Heterotrimeric G proteins couple agonist-activated heptahelical receptors to a variety of effector responses. For the most part, this signaling activity takes place at the plasma membranes of cells. However, G proteins have been detected at a variety of subcellular locations and have been implicated in numerous membrane trafficking activities 12. This suggests that distinct, though not yet determined, mechanisms exist to target G protein subunits to subcellular locations other than plasma membranes (PM). Furthermore, post-transcriptional or post-translational modifications may play a role in directing a G protein subunit to a specific intracellular location.

Unique subcellular localization has been observed for a novel alternative spliced form of the G protein α subunit (Gα), α_i23. This α_i2 splice variant, termed sα_i2, contains a novel 35 amino acid sequence that replaces the C-terminal 24 amino acids of α_i23. sα_i2 has been observed to localize to Golgi membranes and other intracellular membranes 34; in contrast, wild type α_i2 is typically found at PM. Based on this unique localization of sα_i2, a prominent role in membrane trafficking has been proposed for this subunit 3. However, no report to date has supported a functional role for sα_i2 in vesicular transport, although a variety of functional assays have implicated other Gα, including α_s and α_i3, in vesicle trafficking pathways 12.

In spite of its unknown functional significance, sα_i2 provides a potential model for studying mechanisms of subcellular localization of Gα. Clearly, some aspect of the novel splicing is responsible for the difference in subcellular targeting of sα_i2 (i.e., Golgi and intracellular membranes) compared to wild type α_i2 (i.e., PM). To address the mechanism of subcellular localization of sα_i2, two hypotheses were tested herein. First, the possibility was tested that the novel 35 amino acid sequence found at the C-terminus of sα_i2 functions as a specific Golgi or intracellular membrane targeting motif. A second possibility was suggested by considering the location of the splice site. Several studies have shown that deletions or amino acid substitutions in the extreme C-terminal region of Gα creates a protein defective in guanine-nucleotide binding 5678910. Rapid release of guanine-nucleotides can result in an unstable Gα 910. This leads to the prediction that sα_i2 is also unstable due to a disruption in the structure of its C-terminus caused by the novel splice sequence. Thus, the second possibility is that failure of sα_i2 to localize at the PM but instead localize at intracellular membranes is the result of a general defect in protein stability. Results presented herein are consistent with this second hypothesis.

α_i2 and sα_i2 differ only at their extreme C-termini. Alternative splicing results in a novel 35 amino acids in sα_i2 that replace the C-terminal 24 amino acids of α_i2 (Figure 1). In the studies presented here, both α_i2 and sα_i2 contain an internal EE epitope, as used previously for α_i21112. The use of the EE epitope allows the direct comparison of subcellular localization and expression of α_i2 and sα_i2, and facilitates immunoprecipitation of the two expressed proteins. This is essential since suitable antibodies to detect and immuno-isolate sα_i2 are not readily available. Attempts to use a previously described polyclonal antibody directed against the C-terminus of sα_i234 were unsatisfactory in our hands. In particular, attempts to detect endogenous sα_i2 by immunofluorescence failed due to faint detection that showed drastically different patterns of staining depending upon fixation technique.

The subcellular localization of expressed α_i2 and sα_i2 was compared in COS and BHK cells (Figure 2). In both cell types, α_i2 was detected at cellular plasma membranes as highlighted by bright staining at the cell periphery (Figure 2A and 2C). In contrast, sα_i2 was not detected at plasma membranes, but instead typically displayed an intracellular and perinuclear staining pattern (Figure 2B and 2D), consistent with localization to intracellular membranes as previously described 34. The subcellular localization of sα_i2 showed overlap in immunofluorescent staining with the Golgi protein, β-COP (Figure 3A and 3B) and the endoplasmic reticulum protein PDI (Figure 3C and 3D); localization of sα_i2 exclusively to Golgi was not observed, even at the lowest levels of detection. Subcellular fractionation confirmed that sα_i2 was localized, at least partly, to membranes (Figure 2E). After lysis of cells in hypotonic buffer followed by high-speed centrifugation, α_i2 was found predominantly in the particulate fraction from COS (Figure 2E, lanes 3 and 4) or BHK cells (Figure 2E, lanes 9 and 10). A majority of sα_i2 was also detected in the particulate fraction from both COS (Figure 2E, lanes 5 and 6) and BHK cells (Figure 2E, lanes 11 and 12), although some was also found in the soluble fraction. As a comparison, a non-palmitoylated mutant of α_q, in which cysteines 9 and 10 have been substituted with serines, is found predominantly in the soluble fraction (Figure 2E, lanes 7 and 8), as described previously 13. Surprisingly, in comparison to α_i2, sα_i2 was always detected in fewer cells, by immunofluorescence analysis, and showed lower levels of protein expression, by western blot analysis of subcellular fractions (Figure 2E) or whole cell lysates (not shown). These expression differences suggested that sα_i2 was either less efficiently expressed or more rapidly degraded compared to α_i2, as described below.

