Selective Antagonism of the NPY Y5 Receptor Does Not Have a Major Effect on Feeding in Rats

NPY, human PYY, [cPP^1–7,NPY^19–23, Ala³¹,Aib³²,Gln³⁴]-hPP, and sibutramine were purchased from Bachem (Saffron Walden, Essex, U.K.), Sigma Aldrich, Tocris Cookson (Bristol, U.K.), and St. Andrews ChemTech (Fife, U.K.), respectively. NPY5RA-972 (31) was synthesized within the Medicinal Chemistry laboratories at AstraZeneca (Alderley Park, U.K.).

Hi5 insect cells transiently (48 h) transfected (baculovirus-infected) with either rat or human NPY Y5 receptor were used for preparation of NPY Y5 membranes. Human NPY Y1, NPY Y2, and NPY Y4 membranes were prepared from SK-N-MC, KAN-TS, and stable NPY Y4 expressing CHO cells, respectively. Membranes were prepared by sonication (3 × 15 s) in ice-cold hypotonic buffer (12.5 mmol/l Tris, 1.25 mmol/l EDTA, 2.5% sucrose [pH 7.4]) containing complete protease inhibitor tablets (1/100 ml; Roche Molecular Biochemicals). The lysate was layered onto a 41% sucrose cushion and centrifuged at 100,000g for 1 h. The membrane layer was harvested and stored (in 50 mmol/l Tris, 5 mmol/l EDTA, 10% sucrose [pH 7.4]) at −80°C until use.

Binding assays were performed in presiliconized, round-bottomed, polypropylene 96-well plates (Corning Costar). Compounds were dissolved in DMSO and diluted (20-fold) in binding buffer (50 mmol/l HEPES, 2.5 mmol/l CaCl₂, 1 mmol/l MgCl_2, 0.5% BSA [pH 7.4]). Each incubate had 0.01 μCi of [¹²⁵I]porcine peptide YY (Amersham Pharmacia Biotech) added to 10–80 μl of membranes (sufficient to give a specific binding of 1,500 cpm) and 10 μl of compound solution. After a 2-h incubation at room temperature, incubates were filtered onto Canberra Packard GF/C 96-well filter plates (Packard Instruments) pretreated with 0.5% polyethylenimine using a Brandel harvester. The plates were washed twice with wash buffer (binding buffer + 0.5 mmol/l NaCl). The filter plates were dried overnight and counted on a Canberra Packard Top Count (Packard Instruments) after addition of 20 μl of Microscint 40 (Packard Instruments) to each well.

Recombinant HEK293 cells stably expressing the rat NPY Y5 receptor and also a reporter cassette consisting of a cyclic AMP driven β-galactosidase gene were used. The cells were grown in T175 flasks to 80–90% confluence in Dulbecco’s modified Eagle’s medium (DMEM) plus 10% FCS, 1% penicillin/streptomycin, and 1% l-glutamine. After harvesting, cells were washed and resuspended in indicator-free DMEM. Cells (2 × 10⁴ cells in 50 μl) were preincubated (in triplicate) with compounds (in 20 μl) for 5 min at room temperature in flat-bottomed polystyrene 96-well plates (Corning Costar). NPY (20 μl) was added to each of the wells for a 5-min incubation at room temperature. Ten microliters of 10 μmol/l forskolin (Sigma Aldrich) was added to start the incubation of 3.75 h at 37°C. After incubation, the plates were cooled at room temperature before the addition of 50 μl of the chlorophenol red-β -d-galacto-pyranoside cocktail (CPRG) (0.6% SDS, 1.5 mmol/l CPRG (Roche Molecular Biochemicals), 24 mmol/l Na₂HPO₄(2H₂O), 40 mmol/l NaH₂PO₄(2H₂O), 10 mmol/l KCl, and 1 mmol/l MgSO₄(7H₂O). After an overnight incubation at room temperature, the absorbance of each plate was read on a Tecan SpectraFluor at 570 nm.

Male Wistar and obese (fa/fa) Zucker rats (Zucker_fLIS/Alpkh) were obtained from the animal breeding unit at AstraZeneca. Sprague-Dawley rats were obtained from Charles River (Kent, U.K.). Unless otherwise stated, animals were housed individually in a temperature- and humidity-controlled environment with a 12-h light:12-h dark cycle (lights on 06:00 h) and provided with standard rat diet (RM1, 2.61 kcal/g; Special Diet Services, Essex, U.K.) and water ad libitum.

