Abstract
GPR35 is a G protein-coupled receptor recently “de-orphanized” using high-throughput intracellular calcium measurements in clonal cell lines expressing a chimeric G-protein α-subunit. From these screens, kynurenic acid, an endogenous metabolite of tryptophan, and zaprinast, a synthetic inhibitor of cyclic guanosine monophosphate-specific phosphodiesterase, emerged as potential agonists for GPR35. To investigate the coupling of GPR35 to natively expressed neuronal signaling pathways and effectors, we heterologously expressed GPR35 in rat sympathetic neurons and examined the modulation of N-type (CaV2.2) calcium channels. In neurons expressing GPR35, calcium channels were inhibited in the absence of overt agonists, indicating a tonic receptor activity. Application of kynurenic acid or zaprinast resulted in robust voltage-dependent calcium current inhibition characteristic of Gβγ-mediated modulation. Both agonist-independent and -dependent effects of GPR35 were blocked by Bordetella pertussis toxin pretreatment indicating the involvement of Gi/o proteins. In neurons expressing GPR35a, a short splice variant of GPR35, zaprinast was more potent (EC50 = 1 μM) than kynurenic acid (58 μM) but had a similar efficacy (approximately 60% maximal calcium current inhibition). Expression of GPR35b, which has an additional 31 residues at the N terminus, produced similar results but with much greater variability. Both GPR35a and GPR35b appeared to have similar expression patterns when fused to fluorescent proteins. These results suggest a potential role for GPR35 in regulating neuronal excitability and synaptic release.
G protein-coupled receptors (GPCRs) comprise one of the largest families of cell surface proteins and represent a major target for both current therapeutic agents and drugs under development. Many cDNA clones are predicted to code for GPCRs based on high sequence similarity, especially in the transmembrane domains, to established GPCRs. For such orphan GPCRs, the identification of agonists, antagonists, and signal transduction pathways represents a major effort by industry for the discovery of novel drug targets (Stadel et al., 1997). GPR35, an orphan GPCR first discovered during a human genomic DNA screen, shares limited sequence homology with purinergic P2Y receptors, nicotinic acid receptor HM74, lysophosphatidic acid receptor GPR23, and an orphan receptor, GPR55. The highest levels of GPR35 mRNA were found in immune and gastrointestinal tissues with only limited expression in lung and neuronal tissues (O'Dowd et al., 1998; Wang et al., 2006). Subsequent to the initial description of GPR35 (now denoted GPR35a), Okumura et al. (2004) isolated a splice variant (GPR35b) from a human gastric cancer cDNA library that coded for an additional 31 amino acids at the N terminus. GPR35a and GPR35b mRNA levels were up-regulated in gastric cancer tissue, suggesting a role for both splice variants in malignant transformation. However, in the absence of an identified agonist, the functional role GPR35 plays in either normal or pathological states remains enigmatic.
Recently, using high-throughput technology, kynurenic acid (KYNA) (Wang et al., 2006) and zaprinast [1,4-dihydro-5-(2-propoxyphenyl)-7H-1,2,3-triazolo[4,5-d]pyrimidine-7-one] (Taniguchi et al., 2006) were identified as putative agonists for GPR35 by monitoring elevation of intracellular [Ca2+] in cell lines coexpressing GPR35a and chimeric G protein α-subunits. Kynurenic acid, an intermediate metabolite of tryptophan, is often used as a nonspecific ionotropic glutamate receptor antagonist. Kynurenic acid is also a noncompetitive antagonist of the α7 nicotinic acetylcholine receptor that inhibits synaptic transmission in cultured rat hippocampal neurons via both pre- and postsynaptic mechanisms (Hilmas et al., 2001). As a naturally occurring metabolite, KYNA potentially serves as an endogenous agonist for GPR35. The other identified agonist, zaprinast, is a synthetic inhibitor of cGMP-specific phosphodiesterases (PDEs), especially PDE5 and PDE6. As a compound closely related to sildenafil (a clinically effective drug for erectile dysfunction), zaprinast has been used as diagnostic tool to identify signaling pathways involving cGMP generation and degradation (Gibson, 2001). However, zaprinast-mediated intracellular Ca2+ mobilization via GPR35a activation was found to be independent of cGMP degradation (Taniguchi et al., 2006).