The C-terminal 35 amino acids of sα_i2, which replace the C-terminal 24 amino acids of α_i2 (Figure 1), have been proposed to function as a Golgi targeting signal 34. To test the potential importance of this C-terminal region in localization of sα_i2 to intracellular membranes and prevention of sα_i2 from localizing to PM, a deletion mutant of α_i2 was constructed consisting of amino acids 1–331. In sα_i2, the C-terminal splice occurs just after amino acid 331 (Figure 1), and thus the novel 35 amino acid splice sequence is removed in α_i2(1–331). Immunofluorescence microscopy indicates that α_i2(1–331) (Figure 4B) displays a subcellular localization pattern very similar to sα_i2 (Figure 4A). This result is consistent with the alternative hypothesis that a disruption of α_i2 at the C-terminus creates a protein unable to localize at the PM.

If the unique C-terminal 35 amino acid sequence of sα_i2 functions as a Golgi targeting signal, it is important to test whether it is sufficient to target a heterologous protein to intracellular membranes. When the 35 amino acid sequence was fused to the C-terminus of GFP (GFP-sα_i235aa), this unique sα_i2 sequence failed to direct GFP to intracellular membranes (Figure 5B); GFP-sα_i235aa was diffusely distributed throughout the cytoplasm and nucleus, similar to what is observed for GFP alone (Figure 5A). On the other hand, when the first ten amino acids of α_i2 are fused to GFP to create α_i2(1-10)-GFP, fluorescence is detected strongly at the PM (Figure 5C), indicating the N-terminus of α_i2, which contains the sites for myristoylation and palmitoylation 14, can be sufficient for directing PM localization of a normally cytosolic protein. In agreement with these results, the sα_i2 C-terminal sequence failed to redirect the localization of a secreted or a nuclear protein 4. Thus, not only is the unique sα_i2 sequence unable to override other localization signals, it is unable to target a cytoplasmic protein to intracellular membranes. These results are consistent with a lack of a specific role for the sα_i2 sequence in Golgi membrane targeting.

The above experimental results failed to support the hypothesis that the unique 35 amino acid sequence from sα_i2 serves as a specific Golgi or intracellular membrane targeting motif. Thus, the following experiments concentrated on addressing the alternative hypothesis: sα_i2 is unstable and rapidly degraded due to disruptions in the C-terminus, and such instability results in a failure of sα_i2 to reach its proper place at the PM.

The ability of α_i2 and sα_i2 to undergo N-terminal lipid modifications was compared. Myristoylation occurs co-translationally at the extreme N-terminal glycine, while palmitoylation occurs post-translationally at a cysteine immediately after the myristoylated glycine 15. The subcellular site where palmitoylation occurs has not been well defined, although in one model palmitoylation occurs at the PM 16. Cells expressing α_i2 or sα_i2 were metabolically labeled with ³H-palmitate or ³H-myristate. α_i2 and sα_i2 were immunoprecipitated from cell extracts, and incorporation of radio-label was analyzed by SDS-PAGE followed by fluorography (Figure 6A). No palmitoylation of sα_i2 was detected (Figure 6A, lane 3), although α_i2 was strongly palmitoylated (Figure 6A, lane 2). In contrast, both α_i2 and sα_i2 incorporated similar levels or myristate (Figure 6A, lanes 5 and 6). The similar level of myristoylation was particularly surprising since total sα_i2 protein was much lower than α_i2 as determined by immunoblotting of the immunoprecipitates (Figure 6B, compare lanes 5 and 6). Myristoylation is a co-translational modification, and levels of radio-labeling should thus reflect synthesis of new protein. Accordingly, the myristoylation results (Figure 6A and 6B, lanes 5 and 6) strongly suggested that α_i2 and sα_i2 are translated equally well, as evidenced by similar levels of radiolabeled myristate incorporation during the labeling period, but sα_i2 is more rapidly degraded, as seen by reduced steady-state levels of immunoprecipitated protein. The lack of palmitoylation of sα_i2 is consistent with at least two, not necessarily mutually exclusive, possibilities. sα_i2 may fail to undergo palmitoylation because it is not stable long enough for post-translational palmitoylation to occur or because it fails to reach a PM site of palmitoylation. Alternatively, the novel C-terminal sequence of sα_i2 may somehow prevent direct recognition by a palmitoyl transferase.

