Abstract
IFN-γ orchestrates a potent antimicrobial host response. However, the underlying molecular basis for this immunological defense system is largely unknown. In a systematic approach to identify IFN-γ-regulated host effector molecules, a notable number of transcripts with consensus GTP-binding motives were obtained. Further extensive transcriptome and genome analyses identified five novel family members of murine guanylate-binding proteins (mGBPs) now designated mGBP6, 7, 8, 9, and 10. Moreover, in this study, all 10 mGBP members (mGBP1–10) were extensively characterized. mGBPs are selectively up-regulated in vitro by a set of proinflammatory cytokines and TLR agonists as well as in vivo after Listeria monocytogenes and Toxoplasma gondii infection. After IFN-γ stimulation, mGBP1, 2, 3, 6, 7, and 9 are associated with intracellular Toxoplasma parasites and, interestingly, virulent Toxoplasma interfere with mGBP recruitment. Taken together, mGBPs comprise an important set of host defense molecules.
Interferons and TNF factors are essential for control of intracellular pathogens (1, 2, 3). Moreover IFN-γ has been shown to be important for survival in polymicrobial sepsis (4). IFNs regulate the expression of several hundred genes, including three families of GTPases: Mx proteins, p47 GTPases, and p65 guanylate-binding proteins (GBPs)4 (5, 6). The p65 GBPs and the immunity-related GTPases (IRGs), formerly named as p47 GTPases, are among the most abundantly activated transcripts induced by IFN-γ. Together with type I IFN-induced Mx proteins, IRGs were described as playing important roles during pathogen infections (7). Mx proteins have been shown to possess a potent antiviral activity against a wide range of RNA viruses, e.g., influenza and vesicular stomatitis virus (VSV). Recent studies have attributed this effect to direct interaction of the protein with viral particles, thereby inhibiting viral replication (8). The IFN-γ induced IRGs have been the subject of extensive studies in recent years (6, 9). Currently 23 IRG genes are known in mice (10). Through the investigation of gene-deficient mouse strains, members of the IRGs, such as Irgm1 (LRG-47), Irgd (IRG-47), and Irgm3 (IGTP), have been established as essential proteins during infection with intracellular pathogens, such as Listeria monocytogenes and Toxoplasma gondii in mice (6, 9). Although IRGs exist in several mammalian species, the gene family appears to be degenerated in humans (10).
In contrast, the 65-kDa GBPs are highly conserved throughout the vertebrate lineage. After the first description of human GBP1 (hGBP1) (11, 12), closely related genes have been found in mice, humans, and several other species (13, 14, 15, 16, 17, 18).
GBPs have been assigned to the superfamily of dynamin-related GTP-binding proteins (19). Although diverse in their primary sequences, dynamin-related GTP-binding proteins have common structural features comprising a large GTP-binding domain, a middle domain, and a GTPase effector domain as well as the common biochemical property of oligomerization-dependent GTPase activity (19).
Despite that 65-kDa GBPs have been known for >20 years, most aspects of their functions remain enigmatic. Recent studies have shown that hGBP1 and murine GBP2 (mGBP2) are involved in cell growth regulation. Overexpression of mGBP2 promotes cell proliferation in NIH 3T3 fibroblasts (20), whereas retroviral expression of hGBP1 inhibits proliferation of human endothelial cells (21). Although these findings seem contradictory at first, the effects of mGBP2 and hGBP1 mirror the respective proliferative activities of IFN-γ on these cell lines. hGBP1 also has been shown to negatively regulate expression of matrix metalloproteinase 1 in endothelial cells, leading to reduced invasive capabilities of the cells into collagen I matrices (22). Additionally, hGBP1 expression has been associated with paclitaxel resistance in ovarian cancer cell lines (23).
Furthermore, with respect to immune function, hGBP1 and mGBP2 have been shown to mediate antiviral activities in HeLa cells and NIH 3T3 fibroblasts, respectively, after infection with VSV or encephalomyocarditis virus (24, 25). However, this antiviral activity was significantly lower than that measured for Mx proteins. The viral growth inhibition by mGBP2 was dependent on GTP binding for encephalomyocarditis virus but not for VSV, suggesting different effector mechanisms of the protein depending on the type of virus. However, the biological functions of GBPs in host defense against bacterial and protozoan infections have been elusive up to now.
In an effort to screen for IFN-γ-initiated antimicrobial mechanisms in immune cells, five cDNAs were identified that encode novel members of the 65-kDa mGBP family, extending its number to 10 highly homologous genes. All mGBPs were the subject of extensive mRNA and protein expression analyses demonstrating a strong induction in vitro under proinflammatory conditions and in vivo after infection of mice with L. monocytogenes or T. gondii. Additionally, an enclosure of intracellular T. gondii by distinct mGBP family members could be shown. Interestingly, highly virulent T. gondii were able to interfere with the enclosure by mGBPs. Taken together, this is the first report that indicates a role for mGBPs in host defense in bacterial and protozoan infections.
Materials and Methods
Transcriptome profiling
Synthesis of biotin-labeled cRNA samples was performed using standard protocols supplied by the manufacturer (Affymetrix). Briefly, double-stranded cDNA was synthesized from total RNA using a T7-tagged oligo(dT) primer (Eurogentec). The double-stranded cDNA was directly used in an in vitro transcription reaction with T7 polymerase in the presence of biotin-labeled nucleotides (BioArray High-Yield RNA Transcript Labeling Kit; Enzo Biochem). Biotin-labeled cRNA samples were used for hybridization of Affymetrix MG-U74v2 high-density oligonucleotide arrays. Hybridization, staining, and washing of the arrays were performed according to the manufacturer’s specifications using a Affymetrix GeneChip Fluidics Station 400.
