The inducibly expressed GTPase localizes to the endoplasmic reticulum, independently of GTP binding.

The inducibly expressed GTPase (IGTP) is representative of a newly identified group of interferon gamma-inducible GTPases, whose functions are currently unknown. We have begun to address the cellular function of IGTP by examining its subcellular distribution and its guanine nucleotide binding status. Using immunofluorescence, electron microscopy, and subcellular fractionation, IGTP was localized predominantly to the endoplasmic reticulum of both RAW 264. 7 macrophages and C127 fibroblasts. In the immunostaining experiments, staining of discrete cytoplasmic structures on the periphery of the endoplasmic reticulum was also evident. Using polyethyleneimine-cellulose thin layer chromatography, the guanine nucleotides that complexed to immunoprecipitated IGTP, in both control and interferon gamma-stimulated cells, were 90-95% GTP and 5-10% GDP, suggesting that the protein was in an active state. A mutant IGTP protein was created that had no detectable complexed GTP, and in both subcellular fractionation and IGTP-green fluorescent protein fusion studies, this mutant also localized to the endoplasmic reticulum. These results suggested that the GTP binding status of IGTP is independent of its capacity to localize to the endoplasmic reticulum. Given these results, we propose that IGTP is representative of a new family of endoplasmic reticulum GTPases that may be involved in protein processing or trafficking.

Interferon ␥ (IFN␥) 1 is a pleiotropic cytokine that regulates a wide variety of immunological and inflammatory responses (for review see Ref. 1). While production of IFN␥ is limited to T cells (2) and natural killer cells (3), its receptor is found on almost all cell types where it elicits diverse physiological responses. For example, in endothelial cells, IFN␥ increases the expression of major histocompatibility class I molecules and the cell adhesion molecules, ICAM-1 and ELAM-1, thereby promoting recruitment of immune cells to areas of inflammation (4). In macrophages, on the other hand, IFN␥ induces expression of major histocompatibility class II molecules (5), thus increasing Fc receptor-mediated phagocytosis (6) and mediating removal of neoplastic cells and parasitically and virally infected cells (7).
In addition, IFN␥ is thought to be involved in many pathological responses; in multiple sclerosis for instance, its production exacerbates disease symptoms (8) and increases the relapse rate (9).
The potent immunomodulating properties of IFN␥ have been utilized in a number of cancer treatment regimens that are currently used clinically or are under development. In particular, the interferons in combination with cytotoxic drugs have been quite effective in treating hairy cell leukemia (10) and chronic myelogenous leukemia (11). Preliminary results using IFN␥ for treatment of renal cell carcinoma (12), breast carcinoma (13), melanoma (14), and glioma (15) have also been promising. However, the molecular mechanisms that underlie these therapeutic effects as well as those underlying the normal physiological responses of IFN␥ are poorly understood and are the subject of active investigation.
The inducibly expressed GTPase (IGTP) is a recently identified 48-kDa protein whose expression is markedly up-regulated at the transcriptional level by IFN␥. In cultured macrophages and fibroblasts, IGTP mRNA accumulates within 1 h of exposure to IFN␥, reaching peak levels within 3 h and remaining at high levels for at least 48 h (16); in the mouse, the mRNA is highly expressed in thymus, spleen, small intestine, and lung (16). The IGTP protein, which contains three consensus GTP binding motifs, GXXXXGK(S/T), DXXG, and NTKXD (17), and inherent GTPase activity (16), is representative of an expanding group of closely related, IFN␥-regulated GTPases. Also included in this group of GTPases are the proteins encoded by the IRG-47 (18), TGTP (19)/Mg21 (20), and LRG-47 (21) cDNAs. Although their expression patterns imply that they may be important mediators of IFN␥ functions, the cellular and physiological functions of these proteins remain unknown. In the studies described here, we have begun to address the cellular function of IGTP by determining its subcellular localization, using a combination of immunostaining, electron microscopy, and subcellular fractionation procedures. We also examined the influence that the GTPase activity of the protein has on its subcellular localization, by first determining the guanine nucleotides complexed to the protein, and then examining the subcellular localization of a mutant protein that could no longer bind GTP. Based on our results, we propose that IGTP is representative of a family of IFN␥-regulated GTPases that localize to the endoplasmic reticulum (ER) and may be involved in protein processing or trafficking. quence encoding the FLAG amino acids DYKDDDDK would be inserted immediately downstream of the ATG start codon. This was accomplished using one primer that flanked the NcoI site at the ATG start codon (5Ј-GGCGCCATGGACTACAAGGACGACGACGACAAGGATTT-AGTCACAAAGTTGCCACAAAATATCTGG-3Ј) and contained the sequence encoding the FLAG amino acids, and a second primer that flanked the PflMI site 460 bases downstream of the ATG (5Ј-CGGGC-TAAAACAGGATATCAGG-3Ј); following amplification of this segment, the NcoI-PflMI-digested product was subcloned into NcoI-PflMI-digested pBS/IGTP. The products were sequenced to verify the insertion.
