INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Heterotrimeric G proteins are classically known to couple cell surface receptors to various membrane-bound effectors (Gilman, 1987; Hamm and Gilchrist, 1996). Each heterotrimer consists of an  subunit, which exchanges GDP for GTP upon activation by a receptor, and subunits, which are tightly bound together. The  subunits attach tightly to membranes through posttranslational lipid modifications and binding to the relatively hydrophobic complex. Different  subunits segregate into specific subdomains of the plasma membrane (Stow et al., 1991) and are also associated with the membranes of different intracellular organelles (Jones, 1994; Denker et al., 1996; Stow and Heimann, 1998). Within the cell, G proteins are involved in vesicular transport between the endoplasmic reticulum and the Golgi (Beckers and Balch, 1989; Hidalgo et al., 1995), through the Golgi stack (Melançon et al., 1987; Leyte et al., 1992; Stow and Heimann, 1998), and other membrane-trafficking steps (Bomsel and Mostov, 1992). The functional consequences of receptor-G protein-effector systems at selective membranes presumably require the proper targeting of G proteins. The strong homology between  subunits makes their geographic diversity a puzzling trafficking problem.

Mutagenesis studies on targeting signals for G_i2 and G_i3 (de Almeida et al., 1994) and the discovery of splice variations of  subunits that change their intracellular localization provide clues to this problem. XLs (extra-large s), a splice variant of G_s, is particularly interesting because it is found on Golgi membranes in cells with both regulated and constitutive pathways of protein secretion (Kehlenbach et al., 1994a,b). This protein, identified as a cholera toxin substrate, is identical to G_s except for exon 1, at which a 347-amino acid amino terminus replaces 47 amino acids in G_s. The XL portion has proline-rich and cysteine-rich regions and areas of EPAA and AARA repeats. Another  subunit splice variant, sG_i2, also resides on Golgi membranes and is identical to G_i2 except for the last exon, which encodes a different 35-amino acid, proline-rich carboxy terminus (Montmayeur and Borrelli, 1994).

The signals involved in the Golgi targeting of peripheral membrane proteins, including G proteins, are poorly understood (reviewed by Stanley, 1996; Gleeson, 1998; Munro, 1998). Sorting signals on intrinsic membrane proteins act to retain these proteins in the Golgi apparatus or to retrieve them from the cell surface or other organelles. The transmembrane domain is critical for retention in the Golgi by proposed mechanisms that include oligomerization and sorting based on the length of the transmembrane domain. Retrieval signals such as the di-leucine motif are found on the cytoplasmic domains of integral membrane proteins. For peripheral membrane proteins at the Golgi, some transiently associate with ADP-ribosylation factor proteins leading to translocation to Golgi membranes. For several coiled-coil proteins, a Golgi-targeting domain, the GRIP domain, has been identified in their carboxy termini (Barr, 1999; Kjer-Nielsen et al. 1999; Munro and Nichols, 1999). For others, regions have been identified that are crucial for localization. Endothelial nitric oxide synthase (eNOS), SCG10, and glutamate decarboxylase have Golgi-localizing regions at their amino termini that also contain sites for two or more lipid modifications (Solimena et al., 1994; Di Paolo et al., 1997; Liu et al., 1997).

Palmitoylation, the reversible addition of palmitate to cysteine residues by a thioester bond, occurs on the amino terminus of G protein  subunits (Linder et al., 1993) and has several functions. Palmitoylation increases receptor-G protein coupling and the affinity of the  subunit for the complex (Iiri et al., 1996; Ponimaskin et al., 1998). For G_s, receptor activation causes a rapid turnover of palmitate, suggesting a role in signaling (Degtyarev et al., 1993b; Mumby et al., 1994; Wedegaertner and Bourne, 1994). The importance of palmitoylation in gross membrane attachment is controversial (Wedegaertner et al., 1993; Huang et al., 1999), but it may play a more subtle role in targeting G proteins to membrane microdomains (Arni et al., 1998; Melkonian et al., 1999). XLs does not have the acylation site found on G_s but does have a cysteine-rich domain containing six cysteines that are potential sites for palmitoylation.

