|Home | About | Journals | Submit | Contact Us | Français|
G protein-coupled P2Y receptors (P2Y-R) are activated by adenine and uracil nucleotides. The P2Y14 receptor (P2Y14-R) is activated by at least four naturally occurring UDP sugars, with UDP-glucose (UDP-Glc) being the most potent agonist. With the goal of identifying a competitive antagonist for the P2Y14-R, UDP was examined for antagonist activity in COS-7 cells transiently expressing the human P2Y14-R and a chimeric Gα protein that couples Gi-coupled receptors to stimulation of phosphoinositide hydrolysis. UDP antagonized the agonist action of UDP-Glc, and Schild analysis confirmed that the antagonism was competitive (pKB = 7.28). Uridine 5′-O-thiodiphosphate also antagonized the human P2Y14-R (hP2Y14-R) with an apparent affinity similar to that of UDP. In contrast, no antagonist activity was observed with ADP, CDP, or GDP, and other uracil analogs also failed to exhibit antagonist activity. The antagonist activity of UDP was not observed at other hP2Y-R. In contrast to its antagonist action at the hP2Y14-R, UDP was a potent agonist (EC50 = 0.35 μM) at the rat P2Y14-R. These results identify the first competitive antagonist of the P2Y14-R and demonstrate pharmacological differences between receptor orthologs.
P2Y receptors (P2Y-R) are members of the superfamily of G protein-coupled receptors (GPCRs) and are activated by adenine and uridine nucleotides and nucleotide sugars. Atleast eight receptors comprise the P2Y-R family. P2Y1, P2Y2, P2Y4, and P2Y6 receptors are coupled to Gq and activate phospholipase C, whereas P2Y12, P2Y13, and P2Y14 receptors couple to Gi, leading to the inhibition of adenylyl cyclase and activation of ion channels (Burnstock, 2006). The P2Y11-R uniquely couples both to Gq to activate phospholipase C and to Gs to stimulate adenylyl cyclase (Communi et al., 1997; Qi et al., 2001).
The human P2Y14-R (hP2Y14-R) was identified as the eighth legitimate member of the P2Y receptor family (Chambers et al., 2000). UDP-glucose (UDP-Glc) was proposed to be the endogenous agonist for the P2Y14 receptor (P2Y14-R), with UDP-galactose, UDP-glucuronic acid, and UDP-N-acetylglucosamine acting as less potent P2Y14-R agonists. P2Y14-R mRNA was detected in a broad range of human tissues including placenta, stomach, intestine, adipose, brain, lung, spleen, and heart, and it was also found in specialized cells such as circulating neutrophils (Chambers et al., 2000; Scrivens and Dickenson, 2006). The rat and mouse P2Y14-R exhibit 80 and 83% amino acid identity to the human receptor, and both rodent orthologs are activated by known agonists with a similar profile to that of the hP2Y14-R (Freeman et al., 2001).
Cellular UDP-Glc is released in a constitutive manner into the medium of a broad range of cell types. Although most extracellular nucleotides are metabolized quickly, UDP-Glc accumulates in the medium of several cell lines (Lazarowski et al., 2003). The mechanisms of UDP-Glc release and extracellular metabolism remain unclear. Observations of UDP-Glc-promoted signaling were reported in multiple types of immune cells (Fumagalli et al., 2003; Skelton et al., 2003; Müller et al., 2005; Scrivens and Dickenson, 2005), suggesting that the P2Y14-R may have a yet-to-be-defined role in the regulation of immune system homeostasis.
Characterization of the P2Y14-R has been slowed by the lack of a selective competitive antagonist. We have identified and developed novel, selective ligands for several P2Y receptors that have proven useful for pharmacological resolution of molecularly defined P2Y-R in cells and tissues (Boyer et al., 1996; Houston et al., 2006, 2008; Jacobson et al., 2006). Accordingly, we are interested in identifying a selective antagonist for the P2Y14-R.
