|Home | About | Journals | Submit | Contact Us | Français|
Ebola virus initially targets monocytes and macrophages, which can lead to the release of proinflammatory cytokines and chemokines. These inflammatory cytokines are thought to contribute to the development of circulatory shock seen in fatal Ebola virus infections. Here we report that host Toll-like receptor 4 (TLR4) is a sensor for Ebola virus glycoprotein (GP) on virus-like particles (VLPs) and that resultant TLR4 signaling pathways lead to the production of proinflammatory cytokines and suppressor of cytokine signaling 1 (SOCS1) in a human monocytic cell line and in HEK293-TLR4/MD2 cells stably expressing the TLR4/MD2 complex. Ebola virus GP was found to interact with TLR4 by immunoprecipitation/Western blot analyses, and Ebola virus GP on VLPs was able to stimulate expression of NF-κB in a TLR4-dependent manner. Interestingly, we found that budding of Ebola virus VLPs was more pronounced in TLR4-stimulated cells than in unstimulated control cells. In sum, these findings identify the host innate immune protein TLR4 as a sensor for Ebola virus GP which may play an important role in the immunopathogenesis of Ebola virus infection.
Ebola virus and Marburg virus comprise the Filoviridae family and represent important human pathogens and potential agents of bioterrorism. Currently there are no approved vaccines or specific treatments available to prevent or treat filovirus infections. The filoviruses are the cause of severe hemorrhagic disease in humans (7). Ebola virus initially targets monocytes/macrophages and dendritic cells (DCs), which can lead to the release of proinflammatory cytokines and chemokines (3, 7). A better understanding of the physical and functional interactions between Ebola virus proteins and cellular factors regulating the host innate immune response may reveal novel insights into the pathogenesis of Ebola virus and offer new strategies to inhibit Ebola virus replication.
The VP40 matrix protein of Ebola virus is a key structural protein critical for budding virus-like particles (VLPs) and virion egress. Interactions between late budding domains of VP40 and specific host proteins facilitate efficient release of VLPs and infectious virus. Viral proteins other than VP40 also contribute to efficient budding of VLPs. Ebola virus glycoprotein (GP), when coexpressed with VP40, is incorporated into budding VLPs and enhances VLP egress (15), possibly by antagonizing the function of host proteins (12).
Several studies have reported the induction of an innate immune response following infection or stimulation of macrophages/monocytes and DCs with Ebola virus or VLPs, respectively (2, 31). For example, incubation of Ebola virus VP40+GP VLPs with DCs led to the induction of interleukin-6 (IL-6), IL-8, NF-κB and ERK1/2 (18, 31). The triggering mechanism by which Ebola virus VLPs stimulate cytokine production is unknown. Here, we present evidence that Ebola virus VLPs stimulate induction of proinflammatory cytokines as well as SOCS1 (a ubiquitin ligase and negative feedback regulator of cytokine production) by interacting with host Toll-like receptor 4 (TLR4). Importantly, Ebola virus VP40+GP VLPs, but not VP40 VLPs, induced cytokine and SOCS1 expression in a TLR4/MD2 dependent manner both in a human monocytic cell line (THP-1 cells) and in 293T cells expressing a functional TLR4/MD2 receptor. These results indicate that the stimulation of TLR4 by Ebola virus envelope GP results in an innate host response, induction of SOCS1 protein, and potential enhancement of virus egress.
Human 293T cells were cultured in Dulbecco modified Eagle medium with 10% fetal bovine serum. The monocytic leukemia cell line THP-1 was grown in RPMI 1640 with 0.05 mM 2-mercaptoethanol and 10% fetal calf serum. Plasmids expressing wild-type (WT) VP40 and Ebola virus GP have been described previously (16). Plasmids expressing TLR4-N and TLR4-J were kindly provided by Susan Ross (University of Pennsylvania) (22). Plasmids expressing pBIIX-luc(NF-κB) and pRL-TK Renilla were kindly provided by Michael J. May (University of Pennsylvania), and plasmid phINFB-luc was kindly provided by Paula Pitha (Johns Hopkins University). HEK293-TLR4/MD2 cells were kindly provided by Katherine Fitzgerald (University of Massachusetts). Control HEK293 cells are not the parental cells for the HEK293-TLR4/MD2 cell line and do not contain the same antibiotic resistance genes. RP105-His and MD1-HA expression plasmids were kindly provided by Christopher L. Karp (University of Cincinnati College of Medicine). Proteinase K was purchased from Sigma-Aldrich (St. Louis, MO).
