Specific Antiviral Activity Demonstrated by TGTP, A Member of a New Family of Interferon-Induced GTPases¹

We have previously described the cloning of a cDNA encoding a putative high m.w. GTPase designated TGTP. TGTP’s expression was induced rapidly after TCR cross-linking and appeared to be restricted to T cells in ex vivo tissues (1). The same cDNA was also cloned by Lafuse et al. from a library derived from IFN-γ-treated macrophages (2). These authors designated their clone Mg21 and demonstrated that it was an immediate-early (IE)⁴ gene of IFN-γ in macrophages. TGTP/Mg21 had one relative in the sequence databases designated IRG-47. IRG-47 was described as an IFN-γ-inducible gene encoding a protein with a putative GTP-binding domain. IRG-47 was expressed predominantly in cell lines of B cell and fibroblast origin (3). Since these publications, two more IFN-inducible relatives have been described in macrophages, including LRG 47 and IGTP. LRG-47 was shown to be induced by LPS and both type I IFN-αβ and type II IFN-γ (4). IGTP was induced by LPS and IFN-γ and its GTPase activity were confirmed in vitro (5). Of the data so far reported for this emerging gene family, members share approximately 47 kDa size, 25 to 40% amino acid identity, GTPase signature motifs, and IFN-γ inducibility. Indeed TGTP/Mg21, IGTP, and IRG-47 were shown to be IE genes of IFN-γ in macrophages and B cell lines, respectively. Expression of these members was not induced by other cytokines tested, including IL-2, -4, -10, and TNF-α, (TGTP/Mg21) (2), or IL-1, -2, -4, -6, TNF-α, and granulocyte-macrophage-CSF (LRG47) (4), reinforcing the view that these molecules function primarily in response to IFN-γ and therefore contribute to the innate response to bacterial infections. However, the function of this family of IFN-γ-induced molecules remains unresolved.

The importance of IFNs in host defense against viral and bacterial pathogens was clearly demonstrated by the increased susceptibility to viral and bacterial challenge in mice lacking the capacity to respond to type I (6) and type II (7, 8) IFNs, respectively (9). IFN-γ is known to influence the expression of over 240 genes that fall into approximately 37 functional categories (10). It has been proposed that this extensive response to IFN-γ reflects the induction of four major genetic programs that collectively regulate immunity and promote the elimination of infectious agents (7, 8, 10). These programs include the regulation of cytokine networks governing T and B cell differentiation, activation of phagocytic processes in macrophages and neutrophils, enhancement of Ag presentation, and direct antiviral activity (10). The genetic/functional complexity of the response to IFN-γ is impressive, but is also a confounding factor in the assignment of function to novel IFN-induced genes.

In this report, we confirm that TGTP is indeed a GTPase. We show that despite the rapid kinetics of TGTP RNA induction observed after TCR cross-linking (2 to 3 h) (1), the induction is nevertheless mediated by IFN-γ. IFN-γ is one of the most rapidly induced IE genes induced upon TCR stimulation (11) and, as in macrophages (2), TGTP is an IE gene of IFN-γ in T cells. Several observations directed our investigations toward the possible association of TGTP with viral defense. First, TGTP was found to be IFN inducible in cells lines of diverse lineages, including those derived from both hemopoeitic and non-hemopoeitc tissues; TGTP expression was not limited to phagocytic populations or T/B lineages. Second, although exogenous IFN-αβ was a weak inducer of TGTP, lipofection of plasmid DNA or poly(I:C) efficiently induced high levels of TGTP and the induction was mediated by IFN-αβ. IFN-αβ is a primary mediator of antiviral defense mechanisms. Third, difficulties in obtaining stable transfectants of TGTP suggested that expression of TGTP protein was tolerated poorly. This was consistent with possible antiviral activity insofar as interference with protein synthesis is an activity displayed by other known innate antiviral mechanisms, including dsRNA-dependent protein kinase (PKR) (12), the 2-5A synthetase and RNase L systems, and dsRNA-specific adensoine deaminase (dsRAD) (13, 14). Fourth, the only other functionally defined IFN-induced GTPase is Mx, an IFN-αβ-induced 70 to 80-kDa molecule with potent antiviral activity against a set of negative strand RNA viruses (15). We therefore examined whether TGTP transfection conferred a virus-resistant phenotype using the negative strand RNA virus, vesicular stomatitis virus (VSV). We obtained evidence that forced expression of TGTP conferred an antiviral state for the negative strand RNA virus, VSV, but not the DNA virus, Herpes. Thus TGTP functions as a new mediator of innate antiviral immunity and may thereby implicate other members of the aforementioned family of GTPases in viral defense.