We investigated the Golgi targeting of XLs by studying two regions in its XL amino terminus: the cysteine-rich region (CRR) and the proline-rich region (PRR), which bears similarity to the proline-rich carboxy terminus of sG_i2. We wished both to better understand the mechanisms involved in G protein targeting and to test the generality of the Golgi-localization signals suggested for eNOS and sG_i2. We found that the PRR and the cysteines in the CRR were critical for the Golgi targeting of XLs.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cell Culture

COS-7 and HEK-293 cells were grown in DMEM supplemented with penicillin (100 IU/ml), streptomycin (100 µg/ml), and 10% (vol/vol) FBS at 37°C in a humidified atmosphere of 5% CO₂. PC12 cells were maintained in identical conditions except that the serum supplementation was 5% (vol/vol) FBS and 10% (vol/vol) heat-inactivated horse serum.

Plasmid Constructs and Mutagenesis

The cDNA coding the 715-amino acid rat XLs (Kehlenbach et al., 1994a,b) in a pCDNA3.1 (+) plasmid (Invitrogen, Carlsbad, CA) underwent site-directed mutagenesis by a PCR-based method, QuikChange (Stratagene, La Jolla, CA), except that Pwo Polymerase (Boehringer Mannheim, Indianapolis, IN) was used. To generate green fluorescent protein (GFP) fusion proteins, cDNAs encoding the following regions of XLs were amplified from wild-type XLs-pcDNA3.1 or the 2C and 0C cysteine-to-serine mutants with the use of PCR: 1) PRR (amino acids 201-235); 2) PRR plus CRR (amino acids 201-312); and 3) CRR (amino acids 225-312). cDNAs were inserted into the pEGFP N1 vector (Clontech, Palo Alto, CA) at a BamHI site upstream and in frame with the cDNA coding EGFP. Mutations were confirmed with the use of ABI PRISM dye terminator cycle sequencing (Perkin Elmer-Cetus, Norwalk, CT).

Transient Transfection of the Cells

COS-7 cells, grown to subconfluence in 75-cm² tissue culture flasks, were transfected with the use of 2 µg of plasmid DNA per flask and the DEAE-dextran method, as described (Butrynski et al., 1992). PC12 cells, suspended in 0.8 ml of RPMI-1640 at a density of 4 × 10⁶ cells/ml, were transfected by electroporation at 960 µF and 360 V with the use of a Bio-Rad (Hercules, CA) Gene Pulser and 40-50 µg of plasmid DNA. Transfected PC12 cells were seeded into two-well chamber slides in RPMI-1640 supplemented with 10% FBS. The medium was changed to DMEM with 5% FBS and 10% heat-inactivated horse serum after 24 h. Metabolic labeling and immunodetection were performed 2 d after transfection. HEK-293 cells were transfected with the use of the LipofectAMINE PLUS reagent (GIBCO-BRL/Life Technologies, Grand Island, NY).

[³H]Palmitate Labeling and Cell Fractionation

For metabolic labeling, cells were first incubated for 2 h in serum-free DMEM and then in 5 ml of serum-free DMEM containing 500 µCi/ml [³H]palmitic acid (specific activity, 60 Ci/mmol; American Radiochemical, St. Louis, MO) for 50 min. Cells were harvested by scraping in 45 ml of ice-cold PBS and centrifuged at 2000 × g for 10 min, and cell pellets were stored at 70°C. The pellets were lysed, homogenized, and fractionated into particulate and soluble fractions by centrifugation at 125,000 × g for 1 h as described (Degtyarev et al., 1993a). Protein concentrations were determined with the use of the Bio-Rad assay (Bradford, 1976) with immunoglobulin G as the standard.

