The Interferon (IFN)-induced GTPase, mGBP-2

To investigate the function of mGBP-2, a member of the interferon (IFN)-induced guanylate-binding protein family of GTPases, NIH 3T3 fibroblasts were generated that constitutively expressed mGBP-2. mGBP-2 induced a faster growth rate, with the highest expressing clones showing approximately a 50% reduction in doubling time. mGBP-2-expressing cells also grew to higher density and exhibited partial loss of contact growth inhibition, as evidenced by the formation of foci in post-confluent cultures. In addition, mGBP-2-expressing cells showed decreased dependence on serum-derived growth factors. However, they did not lose the requirement for anchorage-dependent growth. Finally, NIH 3T3 cells expressing mGBP-2 formed tumors in athymic mice. An mGBP-2 protein carrying a point mutation (S52N) that reduced GTP binding failed to produce these phenotypes when expressed at the same levels as wild type. The additional finding that IFN-γ treatment of NIH 3T3 cells resulted in an increase in proliferation similar to that observed for mGBP-2 in the absence of other IFN-induced proteins suggests that mGBP-2 may indeed be important for these growth changes.

To investigate the function of mGBP-2, a member of the interferon (IFN)-induced guanylate-binding protein family of GTPases, NIH 3T3 fibroblasts were generated that constitutively expressed mGBP-2. mGBP-2 induced a faster growth rate, with the highest expressing clones showing approximately a 50% reduction in doubling time. mGBP-2-expressing cells also grew to higher density and exhibited partial loss of contact growth inhibition, as evidenced by the formation of foci in post-confluent cultures. In addition, mGBP-2-expressing cells showed decreased dependence on serum-derived growth factors. However, they did not lose the requirement for anchorage-dependent growth. Finally, NIH 3T3 cells expressing mGBP-2 formed tumors in athymic mice. An mGBP-2 protein carrying a point mutation (S52N) that reduced GTP binding failed to produce these phenotypes when expressed at the same levels as wild type. The additional finding that IFN-␥ treatment of NIH 3T3 cells resulted in an increase in proliferation similar to that observed for mGBP-2 in the absence of other IFN-induced proteins suggests that mGBP-2 may indeed be important for these growth changes.
Cellular responses to interferons (IFNs) 1 involve the transcriptional induction of hundreds of genes (reviewed in Refs. [1][2][3][4], with some of the most robustly IFN-induced genes belonging to three families of GTPases (5). For type I IFNs (␣/␤) the most abundantly induced GTPases are the members of the Mx family. Although the function of Mx proteins is not fully understood, some have been shown to have antiviral activity (reviewed in Refs. 6 and 7). The response of cells to type II IFN (IFN-␥) is dominated by the induction of two families of GTPases, the 47-kDa family and the guanylate-binding proteins (GBPs) (5). Little is known about the function of either of these families of proteins, with the exception of the 47-kDa family protein, IGTP, that has been shown to have antimicrobial action against Toxoplasma gondii (8).
The GBPs are a family of 65-67-kDa proteins whose induction by IFNs has been studied in great detail but about which almost no information on their function has been acquired (9 -11). GBPs were originally identified as some of the most abundant proteins in IFN-treated cells (12)(13)(14). Although induced by IFN-␣/␤, their induction is much more robust following exposure to IFN-␥ (15,16). In fact, GBPs may make up close to 2% of the total protein in IFN-␥-treated murine fibroblasts (5). To date, GBPs have been cloned from humans (17), mice (5, 18 -20), rats (21), and chickens (22). In mammals there appear to be two genetically linked genes for GBPs that may have arisen by genetic duplication (18). In mice, there are at least two additional, more distantly related, family members, MAG-2 (19) and mGBP-3 (23). GBPs have only two easily identifiable motifs, a guanine nucleotide-binding site and, for most GBPs, a CAAX site for isoprenoid modification.
