Unique isoform of Galpha -interacting protein (RGS-GAIP) selectively discriminates between two Go-mediated pathways that inhibit Ca2+ channels.

Regulators of G-protein signaling (RGS) proteins constitute a large family of GTPase-activating proteins for heterotrimeric G proteins. More than 20 RGS genes have been identified in mammals. One of these, the Galpha-interacting protein (GAIP), preferentially interacts with members of the G(i)/G(o) subfamily of G proteins in mammalian cells, but its selectivity among members of this subfamily in vitro is limited. Here we report the cloning and functional characterization of a unique cDNA isoform of GAIP, derived from embryonic chicken dorsal root ganglion neurons. Chick GAIP is composed of 199 amino acids, organized into a conserved RGS domain (85% identical to human GAIP), and a unique, short N terminus (only 41% identical, 50% homologous to known mammalian orthologues). Consistent with this unique primary structure, chick GAIP has physiological properties that distinguish it from mammalian GAIPs. We have explored the selectivity of chick GAIP in electrophysiological assays of two G(o)-mediated forms of Ca(2+) channel inhibition produced by gamma-aminobutyric acid in chick dorsal root ganglion neurons, voltage-independent inhibition (mediated by G(o)alpha) and voltage-dependent inhibition (mediated by G(o)betagamma). Dialyzing recombinant chick GAIP in these cells selectively reduced voltage-independent inhibition without affecting voltage-dependent inhibition. Mammalian GAIP, tested under identical conditions in previous studies, demonstrated no selectivity between these two inhibitory processes; thus, our results suggest that the functional specificity of chick GAIP is likely to be determined by its unique N terminus.

Regulators of G-protein signaling (RGS) proteins constitute a large family of GTPase-activating proteins for heterotrimeric G proteins. More than 20 RGS genes have been identified in mammals. One of these, the G␣-interacting protein (GAIP), preferentially interacts with members of the G i /G o subfamily of G proteins in mammalian cells, but its selectivity among members of this subfamily in vitro is limited. Here we report the cloning and functional characterization of a unique cDNA isoform of GAIP, derived from embryonic chicken dorsal root ganglion neurons. Chick GAIP is composed of 199 amino acids, organized into a conserved RGS domain (85% identical to human GAIP), and a unique, short N terminus (only 41% identical, 50% homologous to known mammalian orthologues). Consistent with this unique primary structure, chick GAIP has physiological properties that distinguish it from mammalian GAIPs. We have explored the selectivity of chick GAIP in electrophysiological assays of two G o -mediated forms of Ca 2؉ channel inhibition produced by ␥-aminobutyric acid in chick dorsal root ganglion neurons, voltage-independent inhibition (mediated by G o ␣) and voltage-dependent inhibition (mediated by G o ␤␥). Dialyzing recombinant chick GAIP in these cells selectively reduced voltageindependent inhibition without affecting voltagedependent inhibition. Mammalian GAIP, tested under identical conditions in previous studies, demonstrated no selectivity between these two inhibitory processes; thus, our results suggest that the functional specificity of chick GAIP is likely to be determined by its unique N terminus.
Heterotrimeric GTP-binding proteins (G proteins) 1 initiate a heterogeneous array of cellular functions. The time course of G protein signaling is generally limited by intrinsic GTPase activity of the G␣ subunits and reassociation of the G␣␤␥ heterotrimer. A number of accessory mechanisms capable of altering this time course have been described over the past few years. Most recent is a family of multifunctional proteins called regulators of G protein signaling (RGS) proteins (1-4) that share a well conserved 120-amino acid "RGS domain." This domain interacts directly with G␣-GTP subunits (5) and carries out a key function of RGS proteins to accelerate the intrinsic GTPase activity of G␣-GTP, thereby terminating signaling through reassociation of the heterotrimer (3,6). The GTPase-accelerating (or GAP) activity has been demonstrated to alter the time course of many G protein-dependent cellular functions, including membrane trafficking (7,8), lymphocyte chemotaxis (9), hormone signaling (10,11), rod photoreception (12,13), K ϩ channel activation (14,15), and Ca 2ϩ channel inhibition (16 -19).
As observed with GAPs for monomeric G proteins (20,21), some specificity of interaction has been described between RGS proteins and heterotrimeric G␣ subunits, but the specificity appears to be surprisingly limited for many RGS proteins (22)(23)(24)(25). Given that such assays are often carried out on recombinant proteins in vitro, it is possible that assay conditions give rise to artifactually low specificity. As most RGS proteins contain additional protein-protein interaction domains (e.g. cysteine string, PTB, GGL, DEP, and PDZ motifs), such domains are likely determinants of RGS-target interactions (19,26,27). Studies of RGS proteins in intact cells, particularly primary cells in which G protein-dependent signaling pathways have been well characterized, are likely to shed some light on this issue.
Sensory neurons from embryonic chick have proven to be a valuable model of differentiated cells in which to test specificity of G protein-dependent signaling (17, 28 -30). Acting through metabotropic GABA type B receptors on these neurons, the neurotransmitter GABA inhibits voltage-gated Ca 2ϩ channels through two distinct G o -mediated pathways. One pathway reduces Ca 2ϩ current via G o ␣ and a Src-like tyrosine kinase (28,31). A second pathway reduces current via a direct interaction between G o ␤␥ and the channel (28,32). These two pathways also exhibit distinct biophysical characteristics. The G o ␤␥-mediated inhibition is voltage-dependent (VD) and can be reversed at very positive voltages, whereas the G␣/Src-mediated inhibition is voltage-independent (VI) (28,33).
