Trace amine–associated receptor 1 (TAAR1) promotes anti-diabetic signaling in insulin-secreting cells

Pancreatic β-cell failure in type 2 diabetes mellitus is a serious challenge that results in an inability of the pancreas to produce sufficient insulin to properly regulate blood glucose levels. Trace amine–associated receptor 1 (TAAR1) is a G protein–coupled receptor expressed by β-cells that has recently been proposed as a potential target for improving glycemic control and suppressing binge eating behaviors. We discovered that TAAR1 is coupled to Gαs-signaling pathways in insulin-secreting β-cells to cause protein kinase A (PKA)/exchange protein activated by cAMP (Epac)–dependent release of insulin, activation of RAF proto-oncogene, Ser/Thr kinase (Raf)–mitogen-activated protein kinase (MAPK) signaling, induction of cAMP response element–binding protein (CREB)–insulin receptor substrate 2 (Irs-2), and increased β-cell proliferation. Interestingly, TAAR1 triggered cAMP-mediated calcium influx and release from internal stores, both of which were required for activation of a MAPK cascade utilizing calmodulin-dependent protein kinase II (CaMKII), Raf, and MAPK/ERK kinase 1/2 (MEK1/2). Together, these data identify TAAR1/Gαs-mediated signaling pathways that promote insulin secretion, improved β-cell function and proliferation, and highlight TAAR1 as a promising new target for improving β-cell health in type 2 diabetes mellitus.

Diabetes mellitus is an increasingly prevalent metabolic disease that affects ϳ400 million people worldwide, with future projections reaching nearly 600 million by 2035 (1). Type 2 diabetes (T2DM) 2 still accounts for 90 -95% of cases and is associated with a sedentary lifestyle, poor diet, obesity, metabolic syndrome, and genetic risk factors (2). The pathogenesis of T2DM involves a vicious cycle of escalating insulin insensi-tivity, elevated blood glucose, and rising insulin levels that eventually exhausts pancreatic ␤-cells. This results in ␤-cell dysfunction, decreased insulin output, and ␤-cell death (3). Despite a multitude of available treatments to bolster insulin levels, new disease-modifying agents that improve the health and function of pancreatic ␤-cells will be essential to meaningfully reduce overall disease burden and complications of diabetes (4).
One of the more recent trends in T2DM therapies have focused on incretin hormones (5). Stable incretin mimetics targeting the glucagon-like peptide 1 receptor (GLP-1R) have been found to be effective at improving glycemic control in T2DM, with the added benefit of modest weight loss. GLP-1 homologs may confer an advantage over other antihyperglycemic agents such as insulin analogues, sulfonylureas, and thiazolidinediones, which have been associated with weight gain (5,6). The salutary effects of incretin receptor agonism in ␤-cells appear to derive from G␣ s signaling, which mediates glucose-stimulated insulin secretion (GSIS) and promotes ␤-cell proliferation while reducing stress-induced apoptosis (7,8).
Recently it was found that the G protein-coupled trace amineassociated receptor (TAAR1) is expressed in human pancreatic islets and increases insulin secretion and glucose tolerance in mouse models (9). However, the molecular mechanisms by which TAAR1 directly regulates insulin secretion in ␤-cells are largely unknown, as TAAR1 plays additional roles in gut motility, satiety, and eating behaviors (9,10). Amine agonists such as octopamine, ␤-phenylethylamine, tyramine, and amphetamines (11) exhibit neurotransmitter-like activity, and TAAR1 is expressed in the brain, where it has been shown to decrease binge eating behaviors and stimulate weight loss (9,10). Aside from trace amines, TAAR1 is activated in the periphery by thyroid hormone 3-iodothyronamine (T 1 AM) (12), and a concerted effort is currently underway to produce selective TAAR1 agonists for use in control of impulsive behaviors (13). Here we delineate the downstream effectors of TAAR1 in pancreatic ␤-cell lines and provide evidence that TAAR1 agonists trigger beneficial anti-diabetic signaling pathways that could help lead to their use as anti-diabetic agents. determined whether TAAR1 utilized adenylyl cyclase signaling pathways. The Ins-1 and Min6 ␤-cell lines were found to exhibit cAMP-dependent potentiation of GSIS in response to the TAAR1 agonist T 1 AM. Treatment of cells with MDL-12,330A, an inhibitor of adenylyl cyclase (AC), significantly (p Ͻ 0.001) reduced TAAR1 potentiation of GSIS by 77-82% (Fig. 1, A and  B). The specificity of T 1 AM for TAAR1 was confirmed by the use of the selective TAAR1 antagonist EPPTB (14), which significantly (p Ͻ 0.001) inhibited (70 -82%) potentiation of GSIS by T 1 AM (Fig. 1, A and B). PKA and Epac are downstream effectors of adenylyl cyclase-cAMP signaling and regulate granule exocytosis. Both H89 (a PKA inhibitor) and HJC-0350 (an Epac inhibitor) caused a significant (p Ͻ 0.001) reduction in TAAR1-mediated potentiation of GSIS by 82-93% (Fig. 1A). These data indicate that potentiation of glucose-stimulated insulin secretion by TAAR1 occurs through AC-cAMP dependent signaling pathways and requires both PKA and Epac activity to effect maximal insulin release.