Ault and Broach (2006) recently used a yeast model system in which various nucleotides and nucleotide sugars were examined for their ability to stimulate growth of mutant P2Y14-R-expressing yeast cells in studies focused on the identification of mutant P2Y14 receptors with differential agonist sensitivities. Studies performed using one of these mutant receptors revealed that UDP antagonized UDP-Glc-promoted receptor activation in a concentration-dependent manner (Ault and Broach, 2006). We hypothesized that UDP acts as a competitive antagonist at the wild-type P2Y14-R and, therefore, used a transfected COS-7 cell system to investigate UDP activity at the human and rat P2Y14-R. In this study, we show that UDP is a selective and competitive antagonist of the hP2Y14-R. Thus, signals emanating from extracellular UDP apparently occur as a consequence of activation of the P2Y6-R as well as through antagonism of the P2Y14-R. We were surprised to find that UDP is a potent full agonist at the rat P2Y14-R.
UDP-Glc, UDP, ADP, CDP, GDP, and 2-methyl-thio-ADP were purchased from Sigma-Aldrich (St. Louis, MO). Diuridine 5′,5′-triphosphate (UP3U) was synthesized according to the methods detailed in Pendergast et al. (2001). The source of diuridine 5′,5′-tetraphosphate (UP4U) was as previously reported (Ivanov et al., 2007). ATP and UTP were purchased from GE Healthcare (Piscataway, NJ). [32P]PPi was synthesized as described previously (Lazarowski et al., 2003). Uridine 5′-O-thiodiphosphate (UDPβS) as well as a mammalian expression vector for ecto-nucleotide pyrophosphatase/phosphodiesterase-1 (ENPP1) were generous gifts from Dr. José Boyer of Inspire Pharmaceuticals (Durham, NC).
COS-7 cells were grown on 12-well culture dishes and maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum and 4 mM L-glutamine at 37°C in a 10% CO2 environment. FuGENE 6 (Roche Applied Science, Indianapolis, IN) was used as the transfection reagent following the manufacturer’s protocol. Cells were transfected 48 h before assay with pcDNA3.1 expression vectors encoding either the human or rat P2Y14 receptor with an N-terminal hemagglutinin epitope tag. The expression vector for the hP2Y14-R was obtained as reported previously (Lazarowski et al., 2003). Transfections also included pcDNA3.1-Gαq/i, a vector that directs expression of a chimera of Gαq containing the last five amino acids of Gαi. This chimeric G protein promotes activation of phospholipase C through Gαi-coupled receptors (Coward et al., 1999). The levels of basal inositol phosphates increased markedly in COS-7 cells upon expression of human or rat P2Y14-R and Gαq/i. Because we previously illustrated that UDP-sugars are basally released by various cell types (Lazarowski et al., 2003), in some experiments pcDNA3.1-expressing ENPP1 was cotransfected with the goal of hydrolyzing extracellular nucleotide sugars and potentially decreasing inositol phosphate accumulation in the absence of added P2Y14-R agonists.
Cells were labeled 8 to 18 h before assay with 0.5 to 3 μCi/well myo-[3H]inositol (American Radiolabeled Chemicals, St. Louis, MO) in inositol-free and serum-free DMEM. Assays were initiated with the addition of 10 mM LiCl with or without drugs, and incubations continued for 45 min at 37°C. Reactions were stopped by aspiration of medium and addition of ice-cold 50 mM formic acid. After neutralization with 150 mM ammonium hydroxide, [3H]inositol phosphates were isolated by Dowex column chromatography as described previously (Nakahata and Harden, 1987). Stable cell lines for P2Y1, P2Y2, P2Y4, P2Y6, or P2Y11 receptors were generated in 1321N1 human astrocytoma cells as described previously (Nicholas et al., 1996; Kennedy et al., 2000). Experiments testing the potential activity of UDP at the P2Y2-R and P2Y4-R included hexokinase to eliminate contamination of UTP as previously described in Nicholas et al. (1996). In brief, UDP was treated with 10 U/ml hexokinase in the presence of 22 mM glucose for 1 h at 37°C, and 1 U/ml hexokinase was included in the assay buffer for the duration of the incubation. Variability in cpm of [3H]-inositol phosphate accumulation across experiments occurred due to differences in the amount of [3H]inositol used for labeling and/or the duration of the prelabeling period.
Quantification of UDP-Glc was performed as described previously (Lazarowski et al., 2003). In brief, incubations were in a final volume of 150 μl containing known or unknown amounts of UDP-Glc, 25 mM HEPES, pH 7.4, 0.5 U/ml UDP-glucose pyrophosphorylase from baker’s yeast (Sigma-Aldrich), and 100 nM [32P]PPi (200,000 cpm). Incubations were terminated by addition of 0.3 mM PPi and immediate heating of samples for 2 min at 95°C, and formation of [32P]UTP was quantified by high-performance liquid chromatography as described previously (Lazarowski et al., 2003).