B34 is a mouse IgG1 mAb generated against TGTP-GST fusion protein as described below and is specific for TGTP. XMG 1.2 is a neutralizing anti-IFN-γ Ab (16) and 145-2C11 is a mAb specific for CD3ε (17). Composite anti-IFN-αβ was obtained from Research Diagnostics (Flanders, NJ).

L cells were cultured in Iscove’s modified MEM (I medium) (Life Technologies, Burlington, Ont., Canada) supplemented with 10% FCS; I medium was used for all tissue culture. TGTP was induced in cultured cells by supplementing medium with IFNs. Commercial supplies of IFN-α (ICN, Montreal, Que., Canada) and IFN-γ (PharMingen, Mississauga, Ont., Canada) were used as well as IFN-γ-containing supernatants (SN) from Con A-stimulated spleen cells (Con A SN) generated in house. The TGTP-inducing activity of Con A SN was blocked by the IFN-γ-neutralizing Ab XMG1.2 (data not shown) and was thus considered to be mediated entirely by IFN-γ present in the supernatant.

Mice were immunized with TGTP-GST in Titremax (Cedarlane, Hornby, Ontario, Canada) on day 0 in both hind footpads, boosted with Ag in PBS on days 4, 8, and 12, and lymphocytes from popliteal, inguinal, and paraortic lymph nodes were fused lymphocytes with the non-Ig-producing myeloma cell line, AG865.3, on day 13. After fusion, cells were incubated overnight, harvested, and cultured with I medium, including 10% FCS and 1× hypoxanthine/aminopterin/thymidine in half-area plates 100 μl/well. Wells were scored for single colonies after 1 wk and transferred to 24-well plates in 1× hypoxanthine/thymidine (HT) medium. Supernatants were screened by their capacity to detect TGTP-GST in ELISA assays, and reactivity of putative clones was subsequently confirmed by Western blotting. The B34 Ab was able to detect TGTP by Western blotting and could also be used to quantitatively immunoprecipitate TGTP from cell lysates.

Blots were generated and probed using standard methods. Briefly cells were washed in PBS and lysed at ≤5 × 10⁷/ml on ice in buffer containing 20 mM Tris 7.5, 0.15 M NaCl, 0.5% Triton X-100, 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 174 μg/ml PMSF. Lysates were mixed by rotation for 30 min at 4°C and spun 20 min at 4°C at top speed in an Eppendorf centrifuge. The supernatant was collected and combined with Laemmli sample buffer and loaded into minigels (BioRad, Richmond, CA) with a 4% stacking gel and 10% resolving gel. Resolved proteins were transferred to PVDF (Millipore, Nepean, Ont., Canada) membrane that was subsequently blocked with 5% BSA in TBS (8.76 g NaCl + 6.06 g Tris base (pH 8.0) per liter). Blots were probed with B34 diluted in TBS containing 0.5% BSA and 0.5% Tween 80 (Fisher, Nepean, Ont., Canada) for 60 min. Blots were washed, bound B34 was detected with goat-anti-mouse-Ig-HRP (Southern Biotech, Birmingham, AL), washed, and developed with enhanced chemiluminescence reagent (Amersham, Oakville, Ont., Canada) for autoradiography with Hyperfilm (Amersham) according to the manufacturer’s instructions.

Northern blots were prepared as described previously (1). Cyclohexamide was used at 50 μg/ml to prevent protein synthesis.