Human 293T cells were transfected with the indicated amounts of plasmid DNA by using Lipofectamine in OptiMem (Invitrogen Carlsbad, CA) according to the manufacturer's instructions. VLP purification and budding assays were performed as described previously (15, 21). Protein content was quantitated using the DC protein assay (Bio-Rad, Hercules, CA).
Immunoprecipitation (IP) and Western blot analyses were performed as described previously (21). Anti-Flag and antiactin antisera were purchased from Sigma-Aldrich (St. Louis, MO). Anti-VP40 monoclonal antiserum was kindly provided by Gene Olinger (USAMRIID, Ft. Detrick, MD). Ebola virus anti-GP polyclonal antiserum was kindly provided by Paul Bates (University of Pennsylvania).
HEK293-TLR4/MD2 cells (2 × 105 cells/well on a 24-well plate) were transfected with 0.1 μg pBIIX-luc and 10 ng/well Renilla pTK plasmid. Luciferase activity was measured using the dual luciferase assay system (Promega, Madison, WI) and normalized against Renilla luciferase activity. Data are expressed as the mean relative stimulation ± standard deviation. The data represent an average of a minimum of five independent experiments.
Total RNA was isolated from the indicated samples using the RNeasy miniprep kit (Qiagen Valencia, CA). One microgram of DNase-treated RNA was reverse transcribed to cDNA with oligo(dT) primers and Superscript III (both from Invitrogen). The cDNA was then amplified by PCR using appropriate primers.
First, we sought to confirm that Ebola virus VLPs were capable of inducing production of proinflammatory cytokines as reported previously (32), as well as additional regulators of the host innate immune response, such as SOCS1. Briefly, human THP-1 monocytic cells were treated with VP40 VLPs or VP40+GP VLPs for 6 or 12 h, or with purified lipopolysaccharide (LPS) (as a positive control), and analyzed for expression of RNAs specific for SOCS1 and cytokines tumor necrosis factor alpha (TNF-α), IL-6, and beta interferon (IFN-β) by RT-PCR (Fig. (Fig.1A).1A). There was no expression of SOCS1 or cytokine TNF-α, IL-6, or IFN-β RNA in untreated cells (Fig. (Fig.1A,1A, lanes 1 and 2), whereas strong signals for SOCS1, TNF-α, IL-6, and IFN-β were observed in cells treated with LPS, a known ligand for TLR4 (Fig. (Fig.1A,1A, lanes 3 and 4). Cells treated with VP40 VLPs showed weak IL-6 and SOCS1 signals at 6 h of incubation but virtually no signal after 12 h (Fig. (Fig.1A,1A, lanes 5 and 6). In contrast, cells treated with VP40+GP VLPs showed reproducibly strong SOCS1 signals at both 6 and 12 h of incubation (Fig. (Fig.1A,1A, lanes 7 and 8). In addition, VP40+GP VLPs also induced IL-6, TNF-α, and IFN-β RNA expression to levels similar to those induced by LPS (Fig. (Fig.1A,1A, compare lanes 3 and 4 to lanes 7 and 8). Interestingly, VLPs containing VP40 and a mucin domain deletion mutant of GP (ΔmucGP) were unable to induce expression of SOCS1, IL-6, TNF-α, and IFN-β RNAs to levels similar to those induced by VP40+GP VLPs (Fig. (Fig.1A,1A, compare lanes 7 and 8 with lanes 9 and 10). In addition to activation of SOCS1, IL-6, TNF-α, and IFN-β, treatment of human THP1 cells with Ebola virus VP40+GP VLPs also led to activation of NF-κB and IRF3 (see Fig. S1 in the supplemental material).