GBPs are unlike other GTPases in that they have only two of the three conserved regions of the standard guanine nucleotide-binding site (24,25). Consistent with these primary sequence differences, GBPs display a wider range of guanine nucleotide binding and hydrolysis than many other GTPases. All GBPs examined to date hydrolyze GTP to both GDP and GMP (22,26,27). Recently, two crystal structures for human GBP-1 (hGBP-1) have been solved, one in the absence of bound nucleotide (28) and the second bound to nonhydrolyzable GTP (29). Based on the biochemical properties and structure, hGBP-1 represents a unique class of GTPases but is suggested to be distantly related to the dynamin family of proteins (30,31). These are large GTPases that share the capacity to oligomerize and have an intrinsic high GTPase activity that shows a concentration dependence (28). The IFN-induced Mx proteins are also part of the dynamin family (31).
Also present on GBPs, except the more distantly related MAG-2 and mGBP-3, is a CAAX amino acid motif at their carboxyl termini that encodes a potential site for isoprenyl lipid modification. Isoprenylation is a multistep process that involves attachment of either a C-15 farnesyl or C-20 geranylgeranyl lipid to the cysteine residue of the CAAX (32). Although not all CAAX sequences are utilized in vivo, it has been demonstrated that hGBP-1 (33), rat p67 GBP (34), mGBP-2 (18), and mGBP-1 (35) are lipid-modified in vivo. All of these are modified by the C-20 isoprenoid except hGBP-1, which is farnesylated. Interestingly, it would appear that not all GBPs are uniformly prenylated in all cells, as mGBP-1 has recently been shown to be very poorly isoprenylated (35). We have recently demonstrated that mGBP-2 isoprenylation is necessary for targeting to a population of heterogeneously sized intracellular vesicle-like structures (16). mGBP-1 fails to go to these intracellular vesicle-like structures (16), possibly as a consequence of incomplete prenylation.
By using NIH 3T3 cells as a model system, we examined the functional consequences of expressing the GBP family member mGBP-2 in the absence of other IFN-induced proteins. We report here the first functional studies of murine GBPs that concludes mGBP-2 alters the growth characteristics of murine fibroblasts. Cells constitutively expressing mGBP-2 form foci when allowed to grow to post-confluence. The robustness of this phenotype correlates with the expression level of mGBP-2 and is not observed in cells expressing mGBP-2 with a mutant GTP binding domain. In this mutant mGBP-2, a single amino acid in the P loop has been changed from Ser at position 52 to Asn, a mutation equivalent to the traditional S17N mutation in Raslike proteins that reduces their relative affinities for GTP (36). Cells expressing mGBP-2 also grow at a faster rate and to a higher density. In addition, they have a reduced need for serum-derived growth factors for proliferation. However, these cells fail to grow as colonies in soft agar, suggesting that they have retained anchorage-dependent growth. In addition, they grow as tumors in athymic mice. Whereas IFN-␤ treatment of NIH 3T3 cells has little effect on cell growth, IFN-␥ enhances NIH 3T3 cell proliferation. The magnitude of the IFN-␥-induced reduction in doubling time is comparable with that observed for cells constitutively expressing mGBP-2 in the absence of other IFN-induced proteins. Taken together these data indicate that mGBP-2 functions by contributing to a signaling pathway that controls growth regulation and supports a role for mGBP-2 in IFN-␥-mediated growth regulation. These studies provide information that the highly abundant GBPs can function in regulation of some aspects of growth control by IFNs.