Immunoprecipitation and Immunoblotting

The polyclonal, affinity-purified antibody for G_s (RM) and antibody for G_i (AS), which recognize the carboxy-terminal decapeptide of G_s and XLs and the carboxy-terminal decapeptide of G_i, respectively, were used (Jones et al., 1990). Equal amounts of protein (300-400 µg) from the particulate fractions were incubated with the antibody in 500 µl of solubilization buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% [vol/vol] Triton X-100, 0.2% [wt/vol] SDS, 1 mM EDTA) overnight at 4°C. The immunoprecipitates were recovered by a 2-h incubation with protein A-Sepharose CL-4B (Pharmacia LKB Biotechnology, Piscataway, NJ), washed, separated by SDS-PAGE, and prepared for fluorography as described previously (Jones et al., 1990). Densitometry of the fluorographs was performed with an AGFA Arcus II scanner (Pharmacia LKB Biotechnology) and NIH Image software (http://rsb.info.nih.gov/nih-image/). For immunoblots, 5 µg of protein from each particulate and soluble fraction was separated by SDS-PAGE and transferred to nitrocellulose paper. The proteins were detected with the RM antibody (1 µg/ml) or a polyclonal antibody raised against the GFP (1:3000) (Molecular Probes, Eugene, OR) and ECL (Amersham, Arlington Heights, IL) as described by the manufacturer.

Cholera Toxin Labeling and Detergent Solubilization

Particulate fractions (75 µg of protein) were incubated with 300 mM potassium phosphate, pH 7.0, 10 mM thymidine, 1 mM ATP, 0.1 mM GTP, 10 mM MgCl₂, 1 mM EDTA, protease inhibitors, 6 µg of activated cholera toxin (Ribeiro-Neto et al., 1985), 1 mM DTT, 10 µg of purified brain subunits (Sternweis and Robishaw, 1984), 5 µM NAD, and 10 µCi of [³²P]NAD (specific activity, 1000 Ci/mmol; Amersham) in a volume of 60 µl for 45 min at 37°C. The reaction was stopped, and the samples were prepared for SDS-PAGE as described (Ribeiro-Neto et al., 1985). The gels were analyzed by autoradiography and a Phosphorimager (Molecular Dynamics, Sunnyvale, CA). Solubilization of proteins in the particulate fractions was performed with 1% (vol/vol) Triton X-100 as described previously (Jones and Gutkind, 1998).

Immunocytochemistry and Fluorescence Microscopy

Cells, grown and transfected in two-well chamber slides, were washed four times with PBS at room temperature, fixed and permeabilized by incubation in methanol at 20°C for 3 min, and washed four times with PBS. Cells expressing the GFP fusion proteins were fixed with 2% (wt/vol) paraformaldehyde in PBS for 20 min and permeabilized with 0.1% (wt/vol) Triton X-100 in PBS for 15 min. After incubation for 15 min in the blocking buffer (1% normal goat serum, 0.2% [vol/vol] Triton X-100 in PBS), cells were incubated for 1 h at room temperature in 0.1% (wt/vol) BSA in PBS with the RM antibody at 0.5 ng/ml and with the anti-58-kDa protein monoclonal mouse antibody (Sigma Chemical, St. Louis, MO) at a 1:100 dilution. After washing four times with 0.1% BSA in PBS, cells were incubated for 1 h at room temperature with one of the following secondary antibodies (Jackson ImmunoResearch, West Grove, PA): Cy3-labeled anti-rabbit (1:4000), FITC-labeled anti-mouse (1:100), or Cy3-labeled anti-mouse (1:100) antibody. Cells, washed four times with 0.1% BSA in PBS and once with distilled water, were mounted in Prolong antifade reagent (Molecular Probes) and visualized with the use of a Zeiss (Thornwood, NY) Axioskop microscope equipped for fluorescence microscopy with a 63×, 1.4 numerical aperture Plan-Apochromat oil immersion objective or a Leica (Wetzlar, Germany) LSM confocal microscope. For Golgi labeling of live PC12 cells with BODIPY FL C₅-ceramide (Molecular Probes), the method described by Ktistakis et al. (1995) was used. For treatment with brefeldin A, transfected COS cells were incubated with BODIPY TR ceramide (1 µM, mixed with equimolar defatted BSA in serum-free DMEM; Molecular Probes) for 30 min at 4°C, washed three times with DMEM with 10% FBS, and incubated at 37°C for 30 min. Brefeldin A (Epicentre Technologies, Madison, WI) was added at a final concentration of 5 µM, and cells were incubated for another 15 min, washed three times with PBS, and mounted and visualized as described above.