Previous experiments carried out in the Dunlap laboratory (17) tested human RGS-GAIP on these chick neurons through intracellular application of the recombinant protein; it was found to terminate rapidly both types of G o -mediated inhibition without affecting the G i -dependent form of Ca 2ϩ channel inhibition also present in the cells. This specificity for G o was somewhat surprising in light of the significant GAP activity of human GAIP demonstrated in vitro against most G i ␣ and G o ␣ subunits tested (6,34), underscoring the potential importance of the cellular environment.
We have further explored this issue of GAIP specificity by using a novel GAIP cDNA, cloned from a chick sensory neuron library. We report here that chick GAIP has a short N terminus that is phylogenetically divergent from mammalian orthologues. By using in vitro GTPase assays with recombinant G␣ subunits and electrophysiological recording of G o -mediated Ca 2ϩ channel inhibition, we demonstrate that chick GAIP is more selective than its human counterpart both in vitro and in intact cells. These results suggest that the unique N terminus of chick GAIP confers functional specificity both in vitro and in vivo.

EXPERIMENTAL PROCEDURES
Reverse Transcriptase-PCR-Dorsal root ganglion neurons were dissected from 11-or 12-day-old chicken embryos (Charles River SPAFAS, North Franklin, CT) as described (35), and total RNA was extracted by a phenol/guanidinium thiocyanate method (Tel-Test Inc., Friendswood, TX). Freshly obtained RNA (0.5-1 g) was used as template for reverse transcription using Moloney murine leukemia-virus reverse transcriptase (2.5 units/l), 4 mM dNTP mix, and a 1:1 mix of random hexamer and oligo(dT) (2.5 M) as primer. The reaction mixture was heated at 42°C for 15 min for first strand synthesis. After addition of 1 unit of Taq DNA polymerase and 1 M each of degenerate sense (5Ј-GAG AAC ATG(T) C(T)TC (A)TTC TGG-3Ј) and antisense (5Ј-TAG(T) GAG TCC CG(T)G TGC AT-3Ј) primers, which target the highly conserved RGS domain of mammalian GAIP, PCR was performed using the following cycling protocol: 94°C for 1 min (1 cycle), 94°C for 30 s, 45°C for 1 min, 72°C for 1 min (40 cycles), and 72°C for 5 min (1 cycle). Reaction products of the expected size (240 bp) were subcloned into pCR2.1 (Invitrogen), mapped by restriction digestion, and sequenced. Three independent clones yielded identical partial fragments of the chick GAIP RGS domain.
cDNA Library Screening-An embryonic chick DRG cDNA library (in ZAP II vector) was constructed as described (36). In each round of screening, about 1 million plaques were transferred to Nytran membranes (Schleicher & Schuell) in duplicate. Phage DNA was denatured (0.5 M NaOH, 1.5 M NaCl) and neutralized (0.5 M Tris-HCl, 1.5 M NaCl, pH 8.0). A 240-bp cDNA probe, corresponding to the PCR-amplified fragment of the chick GAIP RGS domain (see above), was labeled with [␣-32 P]dCTP using a random priming kit (Amersham Biosciences) and added to a hybridization solution of 50% deionized formamide, 0.8 M NaCl, 20 mM PIPES, 0.5% SDS, 100 g/ml denatured salmon sperm DNA. After hybridization overnight at 42°C, membranes were washed twice at 58°C for 45 min in wash buffer (0.1ϫ SSC, 0.1% SDS). Doublepositive plaques were isolated, amplified, and subjected to a second round of screening. Individual plaques positive after the second screening were subjected to in vitro excision using the Exassist/SORL system (Stratagene, La Jolla, CA). Each phage gave rise to a phagemid (pBluescript) containing an individual cDNA clone. The presence of a cDNA insert in the phagemid was confirmed by restriction digestion and sequence analysis.
Northern Blot Analysis-Twelve-day-old chicken embryos were dissected, and total RNA was extracted from several tissues as described above. Poly(A) ϩ RNA samples were then obtained with the poly(A) ϩ Tract mRNA Purification System (Promega, Madison, WI). An agarose gel (0.7% agarose, 2.2 M formaldehyde, 20 mM MOPS, 8 mM sodium acetate, 1 mM EDTA, pH 7.0) was loaded with poly(A) ϩ RNA from each tissue (1.5 g); following electrophoresis, the RNA was transferred to a Nytran membrane (Schleicher & Schuell) by alkaline transfer (8 mM NaOH, 3 M NaCl, 1.5 h). A 580-bp PCR product containing most of the coding region and part of the 3Ј-UTR of chick GAIP was labeled with [␣-32 P]dCTP by random priming (see above), and 10 7 dpm/ml of probe were added to the hybridization solution (Northern MAX Prehyb/Hyb Buffer, Ambion, Austin, TX). The blot was hybridized overnight at 42°C and then washed twice (45 min at 60°C) in Northern MAX Wash Buffer (Ambion). After drying, the membranes were analyzed by autoradiography.