Thus far, only artificial G proteins α-subunits (i.e., Gqi5, Gqi9,Gqo5, and Gqs5) have been used to detect transduction signals after GPR35 activation (Wang et al., 2006; Taniguchi et al., 2006). In these high-throughput systems, expression of a chimeric G protein α-subunit switches the coupling of Gi/o and Gs preferring GPCRs to Gq (by altering residues in the C terminus that convey receptor coupling specificity) thereby allowing changes in intracellular [Ca2+] to be used as a readout for receptor activation. Although chimeric Gα-subunits are useful for predicting GPCR coupling specificity, functional discrepancies between chimeric and native G protein-mediated effects have been reported (Kostenis et al., 2005). Therefore, a goal of the present study was to evaluate the function of GPR35 expressed within a context of unmodified G proteins at native expression levels. A second goal was to establish a model system for the future testing of novel agents, such as the endocannabinoid N-arachidonoyl l-serine (Milman et al., 2006), on GPR35 function. GPR35 shares significant sequence similarity with GPR55, an orphan GPCR that putatively binds cannabinoid ligands (Baker et al., 2006) but may not represent a classic cannabinoid receptor (Petitet et al., 2006). In the present study, we examined N-type Ca2+ channel modulation in rat sympathetic neurons after GPR35 activation by KYNA or zaprinast. We found that both compounds inhibited N-type Ca2+ channels in neurons heterologously expressing GPR35. The modulation was PTX-sensitive, suggesting coupling of GPR35 to the endogenous Gi/o protein. We also observed tonic activity of GPR35, which was blocked by PTX treatment. Although there was a population of GPR35b-expressing neurons that did not respond to ligand application, GPR35a and GPR35b displayed no apparent difference in receptor expression pattern.
Materials and Methods
Cloning of GPR35a and GPR35b. GPR35a was amplified from human genomic DNA using PCR and PfuUltra DNA polymerase (Stratagene, La Jolla, CA). Oligonucleotide primers were designed on the basis of AF089087 (GenBank accession number) for subcloning into the mammalian expression vector pCI (Promega, Madison, WI) as follows: forward (MluI site underlined), 5′-GATCACGCGTACCATGAATGGCACCTACAACACCTG-3′; and reverse (NotI site underlined), 5′-GATCGCGGCCGCTTAGGCGAGGGTCACGCACAGAG-3′. Fluorescent protein fusion constructs were made in the plasmid Venus-N1, which contains a variant of the yellow fluorescent protein (Nagai et al., 2002). Venus-N1 was constructed from YFP-N1 (Clontech Laboratories, Mountain View, CA) using Quik-Change Multi (Stratagene) site-directed mutagenesis. For subcloning as a fusion protein into Venus-N1 the following primers were used: forward (BglII site underlined), 5′-GATCAGATCTACCATGAATGGCACCTACAACACCTG-3′; and reverse (HindIII site underlined) 5′-GATCAAGCTTGGCGAGGGTCACGCACAGAG-3′. The blunt PCR products were subcloned into the vector pCRBluntII TOPO (Invitrogen, Carlsbad, CA) for DNA sequencing. Two GPR35a clones were identified: 1) a sequence identical with GenBank accession number AF089087, and 2) an A880C substitution (referenced to the open reading frame) corresponding to a nonsynonymous single nucleotide polymorphism (S to R at residue 294). The latter variant has been reported in mRNA sequences CR541765, BC117453, and BC117455. The R294 variant was the first one cloned into the vectors pCI and Venus-N1. Consequently, the S294 version was generated by QuikChange.. site-directed mutagenesis from the R294 parent construct.
GPR35b was amplified by PCR from human whole brain cDNA (Clontech Laboratories) using PfuUltra polymerase. N-terminal primers were derived from the protein sequence of Okumura et al. (2004) and human genomic sequence AF158748: pCI forward (MluI site underlined), 5′-GATCACGCGTACCATGCTGAGTGGTTCCCGGGCTGTC-3′; and Venus-N1 forward (BglII site underlined), 5′-GATCAGATCTACCATGCTGAGTGGTTCCCGGGCTGTC-3′. C-terminal primers were those used above for GPR35a. The PCR products were subcloned into the vectors pCI and Venus-N1 as above. Sequencing revealed that both GPR35b clones coded for R324 at the residue cognate to position 294 of GPR35a. All constructs were fully sequenced using a CEQ 8000 Automated DNA Sequencer (Beckman-Coulter, Fullerton, CA). Hereafter, the cDNAs cloned into pCI are denoted GPR35a(S294) and GPR35b. The fluorescent protein fusion constructs cloned into Venus-N1 constructs are denoted as GPR35a(S294)-Venus, GPR35a(R294)-Venus, and GPR35b-Venus, respectively.