Quantitative Analysis of the Immunofluorescence

Cells were counted if they were brightly stained with RM antibody or GFP fluorescence, indicating that they had undergone transfection. The cells showing staining that colocalized with the anti-58-kDa antibody staining of a dense, perinuclear area were counted as positive with respect to the Golgi localization of XLs proteins. Transfections were done in parallel, and coded slides were used for this quantitation.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Endogenous XLs Undergoes Palmitoylation in PC12 Cells

G_s undergoes palmitoylation at its amino terminus (Linder et al., 1993). XLs lacks the cysteine residue at the third position of G_s that is critical for the modification (Degtyarev et al., 1993a) but has six cysteines in a cysteine-rich domain of its so-called XL portion (Figure 1A; Table 1). To determine if endogenous XLs is palmitoylated, PC12 cells were metabolically labeled with [³H]palmitic acid and proteins were immunoprecipitated with the RM antibody, an antibody specific for the carboxy terminus of G_s and XLs (Figure 1A). Tritium incorporation was seen in 94- and 42-kDa protein bands, corresponding to the expected sizes of XLs and G_s, respectively (Figure 1B). Immunoprecipitation with the AS antibody, a G_i-specific antibody prepared in the same way as the RM antibody, did not detect the 94-kDa band that is prominent in the fluorograph of the labeled membranes. Under these conditions, the tritium label incorporated into G_s is [³H]palmitate (Degtyarev et al.,1993a), suggesting that XLs is palmitoylated.

View larger version (37K): [in this window] [in a new window]   Figure 1.   (A) Domain organization of XLs. XLs is identical to Gs with respect to exons 2-13. The alternatively spliced amino terminus contains areas with EPAA repeats (residues 37-104), ARAA repeats (residues 103-187), and the proline-rich (residues 205-228) and cysteine-rich (residues 237-304) domains. The RM antibody, which recognizes both XLs and Gs, was generated from the decapeptide sequence at the carboxy termini of these proteins. (B) Incorporation of [3H]palmitate into endogenous XLs. PC-12 cells were incubated with [3H]palmitate, harvested, and separated into particulate and soluble fractions. One milligram of protein from the particulate fractions underwent immunoprecipitation with either the RM antibody or the AS antibody specific for Gi. The immunoprecipitates and 20 µg of particulate fraction protein (lane 1) were analyzed by SDS-PAGE and indirect fluorography. Fluorographs were exposed for 1 mo at 70°C.

View this table: [in this window] [in a new window]   Table 1.  XLs mutations

[³H]Palmitate Incorporation and Membrane Attachment of XLs Mutants in COS-7 Cells

We investigated whether palmitoylation occurred on the cysteines in the CRR by creating a series of cysteine-to-serine mutants by site-directed mutagenesis (Table 1). COS-7 cells were transfected with vectors containing the wild-type and mutant XLs cDNAs and incubated with [³H]palmitate. The G_s and XLs proteins in the particulate fraction were immunoprecipitated with the RM antibody. COS-7 cells express the long and short forms of G_s seen as 45- and 42-kDa bands, respectively, in all the lanes but do not express XLs endogenously (Figure 2A). Cells transfected with wild-type XLs showed a band at 94 kDa strongly labeled with tritium. Mutation of the six cysteine residues in the wild type decreased the tritium incorporation into XLs in a stepwise manner for mutants 4C, 3C, and 2C (Figure 2A). Densitometry readings of the 94-kDa bands for the 4C, 3C, and 2C mutants were 94, 85, and 60%, respectively, that of the wild-type XLs. The 1C and 0C mutants did not show any tritium incorporation. A 64-kDa band seen in the transfected cells was probably the product of a late initiation site at residue 136. This protein band also showed a decrease in tritium incorporation because it contains the CRR.

View larger version (36K): [in this window] [in a new window]   Figure 2.   (A) Incorporation of [3H]palmitate into wild-type and cysteine mutants of XLs. COS-7 cells were transiently transfected with the pCDNA3.1 (+) vector alone (V) or with the cDNA inserts of either the wild-type XLs (WT) or its cysteine mutants (Table 1), incubated with [3H]palmitate, and separated into particulate and soluble fractions. A total of 500 µg of protein from the particulate fractions was immunoprecipitated with the RM antibody, followed by SDS-PAGE and fluorography. The fluorographs were exposed for 1 wk at 70°C. (B and C) Subcellular distribution of the cysteine mutants of XLs. Five micrograms of protein from the particulate and soluble fractions of the transfected cells was separated by SDS-PAGE and underwent immunoblotting with the RM antibody and detection by ECL. (D) Expression of XLs mutants with deletion of the PRR. COS-7 cells were transiently transfected with the pCDNA3.1 (+) vector alone or with the cDNA inserts of either the wild-type XLs or mutants with a deletion of the PRR (PRR) (Table 1). Cells were harvested and separated into particulate (P) and soluble (S) fractions, and proteins were analyzed by immunoblotting as described above. The large arrow points to XLs.