Expression and Purification of Recombinant Chick GAIP-A His 6 tag was attached by PCR to the C terminus of chick GAIP; the resulting construct was directionally subcloned into pET11a (Novagen, Madison, WI), and the vector was used to transform Escherichia coli BL21 (DE3) cells. A 10-ml overnight culture from a single colony was used to inoculate 1 liter of LB medium with ampicillin (100 g/ml) at 37°C. Isopropyl-1-thio-␤-D-galactopyranoside (0.5 mM) induction was performed at A 600 ϭ 0.6 -0.8, and cell cultures were shaken for 4 h prior to harvest. Cells were then centrifuged, resuspended in buffer A (50 mM HEPES, pH 8, 20 mM ␤-mercaptoethanol, 100 mM NaCl, 0.1 mM phenylmethylsulfonyl fluoride), and lysed by addition of lysozyme (0.2 mg/ ml) and sonication. DNase I (5 g/ml) was added to reduce viscosity. The resulting lysate was centrifuged (12,000 ϫ g for 30 min at 4°C), and the supernatant fraction was loaded onto a column containing 1 ml of nickel-nitrilotriacetic acid affinity resin (Qiagen, Hilden, Germany) pre-equilibrated with buffer A. The column was washed with 50 ml of buffer A ϩ 50 mM imidazole, and GAIP was eluted with 5 ml of buffer A ϩ 250 mM imidazole. The eluted protein was dialyzed overnight against 1 liter of buffer B (50 mM HEPES, 2 mM dithiothreitol, pH 8), concentrated using a Centricon 10 column (Millipore, Bedford, MA), and stored at Ϫ80°C. GAIP protein preparations were 80 -90% pure as assessed by Coomassie Blue staining of conventional SDS-PAGE gels.
Single-turnover GTPase Assay-GTPase assays were performed using an automated fast-superfusion system with subsecond resolution (37). Bovine G␣ i1 , G␣ i3 , or G␣ o (2 l, 100 g/ml, Calbiochem) were incubated with 2 l of [␥-32 P]GTP (6000 Ci/mM, 10 Ci/liter) and 6 l of Mg 2ϩ -free HDE buffer (50 mM HEPES, 5 mM EDTA, 2 mM dithiothreitol, pH 8.0) for 30 min at 30°C to allow equilibrium binding. Samples were then stored on ice, and His 6 -tagged recombinant chick GAIP (3 M final concentration) was added to half. The labeled G␣ subunits, with or without GAIP, were retained on a 5-mm diameter HA/GFF/HA filter sandwich and placed in a superfusion chamber accessed by three high speed, solenoid driven valves. Each computer-operated valve controlled delivery of a separate solution to the chamber, allowing very rapid solution exchange. The minimal dead volume of the chamber and the relatively high solution flow rate permitted a time resolution for GAIP activity of at least 60 ms. The samples were superfused either with HDE buffer, to calculate background, or with HDE ϩ 10 mM MgCl 2 to trigger hydrolysis. Superfusate fractions were collected every 0.1 s for a total time of 1.3 s. Scintillation mixture (Hydrofluor, National Diagnostic, Manville, NJ) was added to each sample, and the amount of [␥-32 P]P i released by each sample was determined by liquid scintillation counting. For all experiments, samples were assayed in triplicate and averaged and expressed as a percentage of total radioactivity added to each filter.
Preparation and Electrophysiological Recordings of Chick DRG Neurons-Embryonic chick DRG neurons were grown in culture as described previously under conditions identical to those of Diversé-Pierluissi et al. (17). Briefly, dorsal root ganglia were dissected from 11-or 12-day-old chick embryos (Charles River SPAFAS) and incubated for 30 min at 37°C in Ca 2ϩ -, Mg 2ϩ -free saline containing 0.1% collagenase A (Roche Molecular Biochemicals). After washing away the collagenase, the ganglia were resuspended in ϳ1 ml of culture medium (Dulbecco's modified Eagle's medium, supplemented with 10% heat-inactivated horse serum, 5% chicken embryo extract, 50 units/ml penicillin, 50 mg/ml streptomycin, 1 mM glutamine and nerve growth factor) and mechanically dissociated by trituration through a small-bore, fire-polished Pasteur pipette. A small drop of the cell suspension (ϳ100 l) was placed in the center of a 35-mm tissue culture dish and incubated at 37°C to allow the cells to attach to the substrate. After ϳ1 h of incubation, the plated dishes were filled with culture medium.
Four to 48 h after plating, cultured DRG neurons were employed for electrophysiological recordings of Ca 2ϩ currents using standard tight seal, whole-cell methods (38). Cells were visualized on the stage of an inverted microscope, and the cell culture medium was replaced with (in mM) 133 NaCl, 1 CaCl 2 , 0.8 MgCl 2 , 25 HEPES, 12.5 NaOH, 5 glucose, 10 tetraethylammonium Cl, 6 ϫ 10 Ϫ4 tetrodotoxin, 10 Ϫ2 bicuculline, pH 7.4. All reagents were from Sigma. Recording pipettes were fabricated from microhematocrit tubing (Fisher) and filled with an internal recording solution containing (in mM) 150 CsCl, 5 1,2-bis(2-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid, 5 MgATP, 10 HEPES, pH 7.4. Pipette resistances varied between 0.8 and 1.5 megohms when filled with recording solution. Currents were recorded using a List Biologic (Campbell, CA) EPC-7 patch clamp amplifier, filtered at 3 kHz, digitized at 10 kHz using an ITC16 A/D interface (Instrutech Corp., Great Neck, NY), and stored on a Macintosh G3 PowerPC running Pulse software (HEKA Electronik, Germany). Capacitive transients and series resistances were compensated with the EPC-7 circuitry, and the leak currents were subtracted with a standard P/4 protocol. All recordings were performed at room temperature. Igor software (Wavemetrics, Lake Oswego, OR) was employed for data analysis.