TAAR1 stimulates CREB phosphorylation and the downstream target gene Irs-2 via adenylyl cyclase and PKA
CREB is a key transcription factor that promotes ␤-cell health (15); therefore, we determined whether phosphorylation of the PKA target CREB could be detected following TAAR1 activation. Because glucose itself can stimulate CREB pathways, a time course for CREB phosphorylation in response to T 1 AM was first established in Ins-1 cells with or without exogenous glucose ( Fig. 2A). T 1 AM induced peak CREB phosphorylation in both the presence and absence of glucose as early as 5 min and diminished by 30 min (Fig. 2A). The direct adenylyl cyclase activator forskolin gave similar kinetics of activation of pCREB as T 1 AM with ϳ2-fold higher magnitude in the presence of glucose. Treatment of cells with MDL-12,330A (Fig. 2B) as well as H89 (Fig. 2C) to inhibit AC and PKA, respectively, effectively blocked phosphorylation of the CREB transcription factor. Knockdown of the catalytic subunit of PKA with siRNA also caused a reduction in CREB phosphorylation compared with cells treated with control siRNA (Fig. 2D), consistent with results generated using H89. A key CREB target gene, insulin receptor substrate-2 (Irs-2), which promotes pro-proliferative and anti-apoptotic signaling in ␤ cells (15,16), was significantly (p Ͻ 0.001) up-regulated by 50% in cells treated with T 1 AM (Fig. 2E). Induction of Irs-2 was completely blocked by MDL-12,330A (p Ͻ 0.001) but not LY294002, revealing that adenylyl cyclase, but not phosphatidylinositol 3-kinase (PI3K) is required for Irs-2 gene induction by T 1 AM.

TAAR1 stimulates ␤-cell proliferation by activation of adenylyl cyclase-dependent Raf-MAPK signaling
Ins-1 cells treated with T 1 AM and the positive control forskolin exhibited robust ERK1/2 phosphorylation in both the presence and absence of exogenous glucose (Fig. 3A). ERK1/2 phosphorylation appeared at 5 min and peaked at 15 min but mostly faded by 30 min, similar to the kinetics of pCREB (Fig. 2). ERK1/2 phosphorylation induced by both the TAAR1 agonist and forskolin was blocked by MDL-12,330A (Fig. 3B), indicating that adenylyl cyclase is required for TAAR1-mediated ERK1/2 phosphorylation. TAAR1 ERK1/2 phosphorylation was also reduced by siRNA knockdown of the catalytic subunit of PKA (PKA Cat-␣) compared with cells treated with control siRNA (Fig. 3C). The Epac inhibitors Esi-09 (Epac1/2 inhibitor) and Esi-05 (Epac 2 inhibitor) also attenuated ERK1/2 phosphorylation downstream of T 1 AM (Fig. 3D), indicating that both PKA and Epac are involved in TAAR1-mediated ERK1/2 phosphorylation. The suppressive effects of Esi-09 and Esi-05 on ERK1/2 phosphorylation were confirmed by siRNA knockdown of Epac, which again reduced TAAR1 and forskolin-mediated ERK1/2 phosphorylation (Fig. 3E). The MAP Kinase Kinase Kinase Raf has been shown to be regulated by both Epac and PKA (17,18). Raf and the mitogen-activated protein kinase kinase MEK1/2 were identified as upstream mediators of ERK1/2 phosphorylation, as the pan-Raf inhibitor AZ-628 and the MEK1/2 inhibitor PD98059 completely prevented ERK1/2 phosphorylation in response to both T 1 AM and forskolin (Fig.  3F). Both the B-Raf and C-Raf isoforms were found to be expressed in Ins-1 ␤ cells with 5-fold higher relative B-Raf expression (Fig. 3G).
As ERK1/2 phosphorylation is frequently associated with increased rates of cellular proliferation, we sought to determine whether T 1 AM affected proliferation of Ins-1 cells. Both T 1 AM and forskolin significantly enhanced radiolabeled thymidine incorporation in insulin-secreting cells (Fig. 3H). AZ-628 and PD98059 completely blocked the increases in radiolabeled thymidine incorporation in response to T 1 AM, indicating that Raf/ MEK1/2/ERK1/2 signaling is required for TAAR1 stimulation of cellular proliferation of Ins-1 cells.