Next, we sought to demonstrate more specifically that host TLR4 was the cell surface molecule that could sense and respond to Ebola virus VP40+GP VLPs. To this effect, we utilized HEK293-TLR4/MD2 cells stably expressing a functional TLR4/MD2 complex on their surface (Fig. (Fig.1B).1B). Normal HEK-293 cells (TLR4/MD2 negative) served as a negative control. These cells were incubated with VLPs or LPS as described above. We also confirmed that TLR4 RNA was expressed only in HEK293-TLR4/MD2 cells, and not in control HEK293 cells (Fig. (Fig.1B).1B). While low levels of basal SOCS1 RNA expression were detected in all samples from both cell lines, a strong induction of SOCS1 RNA was clearly evident in HEK293-TLR4/MD2 cells treated with either LPS (Fig. (Fig.1B,1B, lanes 3 and 4), or VP40+GP VLPs (lanes 7 and 8). In sum, these results correlated well with those described above and suggested that Ebola virus WT GP on VLPs was capable of triggering TLR4 and subsequent downstream signaling events.
We next sought to determine whether Ebola virus GP interacted with TLR4 in transfected cells by using an IP/Western blotting approach. Briefly, human 293T cells were transfected with the indicated combinations of plasmids expressing VP40, WT GP, ΔmucGP, TLR4J-Flag, or TLR4N-Flag (Fig. (Fig.2).2). Both TLR4N and TLR4J plasmids encode full-length TLR4 proteins that have identical extracellular domains; however, TLR4J-Flag contains a point mutation in its cytoplasmic domain that attenuates downstream signaling (22). Cell lysates from transfected cells were first immunoprecipitated with anti-Flag antiserum to pull down TLR4, and then the immunoprecipitates were immunoblotted with either anti-GP or anti-VP40 antiserum (Fig. (Fig.2).2). Interestingly, both Ebola virus WT GP and ΔmucGP were coprecipitated with TLR4J and TLR4N (Fig. (Fig.2,2, top panel, lanes 8 to 11). Specificity controls indicated that neither WT GP nor ΔmucGP was precipitated from 293T cells with TLR4 antibodies, nor was Ebola virus VP40 coprecipitated with TLR4J or TLR4N (Fig. (Fig.2,2, third panel, lanes 6 to 11). Lastly, expression of GP (Fig. (Fig.2,2, second panel) and TLR4 (Fig. (Fig.2,2, bottom panel) in appropriate samples was demonstrated by Western blotting. It should be noted that the MD2 component of the TLR4 receptor was not present in these TLR4N- or TLR4J-transfected cells, and thus the ability of Ebola virus WT GP, ΔmucGP, or VP40 to interact specifically with MD2 remains to be determined. These findings suggested that both WT GP and ΔmucGP can interact with TLR4 as determined by IP/Western blot analysis of transfected 293T cells.
We next sought to determine whether a GP/TLR4 interaction stimulated the TLR4 signaling pathway, leading ultimately to NF-κB and IRF-3 activation. To test this, we first used transient-transfection assays in which HEK293-TLR4/MD2 cells or control HEK293 cells were transfected with pBIIX-luc(NF-κB), a reporter plasmid expressing luciferase under the control of tandem NF-κB-responsive promoters. The transfected cells were then treated with LPS (positive control) or VLPs containing VP40 alone, VP40+GP, or VP40+ΔmucGP, and luciferase activity was measured 12 h later (Fig. (Fig.3A).3A). All the VLP samples used for stimulation of transfected cells contained the appropriate Ebola virus proteins (Fig. (Fig.3C).3C). In addition, experiments were performed to ensure that VLP preparations were not contaminated with bacterial LPS (see Fig. S2 in the supplemental material). The levels of luciferase activity in the transfected HEK293 control cells were low under all stimulation conditions (Fig. (Fig.3A).3A). In contrast, stimulation of HEK293-TLR4/MD2 cells (Fig. (Fig.3A)3A) with either LPS or VP40+GP VLPs resulted in strong, reproducible enhancement of luciferase activity. In contrast, VLPs containing VP40 alone or VP40+ΔmucGP were not effective inducers of luciferase activity in HEK293-TLR4/MD2 cells (Fig. (Fig.3A3A).
Since TLR4 signaling is also known to activate IRF3, resulting in the induction IFN-β (13), we performed an identical experiment to determine whether VLPs containing Ebola virus GP could stimulate transcriptional activity from an IFN-β promoter (Fig. (Fig.3B).3B). HEK293-TLR4/MD2 cells were transfected with a reporter plasmid expressing luciferase under the control of an IFN-β promoter and stimulated as described above. Indeed, stimulation with LPS or VP40+GP VLPs, but not with VLPs containing VP40 alone or VP40+ΔmucGP, induced substantially higher luciferase activity above that detected in unstimulated controls (Fig. (Fig.3B).3B). Overall, luciferase expression driven by the NF-κB promoter was greater than that driven by the IFN-β promoter (Fig. 3A and B).