TGTP was digested with the restriction enzyme NcoI that cleaves just 5′ of the initiating ATG codon and again at 145 bp after the polyadenylation signal of the TGTP cDNA. The NcoI fragment was blunt ended with Klenow and ligated to the BamHI linearized, blunt ended, and dephosphorylated pGEX-KT vector (American Type Culture Collection, Manassas, VA). Transformants of DH5α were selected that had incorporated TGTP insertions in either + or − orientations for preparation of fusion protein from bacterial lysates. Bacteria were grown in Luria Bertani broth supplemented with ampicillin to OD₆₀₀ 0.8 to 1.0. IPTG was added to 100 μM and the culture shifted to 26°C for 6 to 10 h. Bacteria were then pelleted and lysed on ice for 30 min in lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 mM PMSF, 1 mg/ml lysozyme, and 100 μg/ml DNase I). The lysate was clarified by ultracentrifugation for 45 min (50,000 × g), aliquoted, and frozen at −70°C. Fusion protein was purified with glutathione Sepharose 4B (Pharmacia, Baie d’Urfe, Que., Canada) according to the manufacturer’s instructions. Briefly, glutathione Sepharose slurry was washed three times in wash buffer (bacterial lysis buffer containing 0.1% Triton and lacking lysozyme and DNase I). Glutathione Sepharose was incubated with 10 volumes of lysate for 60 min with constant gentle mixing. After adsorption, the beads were gently spun and washed three times. Beads were rinsed one to two times with 75 mM HEPES, pH 7.4, 150 mM NaCl, and 0.1% Triton (HNT) and could be stored for short periods on ice in HNT including 10% glycerol. GST and TGTP-GST proteins were eluted from glutathione Sepharose with 10 mM reduced glutathione (Sigma, Mississauga, Ont., Canada) in 75 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM DTT, and 0.1% Triton X-100 for 20 min with constant mixing and then separated from the Sepharose by centrifugation. GST and TGTP-GST were subsequently quantitated by gel electrophoresis against purified albumin control reagents, resuspended to 2 mg/ml, and titrated into the GTPase reaction in MKH buffer, including trace [α-³²P]GTP and 150 mM cold GTP in MKH buffer (20 mM HEPES, pH 7.4, 50 mM KCl, and 4 mM MgCl₂). Five microliters of packed beads were then resuspended in reaction buffer (MKH). Hexokinase (Sigma) in the presence of glucose served as a control for the enzymatic hydrolysis of GTP. The GTPase reaction was conducted at room temperature for 30 min. Longer incubations did not yield greater GTP hydrolysis. Hydrolysis was evaluated by TLC on TLC plates (Aldrich, Milwaukee, WI) in 1 M KH₂PO₄ buffer (pH 3.4) and autoradiography.

L cells were seeded with 5 × 10⁵/well (6-well plates) or 10⁵/well (24-well plates) and cultured overnight. Lipofectamine (Life Technologies) transfections were conducted according to the manufacturer’s instructions. Unless otherwise indicated, three milligrams of DNA per 100 μl OPTI-MEM (Life Technologies) were mixed with 14 μl Lipofectamine/100 μl OPTI-MEM for 30 min. Opti-MEM was then added to a final volume of 1 ml and lipid DNA mixtures were added to subconfluent L cell cultures, 1 ml/well of a six-well tissue culture plate, or 200 μl/well of 24-well plates, and incubated 2 to 3 h at 37°C. The lipid-DNA solution was then discarded and I medium with 10% FCS was added for 24 h unless otherwise indicated. For transient expression, cells were assessed by Western blot. To isolate stable transfectants, Geneticin (Life Technologies) was added to the culture (900 μg/ml) until all untransfected control cells were killed. Cultures were then allowed to recover for 2 to 3 days and cloned at 0.7 cells/well or by colony, picking by hand 4 to 5 days later.

Stocks of VSV Indiana (generously supplied by Dr. Gregor Ried and Dr. Wilf Jeffries; University of British Columbia, Vancouver, Canada) and herpes simplex strain IF (HSV) (generously provided by Dr. Frank Tufaro; University of British Columbia) were diluted in PBS + 1% FCS and applied to monolayers of control L cells and transfectants for 60 min at room temperature. After infection, the diluted solutions of virus were discarded and replaced with 2 ml of either I medium with 2% FCS and 0.9% agarose at 45°C for VSV, or I medium with 2% FCS and 0.1% human IgG fraction II (ICN, Montreal, Canada) for HSV. Plates were incubated for 1 or 3 days, respectively, when plaque formation was evaluated with 0.1% neutral red staining solution in I medium containing 2% FCS) for 2 h at 37°C.