Suspensions of NPY5RA-972 or sibutramine were prepared in hydroxypropylmethylcellulose (HPMC; 0.5% wt/vol), polysorbate 80 (Tween 80; 0.1% wt/vol) in water. Vehicle or NPY5RA-972 was dosed by oral gavage (75 mm × 16-G curved oral dosing cannula; IMS, Cheshire, U.K.) at a volume of 2 ml/kg.

Male Wistar rats (220–270 g) were anesthetized with Halothane and secured in a stereotaxic apparatus (Stoelting, Wood Dale, IL). With the incisor bar set at −3.3 mm below the intra-aural line, a guide cannula (8-mm length; Plastics One, Roanoke, VA) was implanted in the third ventricle at the following coordinates, according to the rat brain atlas of Paxinos and Watson (32): lateral 0 mm; anterior-posterior −3 mm; dorsoventral −8 mm. Cannulae were secured with two screws and dental compomer (Dyract AP, Dentsply, Wright Cottrell, Manchester, U.K.) cured with an ultraviolet light (QHL75, Dentsply). After suturing, a dummy needle was inserted to keep the cannula patent, and rats were allowed to recover from anesthesia. Rats were left to recover for 10–14 days, and all animals used in subsequent experiments had recovered at least to presurgical weight by the day of experimentation.

Peptides were dissolved in sterile water and administered intracerebroventricularly via an injection cannula (projecting 1 mm below the tip of the guide cannula) attached to a length of siliconized tubing (0.51 mm internal diameter) connected to a 25-μl Hamilton microsyringe. Intracerebroventricular (ICV) injections (5 μl) were administered to conscious, unrestrained rats over a period of 60 s.

Male Wistar rats (n = 3) were dosed by mouth (p.o.) with 30 μmol/kg NPY5RA-972, and plasma was obtained from 9–10 whole-blood samples withdrawn over a 24-h period (total volume of blood ∼2 ml, replaced by an equal volume of saline) after dosing. NPY5RA-972 concentrations in plasma were determined by high-performance liquid chromatography-ultraviolet (285 nm) analysis.

In separate experiments, male Wistar rats (350–400 g) were dosed with NPY5RA-972 (10–80 mg/kg p.o.). One hour later, they were terminally anesthetized with sagatal (pentobarbitone, 60 mg/ml, 0.6 ml/rat) and mounted in a stereotaxic apparatus, and a cerebrospinal fluid sample (∼100 μl) was collected using a 25-G butterfly needle inserted through the muscles of the neck and piercing the dura overlying the cisterna magna. Cerebrospinal fluid (CSF) samples that showed any discoloration were discarded; only clear samples were subject to analysis. CSF and plasma samples collected terminally were processed for subsequent analysis of compound concentrations.

Either NPY or [cPP^1–7,NPY^19–23, Ala³¹,Aib³²,Gln³⁴]-hPP was injected intracerebroventricularly into male Wistar rats (n = 5–7) in the free-feeding state during the early light phase (commencing 08:00–10:00 h). Animals were returned to their home cage that contained a preweighed amount of food, and the food that remained at 1, 2, 4, and 6 h was determined on a Mettler Toledo PG2002-S balance (Fischer Scientific, Leicestershire, U.K.), corrected for spillage, and recorded to the nearest 0.1 g.

NPY5RA-972 (1–10 mg/kg) or vehicle was dosed (p.o.) to male Wistar rats (n = 4–7) in the free-feeding state during the early light phase (commencing 08:00–10:00 h). One hour later, these rats received an ICV injection of either [cPP^1–7,NPY^19–23,Ala³¹,Aib³²,Gln³⁴]-hPP (0.2 or 0.6 nmol/rat) or NPY (0.2 or 2.0 nmol/rat). Cumulative food intake was measured as described.

Male Wistar rats (219–267 g, n = 8–12) were fasted from 10:00 h and 23 h later dosed (p.o.) with either vehicle or NPY5RA-972 (1–10 mg/kg). One hour later, preweighed food hoppers were returned to the animals, and cumulative food intake was measured 1, 2, 4, and 6 h later.

