Increased Glucose-induced Secretion of Glucagon-like Peptide-1 in Mice Lacking the Carcinoembryonic Antigen-related Cell Adhesion Molecule 2 (CEACAM2)*

Carcinoembryonic antigen-related cell adhesion molecule 2 (CEACAM2) regulates food intake as demonstrated by hyperphagia in mice with the Ceacam2 null mutation (Cc2−/−). This study investigated whether CEACAM2 also regulates insulin secretion. Ceacam2 deletion caused an increase in β-cell secretory function, as assessed by hyperglycemic clamp analysis, without affecting insulin response. Although CEACAM2 is expressed in pancreatic islets predominantly in non-β-cells, basal plasma levels of insulin, glucagon and somatostatin, islet areas, and glucose-induced insulin secretion in pooled Cc2−/− islets were all normal. Consistent with immunofluorescence analysis showing CEACAM2 expression in distal intestinal villi, Cc2−/− mice exhibited a higher release of oral glucose-mediated GLP-1, an incretin that potentiates insulin secretion in response to glucose. Compared with wild type, Cc2−/− mice also showed a higher insulin excursion during the oral glucose tolerance test. Pretreating with exendin(9–39), a GLP-1 receptor antagonist, suppressed the effect of Ceacam2 deletion on glucose-induced insulin secretion. Moreover, GLP-1 release into the medium of GLUTag enteroendocrine cells was increased with siRNA-mediated Ceacam2 down-regulation in parallel to an increase in Ca2+ entry through L-type voltage-dependent Ca2+ channels. Thus, CEACAM2 regulates insulin secretion, at least in part, by a GLP-1-mediated mechanism, independent of confounding metabolic factors.

CEACAM2 is highly homologous to CEACAM1, the protein product of a distinct gene. Both genes contain nine exons, the seventh of which undergoes alternative splicing to yield different transcripts that are distinguished by a long (-L) or a short (-S) intracellular tail, containing or devoid of conserved phosphorylation sites (5,7). In particular, in the case of CEACAM2, exons 3 and 4 commonly undergo alternative splicing to yield a shorter extracellular domain containing two instead of four immunoglobulin-like loops. Hence, CEACAM2 is mostly expressed as CEACAM2-2L/2S. Although CEACAM2 shares some functions with CEACAM1, such as platelet activation/adhesion (3), it does not appear to directly act as a cell adhesion molecule (5). Moreover, the absence of its transcripts in hepatocytes has not warranted the investigation of its role in hepatic insulin clearance, as was the case for CEACAM1 (8 -10).

Experimental Procedures
Mouse Generation-The generation of Cc2 Ϫ/Ϫ mice was described previously (1,6). Unless otherwise mentioned, male mice (5-7 months of age) were used. All animals were housed in a 12-h dark-light cycle and fed standard chow (Harlan Teklad 2016; Harlan, Haslett, MI) ad libitum at the Division of Laboratory Animal Resources at the University of Toledo College of Medicine. All procedures and animal experiments were approved by the Institutional Animal Care and Utilization Committee.
Glucose Tolerance Test-Male mice (n Ͼ5/genotype, 5 months old) were fasted overnight from 17:00 to 08:00 the following day before being anesthetized and administered glucose via oral gavage (3.0 g/kg BWT). Blood was drawn from the retro-orbital sinus to measure blood glucose and plasma insulin levels at 0 -120 min post-glucose administration. In parallel experiments, anesthetized mice were injected intraperitoneally with 5 g of the GLP-1 receptor antagonist, exendin (9 -39) (American Peptide Co. Inc., Sunnyvale, CA), 20 min before glucose administration.
Hyperglycemic Clamp Analysis in Awake Mice-To assess ␤-cell function in vivo, a 2-h hyperglycemic clamp was performed as described previously (18). Male mice (n ϭ 7/genotype, 7 months of age) were anesthetized with an intraperitoneal injection of ketamine (100 mg/kg BWT) and xylazine (10 mg/kg BWT), and an indwelling catheter was inserted into the right internal jugular vein. Four-to-five days later, mice were fasted overnight before being subjected to hyperglycemic clamp analysis with a continuous infusion of 20% glucose to raise and maintain plasma glucose levels at ϳ300 mg/dl for 2 h. Blood samples (40 l) were collected at 15-20-min intervals over a period of 120 min to measure plasma glucose and insulin concentrations.