TGTP was previously shown to be induced by TCR cross-linking in thymocytes. This induction was both rapid, RNA expression occurring within 2.5 h of stimulation, and intense. The tempo of induction and the apparent T cell-restricted distribution of TGTP in ex vivo tissue suggested that TGTP was an early response gene of T cell activation (1). Lafuse et al. cloned TGTP (designated Mg21) independently from a cDNA library generated from IFN-γ-treated macrophages (2). Their data together with the IFN-γ inducibility reported for other TGTP relatives IRG-47 (3) and LRG-47 (4) led to a reexamination of inducibility of TGTP in T cells by IFN-γ. TGTP induction by IFN-γ and anti-TCR Ab 2C11 were compared in thymocytes and lymph node cells as shown in Figure 1. As noted previously, TGTP expression was low/undetectable in cultured thymocytes (1) but was induced in thymocytes exposed to anti-TCR Ab 2C11. TGTP was also efficiently induced in lymph node cells exposed to 2C11. When IFN-γ was substituted for 2C11 in the culture, TGTP was also efficiently induced. When neutralizing anti-IFN-γ Ab XMG-1.2 was included in the thymocyte cultures exposed to either 2C11 or IFN-γ, TGTP induction was blocked. This pattern of inhibition was observed for lymph node cells as well, but the effects of XMG-1.2 were less pronounced. Thus neutralizing Ab to IFN-γ prevented the TGTP induction in thymocytes induced by anti-TCR Ab. These results demonstrated that in thymocytes TGTP protein was indeed inducible with IFN-γ and, moreover, that the induction of TGTP protein by TCR cross-linking was mediated by extracellular IFN-γ produced during the 24-h culture period. Since TGTP RNA was rapidly induced in thymocytes within the 0- to 2.5-h window after TCR cross-linking (1), and since this induction was mediated by IFN-γ (Fig. 1), IFN-γ-induced expression of TGTP must have been rapid indeed. To examine whether TGTP displayed the properties of an IE gene we examined whether the transcriptional induction of TGTP RNA by IFN-γ utilized preformed transcription factors that could function independent of protein synthesis as had been reported for TGTP expression in macrophages by Lafuse et al. (2). As shown in Figure 2, Northern blotting of RNA from thymocytes treated in the presence or absence of the protein synthesis inhibitor cyclohexamide confirmed that the induction of TGTP RNA by anti-TCR Ab required protein synthesis whereas TGTP induction by IFN-γ did not. These results confirmed that TGTP is an IE gene of IFN-γ in T cells. IFN-γ was able to induce TGTP in all cell lines examined to date, including L cells, fibroblasts (primary embryonic, National Institutes of Health 3T3), LPS-stimulated B cell blasts, T cells, macrophages and macrophage cell lines (J774, P388D1, RAW 264.7), and P815 mastocytoma (data not shown).

Despite the presence of the GTP nucleotide-binding motifs in the predicted amino acid sequence of all members of the aforementioned IFN-γ-induced gene family, the GTPase activity of these molecules has been confirmed for only IGTP (5). To evaluate whether TGTP displayed GTPase activity, the full length cDNA was subcloned into the expression vector pGEX-KT so as to generate a TGTP-GST fusion protein. Recombinant TGTP-GST fusion protein purified by glutathione Sepharose chromatography from bacterial lysates was mixed with [α-³²P]GTP for 30 min. The reaction was then analyzed by TLC to resolve GTP and the hydrolysis product, GDP. As shown in Figure 3, GTPase activity was observed with the TGTP-GST fusion protein but not GST alone, confirming that TGTP exhibits GTPase activity. The high purity of fusion protein generated by glutathione Sepharose chromatography suggested that the GTPase activity observed was intrinsic to TGTP and did not require the additional activity of a GTPase-activating protein.

Although IFNs display some overlapping activities, results obtained with mice lacking the capacity to respond to either IFN-αβ or IFN-γ indicate that IFN-γ is primarily involved with defense against bacterial and parasitic infections whereas IFN-α/β is primarily involved in the elimination of viral infections. Thus, it was of interest to clarify whether TGTP was inducible by IFN-γ alone or also by type I IFNs. Initial experiments exemplified in Figure 4 indicated that IFN-α was a weak inducer of TGTP in L cells; 10³ antiviral units of IFN-α were required to induce levels of TGTP comparable with 1 U of IFN-γ. Similar differences in efficacy were observed for thymocytes and spleen cells (data not shown). Given the apparent differences in potency of IFN-γ over IFN-α, and the possibility that high concentrations of IFN-α might engage IFN-γ signaling pathways (18), these results suggested that TGTP was primarily an IFN-γ response gene and thus likely to be associated with bacterial clearance mechanisms. However, after initial failures to isolate stable TGTP transfectants, we began to investigate use of transient transfection by lipofection and observed that lipofection of irrelevant plasmid DNA efficiently induced TGTP in L cells.