Isolation of Rat SCG Neurons and Microinjection of cDNA. The protocols for dissociation of rat SCG neurons and intranuclear injection of cDNAs have been described in detail previously (Ikeda, 2004; Ikeda and Jeong, 2004). Briefly, adult male Wistar rats were anesthetized with CO2 inhalation and then decapitated (as approved by the Institutional Animal Care and Use Committee). SCGs were isolated from adjacent connective tissues, minced, and put into a flask containing 6 ml of modified Earle's balanced salt solution with enzymes (0.7 mg/ml collagenase D, 0.3 mg/ml trypsin, and 0.05 mg/ml DNase I). The flask with SCG fragments was then shaken at 36°C for 1 h. The neurons were subsequently dissociated by shaking the flask forcefully for approximately 15 times, washed, and resuspended in minimum essential medium (MEM) containing 10% fetal calf serum, 100 U/ml penicillin, and 100 μg/ml streptomycin, plated onto poly-l-lysine-coated tissue culture dishes and placed in a 5% CO2/95% air incubator at 37°C.
All cDNA constructs were stored at -20°C as ∼1 μg/μl stock solution in TE buffer (10 mM Tris and 1 mM EDTA, pH 8). A FemtoJet 5247 microinjection unit and 5171 micromanipulator (Eppendorf, Madison, WI) were used to inject DNA at a pipette concentration of 200 ng/μl. In some experiments, pEGFP vector plasmid DNA was coinjected at a concentration of 50 ng/μl to aid in identifying successfully injected neurons.
Cell Culture, Transfection, and Imaging. HeLa cells were cultured in modified MEM and transfected with cDNAs using fully deacylated polyethylenimine (PEI) (Thomas et al., 2005). A mixture of 1 μg of GPR35a(S294)-Venus and 4 μl of PEI at 7.5 mM or a mixture of 1 μg of GPR35b-Venus and 4 μl of PEI at 7.5 mM was added to 150 μl of Opti-MEM and incubated for 20 min at room temperature. The transfectant mixture was then added to the cell culture dishes.
The transfected HeLa cells were imaged using a 60 × 1.4 numerical aperature oil-immersion objective mounted on a Nikon inverted fluorescence microscope, a 12-bit cooled charge-coupled device camera (Orca-ER, Hamamatsu, Japan), and Volocity 4 software (Improvision Inc., Lexington, MA). SCG neurons were imaged with an Achroplan infrared 63 × 0.90 numerical aperature water-immersion objective mounted on an Zeiss Axioplan 2 microscope (Carl Zeiss, Jena, Germany). The microscope was equipped with a Ti-sapphire laser (Chameleon; Coherent, Santa Clara, CA) for two-photon imaging. Zeiss software release 3.2 was used for image acquisition.
Ca2+ Current Recordings. Whole cell Ca2+ channel currents (ICa) were recorded from rat SCG neurons using the patch-clamp technique as described previously (Guo and Ikeda, 2005). Custom designed software (S5) was used for generating voltage protocols. Traces were digitized and stimuli delivered with an ITC-18 data acquisition interface (InstruTECH, Port Washington, NY). ICa traces were filtered at 2 kHz (-3 dB, four-pole Bessel filter) and digitized at 10 kHz. A double-pulse protocol consisting of two 25-ms test pulses to +10 mV separated by a 50-ms conditioning pulse to +80 mV (Elmslie et al., 1990) was used to evoke ICa. All recordings were performed at room temperature (22–25°C).
Solutions and Chemicals. The external recording solution contained 140 mM methanesulfonic acid, 145 mM tetraethylammonium hydroxide (TEA-OH), 10 mM HEPES, 10 mM glucose, 10 mM CaCl2; and 0.0003 mM tetrodotoxin, pH 7.4, with TEA-OH. The internal solution contained 120 mM N-methyl-d-glucamine, 20 mM TEA-OH, 11 mM EGTA, 10 mM HEPES, 10 mM sucrose, 10 mM HCl, 1 mM CaCl2, 4 mM MgATP, 0.3 mM Na2GTP, and 14 mM Tris creatine phosphate, pH 7.2, with methanesulfonic acid. The osmolalities of the bath and pipette solutions were adjusted with sucrose to 325 and 300 mosmol/kg, respectively. Both KYNA (Tocris Cookson, Ellisville, MO) and zaprinast (Sigma-Aldrich, St. Louis, MO) were freshly prepared before experiments. The KYNA stock solution (1 mM in external recording solution) was sonicated to aid dissolution. Zaprinast was dissolved in dimethylsulfoxide (Sigma-Aldrich) at a concentration of 10 mM. Both drugs were then diluted to the final concentration with external recording solution. PTX (List Biological Laboratories Inc., Campbell, CA) was prepared in H2O at a concentration of 100 μg/ml. MEM, Opti-MEM, penicillin, and streptomycin were purchased from Invitrogen, collagenase D from Roche Applied Science (Indianapolis, IN), trypsin (type TRL) from Worthington Biochemicals (Freehold, NJ), and DNase I from Sigma-Aldrich.