Scanning of the arrays was performed on an Agilent GeneArray Scanner. Scanned raw data images were processed with Affymetrix MAS 5.0 software. Arrays were normalized using invariant set features and the dChip software (www.dchip.org;26). Identification of differentially expressed genes was done with the SAM software (www-stat.stanford.edu/tibs/SAM/;27), using nonparametric, permutation-based methods.
Alignment and phylogenetic analysis
The software ClustalW (28) was used for the alignment and JalView (www.jalview.org) was used for the alignment-layout. The phylogenetic analysis was accomplished by use of the maximum likelihood method and Treepuzzle (www.tree-puzzle.de) for construction of the phylogenetic tree. The Treepuzzle software was run with the option for exact parameter estimates using the neighbor-joining method. Finally, the software Drawtree from the Phylip package (www.phylip.com) was used to plot the tree data. All software was run on a Linux PC workstation.
Cell culture and stimulation
The macrophage cell line ANA-1 (29 L. monocytogenes; provided by T. Hartung, Konstanz, Germany), 100 ng/ml LPS E. coli 055:B5 (Sigma-Aldrich), 1 μM CpG1668 (TIB MOLBIOL), and 1 μg/ml poly(I:C) (Sigma-Aldrich). At the indicated time points, cells were harvested for preparation of RNA. Unstimulated cells served as control.
Real-time RT-PCR analyses
Total RNA from cells or tissues was isolated using TRIzol reagent (Invitrogen Life Technologies) according to the manufacturer’s protocol. First-strand cDNA synthesis was performed using 1 μg of total RNA with Moloney murine leukemia virus reverse transcriptase and oligo(dT) primer (Invitrogen Life Technologies). Real-time RT-PCRs were performed using primers and probes listed in Table I⇓and based on conventional TaqMan Probe finder (TIB MOLBIOL) for mGBP6/10, mGBP8, and mGBP9 or UniversalProbeLibrary (Roche). RT-PCR specificity was verified by using cloned cDNAs as templates.
Primers and probes used to perform real-time RT-PCRs
The PCR primer sets used were designed to be intron overspanning to avoid detection of genomic DNA. All cDNAs were assayed in duplicate and results are expressed relative to expression in unstimulated cells or uninfected mice (2−ΔΔCP), and normalized to β-actin.
In vitro passage of T. gondii
Tachyzoites from T. gondii strain ME49 and BK were maintained by serial passage in confluent monolayers of human foreskin fibroblasts as host cells (HS27, ATCC CRL-1634). After infection of fibroblasts, parasites were harvested from culture supernatant and purified from host cell debris by differential centrifugation (5 min, 50 × g; 15 min, 500 × g) 3 days after infection. Parasites were resuspended in medium, counted, and immediately used for inoculation of the host cells.
Infection of murine embryonic fibroblasts and RAW 264.7 macrophages with T. gondii
T. gondii at a ratio of 50:1. To remove extracellular parasites, cells were washed with PBS.
Generation of mGBP-specific antisera
Rabbits were immunized with specific peptides RQEIEKIKNMPPPRSC (mGBP1), EVNGKPVTSDEYLEHC (mGBP2), CLREEMERTRRKPSLF (mGBP3), CNIKEMKQNGDSLVES (mGBP4), CEQLKANYRQQPGKGTQA (mGBP5), and CGGKSSMNTNSDNSDKVRK (mGBP7; Eurogentec). Polyclonal sera were tested in Western blot analyses with lysates of 293T cells transfected with each mGBP and, additionally, cross-reactivity was excluded by testing each serum on all 10 mGBPs expressed in 293T cells. Subsequently, the specific sera were affinity purified on the corresponding peptide (Eurogentec).
Immunofluorescence analyses
Cells were fixed in 3% paraformaldehyde in PBS for 20 min at room temperature. Cells were permeabilized with 0.02% saponin (Calbiochem-Merck) for 15 min and blocked with 2% normal goat serum (DakoCytomation) for 20 min. For staining of endogenous mGBP1, 2, 3, and 5, anti-mGBP affinity-purified antisera were used at a concentration of 1/200. Anti-SAG1 (Abcam) at a concentration of 1/500 visualized the outer membrane of T. gondii. The cells were incubated with the antisera for 1 h at room temperature. After washing three times with 0.002% saponin in PBS, cells were incubated with secondary antisera for 45 min. As secondary reagents, Cy2-conjugated AffiniPure goat anti-rabbit IgG or Cy2- or Cy3-conjugated AffiniPure goat anti-mouse IgG plus IgM (1/200; Jackson ImmunoResearch Laboratories) were used. After washing two times, nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen Life Technologies). The coverslips were fixed on glass slides with FluoromountG (Southern Biotechnology Associates).
Infection of mice with L. monocytogenes or T. gondii
C57BL/6 mice were purchased from Charles River Breeding Laboratories and maintained in the animal facility of the Heinrich-Heine-University under specific pathogen-free conditions. All procedures performed on animals in this study have been approved by the Animal Care and Use Committee of the local government of Dusseldorf and were in accordance with the German animal laws.
L. monocytogenes infection in C57BL/6 mice was performed by i.p. injection of 0.1 × LD50, and organs were removed at the indicated time points after infection. C57BL/6 mice were injected i.p. with 0.2 ml of PBS containing 20 cysts of the T. gondii strain ME49. The cysts were prepared out of brains of infected CD1 mice as described previously (30). The organs were removed at the indicated time points after infection.