As shown in Figure 5,A, the induction of TGTP was DNA-dose dependent and required lipid; in the absence of Lipofectamine, 100-fold higher doses of DNA failed to induce expression of TGTP. Wells receiving the high doses of DNA exhibited reduced induction of TGTP and cell death. As shown in Figure 5 B, we also observed that lipofection of poly(I:C) efficiently induced TGTP and was, on a mass basis, an even more potent inducer (>10×) than plasmid DNA. Lipofection with higher doses of poly(I:C) was toxic. The induction of TGTP was not likely to be mediated by IFN-γ, given the stromal tissue origin of L cells, and the sources of IFN-γ are generally considered to be T cells and NK cells. To analyze the nature of this induction further, lipofected T cells were washed to remove nonadsorbed lipid:DNA complexes and cultured for 36 h. Supernatants from such cultures were able to induce TGTP in previously untreated L cells demonstrating that a soluble factor was produced by lipofected cells that could trigger TGTP synthesis (data not shown). Furthermore, lipofected cultures displayed elevated expression of class I MHC (data not shown). These results suggested that IFN-αβ might mediate TGTP induction after lipofection since type I IFNs are soluble factors and can induce class I major histocompatibility Ags efficiently.

As shown in Figure 5,C, the induction of TGTP after lipofection was indeed inhibited with anti-IFN-αβ Abs, whereas TGTP induction by IFN-γ was not blocked. The relative contributions and efficacy of IFN-α vs IFN-β in stimulating TGTP induction upon lipofection of L cells was not elucidated, but given the stromal origin of L cells, we considered that the soluble factor induced by lipofected DNA was likely to be IFN-β. The data shown in Figure 5 C are consistent with the possibility that IFN-β is a more efficient inducer of TGTP than IFN-α in L cells. The induction of TGTP by poly(I:C) was similarly dependent on the presence of lipid and the induction was similarly blocked by anti-IFN-αβ (data not shown). These results collectively demonstrated that under appropriate conditions TGTP is produced at high levels in cells exposed to both type I and type II IFNs. The capacity of endogenously produced IFN-αβ to induce TGTP reinforced the possibility that TGTP might participate in the response to type I IFNs, specifically in viral defense. We therefore sought to determine whether cell lines transfected with TGTP exhibited altered sensitivity to virus challenge.

The use of transient transfection as a strategy to evaluate effects of TGTP expression on cell phenotype was complicated by the concurrent induction of, and response to, IFN-αβ produced by lipofected cells (described above). Initial efforts to isolate stable TGTP transfectants failed. TGTP cDNA was cloned into the expression vector pEF-BOS (19) and cotransfected with a neo^r-containing vector at a ratio of 3:1 to 10:1 by lipofection into L cells. TGTP expression was assessed by Western blotting cell lysates or by intracellular staining of TGTP and single cell analysis by flow cytometry. Stable transfectants of TGTP were difficult to obtain despite the fact that 1) pEF-BOS vector-directed TGTP expression could be confirmed, and 2) stable transfectants of both neo and βGal vectors were isolated at high frequency. The vast majority of clones that expressed significant levels of TGTP upon primary screening after G418 selection and cloning lost expression of TGTP. Those that did maintain significant expression long enough for further analysis were unstable and lost TGTP expression over time. One clone (clone 2) expressed relatively high levels of TGTP and was a focus of further analysis. Subclones were isolated and were represented by clones that expressed (e.g., 2.2⁺, 2.3⁺), and those that did not express (e.g., 2.18⁻), TGTP. Clone 2.3⁺ was the highest and most stable TGTP expressor isolated to date. As shown in Figure 6, the level of TGTP expressed by 2.3⁺ was substantially lower than that observed in L cells treated with IFN-γ. The expression of TGTP in transfectants was a consequence of vector-driven transcription and not an artifact of endogenous gene transcription stimulated by IFN since coculture of the TGTP expressors with untransfected L cells failed to induce expression of TGTP in the latter (data not shown).