Plasma GLP-1 Measurement in Response to Oral Glucose-Mice (n Ն6/genotype) were fasted overnight, subjected to an oral glucose administration (3 g/kg BWT), and anesthetized with pentobarbital immediately after glucose administration before their retro-orbital blood was drawn 30 min later to measure blood glucose and plasma insulin levels. For GLP-1 measurement, plasma from each genotype was pooled to determine GLP-1 levels in triplicate in 300-l aliquots, as above.
Fluorescence-activated Cell Sorter Purification of Isolated Islets-Islets were isolated by the intraductal collagenase digestion method (20). After PBS wash, the suspension was passed through a 35-m filter before fluorescence-activated cell sorter (FACS) analysis, based on autofluorescence and size (21). Cells were sorted directly into TRIzol reagent, and the purity of the sorted fractions was determined by real time PCR for insulin and glucagon in each fraction.
Insulin Secretion from Isolated Islets-Purification of islets was conducted on pancreases of 4-month-old mice by collagenase digestion (22). Islets were then resuspended in RPMI 1640 medium containing 10% newborn calf serum and 5.5 mM glucose and cultured overnight at 37°C. Islets were stimulated with glucose (2.8 -16.8 mM) or 20 mM KCl for 1 h at 37°C, followed by centrifugation to assay insulin content by radioimmunoassay in the supernatant. Islets were dissolved in high salt buffer and sonicated three times at 80 watts for 10 s to measure DNA concentration for normalization.
Live Cell Fluorescence and Intracellular Ca 2ϩ Measurements-GLUTag cells were plated on Matrigel-coated coverslips (Corning Glass) 1 day prior to undergoing transfection with scrambled or Ceacam2-specific siRNA oligos, as above. Two to 3 days later, cells were loaded with 2 M Fura-2 AM in physiological saline solution (147 mM NaCl, 5 mM KCl, 2.2 mM CaCl 2 , 1 mM MgCl 2 , 10 mM HEPES, and 1 mM D-glucose) for 1 h at room temperature and washed. Coverslips were mounted, and intracellular Ca 2ϩ measurements were obtained using a Nikon TE2000 microscope.
Live cell imaging was performed using a Polychrome IV monochromator-based high speed digital imaging system (TILL Photonics, Gräfelfing, Germany) ported to a fiber optic guide and epifluorescence condenser and coupled to a Nikon TE2000 microscope. Cytosolic Ca 2ϩ concentration dynamics in Fura-2 AM-loaded GLUTag cells was recorded by alternately illuminating cells with 340 and 380 nm wavelength light focused onto the image plane via a DM400 dichroic mirror and Nikon Super-Fluor X40 oil immersion objective, and fluorescence was obtained through a 525 Ϯ 25-nm bandpass filter (Chroma Technologies, Brattleboro, VT). Acquired images were analyzed using TILLvisION (TILL Photonics) and ImageJ (W. S. Rasband, National Institutes of Health, Bethesda).
Western Blot Analysis-Tissues were lysed in 150 mM NaCl, 50 mM HEPES, pH 7.6, containing protease and phosphatase inhibitor, and the protein concentration was determined by BCA protein assay (Pierce) prior to analysis by 7% SDS-PAGE and immunoprobing with specific antibodies. These include polyclonal antibodies against mouse CEACAM1 (␣-mCC1) (Ab-231) (9) and mouse CEACAM2 (␣-mCC2) (above), in addition to monoclonal antibodies against tubulin (Sigma) and GAPDH (Santa Cruz Biotechnology, Inc., Dallas, TX) to normalize for the amount of proteins analyzed. Blots were incubated with horseradish peroxidase conjugated with anti-rabbit and anti-mouse IgG (Amersham Biosciences) antibodies, and proteins were detected by enhanced chemiluminescence (ECL; Amersham Biosciences).
Statistical Analysis-Data were analyzed with SPSS software using one-way analysis of variance, two-way analysis of variance, or Student's t test, as appropriate. For live cell imaging, Graph Pad Prism 3 (Graph Pad Software Inc., La Jolla) software packages were used. Values were represented as mean Ϯ S.E., and p Ͻ 0.05 values were statistically significant.