Cells expressing high levels of sα_i2, and to a much lesser extent α_i2, often displayed very bright intracellular accumulations of sα_i2 as visualized by immunofluorescence. These large spots were reminiscent of recently described aggresomes, accumulations of misfolded protein that are being targeted for proteasome-dependent degradation 1718. This observation suggested the possibility that intracellular localization of sα_i2 may reflect targeting to a proteasome-dependent degradative pathway. To test this idea further, cells expressing α_i2 or sα_i2 were treated with the proteasome inhibitor acetyl-leucyl-leucyl-norleucinal (ALLN) and changes in localization of the α subunit were analyzed by immunofluorescence (Figure 7). After 1h treatment with ALLN, α_i2 localization is relatively unaffected (Figure 7C), but sα_i2 begins to show a slight increase in distinct intracellular accumulations (Figure 7D). At 4 h ALLN treatment, many cells expressing α_i2 display a decrease in PM localization and an increase in intracellular staining (Figure 7E). α_i2 is typically seen in large bright spots and in a perinuclear/intracellular membrane staining pattern after incubation of cells for 4 h with ALLN. In fact, the localization pattern of α_i2 after 4 h incubation with ALLN (Figure 7E) is similar to the observed localization of sα_i2 in the absence or presence of ALLN treatment (Figure 7B, 7D, and 7F). At 8 h treatment with ALLN, both α_i2 (Figure 7G) and sα_i2 (Figure 7H) are detected almost exclusively as a large bright spot in each expressing cell. Similar results were obtained using another proteasome inhibitor, MG-132. Interestingly, ALLN treatment was relatively ineffective at promoting accumulation of overexpressed α_s in aggresome-like structures (not shown). Thus, the results of Figure 7 are consistent with the proposal that the intracellular localization of sα_i2 is due to its targeting to a degradative pathway, a pathway shared by α_i2.

The subcellular localization of sα_i2 was compared to α_i2A327S. This alanine to serine mutation in the C-terminal region of α_i1 (A326S) and α_s (A366S) has been shown to greatly accelerate GDP release, to accelerate irreversible inactivation in vitro, and to cause rapid turnover of the α subunit in cultured cells 910. Thus, if the observed intracellular localization of sα_i2 is caused by instability of the protein due to disruptions at its C-terminus, then it follows that α_i2A327S should show a similar pattern of localization. Indeed, immunofluorescence microscopy revealed a pronounced intracellular distribution for α_i2A327S (Figure 8E). In addition to intracellular membrane localization similar to sα_i2 (Figure 8C), α_i2A327S was detected weakly but consistently at the PM. It seems likely that the C-terminal disruption of sα_i2 (replacing the 24 C-terminal amino acids) is a more severe disruption than the A327S substitution. Thus, the similarity in localization of sα_i2 and α_i2A327S bolsters the idea that mislocalization of sα_i2 is caused by its inherent instability.

G protein βγ subunits can stabilize wild type and alanine to serine mutant α subunits in vitro10. Thus, the effect of βγ co-expression on subcellular localization of α_i2A327S and sα_i2 was tested. α_i2 localizes to the PM efficiently when expressed alone (Figure 8A) or when co-expressed with βγ (Figure 8B). sα_i2 shows a dramatic change in localization when co-expressed with βγ (Figure 8D), displaying strong PM staining. Not only is sα_i2 targeted efficiently to the PM when co-expressed with βγ, but βγ co-expression increases overall levels of transiently expressed sα_i2 and α_i2, as determined by immunoblotting (Figure 8G), and increases the number of cells expressing sα_i2, as observed by immunofluorescence microscopy (not shown). Similarly, co-expression of βγ efficiently promotes PM localization of α_i2A327S (Figure 8F).