To further prove that TLR4 is engaged by Ebola virus GP, we made use of the TLR4-specific inhibitor RP105. The RP105/MD1 complex is a homolog of the TLR4/MD2 complex that specifically inhibits TLR4 signaling (6). We sought to determine whether expression of RP105/MD1 would block the TLR4 signaling observed in Fig. Fig.3A.3A. To test this, HEK293-TLR4/MD2 cells were transfected with pBIIX-luc(NF-κB) alone or with pBIIX-luc(NF-κB) plus RP105/MD1 (Fig. (Fig.4).4). At 24 h posttransfection, cells were treated with phosphate-buffered saline (PBS), LPS (100 ng/ml), or Ebola virus VP40+GP VLPs (Fig. (Fig.4).4). We detected significantly higher levels of luciferase in HEK293-TLR4/MD2 cells treated with LPS in the absence of RP105/MD1 than those treated in the presence of RP105/MD1 (Fig. (Fig.4).4). Virtually identical results were obtained with HEK293-TLR4/MD2 cells treated with VP40+GP VLPs (Fig. (Fig.4).4). These findings suggest that expression of the TLR4-specific inhibitor RP105/MD1 reduced the strength of TLR4/MD2 signaling initiated by both LPS and VP40+GP VLPs.
We also used a second approach to further prove that binding of Ebola virus GP to TLR4 results in downstream gene induction. HEK293 cells were transiently transfected with pcDNA-based plasmids encoding human TLR4 and MD2 (kindly provided by Stephanie Vogel, University of Maryland) along with pNF-κB-Luc. Control cells received empty pcDNA vector plus pNF-κB-Luc. Our results showed that the HEK293 cells transiently transfected with human TLR4 and MD2 yielded high levels of NF-κB-driven luciferase activity when treated with either LPS or VP40+GP VLPs (see Fig. S3 in the supplemental material). These findings provided further evidence that Ebola virus VP40+GP VLPs can functionally signal through human TLR4 to activate downstream gene products.
Since we know that TLR4 stimulation leads to enhanced expression and activity of host ubiquitin E3 ligases such as SOCS1 and TRAF6 and since host Nedd4 ubiquitin E3 ligase interacts with VP40 to enhance VLP budding, we sought to determine whether TLR4 stimulation might enhance egress of Ebola virus VP40 VLPs. To test this, control HEK293 or HEK293-TLR4/MD2 cells were transfected with WT VP40, and at 24 h posttransfection, cells were treated with either PBS as a negative control or LPS (1 μg/ml) to stimulate TLR4 signaling. VP40 in cell lysates and VLPs was detected by Western blotting (Fig. (Fig.5).5). Stimulation with LPS was confirmed by using RT-PCR to detect induction of SOCS1 RNA (data not shown). In control HEK293 cells, the levels of VP40 in cells and VLPs remained unchanged in both PBS- and LPS-treated samples (Fig. (Fig.5,5, lanes 1 and 2). In contrast, the levels of VP40 in VLPs from HEK293-TLR4/MD2 cells treated with LPS were modestly enhanced compared to those observed in the PBS-treated control (Fig. (Fig.5,5, VLPs, compare lanes 3 and 4). The levels of VP40 in the corresponding cell extracts were identical (Fig. (Fig.5,5, cells, compare lanes 3 and 4). The fact that overall VLP egress appears to be reduced in the HEK293-TLR4/MD2 cells compared to that observed in the HEK293 control cells could be due to the altered genetic composition of the HEK293-TLR4/MD2 cells, plasma membrane alterations due to expression of TLR4/MD2, or perhaps a lower efficiency of transfection in the HEK293-TLR4/MD2 cells. These data suggest that stimulation of TLR4 results in a cellular milieu that modestly enhances egress of Ebola virus VP40 VLPs compared to that of unstimulated controls.