The transfected cells were separated into particulate and soluble fractions followed by immunoblotting to test the membrane attachment of the reduced-palmitoylation mutants (Figure 2, B and C). The wild-type XLs and 4C, 3C, and 2C mutants were primarily in the particulate fraction and had minimal amounts of protein localized to the soluble fraction. However, the 1C and 0C mutants, which did not incorporate [³H]palmitate, were found in both the soluble and particulate fractions. These results indicate that these cysteine residues in the CRR were critical for palmitoylation and facilitate membrane attachment, although palmitoylation must not be the sole factor in membrane attachment.

Deletion of the PRR

XLs is localized primarily to the Golgi (Kehlenbach et al., 1994a). We deleted the PRR in the wild-type XLs and the reduced-palmitoylation 2C mutant to create mutants that would test the role of this region in intracellular localization (Figure 1; Table 1). We chose the 2C mutant because its incorporation of [³H]palmitate was impaired compared with that of XLs (Figure 2A), but it was still membrane attached (Figure 2, B and C). The mutants with a deletion of the PRR (PRR) were slightly smaller than the wild-type XLs but were expressed at levels equivalent to those of the wild-type XLs in COS-7 cells (Figure 2D). The PRR mutants were confined to the particulate fraction (Figure 2D) and showed no defects in incorporation of [³H]palmitate. The 2C and PRR mutants underwent ADP-ribosylation catalyzed by cholera toxin and detergent solubilization with Triton X-100 to the same degree as the wild-type XLs.

Intracellular Localization of XLs Mutants

Localization of the reduced-palmitoylation and PRR-deletion mutants of XLs was determined with the use of indirect immunofluorescence microscopy. We used an antibody specific to a protein associated with the cytoplasmic surface of the Golgi, the 58-kDa protein, to identify the Golgi structure (Bloom and Brashear, 1989). The RM antibody was used for the XLs and G_s proteins. In transfected COS cells, the wild-type XLs was localized to a compact perinuclear area that can be identified as Golgi with the anti-58-kDa protein antibody (Figure 3, a and b). We also found colocalization of endogenous and overexpressed XLs with another Golgi marker, BODIPY FL C₅-ceramide (Ktistakis et al., 1995), in PC12 cells.

View larger version (57K): [in this window] [in a new window]   Figure 3.   Intracellular localization of the wild-type XLs, the reduced-palmitoylation (2C) and PRR-deletion mutants of XLs, and the wild-type Gs. Two days after transfection with the indicated cDNAs, COS-7 cells were fixed and permeabilized with methanol at 20°C. Cells were double labeled with the rabbit polyclonal RM antibody to XLs and Gs and the monoclonal mouse anti-58-kDa antibody as a Golgi marker and subsequently with Cy3-labeled anti-rabbit and FITC-labeled anti-mouse antibodies. Each pair of panels shows an image made with a rhodamine filter and the corresponding image made with a FITC filter.

The reduced-palmitoylation mutants 4C, 3C, and 2C also showed bright, compact perinuclear staining that colocalized with the Golgi antibody staining (data shown for the 2C mutant in Figure 3, e and f). The 1C mutant, which did not undergo palmitoylation, exhibited a diffuse, cytosolic pattern consistent with its increased solubility but that obscured its membrane localization.

Deletion of the PRR in the XL portion of XLs led to a significant decrease in Golgi localization (Figure 3, c, d, g, and h). After transfection with the PRR and 2CPRR mutants, only 19% of 120 cells examined and 14% of 122 cells examined, respectively, showed a bright spot in the perinuclear area that colocalized with the Golgi antibody. In comparison, 50% of 136 cells transfected with the wild-type XLs colocalized with the Golgi antibody staining in the perinuclear region. The distribution pattern of the PRR mutants was comparable to the pattern of overexpressed G_s, with staining predominantly on intracellular membranes and some plasma membrane staining (Figure 3, i and j). The PRR mutants expressed in PC12 cells were primarily found on the plasma membrane, with some staining on other intracellular membranes but without prominent perinuclear staining.

