Characterization of the major bovine brain Go alpha isoforms. Mapping the structural differences between the alpha subunit isoforms identifies a variable region of the protein involved in receptor interactions.

Go is the major G protein in bovine brain, with at least three isoforms, GoA, GoB, and GoC. Whereas alphaoA and alphaoB arise from a single Goalpha gene as alternatively spliced mRNAs, alphaoA and alphaoC are thought to differ by covalent modification. To test the hypothesis that alphaoA and alphaoC have different N-terminal lipid modifications, proteolytic fragments of alphao isoforms were immunoprecipitated with an N terminus-specific antibody and analyzed by matrix-assisted laser desorption ionization mass spectrometry. The major masses observed in immunoprecipitates were the same for all three alphao isoforms and corresponded to the predicted mass of a myristoylated N-terminal fragment. Structural differences between alphaoA and alphaoC were also compared before and after limited tryptic proteolysis using SDS-polyacrylamide gel electrophoresis containing 6 M urea. Based upon the alphao subunit fragments produced under activating and nonactivating conditions, differences between alphaoA and alphaoC were localized to a C-terminal fragment of the protein. This region, involved in receptor and effector interactions, implies divergent signaling roles for these two alphao proteins. Finally, the structural difference between alphaoA and alphaoC is associated with a difference of at most 2 daltons based upon measurements by electrospay ionization mass spectrometry.

G proteins, composed of an ␣ subunit bound to a ␤␥ dimer, mediate the effects of many extracellular ligands that act on specific cell surface receptors (1). G o is the major brain G protein, comprising up to 1% of particulate protein in bovine brain (2,3). The exact function of G o is still unclear, but it has been implicated in the regulation of voltage-gated calcium channels (4 -6); the activation of the mitogen-activated protein kinase pathway (7); the development of neuronal growth cones (8), where it is highly concentrated (9); and the regulation of vesicle trafficking (10 -13).
The role of the ␣ o subunit on the level of the whole animal has also been studied. In Caenorhabditis elegans, the ␣ o protein influences behaviors such as locomotion and reproduction (14,15). Mice lacking the ␣ o gene are afflicted with tremors and seizures (16,17) and show loss of motor control and a propensity to run continuously in circles in a counterclockwise direction (17). In both studies with mice deficient in ␣ o , life span was significantly reduced. Although the precise function of ␣ o in neuronal tissue remains to be defined, in the knockout mice, regulation of Ca 2ϩ channels by opioid receptors in dorsal root ganglion cells is altered (17), as is regulation of L-type calcium channels in heart (16). Thus, the expression of ␣ o is clearly required for normal neuronal function.
One gene codes for the ␣ o subunit, which gives rise in brain to multiple splice variants with two different coding sequences contained in mRNAs ␣ o1 and ␣ o2 (18 -20). Based upon protein characterization, however, there are at least four ␣ o isoforms in bovine brain, ␣ oA , ␣ oB , ␣ oC , and ␣ oD (21)(22)(23). Three of the isoforms, ␣ oA , ␣ oB , and ␣ oC , are purified associated with ␤␥ dimers as G oA , G oB , and G oC , while ␣ oD , an inconsistently observed ␣ o isoform, is not. Immunological evidence and peptide mapping suggest that ␣ oB and ␣ oD are protein products of the ␣ o2 mRNA, whereas ␣ oA and ␣ oC are translated from the ␣ o1 mRNA (23)(24)(25)(26). The functional significance of these different ␣ o isoforms is not yet clear, but the G oA , G oB , and G oC heterotrimers each contain a different assortment of ␤␥ dimers (23).