Western blot analyses
Organs of infected and control mice were homogenized in PBS containing protease inhibitor mixture tablets (complete Mini Protease Inhibitor Cocktail; Roche) and lysed with 2% Triton X-100. After incubation for 15 min on ice, lysates were cleared at 4500 × g for 15 min. Supernatants were again centrifuged at 23,000 × g at 4°C for 15 min. Protein concentrations of the supernatants were measured using the BCA protein assay kit (Pierce). Thirty-five micrograms of protein was separated by SDS-PAGE and transferred to a nitrocellulose membrane (Protran; Whatman). Western blot analyses were performed with mGBP-specific antisera and detected with a goat anti-rabbit-POX secondary Ab (BD Biosciences) and ECL reagent (GE Healthcare).
Generation of expression constructs
The coding regions of mGBP1–10 were amplified from full-length cDNAs out of IFN-γ-stimulated ANA-1 cells using the Expand High Fidelity PCR System from Roche. The PCR fragments were engineered into pDsRed-monomer-N1 and -C1 (Clontech-Takara Bio Europe) in frame with the DsRed-coding sequence. All constructs were verified by sequencing (GATC).
Confocal microscopy
Mounted cells were analyzed using a LSM510 Meta confocal microscope (Zeiss). To avoid cross-talk in the detection of the used fluorophores, multitracking scanning mode was used. Image analyses and processing was performed with the LSM software (Zeiss).
Results
Identification of five novel members of the 65-kDa mGBP family
Gene expression profiles of the murine C57BL/6-derived macrophage cell line ANA-1 (29) stimulated with murine IFN-γ compared with unstimulated cells were examined in triplicates using Affymetrix GeneChip arrays (MG-U74v2). Systematic sequence analysis of 102 genes induced by IFN-γ revealed three transcripts (AI595338.1, AI021374.1, and AI006929.1) that showed high similarities to previously published members of the 65-kDa mGBP family (14, 16, 18, 31) (Table II⇓). In particular, the cDNA clone AI595338.1 showed 100% identity to mpa2l and is now referred to mGBP6 (accession no. BK005759). AI021374.1 and AI006929.1 were annotated in the GenBank database and are now named mGBP7 (accession no. BK005760) and mGBP8 (accession no. DQ295175), respectively. Full-length cloning of mGBP6 and mGBP7 out of cDNA from IFN-γ-stimulated ANA-1 cells and sequencing confirmed the deposited expressed sequence tag sequences. Cloning and sequencing of the complete cDNA of mGBP8 revealed that the corresponding exon 6 of the other mGBP family members was absent. Analysis of the mgbp8 genomic locus corroborated this finding (A.K., C.K., J.W., D.D., C.B.-G., S.B., and K.P., manuscript in preparation). Comprehensive homology and motif searches against public databases (Ensembl and National Center for Biotechnology Information) yielded two additional putative members of the 65-kDa mGBP gene family on chromosome 5. Cloning and sequencing of the corresponding full-length cDNAs confirmed them as novel IFN-γ-inducible family members. These two genes were subsequently termed mgbp9 (accession no. DQ985742) and mgbp10 (accession no. DQ985743) (Table III⇓).
Novel GTPases identified in Affymetrix analysis of IFN-γ-stimulated ANA-1 cells
Synopsis of murine 65-kDa GBP members
Based on amino acid sequences derived from all known and newly identified mGBPs, a sequence alignment was performed using the ClustalW algorithm (28) (Fig. 1⇓a). As shown in the alignment, all GTP-binding motifs are highly conserved among the mGBP family members. Strikingly, the N-terminal parts of all mGBPs show higher identities than the C-terminal parts. Making use of the maximum likelihood method, a phylogenetic analysis of the mGBP family was established, revealing two predominant homology clusters (Fig. 1⇓b). Interestingly, these homology clusters correlate with the genomic localization of mGBPs on chromosomes 3 and 5. Within the cluster on chromosome 3, a subcluster consisting of mGBP1, 2, and 5 is characterized by the substitution of V–L in the G3 GTP binding site and by the existence of a C-terminal isoprenylation site (CaaX), which has been described to enhance membrane binding of proteins (32) (Fig. 1⇓a). The C-terminal parts of mGBP1, 2, and 5 differ clearly from mGBP3 and mGBP7, which form the second subcluster on chromosome 3.
a, ClustalW alignment of amino acid sequences for the ten 65-kDa mGBP members. The alignment-layout was edited with the JalView software. Amino acids identical in all 10 GBPs are marked in dark blue, identical in ≥75% in medium blue, and identical in ≥50% in light blue. Red boxes highlight the conserved main GTP-binding motifs: P-loop (G1), switch I (G2), switch II (G3), and G4. Accession no. of the depicted proteins are: ABO88215 (mGBP1), NP_034390 (mGBP2), AAA86645 (mGBP3), ABO88216 (mGBP4), NP705792 (mGBP5), DAA05845 (mGBP6), DAA05846 (mGBP7), ABB90898 (mGBP8), ABJ51941 (mGBP9), and ABJ51942 (mGBP10). b, Phylogenetic tree of 65-kDa mGBPs. Branch lengths are measured relative to the estimated number of substitutions. The gray boxes enclose genes that are clustered on two loci on chromosomes 3 and 5. Punctuated lines mark two subclusters on chromosome 3. c, Phylogenetic tree of 65-kDa mGBPs, Mx proteins, dynamins, and selected IRGs. Branch lengths are measured relative to the estimated number of substitutions.
