shown in Fig. 6, administration of BBS during the preprandial period significantly decreased ghrelin expression. Confirming previous results, we found significantly lower ghrelin expression in the postprandial period than in the preprandial period. BBS treatment did not cause a further decrease in ghrelin expression in the postprandial condition (Fig. 6).
|
| DISCUSSION |
|---|
|
|
|---|
The goldfish pituitary expresses several SRIF receptor subtypes, with types 2 and 5 (sst2 and sst5) being the most highly expressed (4042, 44). Goldfish sst5 shows a four- to fivefold higher affinity to mammalian SRIF-28 and gSRIF-28 (42) and has similar pharmacological characteristics to mouse sst5 (55), which also binds preferentially to mammalian SRIF-28. It was recently shown that estradiol treatment reduces the in vitro GH responsiveness to SRIF-14 and simultaneously reduces the expression levels of sst2 in goldfish pituitary (12). However, estradiol treatment increases the expression levels of sst5 (9). Thus it seems that sst5 may be specialized in the transduction of gSRIF-28 actions in the pituitary. Taking all of these results together, we consider it plausible that a reduction in gSRIF-28 inhibitory tone occurs during meals as a postprandial control mechanism of GH secretion. However, this hypothesis needs further experimental demonstration.
The mechanisms underlying the control of SRIF gene expression during meals has not yet been established. However, in a recent report (11), our group analyzed the effects of intraperitoneal administration of CCK and BBS, two good candidates that link food intake and GH secretion in goldfish. CCK is widely distributed in both peripheral nerves and endocrine cells in the gastrointestinal tract in mammals, as well as in nonmammalian vertebrates and is involved in many gastrointestinal functions, including gallbladder contraction, pancreatic enzyme secretion, stimulation of gastrointestinal motility, and inhibition of gastric emptying (1, 14, 23, 32, 76). In mammals, CCK acts as a satiation factor at the levels of the gut and centrally in specific brain regions (61, 66). Intraperitoneal or intracerebroventricular injection of CCK acutely suppresses food intake in goldfish, supporting the action of CCK as a satiety factor in fish (27) and stimulates GH release (24). Furthermore, CCK administration reduces the forebrain expression of PSS-I, but it does not affect PSS-II or PSS III levels (11). Taking these and the present results into account, it is possible to conclude that CCK may not be involved in the postprandial control of GH secretion. Moreover, variations in CCK mRNA levels in goldfish brain occur 2 h after feeding (60), well after the postprandial changes in GH release have occurred. In line with this interpretation, Talbot and Higgins (67) found that, in trout, the gallbladder content (bile acid) is reduced around 36 h after initiation of feeding as a consequence of gallbladder contractions stimulated by CCK (1). Therefore, the endogenous postprandial release of CCK occurs several hours after food intake. In conclusion, the postprandial increases in activity of the CCK neuronal system more likely relates to satiation, rather than to control of GH release.
BBS is a tetradecapeptide that was first isolated from the skin extracts of the fire-bellied toad, Bombina bombina (2). The structurally related GRP, a 27-amino acid peptide, has been isolated from several mammalian species, and closely related forms of the peptide are also found in other vertebrates (30, 51, 78). This group of peptides is characterized by a highly conserved COOH terminus, which is important for biological actions. Although BBS and antibodies against BBS have been used in most physiological and immunohistochemical studies, GRP is the peptide expressed in the gut and brain of vertebrates. Exogenous BBS most likely mimics the function of endogenous GRP (49). BBS/GRP peptides are widely distributed in the gastrointestinal tract and central nervous system (50) and have been shown to be potent in suppressing food intake when administered either intraperitoneally or centrally within specific areas of the hypothalamus and/or the hindbrain in mammals (20, 22). In fish, BBS-like peptides have been detected within neurons and endocrine cells of the gastrointestinal tract, cardiovascular system, and brain (Ref. 25 and references cited therein). BBS-like peptide has also been shown to be involved in the regulation of gut motility and visceral activity in several fish species (5, 31). In goldfish, BBS acutely suppresses food intake and also stimulates GH release when administered either intraperitoneally or intracerebroventricularly (25, 26). Furthermore, intraperitoneal administration of BBS reduces PSS-I and PSS-II but not PSS-III gene expression in goldfish forebrain (11). These actions on SRIF neurons likely represent the route of action of BBS/GRP on anterior pituitary hormone release since only low densities of BBS/GRP binding sites are present in the anterior pituitary (29). On the other hand, high densities of binding sites are present in the neurointermediate lobe and the hypothalamus. Even if BBS/GRP actions on the anterior pituitary are mediated by forebrain SRIF neurons, its actions on forebrain SRIF gene expression do not allow a full explanation for the postprandial changes in expression of SRIF precursors shown in the present paper. This is because BBS affected not only PSS-II but also PSS-I.