Data Analyses and Statistics. Igor Pro software (WaveMetrics, Lake Oswego, OR) was used to analyze ICa. The data are presented as mean ± S.E.M. with n representing the number of cells tested. The percentage of ICa inhibition was calculated as (I1 - I2)/I1 × 100, where I1 and I2 were the prepulse ICa values in the absence or presence of drug application. The prepulse and postpulse ICa values were measured 10 ms after initiation of the test pulse (+10 mV). The facilitation ratio was determined as the ratio of postpulse to prepulse ICa in the absence or presence of drug application. The concentration-response curves were fit with a Hill equation: I = Imax/{1 + (EC50/ [agonist])c}, where I, Imax,EC50, [agonist], and c are percent inhibition, maximum inhibition, concentration producing a half-maximum response, agonist concentration, and Hill coefficient, respectively. Prism 4 software (GraphPad Software, Inc., San Diego, CA) was used for statistical comparisons determined by the unpaired t test, one-way ANOVA followed by Newman-Keuls test, repeated-measures ANOVA, or Pearson correlation, as noted. P < 0.05 was regarded as significant.
Results
Kynurenic Acid-Induced ICa Inhibition via GPR35a Activation. To examine the coupling of GPR35a with endogenous G proteins and effectors, we used sympathetic neurons as surrogate hosts for heterologous expression. We first examined the effect of KYNA application on ICa recorded from uninjected SCG neurons (Fig. 1A) to establish whether the preparation possessed a suitable null background. ICa was evoked at 0.1 Hz using a double-pulse voltage protocol (Fig. 1A, left) in solutions designed to isolate ICa. Under these conditions, the major component of current (65–90%) originated from ω-conotoxin GVIA-sensitive N-type Ca2+ channels (CaV2.2) (Ikeda, 1991; Zhu and Ikeda, 1994). The facilitation ratio (Fig. 1, □), defined as the ratio of ICa evoked in the second test pulse (postpulse, Fig. 1, ○) to ICa in the first test pulse (prepulse, Fig. 1, •) was used as a measure of Gβγ-mediated ICa modulation (Ikeda, 1991, 1996). The basal facilitation ratio (BFR), i.e., facilitation ratio in the absence of drug application, was 1.23 ± 0.03 (n = 4). Application of KYNA (300 μM) for 30 s (Fig. 1A, right) did not alter ICa (Fig. 1, A and D), indicating a lack of functional native GPR35 expression in SCG neurons. Likewise, application of KYNA (300 μM) to SCG neurons expressing EGFP produced no significant ICa inhibition (0.1 ± 0.8%, n = 4). In contrast, ICa recorded from a neuron previously injected (approximately 18–24 h) with GPR35a(S294) cDNA (Fig. 2B, left) displayed two characteristic properties of voltage-dependent ICa inhibition mediated via Gβγ-subunit: 1) slowed activation in the prepulse; 2) partial relief of inhibition by a depolarization conditioning pulse (Ikeda, 1996; Herlitze et al., 1996; Ikeda and Dunlap, 1999) as shown by an increase in the facilitation ratio (Fig. 1B, □). In the presence of KYNA (300 μM), mean ICa was inhibited by 38 ± 6% (n = 6, Fig. 1D, ) and the mean facilitation ratio was increased from 1.44 ± 0.09 to 2.11 ± 0.29 (n = 6). Figure 1B (right) illustrates the time course of KYNA-induced ICa inhibition. As for other lipid-soluble compounds (Guo and Ikeda, 2004), KYNA produced a slowly developing inhibition requiring approximately 60 s to reach a steady-state that slowly recovered after removal of the drug (neurons were superfused with recording solution before, during, and after drug application). The kinetics of changes in facilitation ratio are also evident in Fig. 1B (right, □). To provide a visual indicator for GPR35a(S294) expression, the C terminus of the receptor was tagged with a fluorescent protein, Venus, an improved variant of enhanced yellow fluorescent protein (Nagai et al., 2002). Because the C terminus of many GPCRs is crucial for receptor trafficking and signal transduction, we tested whether fusion of the fluorescent protein affected the function of GPR35a(S294). Results obtained following expression of GPR35a(S294)-Venus were similar to those illustrated in Fig. 1B. The mean ICa inhibition was 51 ± 5% (n = 5), a value not significantly different from the inhibition obtained with GPR35a(S294) expression (Fig. 1D, ). When examined with fluorescence microscopy, neurons injected with GPR35a(S294)-Venus cDNA displayed a granular fluorescence pattern and nuclear exclusion (shown later).