An ␣ oC -like protein has been observed by a large number of groups (21)(22)(23)(27)(28)(29)(30). Recently, we have found that it is a relatively abundant protein in brain, constituting about a third of the G o of cerebral cortex. 1 Several studies have tried to identify the way in which ␣ oC differs from its immunologically related ␣ oA protein (23-26) but without success. One suggestion is that ␣ oC differs from ␣ oA at the N terminus and that it might be an unmyristoylated form of the protein (29,30). Such speculations are made more plausible by the fact that the related ␣ t protein is heterogeneously modified at the N terminus (31,32). In this paper, mass spectrometry of intact ␣ subunits and limited tryptic proteolysis studies, in conjunction with urea/SDS-PAGE 2 and matrix-assisted laser desorption ionization (MALDI) and electrospray ionization (ESI) mass spectrometry, were used to determine the region of the structural differences between ␣ oA and ␣ oC . Surprisingly, the masses of the ␣ oA and ␣ oC proteins do not differ significantly despite their different mobilities on urea/SDS-PAGE. The N termini of ␣ oA , ␣ oB , and ␣ oC are identical, but a region located at the C terminus was different between ␣ oA and ␣ oC , analogous to the region differing between ␣ oA and ␣ oB . This structural difference is localized to a part of the protein suggesting differential interactions of G oA and G oC with receptors or effectors.

EXPERIMENTAL PROCEDURES
Purification of G Protein Isoforms-G proteins were purified from bovine brain using a modification (33,34) of the method of Sternweis and Robishaw (2). Isoforms of G proteins were purified using a Mono Q anion exchange column with a 0 -300 mM NaCl gradient (23). G protein subunits were separated in the presence of aluminum, magnesium, and fluoride as described previously (33,34).
Production of Site-specific Antibodies-Synthetic peptides corresponding to the ␣ o N terminus, GCTLSAEERAALERSKAIEKNLKE (antiserum AON, residues 2-25) and the C terminus, AKNLRGCGLY (antiserum AOC, residues 345-354) were coupled to keyhole limpet hemocyanin with glutaraldehyde and injected into rabbits, using the method described in Green et al. (35). The AO1 antibody, specific for the protein products of the ␣ o1 splice variant, was described previously (23).
SDS Polyacrylamide Gel Electrophoresis and Immunoblotting-Mobility of proteins and proteolytic fragments was determined using SDSpolyacrylamide gel electrophoresis according to the methods of Laemmli (37). Gels contained 11% acrylamide, 0.29% bisacrylamide without or with 6 M urea (28). Protein bands were stained with Coomassie Blue or silver (39). Immunoblots were performed by the method described in Towbin et al. (40); proteins were transferred onto nitrocellulose using a Bio-Rad semidry transfer apparatus using methanol-free buffer. Immunoreactivity was visualized using ECL reagents (NEN Life Science Products).
Immunoprecipitation-Approximately 5 l of Protein A immobilized on Sepharose 6MB (Protein A-Sepharose) was washed twice with 500 l of 50 mM Tris, pH 8.0, 150 mM NaCl (TBS), blocked in the above buffer with 2.5 mg/ml bovine serum albumin for 30 min, and washed four more times with TBS. Antiserum (1:25) in 300 l of 20 mM Tris, pH 8.0, 1 mM EDTA, 1 mM DTT, 20 mM NaCl, and 0.01% Thesit (TEDNT) was incubated with 5 l of beads at room temperature for 1 h. Excess antibody was removed by washing five times with 500 l of TEDNT. Proteolytic digests were diluted to approximately 150 l with TEDNT and incubated with prepared beads for 1 h at room temperature. Unbound peptides were removed by washing twice with 500 l of TEDNT and twice with 500 l of water. Bound peptide was released by mixing an approximately equal volume of n-propanol/water (1:1), containing 0.1% trifluoroacetic acid with the beads.
MALDI Mass Spectrometry-Analysis was performed using a Voyager-DE MALDI mass spectrometer (PerSeptive Biosystems). One l of the n-propanol/water/trifluoroacetic acid mixture containing the peptide was combined with 1 l of matrix, 50 mM ␣-cyano-4-hydroxycinnamic acid (Aldrich) in 70% acetonitrile, 0.1% trifluoroacetic acid. Samples were spotted on a sample plate and allowed to crystallize at room temperature. Masses were internally calibrated by including peptides or proteins of known mass (in daltons) (renin substrate, 1760.1; insulin, 5734.5; cytochrome C, 12,361; ␤-lactoglobulin B, 18,277) with the samples. Typically, 256 laser shots were averaged to produce a mass spectrum.