The analysis of the effect of βγ was extended by examining the localization of α_i2 or sα_i2 after co-transfection with βγ and after treatment with ALLN. In contrast to Figure 7, a marked difference was observed between α_i2 and sα_i2 after ALLN treatment. When βγ is co-expressed with α_i2, PM localization of α_i2 is observed, as shown in Figure 8B, and incubation with ALLN for 8 h shows little effect on PM localization of α_i2 (Figure 9A). When βγ is co-expressed with sα_i2, PM localization of sα_i2 is observed, as shown in Figure 8D, but incubation with ALLN for 8 h causes an increase in bright intracellular accumulations of sα_i2 (Figure 9B). Taken together, results with βγ co-expression are consistent with the ability of βγ to promote PM targeting of Gα, and further suggest that sα_i2 is less effectively stabilized than α_i2 by βγ, as evidenced by ALLN-induced accumulation in intracellular aggregates (Figure 9B).

Lastly, ³⁵S-methionine pulse-chase labeling studies directly confirmed that sα_i2 was more rapidly degraded than α_i2 (Figure 10). After transient expression of the α subunits, cellular proteins were radiolabeled with a brief pulse of ³⁵S-methionine followed by a chase. Immunoprecipitation of the α subunit after increasing times of chase, followed by SDS-PAGE and fluorography revealed that radiolabeled sα_i2 was rapidly lost. α_i2 consistently displayed a 3–4 fold longer half-life compared to sα_i2, while α_i2A327S was intermediate. In the experiment presented in Figure 10, the t_1/2 was determined to be approximately 90, 180, and 360 min for sα_i2, α_i2A327S, and α_i2, respectively. Although this trend was consistent, the exact t_1/2 showed some variability from experiment to experiment, probably due to varying levels of transient overexpression. It is likely that the difference in turnover of sα_i2 and α_i2 is even more profound when the proteins are expressed at lower levels. Consistent with this idea, sα_i2 was refractory to stable expression even though cells stably expressing α_i2 were easily selected (not shown). The inability to isolate cells stably expressing sα_i2 can be attributed to its rapid degradation. Regardless, ³⁵S-methionine pulse-chase labeling studies (Figure 10) demonstrate that sα_i2 is indeed more rapidly degraded than α_i2.

In addition to their well characterized role at the PM in coupling activated GPCRs to effector proteins, heterotrimeric G proteins have been implicated in a variety of other cellular functions, including membrane trafficking pathways. A splice variant of α_i2, termed sα_i2, has been proposed as a candidate to regulate vesicle transport due to its localization at intracellular membranes. The studies described in this report were undertaken to define mechanisms that underlie subcellular localization of sα_i2. The novel 35 amino acids found in sα_i2 appear not to function as a specific Golgi or endoplasmic reticulum targeting sequence. Instead, the results in this study support the proposal that sα_i2 fails to target to the PM because it is an unstable protein. The following results support this hypothesis: 1) sα_i2 and α_i2 are equally labeled by a pulse of ³H-myristate, although much less sα_i2 protein is detected; 2) sα_i2 displays a propensity to localize to potential aggresome-like structures, and this localization is greatly enhanced by proteasome inhibitor treatment; 3) the α_i2A327S mutant, previously shown to be unstable and defective in guanine-nucleotide binding, shows a similar pattern of subcellular localization (i.e., intracellular membranes rather than PM); 4) βγ over-expression increases expression of sα_i2 and promotes PM localization of sα_i2 and α_i2A327S, but βγ co-expression does not prevent sα_i2 localization to potential aggresome-like structures when cells are treated with proteasome inhibitors; and 5) pulse-chase analysis indicates that sα_i2 is rapidly degraded.