The homology cluster located on chromosome 5 consists of mGBP4, 6, 8, 9, and 10. These proteins are highly conserved over almost the entire sequence, especially with mGBP6 and mGBP10 differing only in 14 aa. An extended homology analysis of Dynamin-related mGBPs and Mx proteins, dynamins, and IRGs is shown in Fig. 1⇑c and the sequence alignment in supplementary Fig. 1.5
Induction of mGBP family members
To independently confirm the induction of mGBP6, 7, and 8 found in the microarray experiments and to determine the inducibility of mGBP9 and mGBP10 by IFN-γ, mRNA expression levels were measured by real-time RT-PCR using specific probes and primers (see Materials and Methods). Because of the very high homology between mGBP6 and mGBP10, it was not possible to distinguish between the two transcripts in a satisfactory manner by real-time RT-PCR. However, RT-PCR analyses of the full-length cDNA of mGBP6 and mGBP10 identified their inducibility by IFN-γ in ANA-1 cells unequivocally, as confirmed by cloning and sequencing of the resulting amplicons (data not shown). Thus, RT-PCR conditions were established for combined detection of mGBP6 and mGBP10 transcripts with equal efficiencies (data not shown). Therefore, one induction value for mGBP6 and mGBP10 (mGBP6/10) is depicted in the following analyses.
As shown in Fig. 2⇓, an induction of mGBP1–10 transcripts in IFN-γ-stimulated ANA-1 cells as compared with untreated cells was observed in a time-dependent manner. All mGBPs were readily induced after 2 h of IFN-γ stimulation and their expression levels constantly increased up to 16 h of stimulation. Induction ratios of mGBPs by IFN-γ varied over a wide range: highest induction could be demonstrated for mGBP2 and mGBP5, whereas mGBP9 was induced the least 10-fold as compared with untreated cells.
Kinetics of mGBP induction upon IFN-γ stimulation. ANA-1 macrophages were stimulated for 2 (□), 6 (▦), and 16 (▪) h, RNA was prepared, and real-time RT-PCR was performed as described in Materials and Methods. iNOS and GTPBP1 were used as positive and negative controls, respectively. All cDNAs were assayed in duplicate and expression levels were normalized to β-actin. Induction of mRNA is depicted as ratio values comparing stimulated to unstimulated cells. One of three experiments is shown.
Interestingly, in contrast to previous studies, we were able to detect induction of mGBP1 mRNA in ANA-1 cells, which was described to be expressed only in some mouse strains, e.g., BALB/c (16, 31, 33), but not in C57BL/6, which is the source of the ANA-1 cell line (29). The specificity of the PCR was confirmed by using cloned cDNAs of mGBP1 as positive control and its closest homolog mGBP2 as negative control.
Taken together, our data confirm the newly identified mGBP6, 7, 8, 9, and 10 as novel members of the 65-kDa family of IFN-γ-inducible GBPs.
Influence of TLR agonists and proinflammatory cytokines on mGBP expression
Previous studies on mGBP1, 2, and 5 showed that these mGBPs are not induced exclusively after IFN-γ stimulation. Other proinflammatory cytokines, which are important for host resistance to intracellular pathogens, such as IFN-β, IL-1β, TNF-α (1, 3, 14, 34, 35, 36), as well as the TLR 4 agonist LPS have been shown to induce expression of mGBP1, 2, and 5 (14, 16). Based on the discovery of the new mGBP family members, a comprehensive study on the regulation of all mGBP genes was conducted in ANA-1 cells using a selected set of TLR agonists and cytokines in addition to IFN-γ (Fig. 3⇓).
Induction of mGBPs after stimulation with a selected set of cytokines and TLR agonists. ANA-1 macrophages were stimulated for 16 h with the indicated cytokines or TLR agonists and real-time RT-PCR analyses were performed. As controls for the stimulation conditions, IL-12p40 and iNOS mRNA induction was determined.
IFN-β increased the expression of all mGBP members except for mGBP1. The highest IFN-β-mediated transcriptional response could be determined for mGBP2 and mGBP6/10. TNF-α was only able to induce mGBP4. IFN-γ and TNF-α costimulation had a synergistic effect on the induction of mGBP6/10, whereas only some cooperation for mGBP2, 4, 5, and 7 could be detected. IL-1β showed only a marginal effect on mGBP expression in ANA-1 cells, except for mGBP6/10. Furthermore, LPS-induced expression was restricted to mGBP2 and to mGBP6/10. Immunostimulatory CpG oligodeoxynucleotide induced mGBP1 and mGBP6/10, without affecting expression of the other family members. The other tested TLR agonists poly(I:C) and lipoteichoic acid showed no effect on mGBP expression levels, as did the negative control IL-2. GTPBP1 was found to be highly homologous in the N-terminal part to mGBPs (data not shown). GTPBP1 was previously described as a GTPase induced in human THP-1 monocytes by IFN-γ (37). However, no inducibility in murine ANA-1 cells by IFNs or TLR agonists was found. As controls for efficient stimulation of the ANA-1 cells, IL-12p40 and inducible NO synthase (iNOS) induction were quantified.
These results indicate that the regulation of mGBP1–10 by IFN-γ and IFN-β is similar throughout the gene family, yet with pronounced quantitative differences. However, the patterns of response to other proinflammatory cytokines and TLR agonists are heterogeneous. Interestingly, among the IFN-γ response genes, mGBPs appear to be massively transcribed as compared with IL-12p40 and iNOS.