Free-feeding male Wistar rats (240–306 g, n = 8 per group) or male obese Zucker rats (472–492 g, n = 7–12) were dosed p.o. with either vehicle or NPY5RA-972 (1–10 mg/kg) 1 h before the onset of the dark-phase (18:00 h) and cumulative food intake measured from 18:00 h for either 2, 4, 15, and 24 h or 6, 14, and 24 h, respectively.

To ascertain a suitable twice-daily dosing regimen for chronic studies that provided 24-h blockade of NPY Y5 receptors in brain, we determined the effects of 12-h predosing of NPY5RA-972 on food intake induced by [cPP^1–7,NPY^19–23,Ala³¹, Aib³²,Gln³⁴]-hPP. Male Wistar rats (n = 6/group) were dosed with NPY5RA-972 (1–10 mg/kg) at 21:00 h and received an ICV injection of [cPP^1–7,NPY^19–23,Ala³¹,Aib³²,Gln³⁴]-hPP (0.6 nmol/rat) 12 h later. Food intake for the 2 h after [cPP^1–7,NPY^19–23,Ala³¹,Aib³², Gln³⁴]-hPP injection was determined as described above. As a positive control, an additional group of rats were administered NPY5RA-972 (3 mg/kg p.o., n = 5) 1 h before [cPP^1–7, NPY^19–23,Ala³¹,Aib³², Gln³⁴]-hPP.

In chronic studies that used previously unmanipulated Wistar rats, NPY5RA-972 (10 mg/kg, n = 7) or vehicle (n = 6) was dosed (p.o.) twice daily (at 06:00 h and 16:00 h). For 3 days before commencement of dosing with compound, rats were accustomed to handling and dosing with vehicle only. Food intake and body weight were measured daily. A positive control group (n = 7) receiving sibutramine was included in the chronic studies. Sibutramine is a marketed appetite-suppressant drug that acts principally by inhibiting the reuptake of serotonin and norepinephrine (33). Sibutramine was dosed at 3 mg/kg once daily (at 16:00 h), a dose that was previously proved to be effective in our and others’ hands (34), and is the maximum dose we have found that avoids marked behavioral disruption. Sibutramine-treated rats also received a dose of vehicle (at 06:00 h) to maintain consistency of procedures across all groups.

In diet-induced obese studies, male Wistar rats were fed a high-energy diet (D12266B, 4.41 kcal/g; Research Diets). This rodent diet is designed to distinguish obesity-prone and obesity-resistant phenotypes and is similar to that used by Levin and colleagues (35,36) in their studies of susceptibility and resistance to obesity. Energy intake in rats fed a high-energy diet is expressed as kcal/d.

Unless otherwise stated, results are presented as mean ± SE. In vitro dose-response curves were calculated and plotted using a nonlinear least squares curve-fitting routine within Origin 5 (Microcal Software). All animal experiments were based on a between-individual design with each animal being tested on a single occasion in a single experiment. Feeding and body weight data were analyzed by one-way ANOVA followed by Bonferroni multiple comparison test. A two-tailed probability of P < 0.05 was considered statistically significant.

NPY5RA-972 bound with high affinity to both the human and the rat Y5 receptor (half-maximal inhibitory concentration [IC₅₀], 3.1 ± 1.0 nmol/l and 9.3 ± 1.2 nmol/l, respectively; Table 1). In contrast, 10 μmol/l NPY5RA-972 displayed no binding to human NPY Y1, NPY Y2, or NPY Y4 receptors and showed at least 1,000-fold selectivity for NPY Y5 in a commercially available panel (MDS Pharma Services) of 129 different bindings assays (including assays for NPY, muscarinic, serotonergic opioid, melanocortin, galanin, cholecystokinin, cannabinoid, neurotensin, glucagon-like peptide receptors, and serotonin transporter).

In a functional reporter assay for NPY Y5 antagonism, NPY5RA-972 completely reversed the suppressive effect of 10 nmol/l NPY on 1 μmol/l forskolin-stimulated HEK 293 cells stably expressing the rat Y5 receptor (Table 1). NPY5RA-972 displayed no agonist activity as indicated by a lack of effect of 10 μmol/l of the compound in the absence of NPY in this reporter assay (data not shown).