TAAR1-MAPK signaling in insulin-secreting cells requires both calcium influx and intracellular calcium release
Surprisingly, we found that addition of extracellular calcium chelator EGTA to Ins-1 cells completely blocked ERK1/2 phosphorylation in response to T 1 AM and forskolin (Fig. 4A), indicating that calcium influx was critical for downstream MAPK signaling initiated by G s -AC/cAMP. Intracellular calcium release from internal stores via IP 3 receptors was also required

TAAR1 promotes anti-diabetic signaling
for ERK1/2 phosphorylation, as the 2-APB antagonist completely blocked ERK1/2 phosphorylation to both T 1 AM and forskolin (Fig. 4B). Likewise, reloading of intracellular calcium stores was partially required for activation of pERK1/2, as blockade of the sarco/endoplasmic reticulum Ca 2ϩ -ATPase (SERCA) pump with thapsigargin was able to reduce ERK phosphorylation in response to these agents (Fig. 4A). PLC-␤, a common mediator of calcium signaling through G q -and G i -coupled receptors, however, was not required for TAAR1-initiated ERK1/2 phosphorylation, as the U73122 PLC-␤ inhibitor had no effect on T 1 AM or forskolin-mediated ERK1/2 phosphorylation (Fig. 4B). U73122 did, however, effectively inhibit calcium signaling in response to LIGRLO (Fig. S1), an agonist of protease-activated receptor 2 (PAR2), a known G q -coupled receptor expressed in the pancreas (19), confirming the efficacy of U73122 in these insulin-secreting cells.
Ca 2ϩ/ Calmodulin (CaM) kinase-dependent signaling has been shown previously to regulate Raf/MEK/ERK pathways in certain cell types (20), providing a possible link between intracellular calcium and downstream ERK1/2 phosphorylation. Interestingly, we found that treatment of Ins-1 cells with KN-93, a potent inhibitor of CaM kinase II (CaMKII), blocked ERK1/2 phosphorylation in response to T 1 AM and forskolin (Fig. 4C). In contrast, the divergent calmodulin-regulated kinase Ca 2ϩ /calmodulin-dependent protein kinase kinase II appeared to play a MAPK-suppressive role, as the STO-609 inhibitor enhanced both the basal and agonist-stimulated pERK1/2 signal (Fig. 4C).

TAAR1 activation triggers calcium influx and intracellular calcium release via cAMP-mediated signaling
Consistent with the effects of the various calcium pathway inhibitors on modulating ERK1/2 phosphorylation, we were able to directly detect a significant, rapid, intracellular calcium flux signal in response to T 1 AM, as well as the cAMP-stimulating positive control forskolin (Fig. 4, D and H). The TAAR1 calcium flux signal depended on extracellular calcium, as reduction of physiologic levels (1.5 mM) of extracellular calcium or introduction of EGTA caused up to a 75% attenuation (p Ͻ 0.001) in calcium flux (Fig. 4, D and E). The residual 25% calcium signal could be completely blocked by inclusion of thapsigargin (Fig. 4, D and E), which inhibits reloading of calcium into endoplasmic reticulum stores. This TAAR1-mediated calcium release occurred through cAMP-dependent signaling pathways, as MDL-12,330A significantly (p Ͻ 0.001) reduced calcium signaling in response to T 1 AM (Fig. 4, F and I). Consistent with the effects on ERK1/2 phosphorylation above, 2-APB also significantly (p Ͻ 0.001) inhibited the T 1 AM-in-