Up-regulation of mGBP members in vivo after L. monocytogenes infection
IFN-γ is essential for survival of L. monocytogenes-infected mice (2). Previously, we could show that mGBP2 and mGBP4 mRNAs were induced in the liver of C57BL/6 mice 24 h after L. monocytogenes infection (14). Now, with more members of the mGBP family identified, the patterns of mRNA and protein expression in spleen and liver of C57BL/6 mice infected i.p. with L. monocytogenes were subjected to a comparative analysis.
mRNA expression of mGBP1–10 as well as the positive controls iNOS and IFN-γ were measured by real-time RT-PCR during the course of infection (Fig. 4⇓). Only slight induction of some mGBPs was observed 8 h postinfection, which was consistent with the IFN-γ levels in liver (Fig. 4⇓a) and spleen (Fig. 4⇓b). All mGBP members were clearly up-regulated after 24 and 48 h of L. monocytogenes infection in these organs. Confirming our results with ANA-1 cells, mGBP1 mRNA induction was measured in C57BL/6 tissues and showed a remarkably different expression kinetic compared with mGBP2 in the spleen, which provides clear evidence for the specificity of the PCR conditions used.
Expression of mGBPs in L. monocytogenes-infected mice. C57BL/6 mice were infected i.p. with L. monocytogenes (0.1 × LD50). After 8, 24, 48, and 72 h of infection, organs were removed and split for measurement of RNA and protein amounts. RNA from liver (a) and spleen (b) was prepared, reverse-transcribed, and real-time RT-PCR was performed as described. All cDNAs were assayed in duplicate and expression levels were normalized to β-actin. Mean values with SD (n = 3 mice) are depicted as fold induction comparing infected to uninfected mice. Protein lysates from liver (c) and spleen (d) were prepared and Western blot analyses with anti-mGBP1, anti-mGBP2, anti-mGBP3, anti-mGBP5, and anti-mGBP7 Ab were performed (molecular mass: 65 kDa). β-actin served as a loading control (molecular mass: 42 kDa). One of three experiments is shown.
In accordance with the in vitro stimulation data, substantial quantitative differences of transcription induction throughout the mGBP family were observed. To address the possibility that these differences are due to diverse constitutive expression levels of mGBP members, the respective absolute protein amounts in liver (Fig. 4⇑c) and spleen (Fig. 4⇑d) of C57BL/6 mice after L. monocytogenes infection were analyzed. For this purpose, peptide-affinity purified polyclonal sera of rabbits immunized with unique peptides from mGBP1, 2, 3, 5, and 7 were generated. In both organs, the quantity of mGBP1, 2, and 7 proteins was dramatically increased after infection. Unexpectedly, mGBP3 and mGBP5 were found to be constitutively expressed in uninfected mice. mGBP3 protein amounts remained steady, which is consistent with the in vivo RT-PCR data where only a minor induction of mGBP3 mRNA after 24 h was detectable (Fig. 4⇑, a and b). Western blot analyses with anti-mGBP5 Ab revealed a specific splice variant of ∼68 kDa in the liver 24 h after infection, whose size corresponds to the published splice variant of mGBP5a (16). Relative amounts of this splice variant mGBP5a were even increased after 48 h. These data were confirmed by real-time RT-PCR using splice variant-specific primers and probes (data not shown). However, this splice variant could not be detected in the spleen of L. monocytogenes-infected mice. Although an Ab specific for the C-terminal part of mGBP4 was generated, no protein could be detected in Western blot analyses of infected mouse tissues (data not shown). Further sequence analyses of cDNA clones of mGBP4 derived from infected mouse tissues and, additionally, stimulated ANA-1 cells revealed two differentially spliced mGBP4 transcripts (57). One of these splice variants resulted in a premature stop codon at 312 bp (accession no. EF494424). The second variant (accession no. EF494423) was not expressed in amounts detectable in Western blot analyses (data not shown). Due to the high homologies of mGBP6, 8, 9, and 10, no specific antisera could be generated, because the selected unique peptide sequences were only weakly immunogenic.
These experiments clearly show that the mGBPs are highly up-regulated on both mRNA and protein levels, indicating an important role during L. monocytogenes infection.
Induction of mGBPs in response to T. gondii infection
This and previous publications (14, 16) suggest an important role of mGBPs during bacterial infections. Analyses of hGBP1 and mGBP2 revealed a modest antiviral activity (24, 25). To our knowledge, no data are available for the in vivo regulation of mGBPs in response to protozoan infections. Thus, mRNA expression patterns of mGBPs upon T. gondii infection in C57BL/6 mice were measured in lung (Fig. 5⇓a) and spleen (Fig. 5⇓b) after 5, 7, and 12 days. The course of infection was monitored by determination of iNOS and IFN-γ mRNA induction. In the lung, expression of all mGBP members was increased at day 5 of infection. The mRNA levels of mGBP1, 2, 4, 5, and 6/10 were further increased at day 7, followed by a decrease of expression observed for mGBP4 and mGBP5 at day 12. Expression of mGBP7, 8, and 9 dropped constantly from day 5 after infection. In the spleen, both mGBP1 and mGBP2 were significantly induced at day 5 with further increase at day 7. Induction of mGBP3 to mGBP8 was less pronounced and no induction was measurable for mGBP9. Interestingly, mGBP1 induction was most prominent throughout all family members upon infection, whereas the effect of T. gondii infection on mGBP9 expression was rather weak.
Expression of mGBPs in T. gondii-infected mice. C57BL/6 mice were infected i.p. with T. gondii (20 cysts of strain ME49) and 5, 7, and 12 days after infection organs were removed. RNA from lung (a) and spleen (b) was prepared and real-time RT-PCR was performed. Protein lysates from lung (c) and spleen (d) were prepared as described in Fig. 4⇑.
Subsequently, expression of mGBP1, 2, 3, 5, and 7 proteins was detected with specific antisera. Expression of all five GTPases was induced during infection with T. gondii in lung (Fig. 5⇑c) compared with uninfected C57BL/6 mice. In the spleen (Fig. 5⇑c), mGBP1, 2, and 7 were readily induced upon infection, whereas mGBP3 and mGBP5 were constitutively expressed and showed only minor up-regulation during infection. The splice variant of mGBP5 that occurred upon infection with L. monocytogenes could not be found after T. gondii challenge in lung, spleen (Fig. 5⇑, c and d), and liver (data not shown). Whether this splice variant is pathogen-specifically induced has to be further addressed.