The plasmids encoding IGTP and IGTP(S98N) fused at their C termini to green fluorescent protein (GFP) were created using the polymerase chain reaction to amplify the coding regions of IGTP and IGTP(S98N) with primers (5Ј-AAAGCTAGCCGATTTAGTCACAAGTT-GCCACAAAATATC-3Ј and 5Ј-AAAGCTTAGCGTGAATTTCGGGAGG-GAGGACAGAGTCCTT-3Ј) that contained NheI restriction sites. The NheI-digested products were then subcloned into the NheI site of pFRED25, 2 a GFP expression plasmid containing a cytomegalovirus promoter and the bovine growth hormone polyadenylation signal.
Transfection-For transient expression assays, plasmids were introduced into cells using a standard calcium phosphate transfection method (23). For cells grown on 100-mm culture dishes, 20 g of plasmid DNA was precipitated in 1 ml of calcium phosphate solution that was then added to 10 ml of culture medium; for culture dishes of different sizes, the amount of DNA used and the volume of the calcium phosphate solution was changed proportionally.
Immunofluorescence Assays-C127 cells were grown on poly-L-lysine-coated, four-well glass chamber slides (12-550-51 SuperCell culture slides; Fisher), until they reached about 50% confluency. Some cells were then exposed for 4 h to 100 units/ml mouse interferon ␥ (Boehringer Mannheim) that was added directly to the growth medium. The cells were fixed with 4% (v/v) formaldehyde in PBS for 15 min, permeabilized with 0.2% (v/v) Triton X-100 in PBS for 10 min, and blocked with 10% (v/v) goat serum, 0.1% (v/v) polyxyethylene sorbitan monolaurate (Tween 20) in PBS for 1 h. The cells were then incubated for 30 min with affinity-purified rabbit polyclonal anti-IGTP (16) diluted to 10 g/ml in blocking solution, followed by an additional 30-min incubation with fluorescein isothiocyanate-conjugated anti-rabbit antibody diluted to 1:1000 in blocking solution (Boehringer Mannheim); between all incubation steps, the cells were washed several times with PBS. After they were mounted, the cells were observed and the images were analyzed using a Zeiss LSM310 confocal laser-scanning microscope; the microscope was configured with a 25-milliwatt argon, internal HeNe laser with the appropriate line (488) for fluorescein isothiocyanate excitation. Normaski images were made using the 543 green laser and appropriate polarizer lenses. Photographs were taken using a Sony color video printer UP5200 MD Mavigraph, Focus Graphics 4700.
Electron Microscopy-C127 and RAW cells were fixed in 4% paraformaldehyde (v/v) and 0.25% (v/v) glutaraldehyde in PBS for 30 min at room temperature. Following four 20-min washes with PBS, the cells were dehydrated in a series of graded ethanol solutions and embedded in LR White waterphilia resin and polymerized at 48°C. Thin sections were then cut in LKB Nova and lifted with formvar-coated 200 mesh gold grids.