The two components of GABA-mediated inhibition of N-type calcium channels were separated using voltage protocols described in our previous work (17,28). Briefly, Ca 2ϩ currents were evoked by a 20-ms test pulse to 0 mV in the presence or absence of GABA. Currents were measured at the time that control currents peaked (T p ) and at the end of the step (T e ), and modulated currents were normalized to control and expressed as a percentage. The difference between control and modulated current at T p constitutes the total GABA-induced inhibition. The difference between the current at T p and T e represents the VD portion of the inhibition produced by GABA; for convenience, the remaining current was taken as an estimate of voltage-independent inhibition. Although the VI component is contaminated by a portion of VD inhibition not effectively reversed during the test pulse, this contamination is likely no greater than 10% (35).
Ca 2ϩ currents were corrected for rundown by measuring Ca 2ϩ currents as a function of time in control cells in the absence of neurotransmitter. Analysis was confined to data from cells that exhibited a rundown of less than 1%/min.
Solutions and Drug Delivery-Drug and recording solutions were prepared from stock solutions immediately before the experiment; they were delivered rapidly (Ͻ1 s) via a pressurized system of multiple converging quartz tubes (140 m internal diameter) positioned in the vicinity of the cell of interest. For all experiments, GABA was diluted into the extracellular solution at a working concentration of 100 M, although GAIP was diluted into the patch pipette solution to a final concentration of 300 nM.
Statistics-All results are expressed as means Ϯ S.E. Statistical differences were analyzed using either paired or unpaired Student's t tests, as appropriate, and differences were considered significant when p Ͻ 0.05.

RESULTS
Cloning of Chick GAIP-RNA from chick DRG neurons was used as template for reverse transcriptase-PCR with degenerate primers derived from regions of the RGS domain that are highly conserved among RGS proteins (39). Along with a number of other RGS sequences, we obtained three independently cloned PCR products that encoded a 240-bp cDNA with 85% identity to human GAIP (6), suggesting that we had amplified the RGS domain of chick GAIP. In order to obtain the fulllength GAIP sequence, an embryonic chick DRG cDNA library was screened using a radioactive probe synthesized from the identified 240-bp PCR product. Three positive clones were isolated from the screening and sequenced. Two of them were cDNAs of identical length and contained the entire coding sequence of the target gene (Fig. 1).
Chick GAIP cDNA consists of 2164 bp, with a putative translation initiation codon at nucleotide 277 and a TAA stop codon at nucleotide 874. An in-frame stop codon occurring at nucleotide 175 and a satisfactory Kozak consensus sequence at nucleotide 271 confirm the ATG at bp 277 as the initiator methionine (40). The 3Ј-untranslated region does not contain a typical polyadenylation signal, AATAAA, but ends with an uninterrupted sequence of 13 adenosines. In addition, two TTTTGT motifs are present in the 3Ј-untranslated region (not shown). These motifs are commonly present in immediate early genes and are thought to play a role in transcriptional activation (41,42). The open reading frame consists of 597 bp, predicting a 199-amino acid, 23,321-Da protein (Fig. 1A). In agreement with human GAIP, the deduced primary sequence of chick GAIP contains a C-terminal RGS domain (Fig. 1A, double underlined) and an N-terminal cysteine-rich (CR) region (shaded). In addition, chick GAIP contains several putative phosphorylation sites for casein kinase II and one putative N-glycosylation site (not shown). The coding sequence was 66% identical to human GAIP and Ͼ50% identical to GzGAP and RGS17 (two more distant members of the GAIP family), confirming that the chick clone is a GAIP orthologue (Table I) (43). Whereas human and rat GAIPs have a similar evolutionary profile, chick GAIP branches off from a common precursor at an earlier point, reflecting a higher number of evolutionary changes in its primary structure (Fig. 1C).
The homology between chick and mammalian GAIPs is limited to the residues in the RGS and CR domains. The initial portion of the N terminus, from the initiator methionine to the beginning of the CR domain, is particularly divergent from mammalian GAIPs (Fig. 1B). The methionine at position 1 of human GAIP is a threonine (indicated with a black diamond) in chick GAIP, moving the start site of chick GAIP 66 nucleotides downstream. In addition, chick GAIP contains a 15-bp insertion not present in mammalian GAIPs immediately 5Ј to the CR domain. Because the N terminus has been shown to play an important role in the localization and biological activity of some RGS proteins (44 -47), the cloning results suggested that chick GAIP may have unique functional properties. The C terminus of chick GAIP also exhibits a few modifications, including a single residue deletion immediately following the RGS domain. Taken together, these changes give rise to a predicted chick protein that is 18 amino acids shorter than its human counterpart (Fig. 1A).