Cells were seeded in 12-well plates at 5 × 104 cells/well 3 days before assay and transfected with mammalian expression vectors as described above. Cells were fixed with 0.4 ml of 4% paraformaldehyde for 30 min at room temperature, washed twice with 1 ml of Hanks’ balanced salt solution plus Ca2+/Mg2+, and incubated for 30 min at room temperature with 0.4 ml of DMEM plus 50 mM HEPES, pH 7.1, and 10% fetal bovine serum. Cells were incubated with mouse HA.11 monoclonal antibody at a 1:1000 dilution in 0.4 ml of medium for 1 h at room temperature. After two washes with 1 ml of Hanks’ balanced salt solution plus Ca2+/Mg2+, cells were incubated with [125I]rabbit anti-mouse IgG antibody diluted to 1:500 in 0.4 ml of medium for 2 h at room temperature. After another series of washing steps, cells were solubilized in 0.4 ml of 1 M NaOH overnight and transferred to glass tubes for quantification of radioactivity in a gamma counter.
The rat homolog of the P2Y14-R (rP2Y14-R) was amplified from rat genomic DNA using Pfu polymerase (Stratagene, Carlsbad, CA) with the following primers: (5′-GAGACGCGTCCGACAACACAACAACCACAGAAC-3′) and (5′-AGACTCGAGTTACAAAGTATCTGTGCTTTCC-3′). The primers contained either a MluI (upstream primer) or a XhoI (downstream primer) restriction site, respectively (sites are underlined), to facilitate cloning. The amplification conditions were as follows: 94°C for 5 min; 35 cycles of 94°C for 45 s, 55°C for 45 s, and 72°C for 90 s; and a final extension for 4 min at 72°C. The amplified rP2Y14-R fragment was digested with MluI and XhoI, purified, and ligated into a similarly digested, modified pcDNA3 expression vector, which fuses a hemagglutinin epitope to Asp-2 at the N terminus of the receptor. An individual clone encoding the receptor was sequenced and found to be identical to the published sequence (Freeman et al., 2001).
The results from each experiment are expressed as the mean ± S.E. from triplicate samples and were analyzed using Prism 4.0 software (GraphPad Software Inc., San Diego, CA). All experiments were repeated at least three times with similar results. Schild analysis was performed using EC50 values from the concentration-effect curves for UDP-Glc generated in the absence and presence of increasing concentrations of UDP. The pKB was calculated using the equation: log ([A′]/[A] − 1) = log [B] − log pKB (Arunlakshana and Schild, 1959), where [A′] is the concentration of UDP-Glc necessary to produce 50% of the maximal effect in the presence of antagonist [B], and [A] is the concentration of agonist necessary to produce 50% of the maximal effect in the absence of antagonist. Drug response data presented in Figs. 3A, 5, A–D, and and7B7B are normalized as a percentage of the response observed with a maximally effective concentration (usually 10 μM) of UDP-Glc. Statistics were carried out using the Student’s t test.
To assay the functional activity of the hP2Y14-R, we used COS-7 cells transiently coexpressing the hP2Y14-R and a Gαq/i chimera. The Gαq/i chimera is a Gαq protein in which the last five amino acids at the C terminus have been substituted with those of Gαi. This chimeric Gα subunit is activated by GPCRs that couple to the Gαi family of G proteins and signal through downstream Gαq effectors such as phospholipase C (Coward et al., 1999). Expression of the hP2Y14-R or the Gαq/i chimera alone in COS-7 cells resulted in levels of [3H]inositol phosphate accumulation similar to that observed in cells expressing empty vector alone. [3H]Inositol phosphate accumulation was not changed by the addition of 100 μM UDP-Glc to cells expressing empty vector, the hP2Y14-R, or Gαq/i (Fig. 1). Coexpression of the hP2Y14-R and Gαq/i resulted in increased [3H]inositol phosphate accumulation in the presence of buffer alone, as we previously reported (Lazarowski et al., 2003). Addition of 100 μM UDP-Glc to cells coexpressing the hP2Y14-R and Gαq/i resulted in a 2-fold increase in inositol phosphate accumulation, consistent with agonist-promoted activation of phospholipase C (Fig. 1).