Plasmid Constructions-Construction of the amino-terminal FLAG epitope-tagged version of mGBP-2 was described (16). To make the equivalent of the Ras17Ser to Asn mutant in mGBP-2, a two-step PCR mutagenesis was performed. For the first amplification the template for both products was the plasmid G 1 (18). To amplify the 5Ј region of mGBP-2, the forward primer was 5Ј-ATATGCGGCCGCGCCTCAGAG-ATCCACATGTCG-3Ј, and the reverse primer was 5Ј-GTTCATCAGG-TAGTTTTTGCCTGTGC-3Ј. To amplify the 3Ј region, the forward primer was 5Ј-GCACAGGCAAAAACTACCTGATGAAC-3Ј, and the reverse primer was 5Ј-TGTGAGATCTGGCCTTCAGAGTATAGTGCACT-TCC-3Ј. The products were isolated and purified and added together as templates for the second PCR that used 5Ј-ATATGCGGCCGCGCCTC-AGAGATCCACATGTCG-3Ј as forward primer and 5Ј-TGTGAGATCT-GGCCTTCAGAGTATAGTGCACTTCC-3Ј as reverse primer. The amplified full-length product was digested with NotI and BglI1 and inserted into NotI/BglI1 cut pCMV2FLAG (NH) (gift of Lawrence Quilliam). Sequencing of the S52N mutant revealed a base change in the 3Ј region generated by PCR. This was corrected by digestion of both NB17 and NB4 with HindIII and BglII. The ϳ1100-bp fragment was isolated and ligated into NB17.
Cell Culture and Transfections-NIH 3T3 cells were obtained from the ATCC and maintained in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. Cells were transfected with FuGENE 6 (Roche Molecular Biochemicals) per the manufacturer's instructions. mGBP-2 constructs were transfected with pSV2Neo into NIH 3T3 cells at a ratio of 10 mGBP plasmids per 1 pSV 2 Neo plasmid. Stable transfectants were selected in 400 g/ml G418 (Invitrogen), and clones were screened for GBP expression by Western blot.
Western Blot-Cell lysates were size-fractionated by SDS-PAGE and transferred to Immobilon membranes, and Western blot analysis was performed as described (16,18). Detection was performed using the ECL system (Amersham Biosciences).
Immunofluorescence-Cells on coverslips in 6-well dishes were processed for indirect immunofluorescence as described (16).
Focus Formation Assay-Focus formation was performed as described (37), with modification. Cells were plated at 5 ϫ 10 5 cells per 6-cm dish. After 7 days the cells were washed gently, fixed in 10% methanol, 10% acetic acid, and stained in 1% crystal violet. Each assay was performed in triplicate and at least 3 times.
Growth in Low Serum-Analysis of growth in low serum was performed as described (37). Cells were plated at 1000 cells per dish in 6-cm dishes (in triplicate) and allowed to adhere overnight in 10% serum. The next day the media was changed to Dulbecco's modified Eagle's medium containing 0, 1, 2, 5, or 10% serum. After 14 days cells were washed gently with phosphate-buffered saline and stained with crystal violet. For several experiments the cells were counted using a hemocytometer at the end of the 14 days rather than stained.
Growth Rate and Saturation Density-Cells were plated at 1-2 ϫ 10 5 cells per dish in Dulbecco's modified Eagle's medium containing 10% serum. Cells were trypsinized and viable cells counted by hemocytometer daily. For growth rates of IFN-treated cells, NIH 3T3 cells were allowed to adhere for 2 h before the addition of 500 units/ml murine IFN-␥ or 1000 units/ml murine IFN-␤. Media and IFNs were replaced after 3 days and every other day after that. All assays were performed in triplicate. Two different lots of murine IFN-␥ (Calbiochem) were used and both elicited enhanced growth. To calculate doubling times, a region of the growth curve that most closely approached linear was chosen. By using values of x and y within the linear portion, the time it took to double the number of cells was calculated.
Binding to Nucleotide Agaroses-GTP-agarose (Sigma) was equilibrated and made into a 50% slurry in 20 mM Tris-HCl, pH 7.0, 150 mM NaCl, 5 mM MgCl 2 , 0.01% Triton X-100 (Binding buffer). These experiments are a modification of previous analyses by Cheng et al. (13). Labeled mGBP-2 and S52N mGBP-2 were generated using the Promega TNT kit and [ 35 S]methionine. The two constructs generated equal amounts of labeled proteins. Equal amounts of TNT lysate (2 l) were added to 30 l of packed beads and 70 l of binding buffer. After 1 h at 4°C, the beads were washed three times with 1 ml of cold binding buffer and eluted with SDS-PAGE sample buffer and incubation at 100°C for 3 min. Samples were run into SDS-PAGE gels, dried, and analyzed by PhosphorImager. Quantification was performed using the ImageQuant for Macintosh program version 1.2.