The initial investigation of antiviral activity of TGTP was partly motivated by the previous precedent, Mx. Mx proteins of Mus musculus (Mx1) and Homo sapiens (MxA) have been identified as IFN-αβ-induced GTPases shown to possess intrinsic antiviral activity against certain negative strand RNA viruses such as influenza and VSV (15, 20, 21). To evaluate the potential role of TGTP in viral defense, TGTP transfectants were challenged with either the negative strand RNA virus, VSV, or the DNA virus, HSV. Plaque formation was scored after 1 or 3 days, respectively, according to standard methodology. When TGTP transfectants were challenged with VSV and HSV, as shown in Figure 7, clone 2.2⁺ and 2.3⁺ displayed a VSV^r HSV^s-sensitive phenotype, whereas clone 2.18⁻ and untransfected L cells supported plaque formation by both viruses. Other transfectants of TGTP-pEF-BOS expressing lower levels of TGTP were more susceptible to plaque formation by VSV. TGTP expression levels on 2.2⁺ and 2.3⁺ diminished with time in culture and the reduction in TGTP expression was paralleled by an increase in VSV susceptibility (data not shown). The correlation between relatively high TGTP expression and VSV^r was reproduced in two additional independent clones. VSV^r properties were not observed among TGTP⁻ or TGTP^low clones analyzed. Finally, the expression of relatively high levels of TGTP, although still lower than that observed after IFN-γ treatment, was also associated with a reduced growth rate that was most evident as cells reached confluence. Since a small proportion of L cell clones not expressing TGTP displayed similar low growth rate, the significance of this observation is presently unresolved.

The major findings described above relate to the specific inhibition of VSV plaque formation by TGTP. The induction of TGTP in response to DNA/RNA lipofection, the wide tissue distribution of TGTP in response to IFN, and the antiviral precedent of the GTPase Mx fostered the notion that TGTP might serve an antiviral function. Ascribing antiviral function for TGTP represents an important advance for three reasons. First, defining antiviral activity for TGTP has functional implications for the gene family of IFN-induced GTPases comprised of IRG-47, LRG-47, IGTP, and TGTP/Mg21, whose function has been obscure. Second, defining antiviral activity for TGTP reveals a novel mediator of antiviral action and places TGTP among a small group of molecules with known antiviral activity (see below). The identification of conserved antiviral mechanisms offers access to insight into innate defense against viral infection and may be of importance for development of effective antiviral strategies. Third, the antiviral activity of TGTP we described was observed in mouse-derived L cells challenged with VSV that poses a limited clinical threat. It is possible, however, that TGTP, like MxA, could mediate antiviral action against other viruses of more immediate clinical interest.

The mechanism of TGTP’s activity against VSV has not been evaluated experimentally. As TGTP expression declined with time in culture, plaque formation in L cell transfectants became progressively more pronounced. The apparent instability of TGTP expression together with the reduced growth rate of TGTP high expressors may be relevant to the VSV^r phenotype as most known antiviral mechanisms have the capacity to inhibit protein synthesis. However, it could be argued that the expression of TGTP, or indeed of any transfected gene, might adversely affect cell homeostasis with concomitant deleterious effects that might in turn nonspecifically interfere with VSV replication or cell growth. Several arguments can be made against the possibility that the VSV inhibition observed was a consequence of nonspecific effects of TGTP protein “toxicity.” First, levels of TGTP expressed in “stable” transfectants were in all cases substantially lower than that observed in IFN-treated cells. Thus, the levels of expression directed by pEF-BOS were not likely to be “toxic” for reasons of nonphysiologic excess production. Second, there appears to be a level of specificity in the influence of TGTP insofar as HSV replication was not observably affected in 2.3⁺ transfectants that were resistant to VSV. One might have expected that a “toxic” perturbation in cell integrity engendered by TGTP expression would interfere with HSV and VSV equally. Finally, there is precedent that IFN-αβγ-induced activities can impede both cell growth (22) and viral replication, including the action of PKR (12) and 2-5A/RNase systems (23, 24). Furthermore, these antiviral agents also have a degree of viral specificity as they can display greater inhibitory activity against some viruses more than others (13, 25, 26).