TAAR1 promotes anti-diabetic signaling
duced calcium signal (Fig. 4, G and I), indicating critical involvement of the IP 3 receptor in TAAR1-mediated calcium signaling. We ruled out G q as being involved in the TAAR1evoked calcium signal, as the potent and selective G q inhibitor YM-254890 had no effect on the T 1 AM calcium response (Fig.  4, J and K). In contrast, YM-254890 completely blocked calcium flux in response to LIGRLO (21), an agonist of the G q -coupled PAR2 (Fig. 4, L and M). Together, these data support a mechanism whereby TAAR1 stimulates G␣ s /cAMP-dependent cal-cium release from internal stores, and this initial calcium flux leads to further calcium influx from external sources.

Small molecule RO5256390 stimulates TAAR1 signaling in insulin-secreting cells
RO5256390 was recently discovered and characterized as a specific agonist of TAAR1 for potential use as a central nervous system-acting drug with potent in vivo activity in rodents and monkeys (13). We sought to determine whether this new small-
As shown in Fig. 5F, RO5256390 induced striking potentiation of GSIS compared with T 1 AM. Neither RO5256390 or T 1 AM induced insulin secretion under low-glucose conditions (Fig. 5F). This demonstrates that the trace amine receptor behaves similarly as the peptide incretin receptors GLP-1R and GIPR, which potentiate GSIS via G␣ s -cAMP signaling but do not trigger insulin secretion under low-glucose conditions. Furthermore, inhibition of cAMP signaling pathways with MDL-12,330A or H89 inhibitors had no significant effect on GSIS in the absence of TAAR1 agonist treatment (Fig. 5F).
TAAR1 stimulates calcium flux, therefore it is possible that it could also have a direct effect on GSIS by increasing calciumdependent insulin-vesicle release, as has been shown with the G s -coupled GLP receptor (22). We determined that the IP 3 receptor antagonist 2-APB significantly (p Ͻ 0.001) reduced TAAR1-dependent insulin secretion in the presence of high glucose (Fig. 5F), underscoring the essential role that calcium flux from internal stores plays in potentiation of GSIS by TAAR1 agonist.