Cell Cycle Analysis-Actively growing cells were trypsinized and analyzed for cell cycle by staining cells with propidium iodide and flow cytometry analysis using Modfit (Becton Dickinson, San Jose, CA).
Tumor Formation in Athymic Mice-Three-to four-week-old NCR male athymic nude homozygous (nu/nu) mice (Taconic Farms) were irradiated with 20 radians followed by intradermal injection of 2 million control 20 or clone 3 cells into flanks in the mid-axillary line. Tumor sizes were measured using the formula for prolate spheroid as follows: volume ϭ (4/3)xa 2 b, where 2a ϭ minor axis and 2b ϭ major axis. Eight control and eight clone 3 injection sites were followed for tumor formation, and the results are presented as mean tumor volume Ϯ S.E.

Generation of Stable Transfectants Expressing mGBP-2 in NIH 3T3
Cells-To begin to understand the function of the robustly IFN-induced GTPase, mGBP-2, NIH 3T3 cells were transfected with FLAG epitope-tagged mGBP-2 (16), and stable cell lines were established that expressed mGBP-2 at a variety of steady state levels. To be confident that the phenotype we observed was not the consequence of variations in the properties of one particular clone of NIH 3T3 cells, two separate transfections and clonal isolations were performed using two different batches of NIH 3T3 cells. The highest expressing cell lines were clones 3 and 4 from the first isolation. The control clone chosen for further analysis from this screen was control 20. A second isolation of clones was performed using a new batch of NIH 3T3 cells at lower passage level. The two highest expressing clones from this screen were clone 1 and clone 11. The control clone chosen from the second screen was control 16. The levels of expression of mGBP-2 in these transfectants were compared with those of NIH 3T3 cells treated with IFN-␥ or IFN-␤ for 24 h (Fig. 1A). As observed previously,
To examine the role of a wild type GTP binding domain in mGBP-2 function, a mutant mGBP-2 was generated that contained a single point mutation in the first conserved region of the tripartite GTP binding domain (P loop). This mutation (S52N) is directly comparable with the serine to asparagine point mutation at position 17 of Ras (S17N), a mutation that has been made in a wide variety of GTPases and has often generated GTPases with reduced affinity for GTP (36). This serine is involved in binding to magnesium ions in the nucleotide-binding pocket in members of the Ras family and in hGBP-1 (28,29). The FLAG epitope-tagged S52N GTPase mutant was used to generate stable cell lines. As shown in Fig. 1A, the clone expressing S52N mGBP-2 that was used for these studies expresses the mutant mGBP-2 at levels comparable with the wild type mGBP-2 expression of clone 3 (91 Ϯ 18%).
To verify that this point mutation did indeed generate a mGBP-2 protein with altered nucleotide binding, the ability of GBPs to bind to GTP-immobilized agarose was exploited (13). Equal amounts of labeled in vitro transcribed and translated mGBP-2 and S52N mGBP-2 were incubated with GTP-agarose, and the relative levels of bound protein were analyzed by PhosphorImager analysis (Fig. 1B). Wild type mGBP-2 bound to GTP-agarose more robustly than did the S52N mGBP-2. The amount of S52N mGBP-2 bound to GTP-agarose was 44.3 Ϯ 8.4% (n ϭ 4) that of wild type mGBP-2, supporting the concept that this mutation reduces the protein's affinity for GTP.