HPLC and Mass Spectrometry of Intact ␣ o Isoforms-A 4.6 ϫ 30-mm Aquapore phenyl column (Brownlee) was used to purify 0.25 nmol (10 g) of ␣ oA and ␣ oC . Buffer A was 90% aqueous, 10% acetonitrile with 0.1% trifluoroacetic acid, and buffer B was 90% n-propanol, 10% acetonitrile with 0.1% trifluoroacetic acid. Proteins eluted with a linear gradient of 2-98% B over 20 min at 0.5 ml/min, and absorbance at 214 nm was measured. Five percent of the eluate was analyzed by a Finnigan LCQ ion trap with an ESI source, while the other 95% was collected as fractions at 1-min intervals.
HPLC and MS/MS of Limited Tryptic Digests-Proteolytic fragments of ␣ OA and ␣ OC were separated on a 4.6 ϫ 30-mm Aquapore phenyl column (Brownlee). Buffer A was 90% aqueous, 10% acetonitrile with 0.1% trifluoroacetic acid, and buffer B was 75% acetonitrile, 25% isopropyl alcohol with 0.095% trifluoroacetic acid. A linear gradient of 2-98% B over 40 min at 0.5 ml/min was used, and absorbance at 214 nm was measured during the gradient. Five percent of the eluate was analyzed by a Finnigan LCQ ion trap with an ESI source, while the other 95% was collected as fractions at 1-min intervals. Tandem MS was performed by selection and fragmentation of the m/z corresponding to the [M ϩ 2H] 2ϩ ion of the ␣ o N-terminal tryptic fragment 2-15. A 2-m/z unit window was used in precursor ion selection. MacBioSpec version 1.0.1 (PE SCIEX Instruments, Thornhill, Canada) was used to generate predicted ion m/z values. Fig. 1A shows a map of the ␣ o protein, its known sites of modification, and the approximate sites of epitopes for the antibodies used here. Three G o variants were isolated from bovine brain G proteins by their differential elution on FPLC (23). The Coomassie Blue-stained proteins are shown in Fig. 1B and immunoblots with site-specific antibodies in Fig. 1C. All three proteins reacted with antisera to the N and C termini of ␣ o , AON, and AOC, respectively, while antibody AO1 recognized ␣ oA and ␣ oC , but not ␣ oB (Fig. 1C). The AO1 antisera (23) was raised to the most variable region of the ␣ o sequence, one indicative of the ␣ o1 mRNA. These data indicate that ␣ oA and ␣ oC are similar at a site that is variable among known ␣ o splice variants and supports the conclusion (23,24,27) that these two proteins arise from a common mRNA and differ by some posttranslational modification.

Characterization of ␣ o Variants Using Site-specific Antibodies-
In the absence of urea, the ␣ o isoforms have the same electrophoretic mobility by SDS-PAGE, traveling at 39 kDa (Fig. 1,  B and C). In the presence of 6 M urea, each of the ␣ o variants has a unique electrophoretic mobility (28) (Fig. 1D). This differential behavior on different kinds of gels is often indicative of subtle structural differences (28,41). This phenomenon was exploited here to identify the location of the structural difference in the proteins.