Discussion
T2DM is part of a growing epidemic of metabolic diseases, and current treatments do not adequately meet the needs of this expanding patient population. Recent strategies to improve glycemic control include enhancing GSIS and pancreatic islet viability by targeting a select group of G protein-coupled receptors that couple to G␣ s (23,24). These include receptors for gut peptide incretins such as GIP (gastric inhibitory polypeptide) and GLP-1 (glucagon-like peptide-1), that stimulate cAMP production via AC activation (25,26). cAMP is a critical second messenger for ␤-cells, as it activates a tightly regulated and complex signaling network that can induce potentiation of GSIS, protect ␤-cells from stress-induced apoptosis, and trigger an expansion of ␤-cell mass to combat ␤-cell failure (27). One of the biggest hurdles in developing treatments that target G␣ scoupled peptide receptors such as GLP-1R and GIPR is the expense and short t1 ⁄ 2 of their peptide hormone agonists (28).
Here we present evidence that the trace amine receptor TAAR1 is coupled to G␣ s signaling pathways in pancreatic ␤-cell lines to cause PKA/Epac-dependent release of insulin, activation of Raf-MAPK signaling, induction of CREB-Irs2, and increased proliferation (Fig. 6). We found that TAAR1 potentiates GSIS through adenylyl cyclase-cAMP activation of both PKA and Epac. In ␤-cells, PKA facilitates vesicle docking and priming for exocytosis by phosphorylation of the key SNARE complex and associated proteins (29,30). PKA also phosphorylates K ATP to reduce channel activity, contributing to membrane depolarization and opening of voltage-gated calcium channels (7). Epac is a cAMP-regulated guanine nucleotide exchange factor that interacts with a variety of intracellular proteins to increase GSIS (31). Epac is recruited to the plasma membrane in response to elevated cAMP in ␤-cells, where it clusters near secretory vesicles and binds to the exocytosis machinery to increase vesicle priming and fusion at docking sites (32). Epac can interact with Rim2 (Rab3-interacting molecule 2), which, together with Rab3, form a GTP-dependent complex between plasma membranes and docked synaptic vesicles to regulate exocytosis (33). Like PKA, Epac also modulates the activity of ATP regulated potassium channels to increase GSIS (34).
We found that TAAR1 stimulated PKA-dependent phosphorylation of CREB at Ser-133 in response to T 1 AM and RO5256390. Induction of TAAR1-CREB signaling by T 1 AM and RO5256390 led to activation of the Irs-2 gene, which has been shown to coordinate the activation of pro-proliferative and anti-apoptotic signaling pathways in ␤-cells (15). However, TAAR1 did not utilize PI3K to induce Irs-2 gene expression. IRS-2 is directly phosphorylated by the insulin receptor, leading to the recruitment and activation of other signaling proteins crucial for insulin signaling. Disruption of IRS-2 has been linked to the development of T2DM, and ␤-cell-specific Irs-2 knockouts have confirmed an essential role in insulin resistance, obesity, ␤-cell mass, and proliferation (36).
We documented a robust increase in ERK1/2 phosphorylation in response to both TAAR1 agonists that involves AC, PKA, Epac, and Raf/MEK1/2 activation (Fig. 6). Both B-and C-Raf mRNA were expressed, with 5-fold higher levels for B-Raf, and both have been shown to regulate MAPK signaling in ␤ cells. In the case of glucose plus GLP-1R agonism, B-Raf, but not C-Raf, was found previously to be required for ERK1/2 phosphorylation (17,37). Conversely, C-Raf knockdown increases rather than inhibits ERK1/2 phosphorylation in response to glucose (17). In terms of the roles PKA and Epac may play in Raf activation, G␣ s /cAMP-stimulated Epac has been shown to activate Rap1/B-Raf and downstream ERK1/2 in neuronal, endocrine, and other cell types (18,38). In thyroid cells, PKA has also been shown to directly activate Raf, which leads to ERK1/2 phosphorylation (39).
CaMKII activity was also necessary for activation of the MAPK cascade in the ␤-cells by both T 1 AM agonist and cAMP. This is consistent with previous work that showed direct CaMKII association and activation of Raf, which led to ERK1/2 phosphorylation in response to outside-in signaling in thyroid cells (20). Therefore, Raf may likely be an important upstream MAPK signaling node in ␤-cells that receives input from both calcium-regulated proteins such as CaMKII in addition to the cAMP-regulated proteins PKA and Epac. Although incretins have been shown to stimulate ERK1/2 activity in ␤-cells, MAPK activation is not required for or related to insulin secretion (40,41) but is linked to increased cellular proliferation rates, which could play a role in compensatory ␤-cell hyperplasia in T2DM (42). Indeed, we observed a Raf/MEK-dependent increase in ␤-cell proliferation in response to TAAR1 agonism, indicating that TAAR1 could potentially improve ␤-cell mass, although this remains to be shown in vivo. Increased proliferation also occurred in response to the cAMP stimulant forskolin, validating the ability of G␣ s /AC pathways to drive ␤-cell proliferation.