Intracellular Localization of mGBP-2 and S52N mGBP-2-We have demonstrated that endogenous mGBP-2 expressed in IFN-␥-treated NIH 3T3 cells is found in a punctate distribution throughout the cell cytoplasm but is also localized to conspicuous vesicle-like structures of heterogeneous sizes and distribution (16) (Fig. 2A). As a prelude to functional assays using the stable cell lines, we first asked whether mGBP-2 targeted correctly when constitutively expressed in the absence of other IFN-induced proteins or whether its localization required the presence of other IFN-induced proteins. Although immunofluorescence localization is not expected to detect subtle differences in protein localization, it can be used to look for gross protein mistargeting or aggresome formation. Indirect immunofluorescence of NIH 3T3 cells stably expressing FLAG epitope-tagged mGBP-2 showed a staining pattern indistinguishable from that of the endogenous mGBP-2 protein in IFN-␥-treated cells (Fig. 2C). Having determined that lipid modification was necessary for vesicle targeting of mGBP-2 containing a wild type GTPase domain (16), it remained to be determined whether the GTP-binding mutant had an intracellular distribution that differed from that of wild type mGBP-2. For small GTPases, such as members of the Rab family, isoprenoid modification is necessary for vesicle targeting, but it is not sufficient. A relatively large proportion of the protein remains in the cytoplasm as a consequence of an isoprenoid-

FIG. 2. Localization of mGBP-2 and the S52N mGBP-2 in stable NIH 3T3 cells. NIH 3T3 cells expressing mGBP-2 (C and D) or S52N
mGBP-2 (E and F) were plated on coverslips, and mGBP-2 was localized by indirect immunofluorescence with anti-FLAG antibody. The localization of the transfected mGBPs was compared with that of endogenous mGBP-2 in IFN-␥-treated NIH 3T3 cells (A and B) detected with rabbit anti-mGBP-2 antiserum. mGBP-2 was distributed in a punctate pattern throughout the cytoplasm of cells and found in larger vesiclelike structures of heterogeneous size, number, and distribution (A and C). An indistinguishable distribution was detected for the S52N mGBP-2 (E and F). Size bar, 50 m.

mGBP-2 Increases Fibroblast Growth
mediated protein-protein interaction. The fraction of Rab proteins associated with vesicles is the GTP-bound or active portion of the protein (38,39). Indirect immunolocalization of the S52N mutant of mGBP-2 was indistinguishable from the wild type mGBP-2, suggesting that mGBP-2 targeting to membranes is independent of GTP binding (Fig. 2E). To date the importance of membrane localization by mGBP-2 has not been correlated with any known function. These observations that mGBP-2 and the GTP-binding mutant show intracellular distributions indistinguishable from the endogenous protein suggest that any differences in the phenotype observed for these cells are not the consequence of gross mistargeting of the respective proteins.
The Expression of mGBP-2 in NIH 3T3 Cells Accelerates Cell Growth-Whereas interferons induce an antiproliferative response in many cells, one notable exception is fibroblasts, where IFN-␥ treatment can promote primary fibroblast proliferation in vitro (40 -46). Consequently, the NIH 3T3 cells expressing mGBP-2 and the GTP-binding mutant were examined for changes in their growth properties. To determine whether there were changes in the rate of cell growth, cell counts were performed subsequent to plating equal numbers of each clone. These cells were followed for 3-4 days after plating. Beyond that time the curves began to plateau and doubling times slowed for some clones, presumably due to cell density. Representative growth curves generated for mGBP-2-expressing cells from both clonal isolations are shown (Fig. 3, A and B). Doubling times were calculated as described under "Experimental Procedures." The doubling time for control 20 (32.2 Ϯ 6.8 h; n ϭ 5) was similar to that observed for control 16 (31.7 Ϯ 3.4; n ϭ 4). The two clones expressing mGBP-2 at levels comparable with IFN-␥-treated NIH 3T3 cells had doubling times The two clones expressing mGBP-2 at lower levels had doubling times of 26.8 Ϯ 3.8 (clone 4; n ϭ 5) and 27.3 Ϯ 8.3 h (clone 11; n ϭ 4). These data show that mGBP-2 expression at the level observed for IFN-␥ treatment of NIH 3T3 cells (clones 3 and 1) shortens the doubling time of NIH 3T3 cells by almost 50% (Fig. 3D). The failure of cells expressing the GTP-binding site mutant of mGBP-2 to grow at these rapid rates (30.38 Ϯ 10.7; n ϭ 5) shows the requirement of GTP binding and/or hydrolysis for modulating cell growth rates (Fig. 3C).
