Regulatory Role of CD38 (ADP-ribosyl Cyclase/Cyclic ADP-ribose Hydrolase) in Insulin Secretion by Glucose in Pancreatic (cid:98) Cells ENHANCED INSULIN SECRETION IN CD38-EXPRESSING TRANSGENIC MICE*

Cyclic ADP-ribose (cADPR) serves as a second mes- senger for Ca 2 (cid:49) mobilization in insulin secretion, and CD38 has both ADP-ribosyl cyclase and cADPR hydro- lase activities (Takasawa, J. Biol. Chem. 268, 26052–26054). Here, we produced transgenic mice overexpressing hu- man CD38 in pancreatic (cid:98) cells. The enzymatic activity of CD38 in transgenic islets was greatly increased, and ATP efficiently inhibited the cADPR hydrolase activity. The Ca 2 (cid:49) mobilizing activity of cell extracts from transgenic islets incubated in high glucose was 3-fold higher than that of the control, suggesting that ATP produced by glucose metabolism increased cADPR accumulation in transgenic islets. Glucose- and ketoisocaproate-in-duced but not tolbutamide- nor KCl-induced insulin se- cretions from transgenic islets were 1.7–2.3-fold

Cyclic ADP-ribose (cADPR) 1 (1) induces the release of Ca 2ϩ from microsomes of pancreatic islets (2,3) and from a variety of other cells (4 -10). Glucose raises the cADPR concentration in islets, and cADPR induces insulin secretion from digitoninpermeabilized islets in vitro (3). We have therefore suggested that cADPR plays a second messenger role in Ca 2ϩ mobilization for insulin secretion (2,3,11). Human lymphocyte antigen CD38 (12,13) has been shown to have both ADP-ribosyl cyclase and cADPR hydrolase activities (11,14). CD38 was found to be expressed in a variety of tissues and cells including pancreatic islets (11,15). We have shown that ATP, which is generated during glucose metabolism in islets, inhibits the cADPR hydrolase activity of CD38, thereby increasing the accumulation of cADPR because of decreased destruction (11). CD38 is thus thought to play a central role in glucose-induced insulin secretion in islets. In the present study, we produced transgenic mice overexpressing CD38 in islets and analyzed the subcellular localization of expressed CD38, changes in Ca 2ϩ mobilizing activity, and glucose-induced insulin secretion.

Construction of Rat Insulin II Promoter/Human CD38 Hybrid
Gene-A rat insulin II promoter previously reported to be active in pancreatic ␤ cells of transgenic mice (16,17) was employed. The 0.7kilobase pair BamHI-XmaI fragment (17) of the rat insulin II promoter (nucleotides Ϫ695 to ϩ22 in Ref. 18), the 0.9-kilobase pair XmaI-SalI fragment of the human CD38 cDNA (11,13) (nucleotides ϩ58 to ϩ980 in Ref. 13; XmaI and SalI sites introduced by polymerase chain reaction), and the 1.6-kilobase pair SalI-EcoRI fragment of the SV40 intron and polyadenylation signal (19) (SalI site derived from plasmid sequence) were ligated at the XmaI and SalI sites in the correct orientation. The resultant hybrid gene (Ins-CD38; 3.2 kilobase pairs) was separated from the plasmid vector pBlueScript SKϪ (Stratagene) by KpnI and NotI and was microinjected into fertilized eggs as described (17).
Northern Blot Analyses-Northern blot analyses were carried out on total RNA extracted from various tissues as described (17) using a 32 P-labeled human CD38 cDNA probe (EcoRI-EcoRI fragment of 528 base pairs; nucleotides ϩ29 to ϩ556 in Ref. 13). Hybridization signals were scanned with a bioimage analyzer, BAS 2000 (Fuji Photo Co., Ltd., Tokyo, Japan).