Results
Expression of CEACAM2 in Endocrine Pancreas-Semiquantitative RT-PCR analysis revealed that Ceacam2 mRNA is expressed at a ratio of ϳ2:1 in FACS-purified mouse pancreatic non-␤-cells relative to ␤-cells (Fig. 1A, panel i), as opposed to the Ceacam1 transcript, which was predominantly expressed in Values were expressed as means Ϯ S.E. *, p Ͻ 0.05 versus ␤-cells. B, liver tissues were removed from 5-month-old mice (n Ͼ3/genotype) and lysates were analyzed by 7% SDS-PAGE with sequential immunoblotting (Ib) with polyclonal antibody against CEACAM1 followed by re-immunoblotting (ReIb) with ␣-tubulin antibody for protein normalization. C, RNA was extracted from pancreatic islets of these mice and reverse-transcribed using specific Ceacam1 and Ceacam2 primers, followed by PCR amplification and analysis by 1% agarose gel. Gels in B and C represent at least two independent experiments.
Effect of CEACAM2 on Insulin Secretion-Five-month-old male Cc2 Ϫ/Ϫ mice exhibited a 2-fold higher plasma C-peptide level than their Cc2 ϩ/ϩ wild-type counterparts ( Table 1). However, this did not translate into changes in plasma insulin levels, possibly because of the countervailing effect of increased insulin clearance, as measured by steady-state C-peptide/insulin molar ratio (Table 1), and consistent with maintenance of the protein level of hepatic CEACAM1 (Fig. 1B), a main promoter of insulin clearance in liver (8 -10), and its insulin-stimulated phosphorylation in the insulin-sensitive male mouse (6). Changes in plasma C-peptide levels are not likely attributed to a regulatory action by CEACAM1 on insulin secretion from pancreatic ␤-cells because it does not play a significant role in insulin secretion, as it does in insulin clearance (9), and its cDNA level is preserved in pooled islets isolated from Cc2 Ϫ/Ϫ mice, as RT-PCR amplification revealed (Fig. 1C).
Immunohistochemical analysis with anti-insulin, anti-glucagon, and anti-somatostatin antibodies, respectively ( Fig. 2A), revealed normal areas of ␤-, ␣-, and ␦-cells in 6-month-old Cc2 Ϫ/Ϫ mice. Immunostaining with a three antibody mixture (3Ab) to glucagon, somatostatin, and pancreatic polypeptide P (PP) also revealed no difference in the area of cells secreting these hormones between the two groups of mice. Consistently, basal plasma levels of somatostatin and glucagon were normal in Cc2 Ϫ/Ϫ as compared with Cc2 ϩ/ϩ mice (Table 1). Furthermore, glucose (2.8 -16.8 mM) and KCl (20 mM), a membranedepolarizing agent, induced a comparable level of insulin release from pooled islets isolated from null or control mice (4-month-old) (Fig. 2B). Taken together, the data suggest that Ceacam2 deletion from the pancreas does not intrinsically affect ␤-cell area or insulin secretion.
To further test insulin secretory function, we subjected 7-month-old mice (n ϭ 7/genotype) to hyperglycemic clamp analysis. Maintaining a steady hyperglycemic state (at ϳ320 mg/dl) in mutants and controls (Fig. 2C, panel i) required a comparable glucose infusion rate in both groups (Fig. 2C, panel  ii). However, Cc2 Ϫ/Ϫ mice displayed higher plasma insulin levels, suggesting preserved insulin secretory function (Fig. 2C,  panel iii).
The increase in insulin levels in response to glucose does not appear to be compensatory to insulin resistance. Consistent with previous reports on 5-month-old mice (6), a 2-h hyperinsulinemic-euglycemic clamp analysis in overnight-fasted awake mice demonstrated that insulin sensitivity was maintained even at ϳ7-8 months of age (Fig. 3). Basal glucose at the beginning of the clamp (Fig. 3A) and basal hepatic glucose production (Fig. 3B), a measure of rate of appearance at pre-clamp condition, were similar in both groups of mice (Fig. 3B, vertical  and patterned hatchings). During the clamp, at similar glucose levels (Fig. 3A, white and black bars), the glucose infusion rate required to maintain euglycemia was comparable in both groups of mice (Fig. 3C), indicating preserved systemic insulin sensitivity in the mutants, as was the case for 5-month-old male mice (6). Ceacam2 null mutation did not affect the ability of insulin to suppress hepatic glucose production (Fig. 3B, white and black bars) and promote whole body glucose disposal (turnover) (Fig. 3D, black versus white bars), including glucose uptake in gastrocnemius muscle (Fig. 3E). Insulin suppresses hepatic glucose production by inhibiting gluconeogenesis and Five-month-old male mice (n Ն13 for all parameters, except for n Ն4 for GLP-1) were fasted overnight before retro-orbital blood was drawn, and plasma levels of hormones were assayed. To determine basal glucagon level, mice were fasted for 4 h before their retro-orbital blood was drawn. Data are means Ϯ S.E.