To investigate the nature of the hP2Y14-R-dependent [3H]inositol phosphate accumulation in the absence of added agonist, we coexpressed ENPP1 with the hP2Y14-R and Gαq/i in COS-7 cells with the goal of removing any released nucleotide/nucleotide sugar potentially present in the medium. UDP-Glc levels (3.6 ± 1.9 nM) in the bulk medium from COS-7 cells expressing hP2Y14-R and Gαq/i were similar to that of control cells (3.3 ± 0.3 nM). In contrast, coexpression of ENPP1 in cells also expressing hP2Y14-R and Gαq/i resulted in an approximately 78% reduction in UDP-Glc levels (0.8 ± 0.5 nM) compared with control cells. Although expression of ENPP1 alone had no effect on basal [3H]inositol phosphate accumulation, expression of ENPP1 with hP2Y14-R and Gαq/i resulted in an approximately 40% decrease (p < 0.01) in basal [3H]inositol phosphate levels compared with cells expressing receptor and G protein alone (Fig. 2A). Expression of ENPP1 had no effect on surface expression of the hP2Y14-R as quantified with an immunoassay (data not shown; see Materials and Methods), and it did not notably change the concentration-effect curve for added UDP-Glc (Fig. 2B). Thus, we conclude that the hP2Y14-R-dependent elevation of [3H]inositol phosphate levels in the absence of added agonist occurs largely because of autocrine/paracrine release of the P2Y14-R agonist. However, these results do not entirely rule out the possibility that the overexpressed P2Y14-R exhibits constitutive activity in this test system.
Four UDP sugars were identified as agonists at the hP2Y14-R, and neither UTP nor UDP exhibited agonist activity (Chambers et al., 2000). To determine whether UDP is an antagonist at the wild-type hP2Y14-R, we generated a series of concentration-effect curves for UDP-Glc-promoted stimulation of phospholipase C in the presence of increasing concentrations of UDP (Fig. 3A). UDP caused a parallel rightward shift of the UDP-Glc concentration-effect curve, and Schild analysis (Fig. 3B) confirmed that the antagonism produced by UDP was competitive (slope = 1.15 ± 0.06, n = 3). The pKB of UDP for antagonism of the hP2Y14-R was 7.28 ± 0.04.
We also assessed whether UDP exhibited antagonist activity at other P2Y-R stably expressed in 1321N1 human astrocytoma cells. P2Y1-R was maximally activated by 1 μM 2MeSADP (Fig. 4A), P2Y2-R and P2Y4-R were each activated by 3 μM UTP (Fig. 4, B and C), and P2Y11-R was activated by 100 μM ATP (Fig. 4D). Although 10 μM UDP completely blocked a near-maximal concentration of UDP-Glc at the hP2Y14-R (Fig. 3), UDP had no effect at any other P2Y receptors tested. Thus, we conclude that UDP is a selective antagonist at the hP2Y14-R.
To determine whether the antagonist effect of UDP at the hP2Y14-R is specific to the uracil structure, we also tested other nucleotides and nucleotide derivatives as antagonists at the hP2Y14-R. In contrast to the action of UDP, other nucleoside diphosphates including ADP, CDP, and GDP, at concentrations of 10 or 100 μM, did not inhibit UDP-Glc (3 μM)-promoted [3H]inositol phosphate formation (Fig. 5). We also tested whether other uridine-based molecules would antagonize activation of the hP2Y14-R by UDP-Glc. Neither UTP, UP3U, nor UP4U inhibited UDP-Glc activation of the receptor, although each of these nucleotides, when tested alone, produced a stimulatory effect in untransfected COS-7 cells (data not shown). The discovery of antagonist activity of UDP at the hP2Y14-R is also supported by the observation that the UDP analog, UDPβS, inhibited activation of this receptor by UDP-Glc (Fig. 6). The IC50 observed for UDPβS was similar to that determined for UDP under the same assay conditions. From these results, we conclude that UDP seems to be unique among naturally occurring nucleotides in its capacity to inhibit UDP-Glc-dependent activation of the hP2Y14-R.