The sections were washed three times with PBS for 10 min each, incubated in 1% (w/v) bovine serum albumin in PBS for 30 min, and then incubated for 2 h at room temperature with an affinity-purified anti-Grp78 antibody (PA1-014, Affinity Bioreagents, Golden, CO) previously conjugated with 5-nm gold particles. Next, following four 20min washes with PBS and one 30-min incubation with 1% (w/v) bovine serum albumin in PBS, the cells were incubated for 2 h at room temperature with the affinity-purified anti-IGTP antibody (16) previously conjugated with 10-nm gold particles. Following the second incu-bation, the cells were washed four times with PBS for 20 min each, two times with water for 10 min each, and then stained with 2% (w/v) uranyl acetate for 4 min. Control cells were stained with a 5-nm colloidal gold solution without the primary antibody to ensure that background staining was low. Sections were observed and images were recorded with a Philips EM 410 transmission electron microscope.
Subcellular Fractionation-Membrane protein fractions were isolated from RAW 264.7 and C127 cells using a previously established sucrose step gradient fractionation method (24). All steps of the procedure were performed at 4°C, and 0.06 g/ml aprotinin and 0.5 g/ml leupeptin were added to all buffers. In brief, the cells were rinsed three times with PBS, dislodged from the culture dishes by gentle scraping with plastic cells scrapers, and then pelleted by centrifugation at 500 ϫ g and 4°C for 5 min. The cells were gently resuspended in about 20 ml of a hypotonic buffer (10 mM HEPES, pH 7.5, 1 mM MgCl 2 ) and pelleted at 500 ϫ g for 5 min. Next, the cells were resuspended in 1 ml of hypotonic buffer and incubated on ice for 5 min. The plasma membranes of the cells were broken with 20 -25 strokes of a Dounce homogenizer; the integrity of the plasma membrane was monitored by light microscopy. The lysate was then cleared of nuclei by centrifugation at 1000 ϫ g for 5 min. The nuclear pellet was resuspended in 1 ml of hypotonic buffer and subjected to 10 strokes with a Dounce homogenizer; the nuclei were pelleted by centrifugation at 1000 ϫ g, the nuclear pellet was discarded, and the 1000 ϫ g supernatant was combined with the other 1000 ϫ g supernatant. 50% (w/v) sucrose was added to the combined supernatants to a final concentration of 8.5%, and this solution was centrifuged at 10,000 ϫ g for 15 min. The 10,000 ϫ g supernatant was carefully layered on a sucrose step gradient of 0.6-ml layers of 20, 30, and 38% (w/v) sucrose solutions, prepared in hypotonic buffer; the gradient was ultracentrifuged at 100,000 ϫ g for 2 h at 4°C in an SW60 rotor (Beckman, Sunnyvale, CA). Fractions were then isolated as described under ''Results,'' and the amount of protein in each fraction was determined using the Bio-Rad protein assay system. Immunoprecipitation, protein gel electrophoresis, and Western blotting were performed as described previously (16). The anti-IGTP rabbit polyclonal antiserum (16) was used at a 1:500 dilution for immunoprecipitation, and a 1:1000 for Western blotting. The mouse monoclonal M2 anti-FLAG antibody (Kodak, New Haven, CT) was used at 20 g/ml for immunoprecipitation. The goat peroxidase-conjugated anti-rabbit antibody (Boehringer Mannheim) was used at a 1:10,000 dilution for Western blotting. The ECL detection system (Amersham, Buckinghamshire, United Kingdom) was used for Western blot detection.
Detection of Guanine Nucleotides Bound to IGTP-Cell lysates were prepared, protein was immunoprecipitated (16), and the complexed guanine nucleotides were separated by polyethyleneimine-cellulose thin layer chromatography according to previously described methods (25). A buffer of 50 mM Tris, pH 8, 0.5% (v/v) Nonidet P-40, 0.15 M NaCl, 0.02 M MgCl 2 , 0.06 g/ml aprotinin, and 0.5 g/ml leupeptin was used for cell lysis, and a 1:100 dilution of the anti-IGTP antiserum (16) was used for immunoprecipitation. A control sample of [␣-32 P]GTP was hydrolyzed in vitro with immunoprecipitated IGTP, as described previously (16), to produce [␣-32 P]GTP and [␣-32 P]GDP; this was used, in addition to the fluorescent indicator in the TLC plates, to determine the position of GTP and GDP on the TLC plates. Radiolabeled GTP and GDP were quantitated with a BAS 1000 (Fuji, Stamford, CT) phosphor imager.