FIGURE 2. Effect of CEACAM2 on islet area, insulin secretion from isolated islets, and ␤-cell secretory function.
A, pancreatic sections from 6-monthold mice (n ϭ 4/genotype) were fixed and immunostained with antibodies against insulin, glucagon, somatostatin, polypeptide P (PP), and 3Ab (a mixture made up of the last three antibodies). Six sections/mouse were analyzed. Magnification ϫ40. B, islets were isolated from 4-month-old mice (n Ͼ3/genotype) by collagenase digestion followed by centrifugation over Histopaque gradient and culturing overnight in RPMI 1640 medium containing 5.5 nmol/ liter glucose. Insulin secretion was assayed by incubating 10 islets in Krebs buffer containing different concentrations of glucose or 20 mmol/liter KCl for 1 h. The amount of insulin secreted was normalized to DNA content, and values were expressed as means Ϯ S.E. C, awake mice (n ϭ 7, 7-month-old Cc2 ϩ/ϩ (E) and Cc2 Ϫ/Ϫ (•)) were continuously infused with glucose to maintain hyperglycemia (panel i), before glucose infusion rate (mg/kg/min) (panel ii), and plasma insulin (milliunits/ml) levels (panel iii) were determined. Values were expressed as means Ϯ S.E. *, p Ͻ 0.05 versus Cc2 ϩ/ϩ . stimulating net hepatic glucose uptake followed by glycogen synthesis (25). Consistently, whole body glycolysis (Fig. 3F) and glycogen synthesis (Fig. 3G) were comparable in both groups of mice. CEACAM2 Regulates Release of Glucagon-like Peptide-1-Because the distal part of the small intestine harbors cells that produce GLP-1 (12,26), we then reassessed CEACAM2 expression in this tissue and investigated whether its regulation of insulin secretion implicates GLP-1 production and/or release. Immunofluorescence analysis detected CEACAM2 in the villi lining the intestinal segment beginning with distal jejunum (Fig. 4, A-C, green). As reported previously (5), CEACAM2 was also detected in crypt cells (Fig. 4, AЈ-CЈ).
Relative to their wild-type counterparts, 5-month-old Cc2 Ϫ/Ϫ mice released a significantly higher level of GLP-1 (Fig.  5A) as well as insulin during an oral glucose tolerance test (Fig.  5B, panel i). Although their basal blood glucose level was normal, they manifested more tolerance to exogenous oral glucose (Fig. 5B, panel ii), consistent with their insulin sensitivity.
Pretreating with exendin(9 -39), a GLP-1 receptor antagonist, suppressed the effect of Ceacam2 deletion on insulin (Fig.  5B, panel iii) as well as glucose excursion (Fig. 5C, panel iv) in response to oral glucose. Together, this suggests that Ceacam2 deletion regulates insulin secretion, at least partly, via a GLP-1mediated mechanism.
To further examine the effect of CEACAM2 on GLP-1 production and/or secretion, we then reduced its expression in murine GLUTag enteroendocrine cells by means of siRNA transfection. As Fig. 6 shows, reduction of Ceacam2 mRNA (Fig. 6A) and protein (Fig. 6B) levels by ϳ50% significantly induced GLP-1 release into the media of siRNA relative to scrambled-transfected cells (Scr), 3 both basally and in response to 11 mM glucose (Fig. 6C). In contrast, reducing CEACAM2 content did not affect mRNA levels of cellular proglucagon (Fig.  6D). Together, the data point to modulation of GLP-1 secretion from intestinal cells independent of other potential confounding metabolic factors.