Tissue Distribution of Chick GAIP mRNA-The distribution of chick GAIP mRNA was investigated by probing a chick poly(A) ϩ Northern blot with a 480-bp cDNA fragment containing the coding region and part of the 5Ј-UTR of chick GAIP. As shown in Fig. 2 (upper panel), bands of comparable intensity corresponding to a transcript of 2.2 kb can be detected in all tissues tested as follows: DRG neurons, brain, lungs, liver, heart, skeletal muscle, and kidney. The length of the transcript is consistent with the length of the cloned chick GAIP cDNA. Interestingly, a second transcript of 1.5 kb, showing the same expression pattern as the 2.2-kb transcript but a much stronger signal, is also detected by the same probe. Neither band is present when the blot is stripped and re-hybridized with a probe to human glyceraldehyde-3-phosphate dehydrogenase, a gene commonly expressed in all tissues and used as a control for mRNA loading (Fig. 2, lower panel).
In Vitro GTPase Activity of Chick GAIP-The best described property of RGS proteins is their ability to act as GAPs for heterotrimeric G␣ subunits. To test whether chick GAIP exhibited GAP activity and to explore the extent of its specificity against G i/o ␣ subunits in vitro, we employed a novel biochemical assay that allows subsecond resolution of the time course of intrinsic GTP hydrolysis of G␣ and its acceleration by GAIP. Pre-formed complexes between purified, recombinant chick GAIP and bovine [␥-32 P]GTP-bound G␣ subunits were immobilized on glass fiber filters and rapidly superfused with a Mg 2ϩ -containing solution to initiate GTP hydrolysis (and the release of 32 P i into the superfusate). Eluted fractions were collected every 0.1 s, and their 32 P i content was measured with liquid scintillation spectroscopy. As shown in Fig. 3, chick GAIP effectively increased the GTPase activity of all G␣ subunits tested (G␣ i1 , G␣ i3 , and G␣ o ), although the rate and extent of the stimulation differed among the three G␣ subtypes. Surprisingly, the strongest GAP activity was observed for G␣ i1 and not G␣ i3 as has been reported for human GAIP (6,23). Chick GAIP stimulated an average 5.2 Ϯ 1.3-fold increase in the rate of G␣ i1 GTPase activity, producing a maximal stimulation of 12.3 Ϯ 0.4-fold (n ϭ 4, Fig. 3D). In contrast, the GTPase rates of G␣ i3 and G␣ o were augmented only 3.0 Ϯ 0.5-and 1.3 Ϯ 0.4-fold, respectively, producing maximal fold increases of 4.6 Ϯ 1.7 (n ϭ 8) and 1.7 Ϯ 0.2 (n ϭ 4) (Fig. 3, B and D). These data indicate that chick GAIP is a GAP for members of the G␣ i/o family (similar to mammalian GAIP), but in contrast to the latter, chick GAIP is most effective on G␣ i1 .
Effects of Chick GAIP on Ca 2ϩ Channel Gating-Prior to testing the actions of chick GAIP on G protein-dependent modulation of Ca 2ϩ current, we investigated whether chick GAIP modified channel properties under basal conditions. Macroscopic N currents were recorded over the course of 10 -15 min after achieving whole cell access. Pipettes contained either control internal solution or internal solution with 300 nM recombinant chick GAIP. Under the two conditions, there were no significant differences in the current-voltage relationships (Fig. 4A, n ϭ 4), current activation kinetics (Fig. 4B, n ϭ 4), or the conductance-voltage curve (Fig. 4C, n ϭ 4), the latter an indirect measure of the channel open probability. These results indicate that, as for other RGS proteins (16, 17), chick GAIP does not directly modify the gating of N-type Ca 2ϩ channels. .MM, mouse) and other members of the GAIP subfamily of RGS proteins. Cysteine-rich region (stippled bar) and the RGS domain (cross-hatched bar). Sequences were aligned using ClustalX on a Macintosh G3 PowerPC. The chick GAIP 5Ј-UTR immediately upstream of the initial methionine (arrowhead) was translated to compare with the N termini of mammalian GAIPs. Residues boxed and indicated with Ϫ22 aa are deleted in chick GAIP but present in mammalian GAIPs. Residues boxed and indicated with ϩ5 aa represent a 5-amino acid (aa) insertion in chick GAIP and are absent in mammalian GAIPs. C, unrooted tree of GAIP family members, generated by neighbor-joining distance methods from the alignment in B using ClustalX. This tree provides an approximation to the actual evolutionary tree, showing sequence relatedness as a function of branch length, where the branch length is proportional to the amount of change along that branch (sequence distance). Scale bar indicates a 5% change in amino acid sequence.

Chick GAIP Selectively Attenuates VI Inhibition of Ca 2ϩ
Current by GABA-In chick DRG neurons, GABA typically inhibits Ca 2ϩ current in more than 80% of freshly dissociated DRG neurons (48). Ca 2ϩ currents were measured with 20-ms test pulses to 0 mV, and 100 M GABA was applied twice, 30 s after break-in (prior to significant GAIP diffusion) and 10ϩ min after break-in (sufficient time for equilibration of GAIP with the cell interior) (Fig. 5A). This approach allows (i) GABAresistant cells to be eliminated from the analysis (Ͻ20% of cells tested), and (ii) each cell to be used as its own internal control.