TAAR1 promotes anti-diabetic signaling
ERK1/2 can regulate the insulin gene promoter and coordinate responses to endoplasmic reticulum stress (43), indicating other potential aspects of TAAR1-ERK1/2-mediated anti-diabetic signaling pathways to be explored.
By directly measuring intracellular calcium flux, we were able to observe a rapid and robust TAAR1-mediated calcium signal in the Ins-1 ␤-cell line. Consistent with the requirement for upstream CaMKII activation within the MAPK cascade, we discovered that both calcium influx from the outside and release from internal calcium stores via IP 3 receptors was necessary for activation of ERK1/2 by a TAAR1-cAMP-dependent mechanism (Fig. 6). TAAR1-calcium signaling was not affected by YM-254890 or U73122, confirming that G␣ q coupling or PLC-␤ were not involved. These data support the mechanism in Fig. 6, whereby the calcium signal induced by TAAR1-G␣ s / cAMP is initially triggered by intracellular release through IP 3 receptors, followed by activation of calcium channels on the plasma membrane, which provide the majority of the total observed calcium signal. Calcium flux has been shown to be a critical second messenger for various G␣ s -signaling pathways in ␤-cells, most importantly insulin granule exocytosis, which depends in large part on the activity of the aforementioned CaMKII (7,44,45). CaMKII is localized to the insulin secretory granules and binds to synapsin-1 and MAP-2 proteins, which are involved in exocytosis (46). As 2-APB, an IP 3 R antagonist, was also found to eliminate potentiation of GSIS by TAAR1, calcium release from intracellular stores induced by TAAR1 activation also plays a role GSIS potentiation by the receptor. In a pathway that diverges from GSIS, however, TAAR1-calcium signaling also induces MAPK activation. This result is akin to what has been observed for glucose and GLP-1R-G␣ s /AC coupling, which also require a rise in intracellular calcium to initiate MAPK signaling (47). Although glucose enhanced the ability of TAAR1 to signal pCREB and pERK1/2 pathways in the ␤-cell lines, we found that TAAR1 could do this independently of exogenous glucose, showing a divergence from the GSIS pathway.
One of the difficulties of studying TAAR1 in the past has been the lack of specific and efficacious agonists of the receptor. The endogenous agonist T 1 AM is capable of activating the G icoupled adrenergic receptor ␣-2A, albeit with poor affinity (48). Several novel compounds have now been identified as specific, full agonists of TAAR1 (49), including RO5256390 (13). We determined that RO5256390 produced essentially identical but more robust TAAR1 signals as T 1 AM. In the context of metabolic diseases, it is interesting to note that lisdexamfetamine dimesylate-the first Food and Drug Administrationapproved drug for binge eating disorder-is a prodrug that is broken down into the active metabolite d-amphetamine (50,51), a known TAAR1 agonist (52). Furthermore, in animal studies, inhibition of adrenergic receptors, which are believed to be a target of the prodrug's active metabolite, only blocks some of

TAAR1 promotes anti-diabetic signaling
the effects of lisdexamfetamine dimesylate (53), raising the possibility that this amphetamine exerts some of its effects through TAAR1. In this regard, TAAR1 agonism has been shown to reduce monoaminergic signaling in the brain, reducing binge eating and impulsive behaviors, via TAAR1-mediated downregulation of dopamine reward circuits (14,54). The ability of TAAR1 to reduce maladaptive eating behaviors that can contribute to obesity and metabolic disease, coupled with our newly identified beneficial effects of TAAR1 agonism on ␤-cell lines, highlight the therapeutic potential of TAAR1.

Cell culture
The clonally derived Ins-1 (832/3) ␤-cell line that exhibits robust glucose and incretin responsiveness was generously provided by Dr. Christopher Newgard and cultured according to established protocols (55). Briefly, cells were maintained in a humidified incubator at 37°C in RPMI 1640 medium supplemented with 10% fetal bovine serum, 10 mM HEPES, 2 mM L-glutamine, 1 mM sodium pyruvate, 50 M 2-mercaptoethanol, 100 units/ml penicillin, and 100 g/ml streptomycin and subcultured when confluent. The Min6 ␤-cell line was kindly provided by Dr. Melanie Cobb and grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 1 mM L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 55 M mercaptoethanol.

Insulin release from ␤-cell lines
Ins-1 or Min6 cells were seeded in 24-well plates and grown until confluent. On the day of the experiment, cells were washed once with HanksЈ Balanced Salt Solution and further incubated for 2 h in HBSS (114 mM NaCl, 4.7 mM KCl, 1.2 mM KH 2 PO 4 , 1.16 mM MgSO 4 , 20 mM HEPES, 2.5 mM CaCl 2 , 25.5 mM NaHCO 3 , and 0.2% BSA) with 3.5 mM (basal) D-glucose at 37°C. Supernatants were sampled for basal insulin secretion levels, after which agonists were incubated with cells for 2 h at 37°C. Supernatants were harvested, and insulin secretion was measured via insulin ELISA according to the manufacturer's instructions; relative insulin secretion was calculated as stimulated/basal insulin.