CEACAM2 Regulates GLP-1 Secretion through L-type Voltage-gated Ca 2ϩ Channels-Because L-type voltage-dependent Ca 2ϩ channels are implicated in GLP-1 release from GLUTag cells (27,28), we investigated whether CEACAM2 regulates GLP-1 release through voltage-dependent Ca 2ϩ channel (VDCC)-mediated changes in intracellular calcium. To this end, we transfected GLUTag cells with scrambled (Scr) and Ceacam2-specific siRNA oligos (siCc2) before loading them with Fura-2, a calcium indicator, and monitored 340/380 nm fluorescence in response to membrane depolarization evoked by application of physiological saline containing elevated KCl concentrations (10 -50 mM). As shown in Fig. 7, applications of elevated KCl induced concentration-dependent cytosolic Ca 2ϩ rises. Whereas KCl at 10 and 20 mM yielded comparable calcium influx in both control and knocked down cells, 50 mM KCl induced a more pronounced calcium rise in siCc2 than Scr cells, as shown in the ratio images and corresponding line traces (Fig.  7, A and C). In addition, repeated depolarization with 50 mM KCl revealed only a slight reduction in a second cytosolic Ca 2ϩ rise suggesting that the VDCC-underlying Ca 2ϩ entry was not robustly inactivated by our stimulus paradigm, as indicated by FIGURE 3. Insulin sensitivity in awake mice. Seven-to-eight month-old overnight fasted awake mice (n ϭ 11/genotype) were subjected to hyperinsulinemic-euglycemic clamp analysis. A, whole blood glucose. B, hepatic glucose production at basal (vertical hatchings, Cc2 ϩ/ϩ ; patterned hatchings, Cc2 Ϫ/Ϫ ) and clamp (white, Cc2 ϩ/ϩ ; black, Cc2 Ϫ/Ϫ ) conditions. *, p Ͻ 0.05 versus basal. C, steady-state glucose infusion rate during clamp. D, whole body (WB) glucose turnover (R d ). E, glucose uptake in gastrocnemius muscle. F, whole body glycolysis. G, whole body glycogen synthesis (white, Cc2 ϩ/ϩ ; black, Cc2 Ϫ/Ϫ ). Values were all expressed as means Ϯ S.E. in mg/kg⅐min Ϫ1 except for blood glucose levels (mg/dl). the ratio images and respective line traces (Fig. 7, B and D). Moreover, the amount of inactivation of entry was similar between control and knocked down groups.
To further identify the calcium channel that mediated depolarization-evoked intracellular Ca 2ϩ rises, the L-type VDCC blockers, nicardipine and nifedipine at 10 M each, were used to selectively inhibit Ca 2ϩ entry in response to 50 mM KCl application. As Fig. 7, E and F, reveals, elevated KCl failed to induce a rise in cytosolic Ca 2ϩ levels in the presence of nicardipine and nifedipine, respectively. This suggests that enhanced Ca 2ϩ entry in siCc2 cells was mediated by L-type VDCC and that CEACAM2 may regulate GLP-1 secretion by modulating L-type channel expression or activity. Cumulative averaged data for these experiments are shown in Fig. 7G.
Glucose Regulates Ceacam2 Expression-We have previously reported that refeeding following an overnight fast markedly represses Ceacam2 mRNA levels (1). Thus, we investigated whether glucose and/or insulin modulate(s) Ceacam2 expression. To this end, we treated GLUTag cells with 3, 5.5, or 20 mM glucose for 15 or 60 min. As Fig. 8A, panel i, reveals, glucose (5.5 mM) decreased Ceacam2 mRNA levels by ϳ50% compared with 3 mM within 15 min of treatment. Increasing glucose concentration and prolonging the treatment period did not further change Ceacam2 mRNA expression. Consistently, Western blot analysis reveals no effect of glucose (3-20 mM) on CEACAM2 protein level after 15 min of treatment. However, treatment with 20 mM, but not 3 mM, glucose for Ն60 min reduced CEACAM1 protein level by ϳ50% (Fig. 8A, panel ii). In contrast to glucose, insulin (100 nM) treatment for 15 or 60 min failed to modulate Ceacam2 mRNA levels in these cells (Fig.  8B).

Discussion
CEACAM1 and CEACAM2 are highly homologous proteins with differential tissue and cellular distribution (4,29) that predict distinct functions. CEACAM1 enhances insulin action by promoting insulin clearance in liver without affecting insulin secretion despite its significant expression in pancreatic ␤-cells (9). Emerging evidence shows that CEACAM2 regulates insulin action by reducing food intake and modulating energy expenditure (6). Using global Cc2 Ϫ/Ϫ null mice, the current studies unravel a novel role for CEACAM2 in inhibiting insulin secretion, at least partly by a GLP-1-dependent mechanism.
Despite predominant CEACAM2 expression in non-␤ pancreatic islets, its deletion did not affect either the islet areas or basal plasma glucagon and somatostatin levels. Together with preserved ␤-cell secretory function and normal insulin release in response to both glucose and KCl in isolated pooled islets, this points to an extra-pancreatic regulatory mechanism on insulin secretion brought about by Ceacam2 deletion.