Images were also obtained by confocal microscopy that showed the colocalization of XLs with anti-58-kDa Golgi staining (Figure 4A). The confocal images show more clearly that the PRR mutant was found on various intracellular membranes, including a small amount on the Golgi that is seen in the limited colocalization with the Golgi antibody (Figure 4B).

View larger version (21K): [in this window] [in a new window]   Figure 4.   Confocal microscopic localization of the wild-type XLs and PRR-deletion mutant (PRR). COS-7 cells were transfected and stained for immunofluorescence as described in Figure 3 and examined with a confocal microscope. The anti-58-kDa antibody (Golgi) staining is green, and the RM antibody (XLs and Gs) staining is red. The extent of colocalization, assessed by superimposing red and green signals, is yellow.

Membrane Attachment and Intracellular Localization of GFP Fusion Proteins

To further evaluate the roles of the PRR and the CRR as Golgi-targeting signals for XLs, we constructed fusion proteins with these regions and the GFP (Table 2) and examined their membrane attachment and intracellular localization. Immunoblotting of the particulate and soluble fractions of transfected COS cell lysates with an antibody to GFP showed that GFP alone and PRR and PRR+(0C)CRR fused to GFP were primarily soluble proteins (Figure 5). In contrast, PRR+CRR, PRR+(2C)CRR, CRR, and (2C)CRR fused to GFP were primarily found in the particulate fraction, indicating that cysteines in the CRR were needed to direct a soluble protein to the membrane fraction.

View larger version (20K): [in this window] [in a new window]   Figure 5.   Subcellular distribution of GFP fusion proteins. COS cells were transfected with the pEGFP N1 vector alone or with vectors containing the constructs of the PRR and the CRR fused to the amino terminus of the GFP (Table 2). Five micrograms of protein from the particulate (P) and soluble (S) fractions of the transfected cells was separated SDS-PAGE and underwent immunoblotting with a polyclonal antibody to GFP and detection by ECL.

We then examined the intracellular distribution of fusion proteins by direct fluorescence microscopy in live COS and HEK-293 cells (data shown for COS cells in Figure 6). Cells expressing the GFP protein displayed a diffuse fluorescence throughout the cells (Figure 6a), consistent with previous reports demonstrating that GFP is expressed as a cytosolic protein. Diffuse cytosolic staining was also found for the PRR and PRR+(0C)CRR fusion proteins (Figure 6, b and e), consistent with the results from cell fractionation (Figure 5). In contrast, fusion proteins containing PRR together with CRR or (2C)CRR displayed a compact perinuclear localization (Figure 6, c and d) that was consistent with Golgi localization. For [PRR+CRR]-GFP, 87% of 101 cells examined showed bright, perinuclear staining, and for [PRR+(2C)CRR]-GFP, 81% of 110 cells examined showed bright, perinuclear staining. The CRR and (2C)CRR fusion proteins exhibited some perinuclear localization but also a scattered intracellular fluorescence, presumably caused by nonspecific localization of these fusion proteins to intracellular membranes (Figure 6, f and g).

View larger version (29K): [in this window] [in a new window]   Figure 6.   Intracellular localization of GFP fusion proteins. COS cells were transfected with the pEGFP N1 vector alone or with vectors containing the constructs of the PRR and the CRR fused to the amino terminus of the GFP (Table 2). GFP fluorescence was recorded on living cells 2 d after transfection.

The Golgi localization of the PRR+CRR fusion protein was confirmed by observing that GFP fluorescence colocalized with staining for the 58-kDa Golgi protein antibody in fixed COS cells (Figure 7, a and b). In addition, the GFP fluorescence colocalized with BODIPY TR ceramide, a Golgi marker, in live cells (Figure 7, c and d). Treatment of cells with brefeldin A, which disintegrates the Golgi apparatus, abolished the perinuclear staining of the PRR+CRR fusion protein (Figure 7, e and f).