Flow cytometry was used to evaluate the proportion of mGBP-2-expressing cells in each stage of the cell cycle. Comparison of actively growing control 20 and clone 3 cells consistently showed a slight increase in the proportion of mGBP-2 expressing cells in S phase with a concomitant decrease in the proportion of cells in G 1 /G 0 phase compared with control cells (Fig. 4). However consistent these differences were, the magnitude was small and not statistically significant. That mGBP-2 did not result in differences in the proportion of cells in each stage of the cell cycle was confirmed by flow cytometry of control 16 and clone 1. These data suggest that the length of each of the different stages of the cell cycle is proportionally reduced by mGBP-2.
NIH 3T3 Cells Treated with IFN-␥ Have Accelerated Growth Rates-Although the literature contained several examples of IFN-␥-induced proliferation of fibroblasts, these studies were performed with primary fibroblasts. Therefore, the growth of NIH 3T3 cells was examined in the presence or absence of IFN-␥ or IFN-␤ (Fig. 5A). IFN-␤ had no significant antiprolif-erative effect on these cells (time for doubling of 32.5 Ϯ 6.5 h compared with 31.2 Ϯ 0.1 h for untreated controls; n ϭ 3). However, IFN-␥ treatment was growth promoting for NIH 3T3 fibroblasts. The IFN-␥-treated cells grew faster than either untreated or IFN-␤-treated cells (21.5 Ϯ 3.4 h; n ϭ 4) (Fig. 5A). Because the NIH 3T3 cells showed greater variability in growth rates, we also presented the doubling times as percent of control (Fig. 5B). When the doubling times of IFN-␥-treated cells are compared with the untreated cells on a per experiment basis, IFN-␥ treatment reduces the doubling time to 60 Ϯ 4% of untreated cells.
NIH 3T3 Cells Expressing mGBP-2 Grow to Higher Density-We continued evaluation of the growth changes mediated by mGBP-2 by determining whether NIH 3T3 cells expressing mGBP-2 showed density-arrested growth. For these experiments cells were allowed to grow for 5 days after having reached confluence and were examined both by microscopy and subsequent to crystal violet staining of the cell monolayer. All of the control clones grew until they formed a single cell monolayer and then stopped growing (Fig. 6A). However, all of the transfectants expressing mGBP-2 continued to grow and formed foci (Fig. 6A). The number and size of these foci correlated with the level of mGBP-2 expression. The morphology of these foci are shown in Fig. 6B. The cells containing the S52N mutant failed to form foci, demonstrating that a wild type GTP binding domain is required for the formation of foci.
To examine more directly the differences in saturation density, growth curves of the cells expressing mGBP-2 were continued until cell numbers plateaued (Fig. 7). Similar to the correlation observed between expression level and doubling times, the clone expressing mGBP-2 at levels comparable with IFN-␤ treatment (clone 4) reached saturation at cell numbers comparable with control cells (clone 20) (about 3 ϫ 10 6 cells per

mGBP-2 Increases Fibroblast Growth
6 cm dish; Fig. 7A), whereas the highest expressing cells (clone 3) did not stop expanding until the cells had reached a density of about 4 ϫ 10 6 cells per 6-cm dish. S52N mGBP-2 grew at a similar rate to control cells (Fig. 3) but did not growth-arrest until a cell density intermediate between control and high expressing cells. There were, however, exceptions to this correlation. Clone 11 cells that express at low levels grew erratically and on occasion would grow to the same density as the highest expressing clone of that isolate (not shown). NIH 3T3 cells when treated with IFNs grew in a similar manner. The cells treated with IFN-␤ and the untreated cells stopped growing at about 3 ϫ 10 6 cells/6-cm dish, whereas treatment with IFN-␥ resulted in density arrest at closer to 4 ϫ 10 6 cells/dish (Fig. 7B).