GABA inhibited Ca 2ϩ currents to a similar extent in GAIPperfused (38.7 Ϯ 7.9%, n ϭ 5) and control (35.0 Ϯ 5.9%, n ϭ 10) cells when measured 30 s after break-in. In contrast, after 10 min of cell dialysis, GABA-induced inhibition in GAIP-perfused cells was only 22% that measured initially in the same cells. However, the second application of GABA produced a reduced inhibition also in control cells (71% of the initial application, n ϭ 10). This GAIP-independent decrease in inhibition in control cells was likely due to dialysis-induced changes of intra-cellular components essential for the inhibition. Comparing GAIP-treated with control cells at the 10-min time point demonstrated that GAIP significantly (p Ͻ 0.01) reduced GABAmediated inhibition (8.6 Ϯ 1.9% for GAIP-treated versus 25.0 Ϯ 4.8% for control) (Fig. 5C). These results indicate that chick GAIP is able to couple to and regulate the activity of the GABA type B receptor/G␣ o pathway(s) in DRG neurons.
Because GABA inhibits chick DRG Ca 2ϩ channels via two G o -coupled pathways, we explored whether chick GAIP exhibits any selectivity among the two. The two forms of inhibition (VD and VI) were separated with a 20-ms voltage pulse to 0 mV (see "Experimental Procedures"). Currents were measured at the time that control currents peaked (T p ) and at the end of the step (T e ), and modulated currents were normalized to control and expressed as a percentage. The difference between control and modulated current at T p constitutes the total GABA-induced inhibition (Fig. 5C). The difference between the current at T p and T e represents the VD portion of the inhibition pro-  2. Northern blot analysis of chick GAIP mRNA. Northern blot of mRNA purified from various tissues (from 12 day-old chick embryos), blotted with 32 P-labeled probe specific for chick GAIP (upper panel) or human glyceraldehyde-3-phosphate dehydrogenase (GPDH) (lower panel, same blot stripped and reprobed as a control for loading). Size markers at left are in kilobases. Arrows at right indicate sizes of the two identified GAIP transcripts (upper) or glyceraldehyde-3-phosphate dehydrogenase (lower). DRG, dorsal root ganglia; B, brain; H, heart; Lu, lung; Li, liver; SM, skeletal muscle; K, kidney.

FIG. 3. Chick GAIP is a GAP in vitro.
The ability of chick GAIP to accelerate the intrinsic GTPase activity of several G␣ subunits was tested in vitro using an assay that allowed subsecond resolution of GAIP-induced GTPase activity. G␣ i1 (A), G␣ i3 (B), and G␣ o (C) subunits were loaded with [ 32 P]GTP, and P i released into the superfusate was measured in the absence (open symbols) or presence (filled symbols) of 3 M chick GAIP. D, fold changes in P i release as a function of time calculated by subtracting background counts and expressing as percentage of total radioactivity loaded onto the filters. All points are means Ϯ S.E. of numbers of measurements noted in the text (from three separate experiments). duced by GABA (Fig. 5E); the remaining current is an estimate of voltage-independent inhibition (Fig. 5D). The inhibition was approximately two-thirds VI and one-third VD (Fig. 5, B, D, and E), consistent with results obtained using a more precise 3-pulse voltage protocol (28,33).
Chick GAIP Accelerates the Rates of Onset of and Recovery from GABA-mediated Inhibition-If the attenuation of GABAmediated inhibition observed for GAIP-treated cells is due to an acceleration of the intrinsic GTPase activity of G␣ o , then changes in rates of onset and recovery from GABA-induced inhibition would be expected. As predicted by mass action, both forward and reverse rates will depend upon the size of the heterotrimer pool (49). To test this, a 20-ms voltage step to 0 mV was applied at 0.5 Hz before, during, and after superfusion of the cell with 100 M GABA. GABA application rapidly inhibited Ca 2ϩ current (Fig. 6A) with a time course well described by a single exponential and a mean rate of 0.4 Ϯ 0.06 s Ϫ1 (n ϭ 10). GAIP produced a 75% increase in the rate of GABA-induced inhibition (0.7 Ϯ 0.07 s Ϫ1 , n ϭ 5). Once GABA-mediated inhibition reached steady state, the superfusate was switched back to control solution in order to measure recovery from inhibition. The time course for recovery was also well described by a single exponential (Fig. 6B), with an average rate of 0.25 Ϯ 0.03 s Ϫ1 in control cells (n ϭ 5). This rate was 136% faster in GAIP-treated cells (0.59 Ϯ 0.02 s Ϫ1 , n ϭ 10).
As expected from the results above, when GABA-induced inhibition was resolved into its two components, chick GAIP was found to accelerate selectively the onset and recovery of VI inhibition, with the time course of VD inhibition remaining . Calibration bars: 500 pA, 5 ms. C, total GABA-induced inhibition of N currents evoked by test pulses to 0 mV, with recording pipettes containing control internal solution (white bars) and or internal solution with 300 nM chick GAIP (gray bars). Inhibition was measured early (after 30 s) and late (after 10ϩ min) following whole-cell access. The voltage-independent (B) and voltage-dependent (C) components of GABA-mediated inhibition at 0 mV were isolated as described, and the effects of chick GAIP on each component was separately assessed and plotted as in A. Double asterisks indicate statistical significance with p Ͻ 0.01 (Student's two-tailed t test). unchanged (Fig. 6). These results confirm that chick GAIP acts as a selective GAP for the G o pathway underlying VI inhibition.