The inhibitory activities of IFN against VSV are numerous; IFNs are known to block VSV at multiple levels including transcription, translation, and late stages of virus assembly/release (13, 25). The molecular basis of IFN-induced antiviral activities in general are thought to include PKR, 2-5A/RNase L, double-stranded RNA-specific adenosine deaminase (dsRAD), and Mx. PKR is a protein kinase activated by dsRNA that phosphorylates eIF2, thereby inhibiting translation and protein synthesis (12). 2-5A activity is similarly activated by dsRNA and generates poly(A) oligonucleotides that stimulate mRNA degradation by activating RNAse L and thus inhibits protein synthesis (27). dsRAD is also type I/II IFN inducible (14) and may interfere with viral replication via conversion of adenosine to inosine in dsRNA leading to untranslatable messages that are targeted by inosine-specific RNAses (28). In summary, although expression of TGTP in transfected L cells might disturb homeostasis, thereby nonspecifically inhibiting cell growth, and, secondarily, VSV replication, there is precedent for antiviral systems whose actions directly interfere with cellular protein synthesis, which can inhibit cellular growth, and whose actions have a degree of viral specificity.

Mx is not known to exhibit antiproliferative activity or to be regulated directly by dsRNA but is expressed in response to IFN-αβ induced by dsRNA. The antiviral activity of the IFN-induced Mx (15) is also restricted to a subset of negative strand RNA viruses and may be relevant to the antiviral action of TGTP. Mx’s precise mechanism of antiviral action is unresolved despite a decade of investigation since its initial cloning/transfection (20), but is thought to block viral RNA synthesis through association with viral RNA polymerase (15). Mx may therefore stand apart from the aforementioned antiviral mechanisms in that its specificity for a viral product may preserve the host cell protein synthetic machinery. Murine Mx, Mx1, is induced only by type I IFNs and interferes with a relatively narrow range of viruses, including influenza (20), Thogoto, and Dhori viruses (29). In contrast to TGTP, which is induced by IFN in cell lines of diverse mouse origin, murine Mx1 is absent in most common laboratory mouse strains (20).

The human version of Mx, MxA, is induced primarily by type I IFN, but also by IFN-γ. MxA interferes with a larger set of negative strand RNA viruses including influenza (21, 30), VSV (21), Hantaan, Lacrosse, Rift Valley fever, Sandfly, human papilloma, Thogoto, and measles viruses (31). MxB is also expressed in humans (32) and is related to MxA although its antiviral action has not yet been established. What makes Mx antiviral function more enigmatic is its genetic relationship with dynamin and Vps1 (33) that appear to serve housekeeping roles in endocytosis (34, 35) and yeast vacuole protein transport (36), respectively. Because of the genetic relationships between Mx, dynamin, and Vps1 we examined the subcellular distribution of TGTP to determine whether it would colocalize to either endosomal or lysosomal compartments. We observed that TGTP is distributed in a punctate cytoplasmic pattern and on perinuclear vesicular structures, but neither of these structures colocalized with either the endocytic pathway (Texas red-conjugated OVA, 4 h) or the lysosomal compartment (lysosomal glycoprotein-specific lectin from Datura stramonium (lgp-FITC conjugate) (37)) (data not shown). Like MxA, TGTP and IGTP are cytoplasmic and exhibit readily detectable GTPase activity in vitro (15). MxA is localized in the cytoplasm whereas murine Mx1 is localized in the nucleus. Thus the Mx gene family contains substantial diversity of function including both antiviral action and housekeeping roles. It will be of interest to see whether other GTPase family members, LRG-47, IRG-47, and IGTP, share common antiviral activity and to identify the relevant TGTP-associated subcellular compartments.