Taken together, a coordinated mGBP response was induced after T. gondii infection in mice.
Localization of mGBPs in T. gondii-infected cells
Little is known about the molecular functions of mGBPs, particularly with regard to their role in pathogen defense. In consideration of the fact that mGBP proteins were up-regulated after T. gondii infection, we investigated whether mGBPs show a direct interaction with the parasite.
Intracellular immunostainings were performed with the antisera specific for mGBP1, 2, 3, and 5 in murine embryonic fibroblasts (MEFs) stimulated with IFN-γ (Fig. 6⇓, left panel). Each mGBP member showed a discrete vesicle-like structure inside the cytoplasm. This localization was concordant with the subcellular distribution described for mGBP2 previously (38). Staining of RAW264.7 macrophages revealed the same distribution. Without IFN-γ stimulation, no staining in MEFs or RAW264.7 cells was visible (data not shown). Moreover, the subcellular distribution of N-terminal DsRed fusion proteins for mGBP1, 2, 3, and 5 in RAW264.7 macrophages was consistent with the staining of the endogenous proteins (data not shown), which is in contrast to the IRG proteins, where large fluorescent tags impaired multimerization and localization (39). Because antisera for the other mGBP family members could not be successfully established, fusion proteins of mGBP6, 7, and 9 with an N-terminal (Fig. 6⇓, left panel) and C-terminal (data not shown) DsRed-monomer were transfected into RAW264.7 macrophages or fibroblasts, respectively. The N-terminal and C-terminal DsRed fusion proteins of mGBP6, 7, and 9 localized also in discrete vesicle-like structures in IFN-γ-stimulated cells.
Colocalization of 65-kDa mGBP members with T. gondii. Subcellular localization of mGBPs was analyzed by immunostaining of mGBP1, mGBP2, mGBP3, and mGBP5 in MEFs (rows 1–4) and by transfection of mGBP6 and mGBP7-DsRed fusion proteins in RAW264.7 macrophages (rows 5 and 6) and mGBP9-DsRed in MEFs (row 7). Cells were stimulated with IFN-γ for 16 h. Subsequently, cells were infected with T. gondii strain ME49 for 2 h, fixed, and T. gondii were stained with anti-SAG1 and the cell nuclei with DAPI. Glass slides were analyzed by confocal microscopy. Bars, 5 μm.
The effect of parasite infection on the subcellular distribution of mGBPs was determined by IFN-γ stimulation of cells and subsequent infection with the T. gondii strain ME49. Interestingly, already after 30 min of infection, a first accumulation of mGBP1, 2, 3, 6, 7, and 9 was observed around the parasitophorous vacuole (PV). After 2 h. most of the parasites were completely surrounded by the bulk of the mGBPs, whereas only few mGBPs were found in vesicular structures (Fig. 6⇑, right panel). However, mGBP5 showed no colocalization in T. gondii-infected cells neither with endogenous staining nor with a fusion construct. Without IFN-γ prestimulation, no protein colocalization with the PV was detectable for any of the analyzed mGBPs (data not shown).
These data show for the first time that mGBP1, 2, 3, 6, 7, and 9 associate with intracellular T. gondii upon IFN-γ stimulation, suggesting an important function of mGBPs in the defense against T. gondii.
Virulent T. gondii antagonized enclosure by mGBPs
T. gondii type II strain ME49 is known to establish a chronic infection in C57BL/6 mice. In contrast, the type I strain BK leads to rapid lethality in infected mice (40, 41). To elucidate whether the virulence of T. gondii influences the localization of mGBPs, fibroblasts were infected with either the strain of T. gondii ME49 or BK. After 2 h, staining of endogenous mGBPs in T. gondii-infected MEFs revealed a clear difference in localization (Fig. 7⇓a). mGBP1 and mGBP2 readily accumulated at the PV of T. gondii ME49, whereas hardly any accumulation of mGBP1 and mGBP2 could be detected at the PV of T. gondii BK. To quantify this difference, all intracellular T. gondii and the mGBP-associated T. gondii ME49 and BK were counted (Fig. 7⇓b). In ME49-infected cells, ∼50% of parasites were surrounded by mGBP1 and 53% by mGBP2. Strikingly, only 4.5 and 6.7% of virulent T. gondii BK were enclosed by mGBP1 and mGBP2, respectively. This effect was likewise observed for mGBP3 and mGBP6 (data not shown).
Quantification of mGBP1 and mGBP2 colocalization with T. gondii strain ME49 and BK. a, MEFs were stimulated with IFN-γ for 16 h. Then cells were infected for 2 h either with the avirulent T. gondii strain ME49 or the virulent strain BK, fixed, and mGBP1 (upper panel) and mGBP2 (lower panel) were stained with the specific Abs. T. gondii were stained with anti-SAG1 and the cell nuclei with DAPI. Bars, 5 μm. b, T. gondii stained by SAG1 (n = 350) and T. gondii additionally surrounded by mGBP1 or mGBP2 were counted and percentages are depicted. Data of three experiments are included.
These observations clearly indicate that virulent Toxoplasma can interfere with the cellular mGBP host response.