DISCUSSION
Chick GAIP Is a Unique Isoform-We have cloned and characterized a 2164-bp cDNA encoding the full-length sequence of chick GAIP. Chick GAIP is expressed as a 2.2-kb transcript, as determined by Northern hybridization and is detected in all tissues tested, with the highest levels in brain and the lowest in liver, skeletal muscle, and kidney. A second transcript, 1.5 kb in size, was detected by the GAIP probe. This transcript shows the same tissue expression pattern as the 2.2 transcript but is more abundant. Because both bands were consistently observed under high stringency hybridization conditions, in distinct blots, and with different GAIP-specific probes, it is not likely that they are cross-hybridization artifacts. We believe they are more likely to represent two independent GAIP mRNAs. This conclusion is supported by the fact that the two transcripts have the same expression pattern, more abundant in brain, weaker in liver, kidney, and skeletal muscle, although a control gene, glyceraldehyde-3-phosphate dehydrogenase, exhibits a completely different tissue distribution. Furthermore, the length of the 1.5-kb transcript is consistent with that reported for human GAIP (6,50). The difference between the 2.2and 1.5-kb transcripts is mostly likely to reside in the length of the 3Ј-UTR; the probe used in the Northern blot hybridizes to 498 of the 597 nucleotides in the coding region and the first 82 nucleotides of the 3Ј-UTR of chick GAIP. Thus, the 375-bp cDNA sequence 5Ј to the hybridizing region is too short to account for the 700-bp difference in length. The 1.5-kb transcript could be either a GAIP splice variant with a shorter 3Ј-UTR or an edited product of the 2.2-bp species. In both cases, the higher abundance of the 1.5-kb species suggests the process generates a more stable mRNA.
The open reading frame of chick GAIP encodes a polypeptide of 199 amino acids, 18 residues shorter than that reported for human GAIP (6). The RGS and cysteine-rich domains are well conserved. The latter represents a putative palmitoylation site involved in membrane localization (23). In addition, all residues known to be covalently modified in human GAIP are conserved in chick GAIP. Human GAIP is phosphorylated by casein kinase II at Ser-24 (51) and by Erk1/2 kinase at Ser-151 (52). This latter modification has functional consequences: when present, it potentiates the GAP activity of human GAIP (52) and stimulates autophagy in human colon cancer cells (52). The 2 equivalent positions in the chick GAIP sequence, 2 and 133, are indeed occupied by two conserved serines, suggesting that chick GAIP is likely to be regulated identically.
In addition to these identities, however, chick GAIP contains a unique N terminus, containing both a 22-amino acid deletion and a 5-amino acid insertion. The N terminus has been suggested to play a role in localization, specificity, and potency of several RGS proteins, including GAIP. The first 33 residues of RGS4, RGS5, and RGS16 (53,54), the first 35 residues of RGS8 (47), and the first 66 residues of RGS2 (55) adopt an amphipathic ␣-helix conformation that is required for membrane targeting and biological activity of the proteins. Deletion mutagenesis identified in the N terminus of RGS4 a domain that conveys high potency and receptor-selective inhibition of G q signaling (44). Furthermore, a GAIP mutant lacking the N terminus loses its ability to discriminate between G i -and G omediated pathways in chick DRG neurons (17). These reports predict that the N-terminal alterations present in chick GAIP would lead to differences in potency and/or target specificity compared with mammalian GAIP (6,23,34), and our results confirm this prediction.
Chick GAIP Has Unique Target Specificity-The increases in G␣ GTPase activity measured in the presence of chick GAIP (2-12-fold) are lower than the increases reported for human GAIP (40 -100-fold) (17,22,23). Although this could indicate sub-optimal assay conditions (3), a mathematical model (Appendix) argues against this possibility. The model is based on the simplifying assumption that chick GAIP acts exclusively as a GAP. When the model employed increases in G␣ o GTPase activity similar to those observed experimentally, the predictions matched well with the observed experimental data (see Appendix and Fig. 8).
Results of previously published experiments with human GAIP, carried out in the Dunlap laboratory under identical experimental conditions (17), contrast with those reported here for chick GAIP. Human GAIP eliminates both VD and VI inhibition evoked by GABA application, whereas chick GAIP is selective for VI inhibition. As both forms of inhibition require the activation of G o (which has two splice variants) (28,56), our results imply that chick GAIP is capable of discriminating either (i) between two splice variants of the same G protein subunit or (ii) between two pathways mediated by the identical G protein subunit. Some specificity of interaction between RGS proteins and their targets has been achieved in studies in vitro; for example, in GTPase and yeast two-hybrid assays, human GAIP interacts well with G␣ i1 , G␣ i3 , and G␣ o , but only poorly with G␣ i2 , and not at all with G s , G q , G z , and G 12/13 (6,23). In intact systems, however, no RGS protein has been shown to be able to distinguish among pathways involving either the same G protein or different isoforms of the same G protein. These results indicate that, when tested in native cells, chick GAIP not only has different targets but also has a higher degree of specificity than does human GAIP. As the only structural differences between the two GAIP isoforms reside in the N terminus, this region is likely to confer the unique attributes of chick GAIP.