Three single nucleotide polymorphisms of GPR35a have been reported previously (Horikawa et al., 2000). During generation of the cDNA clones, we identified a nonsynonymous substitution, A880C, which codes for the R294 isoform of GPR35a. To examine the function of GPR35a(R294), SCG neurons were injected with a GPR35a(R294)-Venus fusion construct. Figure 1C shows that application of KYNA (300 μM) produced voltage-dependent ICa modulation similar to that observed in GPR35a(S294)-expressing neurons. Average ICa inhibition in GPR35(R)a-Venus-expressing neurons was 36 ± 4% (n = 5; Fig. 1D, ▪).
Zaprinast-Induced ICa Modulation via GPR35a Activation. In addition to KYNA, zaprinast, a synthetic phosphodiesterase inhibitor, has been identified as a putative agonist for GPR35 (Taniguchi et al., 2006). With the strategy used in Fig. 1, we examined the effects of zaprinast application on ICa modulation. Zaprinast (10 μM) did not affect mean ICa inhibition in uninjected neurons (2 ± 1%, n = 4) (Fig. 2B, □) or neurons expressing EGFP (0.6 ± 1.2%, n = 4) providing additional evidence for a lack of functional GPR35 expression in SCG neurons. Like KYNA, application of zaprinast (10 μM) resulted in ICa inhibition in GPR35a(S294)-expressing neurons. Zaprinast-induced ICa modulation was also voltage-dependent, based on the slowing of activation (Fig. 2A, left) and increased facilitation ratio (from 1.52 ± 0.09 to 2.88 ± 0.46, n = 4) (Fig. 2A. right, □). Figure 2A (right) depicts the time course of zaprinast-induced ICa inhibition which was relatively slow in onset, usually taking approximately 60 s to reach a steady state. The washout of zaprinast was even slower, taking several minutes to complete. Mean ICa inhibitions produced by 10 μM zaprinast were 59 ± 8(n = 4), 55 ± 4(n = 4), and 42 ± 4% (n = 4) for neurons injected with GPR35a(S294), GPR35a(S294)-Venus, or GPR35a(R294)-Venus cDNAs, respectively (Fig. 2B). There was no significant difference among zaprinast-induced mean ICa inhibition in the three constructs (P > 0.05; ANOVA).
Concentration Dependence of KYNA- and Zaprinast-Mediated ICa Inhibition.Figure 2C illustrates the concentration-response curves for ICa inhibition generated from GPR35a(S294)-Venus-expressing SCG neurons. Because of the slow drug washout, only a single concentration of agonist was applied to each neuron. The data points represent the mean inhibition (percent) calculated from two to eight cells (as labeled) for either KYNA (•) or zaprinast (○). These data were fit with a Hill equation using a nonlinear least-squares algorithm. From these analyses, the maximum inhibitions, EC50 values, and Hill coefficients were 56 and 59%, 58 and 1.1 μM, and 1.1 and 1.6 for KYNA and zaprinast, respectively. The data indicate that zaprinast was more potent than KYNA but had similar efficacy.
Signaling Pathways Mediating ICa Inhibition. A hallmark of ICa inhibition mediated by the Gβγ arm of the heterotrimeric G protein pathway is voltage dependence. At depolarized membrane potentials, the affinity of the channel for Gβγ is believed to decrease, thereby relieving ICa inhibition. Both the agonist-mediated slowing of current activation and increase in facilitation ratio shown previously (Figs. 1 and 2) are indicative of voltage-dependent Gβγ-mediated signaling (Ikeda and Dunlap, 1999). To examine the voltage dependence of GPR35a-mediated ICa modulation in greater detail, current-voltage (I-V) relationships in the absence or presence of KYNA were generated by evoking families of ICa with 70-ms test pulses over the range -120 to +80 mV from a holding potential of -80 mV. Figure 3A illustrates normalized mean I-V curves obtained in the absence (○) or presence of KYNA (•). ICa amplitude was normalized to the maximal ICa amplitude, which usually occurred at a step potential near +5 mV in the absence of KYNA. ICa inhibition appears most prominent at membrane potentials generating the largest ICa amplitude. To quantify this impression, ICa inhibition for five individual neurons obtained at test potentials of 0, +20, and + 40 mV are plotted in 3B. Although the degree of ICa inhibition varied for each neuron, the magnitude of inhibition consistently decreased with increasing test potential. Consequently, mean inhibition at each potential was significantly different (P < 0.001) as determined by repeated-measures ANOVA.
Although voltage dependence implicates the Gβγ-subunit as the mediator of ICa modulation, the composition of the G protein heterotrimer, especially the Gα-subunit is important for receptor coupling. Previous studies suggested that the Gi/o proteins couple efficiently to GPR35a. Therefore, SCG neurons were pretreated with PTX (500 ng/ml) for 8 to 24 h before agonist application and electrophysiological recording. Pretreatment with PTX completely abolished both KYNA (100 μM) and zaprinast (10 μM)-induced ICa inhibition in GPR35a-expressing neurons (1.9 ± 0.4%, n = 4and 1.4 ± 1%, n = 4, respectively) as illustrated in Fig. 3B. Taken together, these data indicate that Gβγ released from Gαi/o-containing G protein heterotrimers mediates ICa inhibition after agonist binding to GPR35a.