Because pharmacological studies are often carried out with rat or mouse tissues, it is important to assess whether receptor orthologs exhibit pharmacological selectivities similar to those of human P2Y-R. Based on the precedent that ATP acts as an antagonist at the human P2Y4-R but is an agonist at the rat P2Y4-R (Bogdanov et al., 1998; Kennedy et al., 2000), we compared the action of UDP at the rP2Y14-R with its action at the hP2Y14-R. The rat P2Y14-R, which exhibits approximately 80% amino acid sequence identity to the hP2Y14-R and almost 90% identity in the transmembrane regions alone, was reported to display a similar UDP-sugar selectivity to that of the hP2Y14-R (Freeman et al., 2001), but the actions of other uridine nucleotides on the rP2Y14-R have not been reported.
Expression of either the rP2Y14-R or Gαq/i in COS-7 cells had no effect on [3H]inositol phosphate accumulation compared with untransfected cells, but coexpression of receptor and Gαq/i resulted in markedly increased basal accumulation (Fig. 7A). Consistent with other reports (Freeman et al., 2001), UDP-Glc was a potent agonist at the rP2Y14-R (Fig. 7B). Whereas UDP had no effect on inositol phosphate accumulation in wild-type COS-7 cells, in cells expressing Gαq/i alone, or in cells expressing the rP2Y14-R alone (data not shown), concentration of UDP-dependent increases in formation of [3H]inositol phosphates occurred in COS-7 cells coexpressing the rat P2Y14-R with Gαq/i (Fig. 7B). The maximal stimulatory effect observed with UDP was similar to that observed with UDP-Glc, as were the EC50 values of UDP (0.35 ± 0.17 μM) and UDP-Glc (EC50 = 0.28 ± μM). No additivity was observed between UDP and UDP-Glc on rP2Y14-R-promoted [3H]inositol phosphate formation (Fig. 7C).
In this study, we show that UDP is a competitive antagonist at the human P2Y14-R, and this action is receptor-selective because UDP does not inhibit agonist-promoted activation of other hP2Y receptors. Moreover, the activity of UDP at the P2Y14-R is species-dependent because we observed that UDP is a potent, and apparently full, agonist at the rat P2Y14-R.
Chambers et al. (2000) reported in their initial study of the hP2Y14-R that UDP has no agonist activity, and we observed similar results in the studies reported here. Using a reporter system in yeast, Ault and Broach (2006) generated a mutant hP2Y14-R displaying a mutation in intracellular loop 1 and various mutations in several of the transmembrane regions. This mutant, selected for its ability to support growth of yeast at lower concentrations of UDP-Glc than the wild-type receptor, exhibited an enhanced UDP-Glc-stimulated response that was inhibited by UDP, and a KB in the micromolar range was reported. The >20-fold higher potency of UDP observed in our studies probably reflects large differences in the assay systems used. For example, whereas incubations with nucleotide were continued for minutes in the current study, they were continued for hours in assays measuring P2Y14-R-mediated regulation of growth of yeast. Our results illustrate that UDP is a potent, competitive antagonist of the wild-type hP2Y14-R.
Demonstration of antagonist action at the hP2Y14-R suggests that UDP may have broader physiological importance as an extracellular signaling molecule than has been previously appreciated. Both UDP and UDP-glucose are known to be released from cells, although the mechanisms of their release remain unclear. UDP is the most potent and selective agonist of the hP2Y6-R (Lazarowski and Harden, 1994; Communi et al., 1996), and physiological responses attributed to UDP-initiated P2Y6-R-promoted signaling include modulation of interleukin-8 production in monocytes (Warny et al., 2001) and human mature dendritic cells (Idzko et al., 2004). In addition, UDP was observed to induce a positive inotropic effect in mouse cardiomyocytes (Wihlborg et al., 2006) and to promote ion transport in human placental cytotrophoblast cells (Roberts et al., 2006). Our data indicate that potential contributions of the P2Y14-R to responses associated with UDP must be considered. Whereas reported distribution of P2Y6-R mRNA overlaps with that of P2Y14-R mRNA in many cells and tissues, such as lung, heart, placenta, and neutrophils (Communi et al., 1996; Moore et al., 2001), it remains unclear whether the two receptor types are coexpressed in the same cells or in different cells that share the same extracellular space.