Green Fluorescent Protein Assays-Cells were seeded onto 60-mm dishes and transfected with pFRED/IGTP and pFRED/IGTP(S98N). 24 h after transfection, the cells were analyzed using a Zeiss LSM 410 Micro System microscope in the confocal mode, with an argon laser line set at 488 nm.

Subcellular Localization of IGTP-
To determine the subcellular distribution of IGTP, C127 mouse fibroblasts were exposed to control conditions or to 100 units/ml IFN␥ for 4 h to induce expression of the protein, and the cells were used for immunostaining with an anti-IGTP antibody (Fig. 1); this rabbit polyclonal antibody had been raised against the 16 C-terminal amino acids of IGTP and previously was shown to identify a dominant IFN␥-inducible 48-kDa band in both immunoprecipitation and Western blotting experiments (16). In C127 cells, immunostaining was localized to the cytoplasm and was concentrated in the perinuclear region on discrete, somewhat globular structures (Fig. 1), a pattern typical of nuclear envelope/ER localization. The overall staining pattern was similar in control and IFN␥-stimu-lated cells and became more intense following IFN␥ treatment, probably as a result of increased IGTP protein synthesis (Fig. 1). Essentially the same staining pattern was obtained when IFN␥stimulated RAW 264.7 mouse macrophages were stained with the anti-IGTP antibody (data not shown). These results suggested that IGTP was localized to the ER in these cells and, in addition, that IFN␥ affected only the quantity of IGTP present and not the subcellular distribution.
The localization of IGTP in IFN␥-stimulated C127 and RAW cells was also examined by electron microscopy using the anti-IGTP antibody and an antibody that recognized Grp78 (BiP), a resident ER protein (26). In both RAW cells (Fig. 2) and C127 cells (data not shown), IGTP and Grp78 immunoreactivity showed extensive colocalization on ER membranes. In certain cases, the IGTP antibody stained large globular structures whose identity could not be determined (data not shown; see also Fig. 7). These structures seemed too large to be crosssections of the ER; it is possible that they were vesicles that had budded off of the ER.
The subcellular distribution of IGTP was also examined biochemically by isolating plasma, Golgi, and ER membrane pro-tein fractions from IFN␥-stimulated RAW cells, using a previously established sucrose step gradient system (24). A cytosolic cell lysate was prepared from IFN␥-stimulated RAW 264.7 cells in a buffer containing 8.5% (w/v) sucrose, and this solution was layered on steps of 20, 30, and 38% sucrose and centrifuged at 100,000 ϫ g. Five fractions were collected from the top of the gradient, with fractions 1 and 2 coming from the 8.5% sucrose layer, fraction 3 from the combined 20 and 30% layers, fraction 4 from the 38% layer, and fraction 5 from the pellet. It has been shown previously using biochemical markers, that for fractions isolated in this manner, fractions 1 and 2 contain mainly soluble cytosolic and plasma membrane protein, fraction 3 contains Golgi protein, and fractions 4 and 5 contain ER protein (24). For the present study, each fraction was subjected to Western blotting with the anti-IGTP antibody and control antibodies that recognized the Golgi protein ␤-COP and the endoplasmic reticulum protein Grp78 (BiP); as expected, ␤-COP was detected mainly in fraction 3, and Grp78 was detected mainly in fractions 4 and 5 (Fig. 3). IGTP was present predominantly in fraction 5, with lesser amounts in fraction 4, only a trace in fraction 3, and none detectable in fractions 1 and 2 FIG. 1. IGTP immunofluorescence in C127 cells. A, cells grown on glass slides were exposed to control conditions or to 100 units/ml IFN␥ for 4 h. The cells were stained with normal IgG or an anti-IGTP primary antibody and with a fluorescein-conjugated secondary antibody. The staining was analyzed and enhanced with a Zeiss LSM310 CLSM in the confocal mode. The digital images of the fluorescein-stained cells were colorized using the Zeiss glow scale table which colors pixels in decreasing order of intensity as white, yellow, and red. B, a higher magnification view of a cell exposed to IFN␥ and stained with the anti-IGTP antibody is shown. The perinuclear, nuclear envelope/ER pattern is clearly evident. (Fig. 3). In similar experiments using IFN␥-stimulated C127 fibroblasts, IGTP showed the same distribution profile (data not shown). These results supported those of the immunostaining experiments and suggested that the majority of IGTP in macrophages and fibroblasts was not present in the soluble cytosolic pool but was bound to the endoplasmic reticulum.