Conclusions-We have reported the full-length cloning and characterization of chick RGS-GAIP, a unique variant with a FIG. 6. Chick GAIP accelerates the onset and recovery from GABA-induced inhibition. The rates of GABA-induced inhibition of N current (A) and recovery from GABA-induced inhibition (B) were measured in control cells (white bars) and in cells exposed to 300 nM GAIP (gray bars) soon after whole-cell access was achieved (after 30 s) and at least 10 min later (after 10ϩ min). The values for total inhibition as well as the voltage-independent and voltage-dependent inhibition are plotted separately, as marked. Data represent means Ϯ S.E. of four independent measurements. Asterisks indicate statistical significance with p Ͻ 0.05 (Student's two-tailed t test).
shorter N terminus than other previously reported isoforms. The differences in structure correlate with a higher G protein target specificity in vitro and in vivo. A mathematical model of GAIP action (which assumes that GAIP acts exclusively as a GAP) predicts many features that characterize GAIP effects in intact cells. These findings suggest that chick GAIP may be helpful for identifying the structural basis of GAIP target specificity. APPENDIX A quantitative model of G-protein cycling kinetics was developed in an effort to determine how RGS proteins might alter G-protein signaling to effectors. An eight-state model (Fig. 7) was optimized to account for the ability of RGS proteins to alter G-protein activation kinetics as assessed using Ca 2ϩ channel modulation as an indicator of G-protein activation. The relative abundance of each of the eight states was calculated using an iterative numerical routine run in Microsoft Excel (57).
To construct the model, we assumed that, in the absence of GPCR activation, the equilibrium distribution of the heterotrimeric forms of G-protein greatly favors the ground state (i.e. (␣(GDP)␤␥) over the activated state (␣(GDP)␤␥ † )). The reaction catalyzed by GPCR activation assumed a 300-fold increase in both the forward and reverse rate constants. The hypothetical GPCR was given a diffusion-limited agonist association rate constant (10 7 M Ϫ1 s Ϫ1 ) and a dissociation rate constant of 10 s Ϫ1 to give an equilibrium K d for agonist binding of ϳ10 Ϫ6 M (58). The values used for the remaining rate constants are summarized in Table II. The model further assumes that RGS proteins have two effects on G-protein function. First is that RGS binding to the G␣ subunit prevents the binding of the G␤␥ subunit. This assumption is supported by structural data (5) indicating RGS binding to a broad surface of G␣ that has considerable overlap with the G␤␥ subunit interface. Second, RGS acts as a GAP, providing a modest increase in the hydrolytic rate constant for ␣*(GTP)-RGS that is three times that of ␣*(GTP). Finally, the model assumes that the efficacy of the active form of the G␣*(GTP) is independent of RGS binding, i.e. that the active ␣ subunit interacts with effectors equally well whether or not RGS protein is bound to it. This assumption is also consistent with the structural data, which suggest that G␣ interacts with effectors at a portion of the molecule that is distinct from the RGS-binding surface. This assumption may only be valid for particular RGS proteins; however, as RGS9 has been shown to alter G␣-effector coupling (59,60). Fig. 8 illustrates the effects of RGS on G-protein activation produced by a 20-s application of 10 Ϫ6 M receptor agonist (a concentration equal to the K d , producing half-maximal occupancy of the receptor). The time course for the production of ␣*(GTP) is illustrated for three concentrations of RGS (the latter of which are relative and dimensionless). The model makes several predictions consistent with the experimental results reported here. First, there is little effect of RGS on G␤␥ concentration. This is because the amount of ␤␥ present is a balance between the rate of heterotrimer dissociation and reassociation. Because the reassociation of the heterotrimer is a second-order reaction dependent on the concentration of ␣(GDP) and ␤␥, it will be relatively slow. Furthermore, both the dissociation and reassociation steps are independent of the action of RGS, although the reassembly of the heterotrimer is retarded by the competitive interaction between RGS and G␤␥. This finding is consistent with the observed lack of effect of RGS on voltage-dependent (i.e. ␤␥-mediated) modulation (as demonstrated in Fig. 5). Second, moderate concentrations of RGS ("RGS ϭ 1") predictably increase the rate of onset as well as recovery from GPCR activation (as was demonstrated experimentally in Fig. 6). Such increases in the rate of agonistinduced activation have been consistently reported in studies of both Ca 2ϩ (15,16) and K ϩ (61-63) channels. Third, with persistent receptor activation, the level of ␣*(GTP) decays from its peak, consistent with the observed desensitization of agonistinduced inhibition in the presence of RGS. This too is a feature of RGS action on Ca 2ϩ channel modulation (17). Finally, at higher levels of RGS (exemplified by RGS ϭ 10), the overall response to agonist is inhibited, indicating that at higher concentrations RGS acts as a negative regulator of G-protein signaling (as shown here in Fig. 5 and reported previously, e.g. Ref. 17). The model illustrates that this effect is conferred by the accumulation of an ␣(GDP)-RGS complex that is sequestered from the GPCR. Thus, a simple model in which RGS proteins interfere with ␣-␤␥ interactions provides a quantita- Eight-state model that accounts for several aspects of RGS action on GABA-induced inhibition of calcium channels. Assumptions outlined under the "Appendix" and values for the rate constants are noted in Table II.  tive description of RGS action on G-protein-dependent inhibition of Ca 2ϩ channels and offers explanations for the selective effect of GAIP on voltage-independent inhibition as well as its ability to promote desensitization and suppress agonist-induced inhibition at high concentrations.