Analysis of Intact G o ␣ Subunits by ESI-MS-ESI-MS is potentially useful for the analysis of large proteins, capable of generating very precise and accurate mass measurements. Therefore, the ␣ subunits of G oA and G oC were passed over an HPLC phenyl column in-line with a Finnigan LCQ ion trap mass spectrometer. The HPLC elution profile and co-migration of immunoreactive protein with the major UV peak observed in the separation are shown in Fig. 2, along with ESI mass spectra for the ␣ subunits. These proteins are known to contain some kind of structural difference (22,23,(27)(28)(29)(30). Nevertheless, the two proteins had very similar masses by ESI-MS (Fig.  2) and are, in fact, indistinguishable from each other within the accuracy of the instrument. The immunoblots in Fig. 2 (A and  B) and gel of the recovered proteins (Fig. 2C) substantiate that the ␣ oA and ␣ oC proteins retained their inherent structural differences after HPLC separation. Thus, the ␣ oA and ␣ oC proteins do not differ significantly in mass but are nonetheless structurally different from one another.
The data in Fig. 2D allow us to assign a possible structure to the ␣ oA protein. The observed mass was not easily reconcilable with the predicted sequence of the reported bovine brain ␣ o1 cDNA (P08239). It was, however, compatible with an alternative sequence reported by Ovchinnikov et al. (42), in which a Thr replaces Ile 262 in the protein. The 40,147.4-Da observed mass is then compatible with a protein myristoylated at Gly 2 after removal of Met 1 , with no other modifications (predicted mass of 40,145.9 Da). This substitution was verified by both MS/MS and Edman sequencing. 3 Since this mass was found repeatedly in our G protein preparations, this alternative sequence represents the major one observed in the cattle used in our studies.
Analysis of N-terminal Fragments of the ␣ o Proteins-The N termini of G␣ subunits are known sites of variable modification involved in several important interactions. Since N-terminal variability could explain the origin of ␣ oA and ␣ oC , we immunoprecipitated N-terminal fragments of ␣ oA and ␣ oC with the AON antisera to evaluate their modifications by MALDI-MS (Fig. 3). Digestion of all three ␣ o proteins with endoproteinase AspN is predicted to produce N-terminal fragments of residues 2-25 and 2-32 of the proteins. In the MALDI mass spectrum of the digest of G oA before immunoprecipitation, we observed numerous proteolytic fragments (Fig. 3B). Similar results were seen with G oB and G oC (data not shown). Immunoprecipitation of G oA digest with preimmune serum yielded no peaks in the low mass region. In contrast, the spectra of immunoprecipitates of G oA , G oB , and G oC digests with the AON antibody all showed precipitation of a fragment of about m/z 2859 (Fig. 3B). These fragments were all very close in mass to that predicted (2858.4 Da) for an N-terminal AspN fragment of ␣ o in which the N-terminal glycine is myristoylated (Fig. 3A). The other major peak in the spectrum of G oB at m/z 3501 corresponds to the larger myristoylated N-terminal AspN fragment of ␣ oB , 2-32 (Fig. 3B). Although the peak at m/z 3501 could suggest possible conformational differences in the N termini of ␣ oB and the other ␣ o isoforms, it was also variably observed in other immunoprecipitation experiments with G oA and G oC .
The data in Fig. 3B suggest that the fatty acyl group at the N terminus of ␣ o does not account for the difference between ␣ oA and ␣ oC . Close examination of the MALDI spectra (Fig. 3C) did show a low level of heterogeneity in the m/z range of the myristoylated N terminus. One mass peak at 2829, for example, was about 28 Da less than the major peak and could be FIG. 2. Analysis of intact ␣ oA and ␣ OC . Intact ␣ subunits were purified by HPLC and passed in-line to the LCQ mass spectrometer. A, ␣ oA ; B, ␣ oC . Absorbance was at 214 nm for each HPLC separation. Above the tracing is an immunoblot showing the fractions containing ␣ o immunoreactivity with antibody AON. C, analysis of proteins isolated A and B by urea/SDS-PAGE. The gel demonstrates that the material recovered from the HPLC separation and analyzed on the LCQ has the same electrophoretic mobility on urea gels as the material injected. Equivalent volumes of ␣ oA or ␣ oC , isolated from the HPLC separation, or a mixture of the two was run on an 10% SDS gel in the presence of urea. D, ␣ oA ; E, ␣ oC . ESI mass spectra obtained from the LCQ during the separation. Insets, deconvolution mass spectra with predicted masses of the observed ions. Calculated mass and S.E. is based upon the labeled charge states in the spectra on the left. explained by a lauroylated N terminus. The other minor peaks were not easily assigned, and none of these gave large enough signals to consider characterization. These results suggest the possibility that there is minor heterogeneity at the N terminus of ␣ o . This does not explain the difference between ␣ oA and ␣ oC , however, since similar heterogeneity was seen for all of the ␣ o proteins, and the heterogeneity of the N terminus of ␣ o observed here is insignificant compared with that seen for transducin (31,32). Thus, the major ␣ oA , ␣ oB , and ␣ oC isoforms appear to be myristoylated with no other modifications.