IGTP association with the ER raised the possibility that the protein could have been secreted from these cells. This seemed unlikely, given the lack of a signal sequence at the IGTP N terminus (16), the very small amounts of IGTP detected in Golgi protein fractions (Fig. 2B), and the lack of immunodetectable IGTP near the cell periphery (Fig. 1). However, the possibility was addressed by Western blotting of protein isolated from the culture medium of RAW cells exposed to IFN␥ for up to 24 h (data not shown). Although IGTP was readily detected in RAW cell lysates, no secreted IGTP was detected in the cell culture medium (data not shown).
IGTP Guanine Nucleotide Binding-Previously, IGTP was shown to contain GTP binding motifs, as well as GTPase activity (16); however, the distribution of guanine nucleotides that were complexed to the protein in vivo was not determined. To address the guanine nucleotide binding status of IGTP, anti-IGTP antibodies were used to precipitate protein from RAW cells exposed to IFN␥ or control conditions, and the nucleotides that were complexed to precipitated protein were then separated by thin layer chromatography (Fig. 4). Under control conditions, 90.0 Ϯ 5.2% (S.D.) of the detectable guanine nucleotides complexed to IGTP were GTP, and the remaining 10% were GDP (Fig. 4). Following IFN␥ exposure, although the total amount of detected guanine nucleotides increased along with the amount of immunodetectable IGTP protein ( Fig. 4 and data shown), the overall percentage of GTP complexed to IGTP remained about the same at 93.8 Ϯ 0.9% (S.D.). Because many GTP-binding proteins, such as Ras (27), are in an active conformation when they are in the GTP-complexed state, these results suggested that the majority of IGTP in these cells was in an active state and, further, that IFN␥ did not influence the steady-state activity of the protein.
As a control experiment, lysates were prepared from asynchronous NIH/3T3 cells, and the guanine nucleotides bound to immunoprecipitated Ras were also determined using thin layer chromatography (data not shown). About 82% of the Ras-com-plexed guanine nucleotides were GDP, which was in agreement with previously published results (25). This demonstrated that GDP was easily detected under the assay conditions, and it suggested that detection of low levels of GDP-bound IGTP was not a consequence of inappropriate experimental conditions. Subcellular Distribution of an IGTP GTP-binding Mutant Protein-There are numerous ways in which the GTP binding status of IGTP could influence its function; one possibility is that the affinity of IGTP for the ER might be regulated by its GTP binding state. Such is true of the Rab GTPases, which are recruited onto membrane vesicles when in the GTP-complexed state (for review, see Ref. 28). To determine the effect of perturbing the GTP binding of IGTP, a mutant IGTP protein was created by mutating serine 98 in the first GTP binding motif (GDSGNGMS) to an asparagine (GDSGNGMN); in the case of cells. Cells were exposed to 32 P i and to 100 units/ml IFN␥ (ϩ) or control conditions (Ϫ) for 4 h and were then used for immunoprecipitation with preimmune serum or anti-IGTP antiserum. The guanine nucleotides complexed to precipitated protein were separated by polyethyleneimine-cellulose thin layer chromatography, followed by autoradiography. The GTP and GDP signals were quantitated with a phosphor imager. The control sample was [␣- 32  the Ras GTPase, the analogous serine is involved in contacting the Mg 2ϩ cofactor (29), and mutation to an asparagine yields a protein with a greatly reduced affinity for GTP (30). A FLAG amino acid tag (DYKDDDDK) was also placed at the amino terminus of the mutant IGTP(S98N) protein so that exogenously expressed protein could be distinguished from the endogenous protein. To characterize the GTP binding properties of FLAG-IGTP and FLAG-IGTP(S98N), the two proteins were transiently expressed in NIH/3T3 cells, protein was immunoprecipitated with the anti-IGTP antibody, and the complexed guanine nucleotides were separated by thin layer chromatography (Fig. 5). For FLAG-IGTP, the guanine nucleotide distribution paralleled that of the endogenous protein (Fig. 4), with 96% of the detected guanine nucleotides being GTP and the remaining 4% GDP (Fig. 5 and data not shown); the small amounts of GDP were evident only after long exposure of the TLC plate to film, but they were easily detectable with a phosphor imager (data not shown). Conversely, while small amounts of GDP were also detected complexed to FLAG-IGTP(S98N) (data not shown), no GTP above background levels was detected (Fig. 5). Similar results were obtained when the FLAG-IGTP and FLAG-IGTP(S98N) were transfected into HeLa cells (data not shown), and in all cases, levels of expressed FLAG-IGTP and FLAG-IGTP(S98N) protein were approximately the same (Fig. 6 and data not shown). Analogous experiments were performed using an anti-FLAG antibody to immunoprecipitate protein, and in these studies, the detected amounts of GTP and GDP complexed to FLAG-IGTP and FLAG-IGTP(S98N) were essentially the same as those detected using the anti-IGTP antibody (data not shown).