The observation that mGBP-2 can induce the formation of foci and partial loss of contact growth inhibition suggested that mGBP-2 may mediate oncogenic transformation. Other hallmarks of such transformation are the reduced dependence on exogenous growth factors manifested by the ability to grow in reduced serum, the loss of anchorage-dependent growth as assessed by growth in soft agar, and the ability to form tumors in athymic mice. NIH 3T3 cells expressing mGBP-2 are unable to grow as colonies in soft agar, suggesting that they have not lost anchorage-dependent growth (data not shown). To determine whether mGBP-2 conferred the ability to grow in lower serum, cells were plated and allowed to attach for 16 h in the presence of 10% fetal calf serum and then were shifted to 2, 5, and 10% fetal calf serum for 14 days. Both clones 1 and 3 grew better than control cells in all serum levels tested (Fig. 8A). To determine whether a wild type GTP binding domain was re-quired for better growth in low serum, control 20, clone 3, and clone S52N were grown in low serum for 2 weeks, and the numbers of cells present were determined by trypsinization and counting. Again, the mGBP-2-expressing cells grew more robustly than the control cells, and wild type GTP binding domain was required for enhanced growth in low serum (Fig.  8B).
To determine whether mGBP-2 expression could promote tumorigenesis of NIH 3T3 cells, clone 3 and control 20 cells were injected intradermally into athymic mice and monitored for tumor formation. The clones had a lag time of about 2 weeks before tumors were large enough for detection, but after that time the tumors generated by mGBP-2-expressing cells grew rapidly (Fig. 9). The tumors generated by mGBP-2 expressing cells were solid, with no grossly detectable necrosis internally. No macroscopic metastatic lesions on other organs were observed.
This report demonstrates the first phenotype for the robustly induced IFN-induced GTPase, mGBP-2. These data show that mGBP-2 is capable of altering a number of growth parameters. mGBP-2 shortens fibroblast doubling time, reduces the need for serum-derived growth factors, induces growth to higher saturation density and focus formation, and induces tumor formation in nude mice. DISCUSSION We have generated NIH 3T3 cell clones constitutively expressing either the murine IFN-induced GTPase, mGBP-2, or mGBP-2 containing a point mutation in the P loop of the GTP

FIG. 6. NIH 3T3 cells expressing mGBP-2 form foci in culture.
A, equal numbers of control and mGBP-2-expressing NIH 3T3 cells were plated and allowed to grow 5 days post-confluence before staining with crystal violet. The clones expressing mGBP-2 formed foci, with clones expressing at higher levels generating more and larger foci (clones 3 and 1) than those expressing at lower levels (clones 11 and 4). binding domain, S52N mGBP-2. Analysis of these cells indicated that mGBP-2 is capable of altering several parameters of cell growth in the absence of other IFN-induced proteins. First, NIH 3T3 fibroblasts expressing mGBP-2 grew faster and to higher saturation density (Figs. 3 and 7). The extent to which the doubling time of NIH 3T3 cells is shortened by mGBP-2 correlates with its level of expression. Expression of mGBP-2 at levels comparable with those of NIH 3T3 cells treated with IFN-␥ for 24 h induced cells to grow about twice as fast as cells without mGBP-2 (Fig. 3). Flow cytometric analysis of the DNA content of these cells shows no difference in the proportion of cells in G 1 , S, or G 2 /M compared with control transfectants, suggesting that the different phases of the cell cycle are uniformly shortened by mGBP-2 (Fig. 4). Second, the increased density of cell growth observed in mGBP-2-expressing fibroblasts was accompanied by at least a partial loss of contact growth inhibition, as evidenced by the ability to form foci when grown to post-confluence. Foci formation is correlated with mGBP-2 expression level (Fig. 6). Third, the growth advantage of mGBP-2-expressing cells is not abrogated in reduced serum (Fig. 8). Finally, NIH 3T3 cells expressing mGBP-2 show enhanced tumor-forming ability in athymic mice (Fig. 9).