Discussion
In this study, comprehensive analyses of murine 65-kDa GBPs in host defense were performed. In addition to the five known mGBPs, five novel murine 65-kDa GTPases were identified by microarray and in silico analyses. Interestingly, all 10 mGBPs were up-regulated in vitro by IFN-γ, selected proinflammatory stimuli, and TLR agonists as well as in vivo upon infection with L. monocytogenes or T. gondii. Importantly, a set of mGBPs were found to enclose intracellular T. gondii parasites, defining them as key effector molecules at the interface of host pathogen interaction. Strikingly, virulent T. gondii interfered with the accumulation of mGBPs around the PV.
GBPs are characterized by their distinct GTP-binding motives G1-G4. The known 65-kDa mGBPs have canonical G1 and G3 GTP binding sites, but possess an RD in G4 instead of the N/T KXD motif found in other GTPases, such as Rab, Ras, and 47-kDa GTPases (12, 15). Mutational analysis of hGBP1 revealed that in the G4-binding motif of GBPs the amino acids RD are functionally sufficient for GTPase activity (42). The newly identified mGBP6-10 proteins bear the characteristic GTP-binding domains G1-G4 also found in mGBP1–5. All 10 mGBPs are highly homologous in their amino acid sequences, particularly in the N-terminal parts of the proteins. The C-terminal parts of mGBPs are divergent, which suggests nonredundant functions of each mGBP against different pathogens in host defense. Remarkably, mGBPs are encoded by 10 exons, except mGBP8, which lacks a corresponding exon 6, leading to an in frame deletion of 82 aa C-terminally of the GTP-binding motives. Whether this is associated with distinctive features of mGBP8 is not known so far.
After submission of the novel mGBP6–10 to the GenBank database, an in silico study identified a novel mGBP on chromosome 3 (43), termed mGBP6, which corresponds to mGBP7 in this publication. The same study identified a cluster of mGBP pseudogenes on chromosome 5. In contrast, our in silico analyses based on BAC clones (RP23-329M7, RP24-210D14) revealed a mistake in the electronic assembly of the mGBP locus on chromosome 5 within the database. Based on updated genomic information, cDNA cloning, and expression analyses performed in this study, it was now possible to define a new cluster of functional mGBP genes on chromosome 5 comprising mGBP4, 6, 8, 9, and 10. Phylogenetic analysis of the mGBP protein sequences and comparison of genomic data (A. Kresse, C. Konermann, J. Würthner, D. Degrandi, C. Beuter-Gunia, S. Beer, and K. Pfeffer, manuscript in preparation) indicated that this gene cluster on chromosome 5 arose from gene locus duplications. Likely, this situation in mice is a result of recent evolutionary events, since an orthologous GBP cluster has not been found in humans so far (43) and the similarities between mGBPs on chromosome 5 are higher than those on chromosome 3.
The regulation of mGBPs by pathogens and inflammatory conditions was demonstrated by in vitro analyses with ANA-1 cells employing TLR agonists and a selected set of proinflammatory cytokines. As described for the previously published mGBP1–5 transcripts, mGBP6–10 transcripts are highly up-regulated by IFN-γ in comparable time kinetics. For up-regulation of mGBPs, the expression of the transcription factor IFN regulatory factor 1 is required (Refs. 14 , 44 , and 45 ; our observations). Promoter studies of hGBP1 revealed an IFN-stimulated response element and IFN-γ activated site sequence explaining the inducibility by type I and II IFNs (46), which was also shown for mGBP2 (38). Moreover, an NF-κB-binding motif within the promoter region of hGBP1 was shown to be responsible for its inducibility by TNF-α and IL-1β (47). The induction of most of the mGBPs, except mGBP1 and mGBP5, upon IFN-β stimulation indicates IFN-stimulated response element and IFN-γ activation sites, whereas the lack of TNF-α-mediated up-regulation suggests that NF-κB-binding motives are missing. To clarify whether the promoters of the murine GBPs contain these binding motives as well, extensive sequence and EMSA studies are necessary, which is however beyond the scope of this study. A heterogeneous pattern of mGBP expression was observed after IL-1/TLR signaling. mGBP6 and mGBP10 could be induced by IL-1β, whereas only a marginal effect on the other mGBPs was found, which is consistent with previously published data using RAW cells (16). Furthermore, sustained LPS-induced expression was restricted to mGBP2, 6, and 10, suggesting a complex control of mGBP transcription after pathogen- associated molecular pattern recognition.
To elucidate the in vivo role of the differentially induced mGBPs, the regulation of mGBPs during infection was studied in C57BL/6 mice. In the investigated organs of L. monocytogenes- and T. gondii-infected C57BL/6 mice, all 10 mGBP mRNAs were rapidly up-regulated, including mGBP1. In contrast, previous studies showed that after poly(I:C) or LPS injection mGBP1 is only expressed in some mouse strains, e.g., BALB/c, but that a functional allele is absent in C57BL/6 mice (16, 33). Thus, poly(I:C) or LPS injection/stimulation are not sufficient for mGBP1 induction in vitro and in vivo, but full activation during pathogen infection is essential for mGBP1 expression in a C57BL/6 background.
In vivo analysis unambiguously showed induced expression of mGBP1 in all investigated organs of L. monocytogenes- and T. gondii-infected C57BL/6 mice. Specificity of the mGBP1 antiserum was verified by Western blot analysis of mGBP1-transfected cells, and no cross-reactivity was detected with lysates of mGBP2–10- transfected cells. Furthermore, mGBP2, 3, 5, and 7 were strongly up-regulated after L. monocytogenes and T. gondii infection. Surprisingly, the mGBP5-specific Ab detected a second species of mGBP5 in the liver of L. monocytogenes- but not T. gondii-infected mice. This might be due to a pathogen-specific immune response pattern, where Listeria induce additional signals required for differential splicing of mGBP5. Recently, it was shown that differential expression of Iigp2 splice variants correlates with Chlamydia spittaci susceptibility (48). Some mGBPs, such as mGBP3 and mGBP5, were basally expressed in spleen and liver, which suggests differing promoter activities among the mGBP genes. In contrast, in the lung, no basal protein amounts of mGBP3 and mGBP5 could be observed, implying that either mGBP3- and mGBP5-expressing cells migrate to the lung due to the infection or that resident cells up-regulated these mGBPs. Despite the induction of mGBP4 mRNA, protein expression of mGBP4 could not be detected with a specific mGBP4 antiserum. Extensive cDNA cloning from infected mice and genomic analyses revealed one mGBP4 variant with a premature stop codon (57). A second variant was not detectable in quantitative amounts in infected tissues indicating that mgbp4 is not a functional locus at least in C57BL/6 mice.