Subcellular Fractionation and Measurement of NAD ϩ Glycohydrolase Activity-The subcellular fractionation was performed according to the islet fractionation method of McDaniel et al. (22). One thousand islets were homogenized in 400 l of homogenizing buffer (50 mM MES, 1 mM EDTA, and 0.25 M sucrose, pH 7.2). The homogenate was centrifuged at 600 ϫ g for 5 min to yield a pellet containing nuclei (nuclear fraction). Centrifugation of the supernatant at 20,000 ϫ g for 20 min yielded a pellet containing the plasma membrane, secretory granules, and mitochondria (membrane fraction). Centrifugation of the resultant supernatant at 150,000 ϫ g for 90 min yielded a pellet containing the microsome (microsome fraction). The supernatant after 150,000 ϫ g centrifugation yielded a soluble protein (cytosol fraction). NAD ϩ glycohydrolase activity in the fractionated protein was measured as described (23). Briefly, 1 g of each fractionated protein was incubated with 0.2 mM NAD ϩ containing 50 nCi of [ 14 C]NAD ϩ (Amersham Corp.) at 37°C for 10 min. The mixture was applied on a column of Dowex-1 (Bio-Rad Laboratories). Nicotinamide was eluted with 20 mM Tris-HCl (pH 7.5), followed by scintillation counting.
Measurement of Insulin Secretion from Isolated Islets-Islets of transgenic mice (line 18) and control mice were isolated in parallel from 6 -10-week-old litters by collagenase digestion. Twenty islets were incubated for 1 h at 37°C in 1 ml of RPMI 1640 medium containing 10% fetal calf serum and various concentrations of glucose. The medium samples were subsequently assayed for radioimmunoassay of insulin using the insulin radioimmunoassay kit (Amersham Corp.) and rat insulin standards. For the time course experiment, 20 islets were incubated at 37°C in 1 ml of the medium containing 11.1 mM glucose, and medium samples (2 l) collected at 10, 20, 30, and 40 min after the incubation were subjected to radioimmunoassay for insulin. For measurements of insulin secretion by other insulin secretagogues, 10 islets were incubated for 1 h at 37°C in 0.5 ml of the medium containing the lowest concentration of glucose (2.5 mM) and then incubated for another 1 h in the same medium containing 10 mM ketoisocaproic acid (KIC) (Sigma), 0.2 mM tolbutamide (Sigma), or 25 mM KCl (Merck). The medium samples were subsequently radioimmunoassayed for insulin.
Assay of Ca 2ϩ Mobilizing Activity-Five hundred transgenic islets (line 18) or control islets were incubated at 37°C for 15 min in 5 ml of RPMI 1640 medium containing 10% fetal calf serum and 2.5 or 11.1 mM glucose. After the incubation, islet cell extracts (50 l) were prepared as described (3). Release of Ca 2ϩ was monitored by adding the islet extracts (15-20 l) to 3 ml of intracellular medium (3) containing 3 M Fluo 3, a fluorescent Ca 2ϩ indicator, and the rat cerebellum microsome fraction (88 g of protein) prepared as described (3). Assay of Ca 2ϩ mobilizing activity using the mouse islet microsome fraction (5 g of protein) was carried out with 0.6 ml of the intracellular medium (see Fig. 6). Fluorescence was measured at 490-nm excitation and 535-nm emission at 37°C.

Measurement of cADPR Hydrolase Activity in the Presence of Various
Concentrations of ATP-The islet cell homogenate (10 g of protein) of transgenic mice (line 18) was incubated for 20 min at 37°C in 0.1 ml of phosphate-buffered saline (pH 7.4) in the presence of 0 -6 mM ATP and 0.2 mM cADPR containing 5 Ci of [ 32 P]cADPR, prepared enzymatically from NAD ϩ and [ 32 P]NAD ϩ using Aplysia kurodai ADP-ribosyl cyclase. Reaction products were analyzed by HPLC (11,21).

Measurement of Serum Insulin and Blood Glucose Levels in Glucose Tolerance
Tests-Transgenic mice (lines 18 and 56) and their respective nontransgenic siblings were fasted 10 h and then subjected to glucose tolerance tests by an intraperitoneal injection of 1 g of glucose/kg of body weight. Blood samples (100 l) were taken from the tail vein at each point after glucose administration, and the serum samples (25 l) were prepared by centrifugation after incubating blood samples overnight at 4°C to complete coagulation. The serum insulin levels were determined by radioimmunoassay. Blood glucose determinations were made on fresh whole blood (15 l) using the Accucheck II (Boehringer Mannheim). All statistical analyses were performed using Student's t test.

RESULTS AND DISCUSSION
The rat insulin II promoter/human CD38 hybrid gene (Ins-CD38; see "Experimental Procedures") was designed to direct the overexpression of CD38 in pancreatic ␤ cells of transgenic mice. The linearized gene fragment was microinjected into the fertilized eggs of (C57Bl/6J ϫ CBA/J) F 1 mice. 20 out of 94 newborn mice were found to carry the Ins-CD38 transgene, as detected by polymerase chain reaction analyses using primers for the insulin promoter and human CD38 cDNA. In the present study, the six transgenic lines, 18, 30, 49, 56, 60, and 72, were maintained on ICR background and analyzed.