To determine if the S98N mutation affected IGTP localization to the ER, FLAG-IGTP and FLAG-IGTP(S98N) were transiently expressed in HeLa cells, membrane protein fractions were isolated, and the expressed proteins were detected by immunoprecipitation with an anti-FLAG antibody and Western blotting with anti-IGTP antibody (Fig. 6). Both FLAG-IGTP and FLAG-IGTP(S98N) were detected almost exclusively in fraction 5, with very small amounts present in fraction 4. These results indicated that greatly reducing the GTP complexed to IGTP had no effect on targeting the protein to the ER; however, they did not rule out that smaller movements within the ER compartment are influenced by GTP. With long exposures of the Western blot, faint bands that migrated slightly more rapidly than FLAG-IGTP and FLAG-IGTP(S98N) were also detected in fraction 5 ( Fig. 6 and data not shown). The identity of these bands was not determined; they may have been wild-type IGTP that lacked the 9.9-kDa FLAG epitope and was co-precipitated with FLAG-IGTP, or alternatively, they may have been unrelated co-precipitated bands that were recognized by the anti-IGTP blotting antibody.
The localization of IGTP(S98N) was also examined in intact, living cells by transiently expressing IGTP-green fluorescent protein (IGTP-GFP) fusion proteins in HeLa cells (Fig. 7). The IGTP-GFP and IGTP(S98N)-GFP fluorescence patterns (Fig. 7) were very similar to the immunostaining pattern of the endogenous IGTP protein (Fig. 1), with strong perinuclear staining that radiated out into the cytosol but did not extend completely to the cell membrane. Along the periphery of the ER, distinct, globular structures also showed strong staining for both IGTP-GFP and IGTP(S98N)-GFP. The nature of the globular structures could not be determined; they may have been crosssections of ER membranes or perhaps portions of the ER that were budding off. These results again suggested that GTPbinding did not affect localization of IGTP to the ER.