View larger version (51K): [in this window] [in a new window]   Figure 7.   Colocalization of the [PRR+CRR]-GFP with Golgi markers and treatment with brefeldin A. COS cells were transfected with a plasmid containing the cDNA for the [PRR+CRR]-GFP construct (Table 2). Two days after transfection, the cells were either fixed and prepared for immunohistochemistry with the anti-58-kDa antibody (a and b) or incubated with BODIPY TR ceramide (c-f) and brefeldin A (e and f) to label and disrupt, respectively, Golgi membranes. Cells were observed with fluorescence microscopy with the use of a FITC filter for detection of GFP (a, c, and e), a rhodamine filter for detection of the Cy3-labeled secondary antibody (b), or a Texas Red filter for detection of BODIPY TR ceramide (d and f).

Together, these data suggest that PRR could be a Golgi-targeting signal but that cysteines in the CRR were crucial for the attachment of otherwise soluble PRR to the Golgi membranes

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Heterotrimeric G proteins work not only at the plasma membrane to couple cell surface receptors to intracellular effectors but also at the Golgi membrane to regulate vesicular transport (Gilman, 1987; Melançon et al., 1987; Leyte et al., 1992). Consequently, these proteins must be sorted to their appropriate intracellular sites. We found that the Golgi localization of XLs, a G_s splice variant, was determined by a PRR and a CRR in its amino terminus.

The PRR of XLs

Deletion of the PRR in XLs led to a loss of its Golgi localization after expression in PC12 cells with endogenous XLs and in COS cells without the endogenously expressed protein. This protein was still attached to membranes and folded properly, as measured by its ability to undergo ADP-ribosylation catalyzed by cholera toxin. Fusion of the PRR and CRR to the GFP, a cytosolic protein, was sufficient for targeting to the Golgi membranes.

We initially chose to study this region because an alternatively spliced form of G_i2 has a PRR in the carboxy terminus and is found at the Golgi rather than at the plasma membrane, as is G_i2 (Montmayeur and Borrelli, 1994). The sequence of the human form of XLs was identified recently, and the PRR was highly conserved between species (Table 3), although the intracellular distribution of the human XLs is not known.

Proline-rich peptides have unique properties based on the unusual side chain of proline that circles back to the backbone amide position (Williamson, 1994). PRRs serve both structural and binding functions for the diverse group of proteins that contain them. For XLs, the PRR may be a rigid spacer between the amino-terminal domain and the cysteine-rich and _s domains. Deletion of the PRR bridge may then change the global conformation to diminish the interaction with the Golgi membrane and/or a Golgi protein. Alternatively, the PRR may be a site of direct binding to the Golgi. PRRs are found on a number of proteins, and their restricted mobility forms a "sticky arm" to facilitate protein-protein interactions that are rapid and less specific than typical receptor-ligand interactions that use "lock-and-key" binding (Williamson, 1994). This type of interaction is consistent with the results of our findings with a GFP fusion protein, in which the PRR and CRR alone could target the GFP fusion protein to the Golgi membranes.

Could the PRR be a Golgi-targeting signal for other proteins? A review of the literature uncovered several peripheral Golgi proteins that contain PRRs (Table 3). However, the functions of the PRRs and the Golgi-targeting signals (except for dynamin and III spectrin; see below and Table 3) are not known for these proteins. The abundance of proline residues, rather than a consensus pattern, distinguishes these regions. These proteins (except for dynamin) do not have known proline-enriched consensus sequences such as XPPXY or PPLP for WW domain binding, E/DFPPPXD/E for Ena-VASP homology domain 1 domain binding, or RXqPXqP or qPXqPXR for Src homology 3 domain binding (X is any amino acid, and q is a hydrophobic residue) (Chen and Sudol, 1995; Mayer and Eck, 1995; Niebuhr et al., 1997).

The PRR of dynamin has been carefully studied. Members of the dynamin family are high-molecular-weight GTPases involved in the vesiculation of clathrin-coated pits and are found on Golgi membranes as well as the plasma membrane and other intracellular membranes (Jones and Gutkind, 1998). Deletion of the proline-rich carboxy terminus abolishes targeting to coated pits (Shpetner et al., 1996), and an isoform of dynamin lacking the proline-rich carboxy terminus does not colocalize to the Golgi (Kamimoto et al., 1998). SH3 domain-binding regions have been identified in this PRR, but the role of these regions and other areas of the carboxy terminus in the intracellular localization of the different isoforms of the protein has not been reported (Okamoto et al., 1997). Interestingly, alternative splicing of dynamin, like XLs and G_s, can markedly change its intracellular distribution (Cao et al., 1998).