A number of reports have demonstrated that disruptions in the C-termini of Gα cause reduced affinity for guanine-nucleotides and protein instability. The C-terminal 30 amino acids of Gα consist of the β6-α5 loop, followed by the α5 helix and finally several amino acids of flexible structure (Figure 1) 192021. The β6-α5 loop stabilizes the guanine ring of bound GDP or GTP, and mutations in this region affect guanine-nucleotide binding. The A326S mutation of α_i1 and A366S mutation of α_s (cognate to the α_i2A327S used in this study) were shown to cause greatly decreased affinity for GDP 910. Defective binding of GDP leads to more Gα in the empty state (no bound guanine-nucleotide), and this form of Gα is rapidly denatured in vitro, as shown for α_sA366S and α_i1A326S 910. Moreover, α_sA366S was demonstrated to undergo rapid degradation (t_1/2 < 1 h) in stably transfected cells 9. The same mutation in α_t has also been shown to greatly decrease guanine-nucleotide binding 522. Additional amino acids in the critical β6-α5 loop are important for maintaining Gα integrity 7.

Although the α5 helix (Figure 1) does not directly contact the bound guanine-nucleotide 192021, it is clearly important for Gα structure and to maintain the proper orientation of the β6-α5 loop. A recent study identified a number of α5 helix residues in α_t that when changed to alanines increased rates of guanine-nucleotide exchange (i.e., decreased affinity for guanine-nucleotides) 5. Individual mutation of amino acids in α_t, cognate to T330, N332, V333, F337 in α_i2 (Figure 1), increased nucleotide exchange rates 5. Two of these critical amino acids, N332 and V333 in α_i2, are in fact the first two amino acids that are replaced by the novel splicing in sα_i2 (Figure 1). Thus, one might speculate that the novel 35 amino acid sequence in sα_i2 would impair guanine-nucleotide binding and result in a unstable protein. Unfortunately, initial attempts to directly show a defect in guanine-nucleotide binding by sα_i2, using a GTPγ S-dependent trypsin protection assay 23, were unsuccessful due, at least in part, to the inability to solubilize expressed sα_i2 from membranes using a mild detergent (e.g., lubrol/polyoxyethylene 10-lauryl ether).

The data in this report are consistent with a scenario in which intracellular Golgi/ER localization of sα_i2 reflects its rapid degradation rather than specific targeting to a subcellular organelle. Particularly compelling is the similarity in subcellular localization of α_i2A327S and sα_i2 (Figure 8). In addition, the use of proteasome inhibitors (Figures 7 and 9) suggest that sα_i2 is degraded by a proteasome pathway. The observed intracellular membrane (ER) localization and proteasome inhibitor-induced juxtanuclear aggregate accumulation (aggresome) is consistent with that reported for other proteins being degraded by such a pathway 1824. When treated with proteasome inhibitors, overexpressed wild type α_i2 also accumulates in aggresome-like structures (Figure 7), suggesting that Gα may normally be degraded by a proteasome-dependent pathway. However, over-expression of βγ prevents ALLN-induced juxtanuclear accumulations of α_i2 but not sα_i2 (Figure 9). Although little is known regarding degradative pathways for G proteins, sα_i2 and other mutants (e.g., α_i2A327S) may be valuable tools for defining such pathways.

The data presented here argue against a sequence specific Golgi membrane targeting function for the novel 35 amino acid sequence found in sα_i2. The novel sequence of sα_i2 is not sufficient to direct other proteins to Golgi/ER membranes. These 35 amino acids did not change the cytoplasmic localization of GFP (Figure 5). Similarly, a recent report showed that the sα_i2 35 amino acids were unable to retain a secreted protein in intracellular membranes and did not affect localization of a nuclear protein 4. However, in addition to causing rapid degradation, one cannot rule out that the 35 amino acid sequence of sα_i2 functions somehow as part of a Golgi targeting motif in the context of other regions of α_i2, as suggested 4. Other researchers have described a specific Golgi membrane localization of sα_i24; in contrast, the results presented in this report always show a much more disperse subcellular localization of sα_i2 to intracellular membranes consistent with staining of both Golgi and ER. The reason for this difference in localization is unclear. Interestingly, in the other study, the authors found that deletion of a proline-rich sequence corresponding to sα_i2 amino acids 348–359 (Figure 1) changed localization of sα_i2 from Golgi membranes to a more diffuse localization throughout intracellular membranes 4. However, mutation of proline residues to alanines failed to affect sα_i2 localization 4. Thus, little evidence exists to support a specific membrane targeting role for the novel sα_i2 sequence.