The other component of known heterogeneity at the N terminus of G protein ␣ subunits is degree of saturation of the fatty acid (31,32). Such a difference could account for the negligible mass difference between ␣ oA and ␣ oC . There was insufficient resolution in the MALDI spectra (Fig. 3) to differentiate single sites of unsaturation in the myristoyl group. To address this issue, the two ␣ o isoforms were digested with trypsin, and the N-terminal fragments were analyzed by HPLC coupled to a Finnigan LCQ ion trap mass spectrometer. From the data generated with this instrument, an ion chromatogram can be constructed showing the distribution of selected ions of a given m/z within Ϯ0.5 units. This is sufficient resolution to differentiate mono-and diunsaturated fatty acid acylation of peptide fragments. Fig. 4 (A and B) shows total ion chromatograms for the separation of digests of ␣ oA and ␣ oC (top tracing) and analogous selected ion chromatograms for the myristoylated N termini of ␣ oA and ␣ oC with zero, one, and two sites of unsaturation. The total ion currents are heavily dominated by buffer constituents as well as protein fragments. Nevertheless, both proteins have a well defined peak at about 19 min for a selected mass of 859.0, corresponding to an N-terminal fragment with a fully saturated myristic acid (2-15 ϩ C14:0). The assignment of this peak was confirmed by MS/MS sequencing of the peptide from ␣ oC (Fig. 4, C and D).
When ions corresponding to a myristoylated fragment containing one or two double bonds were selected (Fig. 4, 2-15 ϩ  C14:1 and 2-15 ϩ C14:2, respectively), there was relatively little signal. The minor peaks eluting at 21 min co-elute with a major peak in the total ion current chromatogram, suggesting a possible nonspecific signal, and others have found that myristoylated peptides with one and two double bonds in the acyl chain elute earlier than peptides with fully saturated myristic acid by hydrophobic interaction chromatography (31), not later as seen here (Fig. 4). These data indicate that significant amounts of mono-and diunsaturated forms of myristic acid are not incorporated into ␣ oA and ␣ oC , and that heterogeneity of the N-terminal lipid does not explain the difference between ␣ oA and ␣ oC .
Identification of the Region of ␣ oA and ␣ oC with Structural Differences-Trypsin preferentially cleaves an N-terminal 2-kDa fragment from G protein ␣ subunits activated with GTP␥S (36, 43) (Fig. 1A). Removal of this fragment allows direct comparison of the electrophoretic mobilities of the remaining 37-kDa fragments by urea/SDS-PAGE. AON immunoreactivity was completely lost from all isoforms after limited tryptic digestion, confirming loss of the N terminus (Fig. 5A). The AO1 antibody, however, recognized both the intact ␣ oA and ␣ oC isoforms, as well as the 37-kDa trypsin fragments generated by digestion of the GTP␥S-activated proteins (Fig. 5A). Retention of the difference in mobilities by the 37-kDa fragments confirmed that the structural difference between ␣ oA and ␣ oC does not result from differences in their N-terminal modifications.
When G protein ␣ subunits are briefly digested with trypsin in the presence of GDP, the 37-kDa C-terminal fragment is further cleaved into an N-terminal ϳ25-kDa fragment and a C-terminal ϳ17-kDa fragment (Fig. 1A) (36). Immunoblotting of digests after urea/SDS-PAGE with the AOC antibody showed that the 17-kDa C-terminal fragments of ␣ oA and ␣ oC retain the difference in mobility as do the intact ␣ subunits (Fig. 5B), while the 25-kDa fragments have identical mobilities.