[cPP^1–7,NPY^19–23,Ala³¹,Aib³²,Gln³⁴]-hPP is an NPY Y5 selective agonist recently described by Cabrele et al. (20). In broad agreement with Cabrele et al. (20), we found that [cPP^1–7,NPY^19–23,Ala³¹,Aib³²,Gln³⁴]-hPP binds with high affinity to rat and human NPY Y5, has potent agonist activity in the rat NPY Y5 reporter assay, and has a high degree of selectivity for the NPY Y5 receptor (>5,000-fold versus NPY Y1, ∼ 200-fold versus NPY Y2, ∼500-fold versus NPY Y4; Table 1).

The mean (of three rats) time to maximum plasma concentration (T_max) after oral gavage of 30 μmol/kg (=10.5 mg/kg) NPY5RA-972 was 1 h, at which time the total concentration (C_max) was ∼35 μmol/l. At a measured in vitro protein binding of 99.1%, predicted free plasma concentrations of NPY5RA-972 at T_max was 315 nmol/l. The half-life of NPY5RA-972 after oral dosing was 6.4 h, the 24-h area under the curve was ∼ 350 μmol · l⁻¹ · h⁻¹, and even at 24 h after dosing total plasma concentrations were still >3 μ mol/l.

In separate dose-response studies (10–80 mg/kg p.o.), total plasma concentrations of NPY5RA-972 at 1 h increased with dose but did not scale linearly. CSF concentrations of NPY5RA-972 (153 ± 35 nmol/l) after a dose of 10 mg/kg (p.o.) massively exceeded the IC₅₀ of NPY5RA-972 (by 16- to 50-fold; Tables 1 and 2), with higher doses producing even greater CSF concentrations.

[cPP^1–7,NPY^19–23,Ala³¹, Aib³²,Gln³⁴]-hPP injected intracerebroventricularly produced a marked and dose-dependent increase in food intake (Fig. 1A). The maximum dose of this NPY Y5 selective agonist seemed slightly lower (0.6 nmol) than that for NPY (2.0 nmol; Fig. 1A and B). The time course of effect of these two peptides seemed slightly different, with the NPY Y5 selective agonist having a less pronounced initial but more sustained effect (Fig. 1C), a profile identical to what we have observed with the less selective NPY Y5 agonist hPP (data not shown).

NPY5RA-972 (p.o.) inhibited food intake in rats induced by [cPP^1–7,NPY^19–23,Ala³¹, Aib³²,Gln³⁴]-hPP (ICV). NPY5RA-972 (10 mg/kg) completely abolished food intake induced by either submaximum (0.2 nmol) or maximum (0.6 nmol) doses of the NPY Y5 agonist (Fig. 2A). Even at doses as low as 1 mg/kg, NPY5RA-972 produced a marked and significant inhibition of the feeding response to NPY Y5 agonist (Fig. 2B).

In contrast to the abolition of the feeding response to a maximum dose of NPY Y5 selective agonist, NPY5RA-972 (3 mg/kg p.o.) had no effect on food intake induced by a maximum dose (2.0 nmol) of NPY infused 1 h later (Fig. 3A). Furthermore, even at higher doses of NPY5RA-972 (10 mg/kg), this antagonist had no effect on feeding induced by a submaximum dose (0.2 nmol) of NPY (Fig. 3B).

NPY5RA-972 (1–10 mg/kg p.o.) dosed 1 h before presentation of food had no significant effect on food intake for the ensuing 6 h (Fig. 4A). NPY5RA-972 (1–10 mg/kg) also failed to affect the fast-induced feeding response in another strain of rat (Sprague-Dawley, data not shown). Similarly, in free-feeding Wistar rats (Fig. 4B) or obese Zucker rats (Fig. 4C) dosed 1 h before the onset of the dark cycle (main feeding phase), NPY5RA-972 had no significant effect on food intake.

At a dose of 10 mg/kg, NPY5RA-972 (but not lower doses) 12 h predosing markedly inhibited food intake induced by the NPY Y5 selective agonist for 2 h (Fig. 5), indicating that this dose is sufficient to antagonize NPY Y5 receptors in the brain for at least 12–14 h. Therefore, a dosing regimen of 10 mg/kg twice daily was selected for chronic studies.