Tonic Activity of GPR35a. Although agonist occupancy is usually required for GPCR activation, some GPCRs, such as the CB1 cannabinoid receptor, 5-HT2c serotonin receptor, M1–M5 muscarinic acetylcholine receptor, and β2-adrenergic receptor, display tonic activity in the absence of overt ligand (de Ligt et al., 2000). In the course of these studies, we found that the expression of GPR35a significantly increased BFR (Fig. 4A), an indicator of tonic G protein modulation (Ikeda, 1991). The mean BFRs in GPR35a(S294)- and GPR35a(S294)-Venus-expressing neurons were 1.60 ± 0.12 (n = 13) and 1.50 ± 0.04 (n = 60), respectively, compared with 1.25 ± 0.03 (n = 24) in control (uninjected) neurons. The BFR in GPR35a(R294)-Venus injected neurons was 1.35 ± 0.06 (n = 13), which was not significantly different from that of uninjected neurons. In GPR35a-expressing neurons, there was a significant positive correlation between ICa inhibition induced by zaprinast or KYNA (both at supramaximal concentrations) and BFR (Pearson correlation coefficient, r = 0.46, P < 0.05) (Fig. 4B). As BFR is an indicator of tonic G protein activation, the correlation implies that the tonic activity of GPR35a is related to receptor expression level.
To assess whether the tonic activity of GPR35a required coupling to Gi/o proteins, we examined the effect of PTX on BFR in neurons expressing GPR35(S294)a-Venus. Mean BFR decreased from 1.52 ± 0.06 (n = 12) to 1.27 ± 0.02 (n = 8) after PTX treatment (Fig. 4C). The mean BFR between uninjected neurons (Fig. 4C, □) and PTX-treated GPR35(S294)a-Venus-expressing neurons (Fig. 4C, ▪) was not significantly different, suggesting that the tonic activity of GPR35 arose from activation of Gi/o protein.
GPR35b-Mediated ICa Inhibition. In humans, alternative splicing of the GPR35 primary transcript produces both GPR35a and GPR35b mRNA. The latter isoform is predicted to code for an additional 31 additional amino acids at the N terminus of the receptor (Okumura et al., 2004). As functional studies of GPR35b have not been published, we examined the effects of KYNA and zaprinast application on ICa modulation in GPR35b-expressing neurons. For GPR35b-injected neurons, two populations emerged on the basis of the response to agonist application. In four of seven cells, zaprinast (10 μM) produced significant ICa inhibition (>20%) (Fig. 5A, bottom). Similarly, KYNA (300 μM) inhibited ICa by >20% in five of eight cells (Fig. 5A, top). As in GPR35a-expressing neurons, the GPR35b-mediated ICa inhibition displayed characteristic voltage-dependent properties. The mean facilitation ratio increased from 1.70 ± 0.10 to 2.87 ± 0.13 in the presence of 10 μM zaprinast (n = 4) and from 1.38 ± 0.08 to 1.97 ± 0.20 in the presence of 300 μM KYNA (n = 5). Comparison of mean agonist-induced (either KYNA or zaprinast) ICa inhibition for either GPR35a- or GPR35b-expressing neurons revealed similar values when “nonresponding” cells were excluded from the analysis (Fig. 5B). To determine whether a fluorescent tag on the C terminus of GPR35b affected function, we monitored ICa inhibition in GPR35b-Venus-expressing neurons. In these cells, KYNA (300 μM) application produced a typical voltage-dependent ICa inhibition (data not shown) and there was no significant difference in mean ICa inhibition between GPR35b (37 ± 6%, n = 5)- and GPR35b-Venus (38 ± 3%, n = 5)-expressing neurons. Figure 5C illustrates the relationship between mean ICa inhibition in GPR35b-Venus-expressing neurons and KYNA concentration. When responses of <20% (shown as ○) were excluded, the mean data (▴) were well fit (-) with a Hill equation with maximum inhibition, EC50, and Hill coefficient of 49%, 66 μM, and 1.0, respectively. As in GPR35a-expressing neurons, agonist (zaprinast or KYNA)-induced ICa inhibition was significantly correlated to the BFR in GPR35b-expressing neurons (Pearson correlation coefficient, r = 0.70, P < 0.005) (Fig. 5D).