Enhanced insulin excursion in response to oral glucose was accompanied by a higher induction of GLP-1 release in Cc2 Ϫ/Ϫ mice. Suppression of the differential effect of oral glucose on insulin secretion and glucose tolerance by the GLP-1 receptor antagonist provides further evidence that CEACAM2 can regulate insulin secretion through a GLP-1-dependent pathway. Because GLP-1 potentiates glucose-dependent insulin secretion from ␤-cells (11), it is likely that Ceacam2 deletion caused   an increase in glucose-stimulated insulin secretion, at least partly, by inducing GLP-1 release. This is supported by elevated GLP-1 content in the media of GLUTag enteroendocrine cells following siRNA-mediated knockdown of Ceacam2. The mechanistic underpinning of the regulation of GLP-1 secretion by CEACAM2 awaits further investigation, but the positive effect of Ceacam2 deletion on Ca 2ϩ entry via L-type VDCC suggests that CEACAM2 regulates GLP-1 secretion in part by reducing cellular Ca 2ϩ entry, consistent with the role of this process in mediating GLP-1 release from GLUTag cells (27,28).
Moreover, induction of GLP-1 release from GLUTag cells by Ceacam2 down-regulation suggests that elevated plasma GLP-1 levels in Cc2 Ϫ/Ϫ mice are independent of other potential confounding metabolic changes stemming from Ceacam2 deletion.
Of note, higher plasma insulin levels in Cc2 Ϫ/Ϫ than in wildtype mice during hyperglycemic clamp analysis point to the possibility that CEACAM2 can also affect insulin secretion independently of GLP-1. In light of CEACAM2 expression in the ventromedial hypothalamus (1), the glucose-sensing center in the brain (30), it is possible that hypothalamic CEACAM2 regulates insulin secretion centrally (31,32) and independently of GLP-1 release (33). Disturbance in this neuronally mediated mechanism may contribute to the higher excursion of insulin during hyperglycemic clamp analysis in Cc2 Ϫ/Ϫ mice.
We have shown that fasting induces and refeeding represses Ceacam2 mRNA levels (1). Based on these studies, it is reasonable to postulate that reduction of Ceacam2 expression during refeeding could result from the rise in glucose. This, in turn, could elevate GLP-1 and subsequently insulin secretion. To avoid hyperinsulinemia, secreted insulin would induce CEACAM1 phosphorylation to promote receptor-mediated insulin uptake (34) and hepatic insulin clearance (9, 10) to maintain normal insulin metabolism and action on glucose homeostasis. Given that GLP-1 prompts transition into the fasting state (35), this may also constitute a negative feedback mechanism to ultimately recover CEACAM2 levels and limit insulin secretion. Of note, CEACAM2 is expressed in the kidney but not liver (4,29). Whether it plays a role in renal insulin clearance is not known but, given that insulin clearance occurs mostly in the liver where CEACAM1 expression dominates, insulin clearance is not expected to be directly affected by Ceacam2 deletion, and elevation in insulin clearance in Cc2 Ϫ/Ϫ mice would be mediated by CEACAM1 phosphorylation in response to insulin in liver (8 -10) as well as kidney (36).
In conclusion, the data provide evidence for a novel role for CEACAM2 in the regulation of insulin secretion, at least partly, via GLP-1 release. Further studies are required to elucidate the underlying mechanisms, but the current data highlight a relevant role for CEACAM2 in GLP-1 release.
Author Contributions-S. S. G. researched data, designed and coordinated experiments, and wrote the manuscript. G. H. researched and analyzed data and reviewed the manuscript. S. G. L., V. P., S. B., P. R. P., A. M. D., T. D., S. K. R., Z. N. S., Y. L., D. Y. J., and T. K. researched data. S. E. oversaw the immunofluorescence analysis of intestinal cells. R. N. K. oversaw, discussed the localization of CEACAM2 to islets, and revised the manuscript. J. K. K. oversaw and discussed clamp analysis data. D. R. G. oversaw and discussed data of live cell fluorescence and intracellular Ca 2ϩ measurements and revised the manuscript. S. M. N. was responsible for study design, conceptualization, data analysis, results interpretation, and reviewing the manuscript. S. M. N. has full access to all the data of the study and takes responsibility for the integrity and accuracy of data analysis and the decision to submit and publish the manuscript.