Real-time quantitative PCR and RT-PCR
RNA was isolated from cells using the Qiagen RNeasy Mini Kit and reverse-transcribed using Moloney murine leukemia virus reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Real-time PCR analysis was carried out using Lightcycler 480 SYBR Green I (Roche Diagnostics) using the following primers to amplify Irs-2, B-Raf, C-Raf, and Gadph: Irs-2, CGC AAG CAT CGA CTT CTT GTC (forward) and GCC CGC AGC ACT TTA CTC TT (reverse); B-Raf, GGA GCA TAA CCC ACC GTC AA (forward) and AAC AGC TGC TGC TC TCT CTG (reverse); C-Raf, CTG TCG CTG CAC TAC GGG (forward) and TCG TCT TCC AAG CTC CCT GT (reverse); Gapdh, GGC ATC GTG GAA GGA CTC ATG AC (forward) and ATG CCA GTG AGC TTC CCG TTC AGC (reverse). Relative abundance of Irs-2, C-Raf, and B-Raf mRNA was calculated with respect to the housekeeping gene Gapdh.

Western blotting
Ins-1 cells were grown to confluence in 12-well plates, at which point the complete medium was removed and (unless otherwise indicated) replaced with serum-and glucose-free RPMI supplemented with 0.2% BSA. Cells were starved for 2 h at 37°C, and antagonists were added 30 min prior to addition of agonists. Cells were treated with agonists for the specified times at 37°C, after which medium was removed, and cells were harvested in ice cold T-Per lysis buffer supplemented with 1ϫ HALT phosphatase and protease inhibitors. Proteins were resolved in 10% SDS-polyacrylamide gels via gel electrophoresis, transferred to nitrocellulose membranes, and blocked in 5% nonfat milk in Tris-buffered saline with 0.05% Tween for 30 min. Membranes were incubated overnight at 4°C with primary antibodies (phospho-ERK1/2, phospho-CREB) at 1:1000 dilution, followed by horseradish peroxidase-conjugated secondary antibody incubation (1:5000 dilution) for 1 h at room temperature. After detection of phosphoproteins, blots were stripped and reprobed with loading controls (total ERK1/2 and ␤-actin) for normalization. Several total CREB antibodies were tested (Cell Signaling Technology) for normalization of total CREB in Ins-1 cells, but we were unable to reliably detect CREB, and ␤-actin was used.

Intracellular Ca 2؉ measurements
Ins-1 cells were loaded with 2 M Fura-2/AM in phenol redfree RPMI with 0.5% BSA for 30 min at 37°C. Cells were washed and resuspended, and calcium flux was assessed by measuring fluorescence emission at a dual excitation of 340 and 380 nm using an LS-50B (PerkinElmer Life Sciences) spectrofluorimeter as before (56). Unless otherwise specified, antagonists were preincubated with cells at room temperature for 30 min before addition of agonists. The area under the curve (AUC) was quantified using Prism 5.0a and standardized to %max AUC using 10 M T 1 AM as 100%.

TAAR1 promotes anti-diabetic signaling Cellular proliferation
Radiolabeled thymidine incorporation was performed as described previously (35). Briefly, Ins-1 cells were seeded in 24-well plates at 1.5 ϫ 10 5 cells/ml and allowed to attach overnight in complete medium. After 24 h, medium was aspirated and replaced with RPMI, 0.2% BSA, and 1 mM glucose to induce senescence, and cells were incubated at 37°C for 24 h. Medium was then aspirated from cells and replaced with RPMI containing 0.2% BSA/4.5 mM glucose. Antagonists were preincubated with cells 30 min prior to addition of agonists, and cells were returned to the incubator for 24 h. During the last 4 h of incubation, 1 Ci of [ 3 H]thymidine was added to each well. Cells were lifted, washed twice, fixed with 6% TCA, and centrifuged. The final pellet was resuspended in 0.2 N NaOH, added to scintillation fluid, and read for [ 3 H]disintegrations per minute (DPM) counts.

Statistical analyses
Data are expressed as means Ϯ S.D. Comparisons between experimental and control cohorts were performed by one-way ANOVA (for which a global p value was calculated) or t test as appropriate, and the mean of individual groups was compared using Dunnett's or Newman-Keuls post hoc correction. Global p values are numerically indicated in the figures, and post-test analyses between individual groups are indicated by asterisks. Analyses were performed using GraphPad Prism 6.0. Statistical significance was defined as follows: *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; ****, p Ͻ 0.0001.