Ghrelin is synthesized mainly in the stomach and was first identified as the endogenous ligand to the GH secretagogue receptor in mammals (38, 39). It is also a potent stimulus to feeding (79, 80), and both the mRNA expression and the circulating levels are upregulated in fasted conditions (3, 79). Periprandial changes in plasma ghrelin levels in humans suggest ghrelin has a role as meal initiator (15). Ghrelin has also been identified in several nonmammalian species (3337, 56), including goldfish (74). As in mammals, goldfish ghrelin acts as GH secretagogue, both in vivo and in vitro, and stimulates food intake (73, 75). Ghrelin induces an increase in serum GH levels 1530 min after intraperitoneal administration in goldfish (Ref. 75 and present results). In mammals, ghrelin stimulates GHRH-producing neurons and acts as a functional antagonist of SRIF-14 (69, 77). In goldfish, perfusion of dispersed cells with SRIF-14 abolishes the stimulatory effects of a ghrelin pulse (75). Ghrelin in fish, as in mammals, represents a link between food intake and GH secretion. In goldfish, serum ghrelin levels are high before meals and are significantly lower 1 h after food intake (73). Our present results show that intraperitoneal administration of ghrelin had a dual effect on PSS-I and PSS-II gene expression. Thus, although changes in ghrelin release cannot completely explain the postprandial changes in forebrain expression of PSS genes, it may be part of the regulatory mechanism for postprandial GH secretion. In support of this view, we found that ghrelin is able to block the inhibitory effects of BBS on PSS-I gene expression but not on PSS-II. Therefore, the interaction between ghrelin, which increases preprandially, and GRP, which is believed to act shortly after food intake, could account for the postprandial variations of forebrain expression levels of SRIF precursors. In line with this hypothesis, we found that, 3 h before the scheduled feeding time (high serum ghrelin levels), intraperitoneal administration of BBS increased GH, while, concomitantly, it only reduced PSS-II gene expression. This experimental result resembles the natural variations during spontaneous feeding. When BBS was administered to unfed animals 3 h after the scheduled feeding time (high serum ghrelin levels), it was correlated with a reduction in PSS-II gene expression, although no changes in serum GH levels were found; the reason for this behavior is not clear at present. One explanation could be that in these experimental conditions (27-h food restriction) some other mechanisms controlling GH secretion are activated in a similar fashion as during starvation, a condition in which GH is elevated. Interestingly, when BBS was administered 3 h after feeding (low serum ghrelin levels), it was correlated with an increase in serum GH levels and a decrease of both PSS-I and PSS-II gene expression, supporting the interpretation that the preprandial levels of ghrelin blocks BBS effects on PSS-I gene expression.
The regulation of production and secretion of ghrelin from the stomach is not fully understood. Plasma ghrelin levels are upregulated under conditions of negative energy balance, such as starvation, cachexia, and anorexia nervosa, and downregulated during positive energy balance, such as following feeding and obesity (15, 53, 64, 71, 72). Similar to mammals, in goldfish, serum ghrelin levels decrease after meals (73). To date, there are no reports on the hormonal regulation of ghrelin in fish. In humans, it has been shown that insulin may reduce ghrelin secretion (19, 46). Also, SRIF-14 infusions into normal subjects reduce plasma ghrelin levels to 7080% of normal control values (7, 54). Furthermore, perfusion with SRIF-14 or the SRIF agonist octreotide reduces ghrelin secretion in isolated rat stomach, suggesting that this is a direct effect of SRIF on the stomach in a paracrine fashion (46, 65). Our present results show, for the first time, that intraperitoneal administration of BBS reduces ghrelin mRNA expression in the gut. GRP has a role in gastric physiology (49) and also induces satiety and meal termination through vagal and nonvagal afferents (52). Therefore, a concomitant inhibition of the hunger signal ghrelin (15) should be expected. It is known that BBS/GRP induces SRIF-14 secretion from gastric fundus and antrum in mammals (17, 48, 63), and, as mentioned before, SRIF-14 reduces ghrelin secretion (65). Then, it may be that SRIF mediates BBS/GRP actions on ghrelin gene expression at the gut level.
In conclusion, the present results show that an interplay between BBS/GRP and ghrelin controls GH secretion after feeding and occurs at the same time as a reduction in the expression of PSS-II. This interplay is schematically represented in Fig. 7, which shows preprandial and postprandial situations. In preprandial conditions, it was previously shown that ghrelin levels are increased (73); as a consequence, 1 h before the scheduled feeding time, circulating levels of GH levels tend to increase in goldfish (Figs. 1A and 7A). This increase in serum GH levels may represent the actions of increasing levels of ghrelin on GH secretion (75), as well as on PSS-II expression (Figs. 1B and 7A). However, high doses of ghrelin can stimulate PSS-I expression, which if translated into peptide secretion would exert a counterbalance to a GH rise. Therefore, GH secretion would be kept close to normal basal levels. Soon after feeding, BBS/GRP exerts a satiety signal. It should be noted that meal termination or satiety occurs before the digestion and/or absorption of food nutrients, and BBS/GRP may represent one of the physiological mediators of satiety signals between gut and brain (52). Additionally, BBS/GRP inhibits PSS-II expression, and thus serum GH levels increase significantly while at the same time PSS-II expression decreases (Figs. 1 and 7B). The expression levels of PSS-I are not affected during BBS/GRP activity due to the presence of ghrelin (Figs. 4 and 7B). We also demonstrated in the present paper that BBS/GRP peptides inhibit ghrelin transcription in the gut (Fig. 6). The reduction of ghrelin expression by BBS is a likely explanation for the postprandial reduction of serum ghrelin levels, and it may contribute to the satiety effect of BBS/GRP peptides.
|
| GRANTS |
|---|
|
|
|---|
| ACKNOWLEDGMENTS |
|---|
Present address for S. Unniappan: Laboratory of Molecular and Cellular Medicine, Dept. of Cellular and Physiological Sciences, D. H. Copp Bldg., 2146 Health Sciences Mall, University of British Columbia, Vancouver, BC, Canada V6T 1Z3.
| FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
| REFERENCES |
|---|
|
|
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||