The increased growth rate, partial loss of contact growth inhibition, increased ability to grow in reduced serum, and the ability to form tumors in athymic mice, but retention of substratedependent growth, suggest a weak or incomplete form of transformation. This "incomplete" transformation has been observed for other GTPases, most notably members of the Rho family of small GTPases (reviewed in Refs. 47 and 48). The Rho family of GTPases is involved in a wide variety of cellular processes, such as signal transduction, cytokinesis, cell adhesion, migration, and cell proliferation (48 -50). In general, transformation by members of the Rho family is much less robust than observed for Ras and requires the expression of proteins containing activating mutations (51)(52)(53)(54)(55)(56). Even then, some Rho family members fail to induce all of these parameters of transformed growth. For example, expression of activated RhoA in NIH 3T3 cells does not result in focus formation or growth in soft agar but does promote reduced dependence on serum and tumorigenesis (47). Rho family proteins mediate these phenotypes by interaction with and activation of a wide variety of effector proteins (49,50). The mechanism(s) by which mGBP-2 mediates similar changes to those observed for activated Rho family members is at this time completely unknown. Certainly, one possibility is that mGBP-2 functions within one of the pathways used to mediate Ras or Rho family-induced cellular changes. Recently, the rat homolog of mGBP-2 was identified as a putative target of Ras transformation, having been identified as a gene robustly induced subsequent to Ras transformation of rat fibroblasts (57).
GTP binding and hydrolysis is crucial for many, but not all, functions of GTPases. We have shown for mGBP-2 that wild type GTP binding is required for its growth promoting activity as well as the ability to form foci and grow in reduced serum. For example, the antiviral activity of the IFN-induced protein MxA does not require GTPase activity (58).
Although interferons are probably best studied for their antiviral and antiproliferative activities, interferons are not antiproliferative for all cell types. For some cells they have no effect on cell proliferation (43)(44)(45)(46), and for others they can actually be growth-stimulatory (40 -42). IFN-␥ has been shown to be growth-promoting or mitogenic for a variety of human fibroblasts, such as human lung, synovial, or dental pulp fibroblasts. The combined observations that IFN-␥ is growth-stimulatory for NIH 3T3 fibroblasts and that mGBP-2 is capable of the same growth stimulatory activity in the absence of other IFN-induced proteins support a role for mGBP-2 in IFN-mediated cell growth regulation. However, the mechanism(s) by which mGBP-2 mediates its growth-stimulatory effects remains to be elucidated. As mentioned above, one possibility is that mGBP-2 functions within a pathway used by other small GTPases, such as members of the Rho family. The Rho proteins are important regulators for a number of growth-related processes. Certainly, Rho family proteins are involved in regulation of cytoskeleton organization and cytokinesis. They are also necessary for integrin-mediated signals, and in this capacity FIG. 8. Cells expressing mGBP-2 have a growth advantage under low serum conditions. A, equal numbers of starting mGBP-2 and control transfectants were grown in 2, 5, or 10% serum as described under "Experimental Procedures." After 14 days the cells were stained with crystal violet. The mGBP-2 transfectants grew better than control in all serum concentrations tested. B, control transfectant (gray bar), clone 3 (stripped bar), and S52N mGBP-2 (clear bar) were plated in triplicate as in A and after 14 days the cells were counted. The presence of mGBP-2 conferred a growth advantage in all serum conditions tested, whereas cells expressing the S52N GTPase mutant grew to similar to control.
FIG. 9. mGBP-2 renders NIH 3T3 cells tumorigenic. Equal numbers (2 ϫ 10 6 cells per injection) of control 20 (solid diamond) and clone 3 (solid square) cells were injected into athymic mice and tumor size was measured as described under "Experimental Procedures." The results are presented as the mean tumor volume in cubic centimeters Ϯ S.E. (n ϭ 8 tumors) versus days post-injection.

mGBP-2 Increases Fibroblast Growth
are actively involved in transducing cell growth signals. Alternatively, mGBP-2 may up-regulate the expression of growth factor receptors or other components involved in mitogenic signaling. How mGBP-2 mediates these growth changes is actively under investigation.