Characterization of the intact ␣ subunits by ESI-MS (Fig. 2) indicated that the difference between ␣ oA and ␣ oC is associated with a negligible mass change. Our results above (Fig. 5) indicated that the difference between ␣ oA and ␣ oC is contained within the 17-kDa fragment. We therefore generated the 17and 25-kDa fragments from G oA and G oC , isolated them by HPLC, and characterized them by ESI mass spectrometry (Fig.  6). ESI-MS allowed us to determine the precise cleavage sites of the 17-and 25-kDa fragments and showed that the masses were essentially identical for both fragments, as originally suggested from the analysis of the intact proteins (Fig. 2) and despite their difference in electrophoretic mobility (Fig. 5B). These studies localize the difference between the ␣ oA and ␣ oC proteins to the 17-kDa fragment and, considering the mass accuracy of the instrument, the difference between these fragments would be restricted to less than 2 daltons. Shown from top to bottom are MALDI spectra for immunoprecipitate of G oB digest, immunoprecipitate of G oA digest, immunoprecipitate of G oC digest, G oA digest before immunoprecipitation, and G oA immunoprecipitate using preimmune serum. C, expanded MALDI mass spectra of the immunoprecipitate of the ␣ oA N-terminal peptide. The spectra is centered on the m/z 2859 peak recovered from the immunoprecipitation of the ␣ oA N terminus.

DISCUSSION
The ␣ oC and ␣ oA proteins are both major brain ␣ subunits, accounting for approximately 35 and 60%, respectively, of the G o protein present in brain cortex. 1 Available evidence suggests that they both arise from the same ␣ o1 mRNA (21-23, [27][28][29][30] and that they differ by some unknown modification. The evidence for this is immunologic cross-reactivity at sites exhibiting marked variability between G protein ␣ subunits and extensive common sequence. The putative modification differentiating these two proteins gives rise to marked differences in mobility of the proteins on urea/SDS-PAGE, possibly suggesting some substantial difference in their structures. Nevertheless, we have shown by ESI-MS that these two proteins differ by less than 2 Da in mass (Figs. 2 and 6). While not specifically identifying the difference between the proteins, these results set criteria to be met in proposing candidate modifications. Any modifications accounting for these proteins would have to result in only 1-2-Da mass difference between the two proteins.
The Significance of N-terminal Modification of G Protein ␣ Subunits and the Homogeneous Modification of ␣ o -The N terminus of G protein ␣ subunits is thought to have several important functions. It is a site of interactions with ␤␥ dimers (44 -47), a site of interaction with the C terminus of the ␣ subunit (48), a possible site of receptor interactions (49), and a probable site of membrane attachment of the ␣ subunit (46,47,50). Probably affecting all of these activities are the multiple modifications at the N terminus of ␣ o (and other ␣ subunit isoforms). The ␣ o protein is predicted to be myristoylated on an N-terminal Gly after removal of the initiating Met (51), and it can be palmitoylated through a thioester linkage on a Cys located at position 3 (52). This palmitate may turn over in response to receptor activation of the protein (53,54). For other G protein ␣ subunits, the N terminus has been found to be a site of variable modification. In addition to palmitoylation at the N-terminal Cys, arachidonylation has been reported for ␣ i , ␣ q , ␣ z , and ␣ 13 in platelets (55). In the case of transducin ␣, the N-terminal Gly is heterogeneously acylated, with either a lauroyl (C12:0) or a myristoyl (C14:0) group, the latter having varying degrees of unsaturation (31,32). This variability appears to be functionally related to the ability of ␣ to bind ␤␥ (32). Finally, a recent report has identified a previously unknown, and still uncharacterized, modification of ␣ s also located near the N terminus of the protein (56).