Subcellular Localization of GPR35-Venus Fusion Proteins. In contrast with GPR35a-mediated ICa modulation, injection of GPR35b cDNA into neurons produced highly variable responses. We thus examined whether the additional 31 amino acids in the N terminus of GPR35b affected the trafficking or expression pattern of the receptor. GPR35a (Fig. 6, left) or GPR35b (Fig. 6, right) fused to the fluorescent protein Venus were expressed in HeLa cells (Fig. 6, top row) or SCG neurons (Fig. 6, bottom row). HeLa cells were imaged using conventional wide-field fluorescence microscopy, whereas SCG neurons were imaged using two-photon microscopy. In GPR35a/b-expressing HeLa cells, fluorescence appeared in discrete cellular compartments as well as near (or on) the plasma membrane. The focal plane in these images was near the cell/cover glass interface and shows a similar fluorescence pattern for both GPR35 splice variants. Surprisingly, in the GPR35a/b-expressing SCG neurons fluorescence was found throughout the cytoplasm, was excluded from the nucleus, and showed no obvious enrichment near the plasma membrane (“rim staining”) as would be expected (given that both GPR35 constructs were functional). The fluorescence pattern in the cytoplasm was granular and thus different from the even distribution seen when soluble (i.e., nonfusion) fluorescent proteins were expressed. Thus, we are limited to the interpretation that the fluorescence patterns of GPR35a and GPR35b were similar and thus the additional residues found in GPR35b did not overtly alter trafficking.
Discussion
In the present study, we demonstrated that GPR35, when heterologously expressed in sympathetic neurons, couples via native heterotrimeric G proteins to modulate N-type Ca2+ channels. Two recently discovered GPR35 agonists, kynurenic acid and zaprinast, activated GPR35 to produce ICa inhibition that was PTX-sensitive and voltage-dependent. Zaprinast was more potent than kynurenic acid but possessed similar efficacy. In addition, the receptor appeared to activate PTX-sensitive G protein in the absence of overt agonist. Two isoforms of GPR35a and a splice variant, GPR35b, produced similar effects although variability was greater with the latter. Expression of fluorescent protein fusion constructs of GPR35a and GPR35b revealed similar expression patterns from which limited inference could be drawn.
Although gene analysis studies implicate GPR35 in gastric cancer (Okumura et al., 2004), Albright hereditary osteodystrophy (Shrimpton et al., 2004), and type II diabetes (Vander Molen et al., 2005), basic information regarding the pharmacology, physiology, and function of GPR35 is scarce. In humans, high levels of GPR35 mRNA were detected in the immune and digestive systems, including peripheral leukocytes, spleen, small intestine, colon, and stomach with levels reported for brain (O'Dowd et al., 1998; Wang et al., 2006). However, GPR35a cDNA was isolated from a human dorsal root ganglion cDNA (Taniguchi et al., 2006), and, in this study, we cloned GPR35b from human whole brain cDNA. In addition, high throughput in situ hybridization studies indicate substantial levels of GPR35 mRNA (Lein et al., 2007) throughout the mouse brain. Therefore, there is accumulating evidence for the expression of GPR35 in both the central and peripheral nervous systems.
Recently, KYNA (Wang et al., 2006) and zaprinast (Taniguchi et al., 2006) were identified as GPR35 agonists using high-throughput screening methods. Kynurenic acid is a tryptophan metabolite in the kyneurenine pathway. Although serotonin and melatonin, two other tryptophan metabolites, are well known ligands for GPCRs, there is limited knowledge about the function of KYNA. In the central nervous system, KYNA is produced by astrocytes and neurons and exerts a neuroprotective effect. Kynurenic acid is a potential endogenous antagonist of N-methyl-d-aspartate receptors and α7 nicotinic receptors (Stone, 2000; Hilmas et al., 2001) and has a dual action on the α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor (Prescott et al., 2006). At lower KYNA concentrations (nanomolar to micromolar), α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor responses are facilitated, whereas higher KYNA concentrations (millimolar) antagonize the receptor. In the present study, we showed that KYNA inhibited neuronal N-type Ca2+ channels via activation of heterologously expressed GPR35. Thus, endogenous KYNA may play a role in the regulation of neuronal excitability and synaptic release by acting at GPR35. To our knowledge, this is the first study to examine the effect of KYNA on G protein-mediated Ca2+ channel modulation. The concentration of KYNA necessary to activate GPR35 (10 μM–1 mM) is within the range used to probe ionotropic glutamate receptor function (Prescott et al., 2006). However, levels of KYNA in extracellular fluid of the rat striatum is approximately 17 nM (Swartz et al., 1990), approximately 3 orders of magnitude less than the EC50 for GPR35 activation reported here and in a previous study (Wang et al., 2006). Thus, although GPR35 activation should be considered a potential mechanism when interpreting data obtained with KYNA in a similar concentration range, another, as yet unidentified, substance probably serves as the endogenous ligand.