In normal Wistar rats, 3 mg/kg sibutramine produced a marked (∼ 30%) inhibition of food intake on the first day of dosing (Fig. 6A). Consistent with published data, the effects of sibutramine on food intake diminished with time (34,37), although cumulative food intake over the 9-day study was significantly (P < 0.001) lower in sibutramine-treated (213.3 ± 5.7 g) than in vehicle-treated (260.2 ± 3.0 g) rats. Sibutramine also significantly reduced overall body weight gain (vehicle 30 ± 2 g, sibutramine 14 ± 3 g; P < 0.001; Fig. 6B). However, chronic antagonism of NPY Y5 with NPY5RA-972 produced no significant effect on either food intake (cumulative food intake 252.3 ± 6.3 g; Fig. 6A) or body weight gain (27 ± 1g; Fig. 6B) compared with vehicle-treated controls.

Five-week-old Wistar rats were placed on a highly palatable high-energy diet or maintained on normal rat diet for a period of 12 weeks. After 10 weeks on the high-energy diet, only rats that exhibited clear hyperphagia and obesity compared with rats fed a normal diet were selected. Indeed, the calorie intakes (107 ± 4 vs. 72 ± 3 kcal/d) and body weights (596 ± 8g vs. 478 ± 18 g) were markedly higher in this selected obese cohort compared with the age-matched controls that were fed the regular diet. This diet-induced obese (DIO) cohort was randomly divided into three groups that were dosed (p.o.) with vehicle (twice daily), sibutramine (3 mg/kg, once daily) or NPY5RA-972 (10 mg/kg, twice daily). As in the previous experiment, sibutramine produced a significant (P < 0.001) inhibition of energy intake (cumulative energy intake 791 ± 26 kcal) compared with vehicle-treated (1,227 ± 55 kcal) rats, although its effects in DIO animals seemed far greater (Fig. 7A) than in Wistar rats that were fed a regular diet (Fig. 6A). Indeed, sibutramine not only slowed weight gain but also caused significant (P < 0.001) weight loss in DIO rats over the 12-day study period (vehicle 14.0 ± 6.1 g, sibutramine −37.0 ± 7.8 g; Fig. 7B). In contrast, NPY5RA-972 had no significant effect on either calorie intake (cumulative intake 1,199 ± 87 kcal) or body weight (12 ± 9g; Fig. 7).

A number of lines of evidence have suggested that NPY Y5 is a major receptor subtype that mediates the effects of NPY on feeding and has attracted much academic and industrial interest as a possible appetite suppression approach to the treatment of obesity (12, 30,38). Notably, the majority of these data have been generated in rodents. Supportive data in humans has been restricted to a neuroanatomical distribution similar to rodents (39,40) and to preliminary human genetic evidence in a very specific cohort (41). Indeed, other human genetic studies have resulted in negative findings (42,43). The data reported herein identify NPY5RA-972 as a potent and selective NPY Y5 receptor antagonist with good systemic and CNS exposure after oral dosing in rats, suggesting that it is a useful tool with which to explore the role(s) of NPY Y5 in physiological processes. Coadministration studies indicated that NPY5RA-972 prevented feeding induced by the NPY Y5 selective agonist [cPP^1–7,NPY^19–23, Ala³¹, Aib³²,Gln³⁴]-hPP. The most obvious conclusion of these experiments is that NPY5RA-972 after oral dosing occupies sufficient NPY Y5 receptors in the brain to prevent NPY Y5 agonist-induced feeding. The possibility that this apparent agonism-antagonism reflects activities at alternative, non-NPY Y5 feeding-related targets cannot be totally disproved. However, 1) [cPP^1–7,NPY^19–23,Ala³¹,Aib³²,Gln³⁴]-hPP is highly selective for NPY Y5 compared with other known NPY receptors (20 and present study); 2) NPY5RA-972 shows at least 1,000-fold selectivity against a panel of 129 receptors, enzymes, and transporters that included a number of receptors relevant to feeding; and 3) [cPP^1–7,NPY^19–23, Ala³¹,Aib³²,Gln³⁴]-hPP and NPY5RA-972 are of different structural classes (peptidic versus nonpeptidic), optimized for NPY Y5 potency and selectivity by two different groups using completely different chemical strategies, making it highly unlikely that both were unwittingly optimized for activity at alternative identical targets. It seems much more likely that their activities in co-administration experiments reflect agonism and antagonism at the NPY Y5 receptor, leading us to conclude that at the doses used in this study, NPY5RA-972 is an ideal tool to test hypotheses regarding the role of the NPY Y5 receptor in the rat.