The capacity of other molecules, such as the bacterial product LPS, to induce TGTP expression could provide insight into TGTP function. LPS was shown to induce both LRG-47 (4) and IGTP (5). In contrast, LPS not only failed to induce TGTP/Mg21, but blocked TGTP/Mg21 mRNA induction in peritoneal macrophages assessed 6 h after treatment with IFN-γ (2). The failure of LPS to induce TGTP/Mg21 in macrophages was unexpected insofar as previous work has shown that LPS can induce type I IFN in macrophages (38) and fibroblasts (39), and we have shown here that IFN-αβ can induce TGTP in L cells. The failure of LPS to induce TGTP/Mg21 in macrophages (2) and B cells (our unpublished observations) may suggest that its function is less heavily associated with the response to bacterial pathogens. Alternatively, it is also possible that LPS might up-regulate TGTP/Mg21 at a later time point than that seen for either IGTP (3 h) or LRG-47 (4–8 h).

We have shown that TGTP is induced under conditions in which cells are stimulated to produce either IFN-γ (anti-TCR-stimulated T cells) or IFN-αβ (L cells lipofected with plasmid DNA or the RNA analogue poly(I:C). The induction of TGTP in L cells by plasmid DNA or the RNA analogue poly(I:C) required both lipid and nucleic acid, may be perceived as a viral infection, and was mediated by endogenously produced IFN-αβ. IFN-αβ is induced primarily by viral infection and, to a lesser degree, by products of bacteria and parasites (40). Although able to induce some mediators of parasitic defense such as indoleamine 2,3-dioxygenase (IDO), IFN-αβ is most effective at inhibition of viral infection (40). IFN induction with bacterial DNA or poly(I:C) has been previously documented, but the mechanism by which plasmid DNA or poly(I:C) induces IFN is unresolved. Hypomethylated CpG motifs in bacterial DNA appear to be required for either accessing the cytoplasm and/or direct activation of signaling cascades that ultimately promote IFN transcription (41). Whatever the sensor and signaling process, detection of nucleic acids results in transcriptional activation of IFN-αβ in macrophages and dendritic cells (41, 42) or IFN-γ in NK cells, CD4⁺ T cells, and B cells (41, 43, 44). The capacity of lipofection to deliver plasmid DNA and/or poly(I:C) and induce IFN-αβ in L cells demonstrates that this response pathway is available in nonphagocytic, nonhemopoeitic cell lineages. Thus, of the family of IFN-induced GTPases described above, LRG-47 (4) and TGTP/Mg21 are confirmed IFN-αβγ response genes. Among the ≈30⁺ functional categories of genes regulated by IFN, those that respond to IFN-αβγ include those involved in Ag presentation (class I MHC, β₂-microglobulin, chaperone), cell death (Fas/FasL, TNFRI/II), inhibition of protein synthesis (IDO, tryptophanyl tRNA synthetase), adhesion (integrins, intercellular adhesion molecule, vascular cell adhesion molecule), as well as direct antiviral actions (PKR, 2-5A synthetase, dsRAD, MxA) (10). Our data suggest that TGTP be included in the latter group.

One aspect of TGTP induction that remains unaccounted for relates to its function as an IE gene of IFN-γ. The substantial IFN-γ-induced IE expression of TGTP reported here and by Lafuse et al. (2) suggest a prominent role of TGTP in the response to IFN-γ. The results with type I vs type II IFN knockout mice suggest that there exists a division of labor between these IFN species; IFN-αβ generally governs resistance to viral challenge whereas IFN-γ generally governs resistance to bacterial and parasitic infection (9). How to accommodate the IE response to IFN-γ and the putative antiviral activity of TGTP, given the apparent division of labor among the IFNs, is not readily obvious. In this context, it is quite possible that the inhibitory activity of TGTP on VSV plaque formation reflects more general anticellular TGTP function that HSV has managed to circumvent. That some viruses (e.g., vaccinia, Theiler) display significantly greater replicative success in IFN-γ knockout mice relative to wild-type mice illustrates that IFN-γ can contribute significantly to viral defense. Further enhancement of virus susceptibility (e.g., vaccinia) in mice deficient in response to IFN-αβγ vs IFN-αβ alone demonstrates nonoverlapping effects of these IFNs in viral defense (9) and may provide a rational explanation for efficient induction of TGTP by IFN-γ.

We thank Simon Ip for technical assistance and Chris Ong, Martin Chang, Patricia Orchansky, Gregor Reid, Leslie Esford, Angela Dyer, and members of the Biomedical Research Center (University of British Columbia) for helpful discussions. We also thank the following for gift of reagents: Drs. Frank Tufaro (herpes simplex), S. Nagata (pEF-Bos), and Wilf Jefferies (VSV).