Other experiments were performed that addressed the effect of complexed guanine nucleotides on the binding of IGTP to ER membranes. In one experiment, in vitro translated IGTP was preloaded with GDP, GTP, or GTP␥S and then incubated with purified ER membranes, either in the presence of Mg 2ϩ so that the IGTP GTPase would be active, or in the presence of EDTA so that the GTPase would be inactive. Buffer conditions were chosen in which the association of other proteins with purified membranes, such as the association of AP-1 with Golgi membranes (31), had been successfully studied in vitro previously. However, under each assay condition, the amount of IGTP that bound to the membranes remained essentially the same (data not shown). In a second experiment, IFN␥-stimulated RAW cells were used for purification of ER membranes that were thus loaded with endogenous IGTP; the membranes were then incubated in buffers containing GTP, GDP, or GTP␥S and either Mg 2ϩ or EDTA. Again, under none of the experimental conditions was IGTP released from the membranes (data not shown). Taken together, these results support the above finding that binding of IGTP to ER membranes is not regulated by complexed GTP. DISCUSSION Using a combination of immunocytochemical and cellular fractionation methods, we have shown that the majority of IGTP in macrophages and fibroblasts is bound to the endoplasmic reticulum, and very little is present in the soluble cytosolic protein pool. In addition, the protein is predominantly GTPbound and, consequently, is probably in an active state. However, it does not appear that complexing with GTP regulates FIG. 5. Guanine nucleotides complexed to FLAG-IGTP and FLAG-IGTP(S98N). NIH/3T3 cells were transiently transfected with a FLAG-IGTP expression vector, a FLAG-IGTP(S98N) expression vector, or the expression vector without an insert (Control). The cells were then exposed to 33 P i for 4 h, lysates were prepared, and protein was immunoprecipitated with an anti-IGTP antibody. The guanine nucleotides complexed to precipitated protein were separated by polyethyleneimine-cellulose thin layer chromatography, followed by autoradiography. The GTP and GDP signals were quantitated with a phosphor imager. The positions of GTP and GDP indicated at the right were determined as described in the legend to IGTP's association with the ER in general, although our data do not rule out that GTP-binding could regulate small movements of the protein within the ER compartment. Based on these findings, we propose that IGTP is representative of a new family of ER GTPases that potentially could be involved in protein processing or trafficking.
The nature of the association of IGTP with the endoplasmic reticulum still remains to be addressed experimentally. Many resident ER proteins, such as Grp78, contain a C-terminal KDEL sequence that is necessary and sufficient for their retention in the ER (32), yet IGTP lacks this sequence. Binding of endogenous IGTP to ER membranes can be completely disrupted with nonionic detergents and partially disrupted with 1.5 M potassium acetate (data not shown), the latter of which suggests that protein-protein interactions may mediate the association. However, an alternative possibility is that IGTP is covalently modified with a lipid moiety, such as a myristoyl or prenyl group, that mediates binding of the protein to the membrane. Because IGTP lacks both a C-terminal CAAX prenylation sequence (33) and an N-terminal myristoylation sequence (MGXXX(S/A/T)) (34), the nature of a possible lipid moiety is not apparent.
In the IGTP-GFP fusion protein studies and also in the immunocytochemical electron microscopy studies, we observed distinct globular structures that were positive for IGTP, but the nature of these structures could not be determined. Because they were located on the periphery of the ER, one possibility is that they were vesicular structures that had budded off of the ER and were perhaps involved in membrane transport from the ER to the Golgi or to some other cellular location. If this were the case, then the lack of substantial amounts of IGTP in the Golgi and plasma membrane fractions suggests that IGTP dissociates rapidly from these putative vesicles once they reach their destination. Characterization of these structures will be important in determining the function of IGTP.
There are many families of GTPases that associate with the ER or Golgi and are involved in protein processing or trafficking. Included in these proteins are the RAB proteins, a group of over 30 small GTPases that are recruited onto nascent trans-port vesicles and target them to their appropriate membrane acceptors (28), and the ADP-ribosylation factors, a group of at least 13 Ras-related proteins that regulate reversible binding of coat proteins to Golgi membranes (for review, see Ref. 35). These proteins, however, are smaller than IGTP and bear no primary sequence homology to it. One GTPase that does bear limited homology to IGTP is SRP54, the 54-kDa subunit of the signal recognition particle that targets ribosomes with a signal sequence to the ER (36); it is 43% similar to IGTP at the amino acid level, including a methionine-rich area near the carboxyl terminus. Future studies should focus on whether SRP54 has a function similar to that of IGTP or whether IGTP and its relatives comprise a distinct class of protein-processing and -trafficking GTPases.

FIG. 7. Localization of IGTP-GFP and IGTP(S98N)-GFP in HeLa cells.
HeLa cells were transiently transfected with an IGTP-GFP or IGTP(S98N)-GFP expression vector. 24 h after transfection, the living cells were analyzed for protein expression, using a Zeiss LSM 410 microscope in the confocal mode, with the argon laser line at 488 nm. Shown are two representative groups of cells for each expressed protein. Note that the GFP fusion protein staining pattern is identical to the immunocytochemistry staining pattern of the endogenous protein in fixed cells (Fig.  1). Also note the globular pattern clearly evident in the living cells.