βγ was able to promote plasma membrane localization of sα_i2 (Figure 8). βγ prevents irreversible inactivation of α_i1, when measured in vitro at 37°C, and slows the inactivation of α_i1A326S 10. Thus, βγ may stabilize sα_i2 and α_i2A327S allowing βγ-dependent 2526272829 PM localization. These results are also consistent with the possibility that the primary defect in sα_i2 is a reduced ability to interact with βγ; Gα containing mutations in known sites of contact with βγ fail to localize to PM, but over-expression of βγ can rescue their PM localization 2829. However, the crystal structures 3031 show that the C-terminal β6-α5 loop/α5 helix region does not directly contact βγ. Thus, it is more likely that indeed sα_i2 is deficient in binding to βγ, but this effect is a consequence of its instability. sα_i2 may exist in a state that does not interact well with βγ simply because it is in the process of being irreversibly inactivated 910. Similarly, α_i2A327S appears not to be defective in its intrinsic ability to interact βγ 10, and co-expression of βγ shifts α_i2A327S, like sα_i2, from intracellular to plasma membranes (Figure 8).

What is the cellular function of sα_i2? The ability of βγ to promote PM localization of sα_i2 raises the possibility that endogenous sα_i2 is in fact localized to PM where it can interact with GPCRs and effectors. Future studies will test the potential of sα_i2, when co-expressed with βγ, to productively interact with receptors and effectors. However, this possibility is viewed as unlikely for several reasons. First, others have shown that in COS-7 cells endogenous sα_i2 is not detected at PM but only found intracellularly 34, although one cannot rule out that a small but functionally significant fraction of sα_i2 reaches the PM. Second, the extreme C-termini of Gα are critical sites for interaction with GPCRs 32, and the novel C-terminal sequence found in sα_i2 would be predicted to disrupt productive interactions with GPCRs. Third, even though βγ can promote PM localization of sα_i2 in the overexpression system described here, sα_i2 remains less stable than α_i2 (Figure 9), suggesting that such instability may also impair productive interactions with GPCRs.

On the other hand, unknown protein(s) may function to stabilize sα_i2 and help direct it to Golgi membranes. A specific combination of β and γ subtypes may play such a role; however, one thorough study showed a lack of βγ on Golgi membranes of exocrine pancreatic cells even though a variety of Gα were readily detected in the Golgi 33. Other proteins that may specifically promote the intracellular targeting of Gα have not been identified.

It is possible that sα_i2 functions as a short-lived protein regulating some aspect of membrane transport pathways; however, no experiments to directly support or refute this hypothesis have been reported. Instead, it is tempting to speculate that alternative splicing may be a mechanism to regulate cellular levels of α_i2. In this model, certain conditions may favor the formation of sα_i2 compared to α_i2, and the resulting rapid degradation of sα_i2 would decrease the cellular content of α_i2. A precedent for such alternative splicing-dependent regulation of expression has been described for H-ras 34. Alternative splicing occurs in H-ras, and the alternative spliced form is predicted to encode an unstable transcript and a protein product that lacks the ability to oncogenically transform cells. A mutation was identified that abolishes the alternatively spliced form, and this mutation leads to an increase in H-ras expression and transforming ability. Thus, it was demonstrated that alternative splicing is a mechanism that cells use to control expression of H-ras 34. Alternative splicing may similarly play an important role in regulating expression of α_i2. Testing of this idea will require a comparison of sα_i2 and α_i2 transcript levels, and an analysis of whether relative expression is affected in response to physiological changes, such as extracellular stimuli or cell differentiation.

In summary, the results presented here demonstrate that sα_i2, a novel splice variant of α_i2, is rapidly degraded when expressed in cells. Such instability is consistent with known structure-function data regarding the importance of the C-terminus of Gα subunits. The observed intracellular localization of sα_i2 is due, at least in part, to its instability. Moreover, the data presented in this study argue against the novel C-terminus of sα_i2 functioning as a specific Golgi targeting motif.

In addition, similarly to sα_i2, an α_i2 mutant, α_i2A327S, previously demonstrated to be unstable in vitro, displays a defect in plasma membrane localization when expressed in cells. These results suggest that both sα_i2 and α_i2A327S may prove valuable in future studies of mechanisms of degradation of Gα.