Although low doses (1–10 mg/kg p.o.) of NPY5RA-972 inhibited food intake elicited by a selective NPY Y5 receptor agonist, it had no effect on feeding as a result of NPY (ICV), fasting, free-feeding in normal, genetically obese, or dietary obese rats. These findings strongly suggest that the NPY Y5 receptor is not a major physiological regulator of feeding in rats.

A previous study with a different NPY Y5 antagonist, CGP 71683A (30), drew very different conclusions regarding the role of NPY Y5 in feeding in the rat. CGP 71683A produces marked inhibitory effects in a variety of rat feeding models, including the ICV administration of NPY. NPY5RA-972 belongs to a series of carbazole compounds, and indeed within this series we found potent in vitro NPY Y5 antagonists that have impressive hypophagic activity in vivo in a variety of rodent feeding models (data not shown). However, we found that an exemplary compound with impressive hypophagic activity was also effective at inhibiting food intake in NPY Y5-deficient mice (data not shown), indicating that NPY Y5 antagonism was not its primary mode of action. Likewise, it now seems that CGP 71683A has been discredited as an in vivo tool to explore the biology of NPY Y5 as a result of its activity at serotonergic transporters and muscarinic receptors (44) and its efficacy in NPY-deficient (45) and NPY Y5-deficient mice (46). Clearly, it cannot be assumed that demonstrated in vitro pharmacology and in vivo hypophagic effects share a “cause-effect relationship.” In this respect, the utility of receptor knockout mice to define the molecular mode of action of drugs cannot be overemphasized. The mechanism of action by which other NPY Y5 antagonists (47,48) reported to produce marked hypophagic effects in rodents is unknown, but notably no data have been presented to indicate that antagonism of NPY Y5 is their primary mode of action. We believe that the present data show convincingly that NPY5RA-972 potently antagonizes NPY Y5 receptors in the brain but fails to affect feeding in a variety of rat feeding models. Emerging preliminary data with a range of other small molecular weight NPY Y5 antagonists (38, 49–52) are consistent with our findings.

It seems pertinent to revisit the evidence supporting a role for Y5 in feeding in rodents. This primarily consisted of data showing similarities of the pharmacology of NPY Y5 receptors in vitro and the pharmacology of NPY-related peptides on feeding in vivo (17–19,23). Although this conclusion has been disputed (53,54), it is now clear that selective activation of NPY Y5 stimulates feeding in rats (20–22). The present study therefore argues that the hyperphagic effect of selective receptor activation is not necessarily a good predictor of the importance of that receptor in feeding under more physiological circumstances. Studies using antisense oligonucleotides to downregulate NPY Y5 expression have also supported the hypothesis that full expression of this receptor is required to elicit the full hyperphagic effects of NPY and that NPY Y5 is important in the regulation of feeding in normal rats (25–28). The explanation for the differences in findings with antisense and the effects of an antagonist are unknown, but this discrepancy serves to indicate that such antisense studies are not always predictive of the effects of pharmacological manipulation.

The lack of a hypophagic phenotype of NPY Y5-deficient mice clearly is not supportive of a major role for NPY Y5 in regulating feeding in mice (29). Furthermore, studies indicating that NPY Y5-deficient mice have a reduced hyperphagic response to ICV NPY (29) have now been contradicted (55), and such knockout studies equally indicate that NPY Y1 (or other NPY receptor) is a likely receptor subtype mediating the effects of NPY on food intake. Indeed, a recent study of the effects of a selective NPY Y1 antagonist in NPY-treated wild-type, NPY Y1-deficient, and NPY Y5-deficient mice provides compelling evidence for the involvement of NPY Y1 in NPY-induced feeding (55). A number of other groups have shown that selective NPY Y1 antagonism inhibits NPY-induced feeding in rats (12,13,38, 48).

Little is known about the function of NPY Y5 in rodent brain other than its putative role in feeding and its possible mediation of the antiseizure activity of NPY (56). The development of highly potent and selective agonists and antagonists for NPY Y5 should now permit a more detailed investigation of the biology of this receptor.

The technical assistance of Gunn-Britt Forsberg, Britt-Marie Fihn, and Bill Brown and critical evaluation of the manuscript by Drs. David Morgan, David Smith, Lynn Pritchard, and Carl Montague are gratefully acknowledged.