Northern blot analysis using the human CD38 cDNA probe (11) showed that all lines of transgenic mice but not that of the nontransgenic mice expressed human CD38 mRNA in the pancreatic islets (Fig. 1A). Densitometric scanning indicated that the transgenic lines 18, 56, and 60 expressed relatively higher levels of human CD38 mRNA in islets, whereas lines 30, 49, and 72 expressed lower levels of human CD38 mRNA. The human CD38 mRNA expression was not detected in other tissues such as brain, lung, heart, stomach, small intestine, liver, kidney, spleen, and testis in the transgenic mice (Fig. 1,  B and C), indicating that the expression of human CD38 is limited to islets. In immunohistochemistry, islets of the transgenic mice were densely stained for human CD38 (Fig. 2, B and  C). On the other hand, islets of the control mice showed no immunoreactivity for human CD38 (Fig. 2A). In contrast to islets, the pancreatic exocrine cells showed no detectable staining for human CD38 in any of the transgenic and nontransgenic mice.
Next, the pancreatic islet homogenates prepared from islets of transgenic and nontransgenic siblings of lines 18 and 56 were incubated with [ 32 P]NAD ϩ , and the reaction products were analyzed by HPLC. The formation of cADPR in the transgenic mice (for line 18, 1.5 nmol/min⅐mg protein; for line 56, 1.5 nmol/min⅐mg protein) was indeed much higher than in the controls (Ͻ0.05 nmol/min⅐mg protein). The formation of ADPR in the transgenic mice (for line 18, 72.8 nmol/min⅐mg protein; for line 56, 174.4 nmol/min⅐mg protein) was also much higher than in the controls (4.3 nmol/min⅐mg protein). CD38 exhibits both ADP-ribosyl cyclase and cADPR hydrolase activities, and the overall reaction is classified as an NAD ϩ glycohydrolase reaction (11,14). To determine the subcellular distribution of expressed CD38, islet proteins were fractionated by centrifugation (22) and fractionated proteins were assayed for NAD ϩ glycohydrolase activity (Table I). The NAD ϩ glycohydrolase activity in the membrane fraction was greatly increased in the transgenic mice, indicating that expressed CD38 was predominantly localized in this fraction. Significant activities were also detected in the nuclear, microsome and cytosol fractions of the transgenic islets. The distribution of the percentage of total activity in fractions of transgenic mice showed a similar tend- ency to that of the control.
We isolated islets from transgenic mouse line 18 and their nontransgenic litter mates and measured secreted insulin after incubation in medium containing various concentrations of glucose (Fig. 3A). At 6.8 -15.6 mM glucose, the transgenic insulin secretion was 1.7-2.3-fold higher than that of the control. Essentially similar results were obtained using transgenic line 56 (data not shown). Time course experiments at 11.1 mM glucose indicated that at 10 min after the exposure to glucose, the glucose-stimulated insulin secretion was significantly higher in the transgenic islets and progressively increased in a time-dependent manner (Fig. 3B).
We next investigated the effects of other insulin secretagogues on insulin secretion from transgenic and control islets. When the islets were exposed to 10 mM KIC, which, like glucose, generates ATP during the metabolism (24), the transgenic insulin secretion was 1.7-fold higher than that of the control (Fig. 4A). Tolbutamide blocks ATP-sensitive K ϩ channel and facilitates Ca 2ϩ influx through voltage-dependent Ca 2ϩ channels (24) without increasing the islet ATP concentration (25). When the islets were exposed to 0.2 mM tolbutamide, the transgenic insulin secretion was not altered as compared with the control (Fig. 4B). When the islets were exposed to 25 mM KCl, which directly induces cell membrane depolarization resulting in Ca 2ϩ influx (24), the transgenic insulin secretion was also not significantly higher than the control (Fig. 4C).