The subcellular localization of functional mGBPs was analyzed by expression of fluorescent fusion proteins or stainings for endogenous mGBPs in IFN-γ-stimulated cells followed by confocal microscopy. All investigated mGBPs were located in vesicle-like structures distributed throughout the cytoplasm. Coexpression of markers for subcellular structures, such as endoplasmic reticulum (calreticulin), Golgi (human β 1,4-galactosyltransferase), endosomes (RhoB), and (plasma) membranes (neuromodulin), did not colocalize with mGBPs and were therefore unable to identify the character of the mGBP vesicles (data not shown). This is in accordance with extensive studies on mGBP2 (38). In contrast to previous observations (38), mGBP1 was also localized in intracellular vesicles. This can be explained by the presence of a C-terminal isoprenylation sequence CaaX, which is also found in mGBP2 and mGBP5. Interestingly, mutation of the CTIL motif to STIL led to the disintegration of the vesicle-like structure of mGBP2 (38) and mGBP5 (data not shown). Astonishingly, although the other mGBPs lack this putative isoprenylation site, they were distributed in vesicles. Preliminary results indicate that this vesicular distribution is due to heterodimerization (data not shown), which may suggest that mGBP1 or mGBP2 recruit the other mGBPs to these vesicle-like structures.
An infection with L. monocytogenes did not change the subcellular distribution of mGBP2, 6, and 7. In contrast, an infection with T. gondii led to a relocalization of mGBP1, 2, 3, 6, 7, and 9, but not mGBP5. Already 30 min after infection, an accumulation of the mGBPs around the PV was observed. Interestingly, recent reports indicated that members of the IRG family, such as Irgm3 (IGTP), Irgm2 (GTPI), Irgb6 (TGTP), Irgd (IRG-47), and Irga6 (IIGP) accumulate at the avirulent T. gondii ME49 PV in IFN-γ-stimulated astrocytes, suggesting the requirement of a direct GTPase-pathogen interaction for antimicrobial effects to take place (49). In contrast, in bone marrow-derived macrophages, no colocalization of Irgm1 (LRG-47) or Irgm3 with the ME49 PV could be demonstrated (50). However, it has been shown that Irgm1- and Irgm3-deficient mice readily succumb to T. gondii, L. monocytogenes, or M. tuberculosis infections (9, 51, 52). Remarkably, the majority of IRGs have not been conserved in humans (10), whereas seven paralogous genes of mGBPs have been identified in humans (53, 54), which compartmentalize into distinct vesicular structures in human cells (53). However, the molecular role of GBPs in T. gondii infection is not clear. Though the accumulation of the mGBPs seems to be relevant for defense against T. gondii since virulent T. gondii can circumvent the accumulation of mGBPs. Thus, mGBPs are highly induced after microbial infection and associate with the intracellular pathogen T. gondii in a cell autonomous manner. Further studies will be needed to analyze whether mGBPs directly interact with T. gondii or whether unknown molecules act as bridging factors. Currently, gene-targeting studies are performed in our laboratory which will be necessary to clarify the molecular requirements of the individual mGBPs in bacterial, protozoan, and viral infections.
Taken together, mGBPs are highly transcriptionally and translationally induced IFN response genes that appear to be intricately involved in host pathogen interactions, defining them as interesting targets for future immunological and cell biological studies as host effector proteins of the immune response toward intracellular pathogens.
Acknowledgments
We thank Karin Buchholz, Nicole Krafzik, and Stephan Schmidt for experimental assistance; Stefanie Kutsch and Gaby Reichmann for assistance with the infection of mice with L. monocytogenes and T. gondii, respectively; and Heike Weighardt and Stefanie Scheu for critical reading of this manuscript.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by Grants PF259/3-3, FOR729, and Leibniz (to K.P.) and GRK1045/1 (to S.B. and K.P.) from Deutsche Forschungsgemeinschaft and MUGEN (to K.P.).
↵2 D.D., C.K., and C.B.-G. contributed equally to this study.
↵3 Address correspondence and reprint requests to Dr. Sandra Beer or Dr. Klaus Pfeffer, Institute of Medical Microbiology and Hospital Hygiene, Universitaetsstrasse 1, Geb. 22.21, Dusseldorf, Germany. E-mail addresses: sandra.beer{at}uni-duesseldorf.de or klaus.pfeffer{at}uni-duesseldorf.de
↵4 Abbreviations used in this paper: GBP, guanylate-binding protein; IRG, immunity-related GTPase; PV, parasitophorous vacuole; VSV, vesicular stomatitis virus; IRG, immunity-related GTPase; hGBP, human GBP; mGBP, murine GBP; DAPI, 4′,6-diamidino-2-phenylindole; iNOS, inducible NO synthase; MEF, murine embryonic fibroblast.
↵5 The online version of this article contains supplemental material.
- Received June 27, 2007.
- Accepted August 22, 2007.
- Copyright © 2007 by The American Association of Immunologists