Zaprinast is characterized as a selective cGMP-specific PDE5 inhibitor and, as such, is used as a tool to identify signal transduction pathways incorporating generation/degradation of cGMP (reviewed by Gibson, 2001). Therefore, the discovery that zaprinast serves as a relatively potent GPR35 agonist (Taniguchi et al., 2006) affects the interpretation of data in which zaprinast was assumed to solely affect cGMP degradation. For example, zaprinast induced a significant depression of basal synaptic transmission in rat hippocampus attributed to elevation of the intracellular cGMP level (Monfort et al., 2002; Makhinson et al., 2006). However, our data provide evidence for the potential modulation of presynaptic Ca2+ channels by zaprinast acting as a GPR35 agonist. It should be noted, however, that the presence of GPR35 in presynaptic terminals and the coupling of natively expressed GPR35 to Ca2+ channels have not been established. Although a direct modulation of N-type Ca2+ channels by cGMP has been reported (Meriney et al., 1994; Grassi et al., 2004), the zaprinast-induced Ca2+ channel inhibition seen here was not likely to have been due to increased intracellular [cGMP] because 1) zaprinast had no effect on ICa in the absence of GPR35 expression and 2) zaprinast produced ICa modulation with all of the hallmarks of a direct Gβγ effect.
In this study, we also provided evidence that GPR35a and GPR35b were tonically active (i.e., in the absence of overt agonist) when expressed in rat sympathetic neurons (Figs. 4 and 5). After GPR35 expression, the BFR, a sensitive indicator of tonic G protein activation (Ikeda, 1991), was elevated. This elevation was abolished after PTX treatment implicating the activation of endogenous Gi/o proteins. Tonic GPCR activity probably depends on receptor cell surface density (de Ligt et al., 2000; Cotecchia, 2007). For example, increasing the density of β2-adrenergic receptors results in the elevation of basal receptor-mediated generation of cAMP (Chidiac et al., 1994). In this study, there was a positive correlation between BFR and the degree of agonist-induced ICa inhibition, suggesting that receptor number contributes to the tonic activity of GPR35. In addition, these data suggest that receptor levels required for full agonist efficacy also produce tonic activation. Consequently, tonic GPR35 activity may not require artifactual levels of expression and thus contribute to the function of GPR35 within a physiological context. It should be noted, however, that, in the absence of a well characterized neutral antagonist, one cannot exclude the existence of endogenous agonists. Thus, the interpretation that tonic activity equates to constitutive activity arising from nonligand bound receptors remains an open question.
At present, there are no studies comparing the coupling of GPR35 isoforms and splice variants to endogenous signaling systems. We examined two isoforms of GPR35a (S294 and R294) and the splice variant GPR35b, which contains an additional 31 amino acids at the N terminus. In general, the biophysical characteristics of ICa modulation and the pharmacological parameters determined after expression of the various GPR35 cDNAs were similar. The only apparent anomaly was an increased response variability after GPR35b expression. Suspecting a potential difference in cellular trafficking between GPR35a and GPR35b, we examined the fluorescence protein of fusions of GPR35 expressed in either HeLa cells or SCG neurons. Surprisingly, although the fusion proteins were fully functional in terms of ICa inhibition, the vast majority of fluorescence was located in the cytosol (although possibly on internal membranes) of SCG neurons instead of concentrated near the cell surface. Thus, we can infer only that the expression patterns of GPR35a and GPR35b were superficially similar and thus the mechanism underlying the variability in GPR35b responses remains unclear. Taken together, the results of the present study demonstrate the coupling of GPR35 to endogenous G proteins that modulate neuronal Ca2+ channel and thereby provide evidence for a potential role of GPR35 in regulating neuronal excitability and synaptic transmission.
Acknowledgments
We thank Dr. Steven S. Vogel for assisting with two-photon confocal imaging.
Footnotes
-
This work was supported by the intramural program of the National Institutes of Health, National Institute on Alcohol Abuse and Alcoholism, Bethesda, MD.
-
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
-
doi:10.1124/jpet.107.127266.
-
ABBREVIATIONS: GPCR, G protein-coupled receptor; KYNA, kynurenic acid (4-oxo-1H-quinoline-2-carboxylic acid); PDE, phosphodiesterase; PTX, Bordetella pertussis toxin; PCR, polymerase chain reaction; SCG, superior cervical ganglion; MEM, minimum essential medium; EGFP, enhanced green fluorescent protein; PEI, polyethylenimine; TEA-OH, tetraethylammonium hydroxide; ANOVA, analysis of variance; BFR, basal facilitation ratio.
- Received June 14, 2007.
- Accepted October 10, 2007.
- The American Society for Pharmacology and Experimental Therapeutics