We prepared cell extracts from the islets incubated in 2.5 or 11.1 mM glucose. The extracts were assayed for the Ca 2ϩ mobilizing activity from microsomes (3). The Ca 2ϩ mobilizing activity of extracts of transgenic islets incubated in 11.1 mM glucose was 3-fold higher than that of the control extracts (Fig. 5). The Ca 2ϩ mobilization by the islet extracts was abolished when the microsomes had been desensitized by previously releasing Ca 2ϩ in response to authentic cADPR (3), indicating that Ca 2ϩ mobilization by the extracts of transgenic islets after high glucose treatment is cADPR-derived. In contrast, at 2.5 mM glucose, the Ca 2ϩ mobilizing activities of the transgenic and control islet extracts were lower, and there was no significant difference between the two extracts (Fig. 5). The effect of exogenous cADPR on Ca 2ϩ mobilization from control and transgenic islet microsomes was essentially similar (Fig. 6), suggesting that microsome sensitivity to cADPR is not altered by human CD38 overexpression.   We have previously shown that ATP, generated during glucose metabolism in islets, dose-dependently inhibits the cADPR hydrolyzing activity of CD38 expressed in COS-7 cells and increases the accumulation of cADPR (11). In fact, higher concentrations of ATP efficiently inhibited the cADPR hydrolase activity of the CD38 expressed in transgenic islets (Fig. 7). The cellular ATP concentration 2 in transgenic islets increased from 2.3 (at 2.5 mM glucose) to 3.3 mM (at 11.1 mM glucose); this rise in cellular ATP concentrations would inhibit the cADPR hydrolase activity of CD38 and thereby increase the cADPR concentration. It is therefore reasonable to assume that the expressed ADP-ribosyl cyclase/cADPR hydrolase (CD38) in transgenic islets generates the enhanced cADPR accumulation upon stim- ulation by glucose or KIC to increase insulin secretion via the cADPR-mediated intracellular Ca 2ϩ elevation.
Next, 10-h fasted mice were subjected to glucose tolerance tests by an intraperitoneal injection of glucose, and then serum insulin levels at each point after the glucose stimulation were determined. In transgenic mice (lines 18 and 56), glucoseinduced insulin increases were 2-fold higher (at 15 min) than those of the controls (Fig. 8, A and B). Determinations of blood glucose levels at each point in the glucose tolerance test showed that the three lines of transgenic mice (lines 18, 56, and 60), which expressed higher levels of human CD38 mRNA in the pancreatic islets (Fig. 1A), had lower glucose levels than the controls (Fig. 8, C-E). However, transgenic lines 30, 49, and 72, which expressed lower levels of human CD38 mRNA in islets, did not have significantly lower glucose levels after glucose administration than the controls. These results indicate that the expressed ADP-ribosyl cyclase/cADPR hydrolase (CD38) reproducibly and dose-dependently facilitates glucose-induced insulin secretion in vivo and, by doing so, reduces the blood glucose levels of the transgenic mice.
It should be noted that the production of human CD38 in islets did not appear to be deleterious to the health of the transgenic mice; fertility and body weight were indistinguishable from controls. The islets of transgenic mice at 1 year of age appeared morphologically normal and were well stained for insulin (not shown).
The results of the present study suggest that not only Ca 2ϩ from extracellular sources (Ca 2ϩ influx through voltage-dependent Ca 2ϩ channels evoked by glucose-induced cell membrane depolarization; Ref. 24) but also Ca 2ϩ released from intracellular stores (cADPR-induced Ca 2ϩ release from microsomes; Ref. 3) play important roles in regulating the glucoseinduced insulin secretion. In fact, intracellular Ca 2ϩ elevation in the absence of external Ca 2ϩ has been reported recently (26,27).
The present results also indicate that CD38 plays a regulatory role in the glucose-induced insulin secretion. In non-insulin-dependent diabetes mellitus, the glucose-induced insulin secretion is impaired (28) even when pancreatic islets retain significant amounts of insulin (29). Thus, it would be important to determine whether there are qualitative or quantitative differences in the CD38 (ADP-ribosyl cyclase/cADPR hydrolase) in non-insulin-dependent diabetes mellitus ␤ cells. FIG. 8. Serum insulin (A and B) and blood glucose (C-E) levels in glucosetolerance tests. A and B, transgenic mice of lines 18 and 56 and their respective nontransgenic siblings (control) were subjected to glucose-tolerance tests, and the serum insulin levels were determined as described under "Experimental Procedures." n ϭ 10 (lines 18 and 56) and 10 (respective control). C-E, in the glucose tolerance tests, blood glucose levels of lines 18, 56, and 60 and their respective nontransgenic siblings (control) were determined using whole blood taken at the indicated times from the tail vein. n ϭ 8 (lines 18, 56, and 60) and 8 (respective control). Vertical bars indicate S.E. *, p Ͻ 0.05.