PRRs are involved in systems that require rapid and reversible association of several proteins into functional complexes (Williamson, 1994). Two well-characterized examples are the RNA polymerase II preinitiation complex and proteins associated with synaptic vesicles such as synapsins, VAMP-1, and synaptophysins. The PRRs of XLs and possibly other Golgi proteins may bind them into a Golgi-associated "complex." Specificity for Golgi membranes may arise through binding to isoforms of cytoskeletal proteins found at the Golgi (De Matteis and Morrow, 1998), because the binding target of PRRs is often the cytoskeleton (Williamson, 1994). For XLs, the PRR alone was insufficient to target the GFP fusion protein to the Golgi membranes, suggesting that it required another factor, such as acylation, to establish or maintain membrane attachment. Given the number of proteins with PRRs, an additional signal may also be present within the PRR and CRR to direct the protein to the Golgi membranes.

Palmitoylation of XLs

Like most other  subunits, XLs underwent palmitoylation, as demonstrated by its ability to incorporate [³H]palmitate. Unlike other  subunits, in which the putative acylation sites are within the first 18 residues, the likely sites of palmitoylation for XLs are on 6 cysteine residues within a CRR distant from the amino terminus, because mutation of these cysteines blocked [³H]palmitate incorporation. The nonpalmitoylated mutants were partially found in the soluble fraction, indicating that palmitoylation facilitated membrane attachment but was not required. The finding that the GFP fusion protein containing the PRR and the CRR lacking the cysteine residues was soluble suggests an additional site for membrane attachment outside of the PRR and CRR. Whether membrane attachment occurs through subunits, as it does for other  subunits, is uncertain because the regions of  subunits that bind differ between XLs and G_s.

Palmitoylation occurs on many membrane-bound proteins, and for that reason it cannot be a targeting signal by itself. However, palmitoylation is frequently found within or adjacent to a protein sequence that is critical for directing intracellular localization. For XLs, the CRR starts nine residues from the PRR. The importance of acylation in Golgi targeting differs among proteins. For eNOS, acylation was critical for its Golgi localization (Liu et al., 1997). Yet for GAD65, mutation of all six cysteines adjacent to the amino-terminal Golgi-targeting domain did not change its Golgi localization (Solimena et al., 1994). Mutations to prevent palmitoylation in another Golgi protein, SCG10, led to a small increase in solubility and the concomitant difficulties in visualizing the intracellular localization (Di Paolo et al., 1997), as was the case for the nonpalmitoylated mutant of XLs. In this study, the PRR was insufficient for Golgi localization, which required the putative acylation sites in the CRR for targeting of the GFP fusion protein. The proximity of acylation sites to targeting sequences may aid in orienting the targeting signal toward the membrane and preserving the contact. Differences among proteins in the role of palmitoylation in Golgi targeting may be due to differences in the affinity of the targeting sequence for the Golgi membrane.

Intracellular Localization and Function of XLs

The intracellular localization of XLs to the Golgi is based on colocalization with two Golgi markers, an antibody to the 58-kDa protein (Bloom and Brashear, 1989) and BODIPY-TR ceramide (Ktistakis et al.1995), and the loss of perinuclear staining after treatment with brefeldin A, a fungal metabolite that disrupts Golgi membranes. The compact pattern of the staining may indicate that it is localized to a discrete area within the Golgi complex. Some XLs is also found at the plasma membrane (Figure 4A). The presence of XLs only in cells with both regulated and constitutive secretion suggests that it may be involved in membrane trafficking. XLs could cycle between the plasma membrane and the Golgi and use its Golgi-localization signals for targeting to a functional domain in the Golgi involved in vesicle transport.

These results are the first to indicate that a PRR was critical for Golgi localization and suggest that, at least for G proteins, insertion of this region by alternative splicing may be a general mechanism for sorting and specific targeting. More studies are needed to determine if PRRs are a general localization signal for Golgi peripheral membrane proteins and the means and function of this targeting.