Until this study, we had considered it likely that variable  Fig.  1A. A, immunoblot of a urea/SDS-polyacrylamide gel using the AON and AO1 antibodies, before and after trypsin digestion in the presence of Mg 2ϩ and GTP␥S as described under "Experimental Procedures." B, silver-stained urea gel and immunoblot of urea gel with AOC antiserum before and after trypsin digestion in the presence of Mg 2ϩ and GDP as described under "Experimental Procedures." modification of the N terminus of ␣ o would account for the structural differences between ␣ oA and ␣ oC . In fact, it has been suggested that these proteins do differ at the N terminus, as myristoylated and unmyristoylated forms of ␣ o (29,30). Palmitoylation state was another possibility that we considered. Because the N terminus of ␣ is involved in binding ␤␥ dimers (44 -47), these possibilities might have explained the different ␤␥ composition of G oA and G oC (23). Despite all of these presumptive arguments, we were unable to document any difference between ␣ oA and ␣ oC at their N termini. Furthermore, and in contrast to ␣ t , we were unable to identify significant variability in the processing of the N terminus of the ␣ o proteins. Whatever function this variability has in the retina, it is not evident in the abundant G proteins commonly purified from bovine brain (2, 3). Finally, it is important to note that we did not find any evidence for a palmitoylated form of any of the G o ␣ proteins. Other analyses of G o purified from bovine brain have also not observed palmitoylation (57). The lack of this modification may be due to the reducing conditions (1 mM DTT) used in the purification, which could cleave the labile thioester linkage, or there may simply be low levels of palmitoylation in vivo. These studies do not dismiss a possible role of palmitoylation in the function of these proteins, but whatever this role may be, it does not account for the differences between ␣ oA and ␣ oC .
␣ oA and ␣ oC Differ in a Region Involved in Receptor and Effector Interactions-These studies identify the region in ␣ o between residues 210 and 354 as the part of the protein containing the structural difference(s) between ␣ oA and ␣ oC . This is the same region of the ␣ o protein that is different in the ␣ oA and ␣ oB splice variants. Variation in the C-terminal region of the ␣ o isoforms has diverse functional implications, including changes FIG. 6. ESI mass spectra from the LCQ analyses of digests of ␣ oA and ␣ oC . A, map of the ␣ o protein showing the predicted fragments generated by trypsin digestion in the presence of GDP and the corresponding predicted masses for the 25-and 17-kDa fragments (seen by SDS-PAGE). B, analysis of the two peaks from the ␣ oA separation. On the left are the ESI mass spectra showing all charge states of the fragments; on the right are deconvolution mass spectra with predicted masses based upon trypsin cleavage. C, analysis of the two peaks from the ␣ oC separation. On the left are the ESI spectra showing all charge states of the fragments; on the right, the deconvoluted mass spectra are shown. Average and S.E. of estimates are calculated from the multiple charge states in the ESI mass spectra.
in nucleotide binding (58), cellular targeting (59), and interactions with receptors and effectors (60 -62). There is already evidence that differences between ␣ oA and ␣ oB influence these interactions. For example, antisense experiments have shown that the ␣ oA and ␣ oB isoforms couple specifically to muscarinic and somatostatin receptors, respectively (63). In terms of effector specificity, ␣ oB , but not ␣ oA , has been shown to be able to inhibit adenylyl cyclase (64) and to mediate uptake of catecholamines by secretory vesicles (65).
As the ␣ oC isoform is not observed in heart (28) or a number of cultured cell lines (27), it is possible that it is neural specific and involved in such brain-specific functions as learning and memory. Indeed, there is evidence that amyloid precursor protein, critical in the formation of amyloid plaques associated with Alzheimer's disease, functions as a receptor for G o (66,67). Since several functions for G o have been proposed, structural heterogeneity resulting from both alternative splicing and variable modification may allow G o to perform divergent functions. For example, signaling pathways involved in calcium regulation, vesicular transport, and synaptic formation could all have a specific ␣ o isoform.