The actions of extracellular neurotransmitters and hormones are highly regulated by their release, metabolism, and reuptake. The possibility that direct antagonism of GPCR activation occurs by extracellular signaling molecules has been suggested by observations that ATP is a competitive antagonist of the hP2Y12-R (Cusack and Hourani, 1982; Bodor et al., 2003) and of the hP2Y4-R (Bogdanov et al., 1998; Kennedy et al., 2000). The physiological relevance of UDP antagonism at the hP2Y14-R will be important to investigate, as will the idea that UDP simultaneously activates the hP2Y6-R while inhibiting the hP2Y14-R.
Our finding that UDP has agonist activity at the rP2Y14-R was surprising. Study of the rat and human receptors under identical conditions rules out trivial explanations for this observation. These data do not unambiguously rule out the possibility that UDP is a partial agonist at the human receptor under some conditions. However, the agonist versus antagonist action of the nucleotide at the rat versus human P2Y14-R has been observed over a broad range of expression levels of these two receptors. Our activity data suggest that the binding affinity of UDP for the human and rat receptors are in fact quite similar, but development of a radioligand binding assay will be necessary to fully assess this assertion. The fact that the maximal agonist activity of UDP was similar to that of UDP-Glc over a broad range of P2Y14-R expression levels (data not shown) suggests that the intrinsic efficacy of UDP at the rat P2Y14-R is similar to that of UDP-Glc.
The differential activity of UDP observed between the rat and human orthologs of the P2Y14-R shares similarities to the actions of ATP at the P2Y4-R. Kennedy et al. (2000) compared the ligand selectivities of the rat and human P2Y4 receptors under conditions that minimize the effects of released nucleotides and of extracellular bioconversion of nucleotides, and they observed that ATP is an agonist at the rat P2Y4-R and a competitive antagonist at the human P2Y4-R. Residues in the second extracellular loop of the P2Y4 receptor are the primary determinants for the agonist versus antagonist activity of ATP between the two species orthologs (Herold et al., 2004). Like the P2Y4-R, the P2Y14-R shares approximately 80% amino acid sequence identity between the rat and human orthologs, with 90% identity when analysis is restricted to transmembrane regions only. Recent work by Ko et al. (2007) defines the human P2Y14 receptor through structure-activity studies in conjunction with molecular modeling studies. Residues in the second extracellular loop of the hP2Y14-R are predicted to interact with the diphosphate moiety and hydroxyl groups on the hexose ring of UDP-Glc. Comparative modeling of the rat P2Y14-R and receptor mutagenesis directed from these predictions may identify key domains responsible for agonist efficacy at the hP2Y14-R.
The mouse P2Y14-R has been cloned and is reported to be activated by the same UDP-sugar agonists as the human and rat receptors (Freeman et al., 2001). The rat and mouse P2Y14-R share 89% overall amino acid sequence identity and are essentially identical in their transmembrane-spanning domains and the second extracellular loop. Thus, we anticipate that the agonist action of UDP observed with the rat receptor will be similarly observed at the mouse P2Y14-R.
The finding that UDP acts as a competitive antagonist at the hP2Y14-R provides an excellent template for rational synthesis of antagonist analogs that exhibit high affinity at the hP2Y14-R. Our structure-activity studies of UDP analogs at the hP2Y6-R (Besada et al., 2006) also provide potential avenues for development of P2Y14-R antagonists that do not act as ligands for the P2Y6-R. Synthesis of a hydrolysis-resistant competitive antagonist for the P2Y14-R is an obvious goal, as is a high-affinity radiolabeled antagonist.
UDP-glucose, UDP-galactose, UDP-glucuronic acid, and UDP-N-acetylglucosamine were previously identified as full or partial agonists at the human and rodent P2Y14-R (Chambers et al., 2000; Freeman et al., 2001). The pharmacological selectivity for the P2Y14-R is now broadened with the finding that UDP also acts at this receptor. The identification of UDP as a competitive antagonist for the hP2Y14-R provides new insight into the physiological regulation of this receptor, and it should be of pharmacological importance in delineating the functional roles subserved by this signaling protein.
We appreciate the excellent technical assistance of Sonia de Castro for the synthesis of UP3U.
This work was supported by National Institutes of Health Grants GM38213, P01-HL034322, and the National Institutes of Health Intramural Research Program of National Institute of Diabetes and Digestive and Kidney Diseases.
U.S. Government work not protected by U.S. copyright