Innate immune responses to virus infection provide a critical first line of defense for the host. The interplay between Ebola virus proteins and the host innate immune system has been an area of great interest, since the outcomes of these virus-host interactions can directly influence virus replication, pathogenesis, and immune evasion (1, 5, 23). Here we present evidence of a previously undescribed interaction between host TLR4 and Ebola virus GP. Our results suggest that the triggering mechanism by which Ebola virus VLPs stimulated expression or proinflammatory cytokines involves TLR4, which is abundantly expressed on resting monocytes/macrophages, polymorphonuclear leukocytes, and DCs (19). Indeed, we found that Ebola virus VLPs containing VP40+GP, but not those containing VP40 alone or VP40+ΔmucGP, were able to induce expression of proinflammatory cytokines and SOCS1 from human THP-1 cells. Surprisingly, ΔmucGP was able to bind TLR4 in cotransfection experiments but was unable to functionally trigger TLR4 when expressed on VLPs. Although the reason for this remains to be determined, one possibility is that the functional oligomerization of TLR4/MD2 required for downstream signaling is not initiated as efficiently by VLPs containing VP40+ΔmucGP as it is with VLPs containing VP40+WT GP. We also demonstrated that RP105/MD1, a TLR4/MD2 homolog and specific inhibitor of TLR4 signaling (6), disrupted NF-κB production as measured by luciferase activity (Fig. (Fig.4).4). It should be noted that VLP preparations were boiled and/or treated with proteinase K before their addition to the cells as a control to confirm that stimulation was not due to contaminating bacterial LPS in the VLP preparations but rather was due to an Ebola virus GP/TLR4 interaction (A. Okumura and R. N. Harty, data not shown). As further proof of this novel interaction between GP and TLR4, we utilized an IP/Western blotting approach, which revealed a physical interaction between these two proteins.
A growing number of reports have implicated TLR4 as being biologically relevant and responsive to viral proteins, including those of vesicular stomatitis virus (VSV) (8), mouse mammary tumor virus (4, 22), and respiratory syncytial virus (10, 14, 17). Interestingly, VSV G has been shown recently to induce TLR4-specific signaling pathways; however, stimulation of TLR4 by VSV G protein did not result in activation of NF-κB in myeloid DCs but rather resulted in induction of IFN-β (8). Since Ebola virus VLPs closely mimic the overall morphology of infectious virions, it is possible that Ebola virus virions will be able to trigger TLR4 as well. On the other hand, infectious virions with their full complement of viral proteins may lead to altered outcomes following infection of TLR4-expressing cells, such as DCs. Indeed, Ebola virus has been reported to elicit an impaired immune response following infection of immune cells (3). Thus, the relevance of these data obtained using VLPs to those obtained with live Ebola virus remains to be determined.
Ebola virus GP has been reported to interact with a variety of cell surface molecules to gain entry cells (29, 33). For example, calcium-dependent lectins (27) and β1 integrin (30) have been implicated in mediating Ebola virus entry. More recently, the TAM family of proteins, Tyro3, Axl, and Mer, were reported to be involved in mediating entry of Ebola virus (26). Interestingly, TAM receptor signaling was shown in a separate report to induce expression of SOCS1, leading to subsequent negative regulation of surface-expressed TLR receptors and TLR-induced cytokine-receptor cascades (24). Since we observed an increase in SOCS1 expression induced by Ebola virus GP on VLPs, it would be of interest to determine whether Ebola virus GP-mediated binding to TAM receptors on macrophages/monocytes could stimulate expression of SOCS1.
SOCS1 regulates the innate immune response by controlling and limiting the proinflammatory response through negative feedback inhibition of cytokine receptors (34). Indeed, the CIS-SOCS family of proteins, as well as other SOCS box-containing molecules, can function as E3 ubiquitin ligases (34). Expression of SOCS1 can alter the activity of macrophages and DCs, affecting their ability to differentiate and defend against invading pathogens. Interestingly, we observed that LPS stimulation of TLR4-expressing cells resulted in enhanced budding of VP40 VLPs over that observed in control cells (Fig. (Fig.5).5). Strong et al. recently reported that LPS stimulation of cells persistently infected with Ebola virus led to enhanced production of virus (28). It is tempting to speculate that TLR4 triggering by LPS leads to the induction of ubiquitin ligases such as SOCS1 and TRAF6, which may then promote VLP budding due in part to their ubiquitination activity. Indeed, ubiquitination and host ubiquitin ligases such as Nedd4 are known to play important roles in promoting Ebola virus VP40-mediated budding (11, 20, 34).
Recently, SOCS1 has been shown to positively regulate the late stages of human immunodeficiency virus replication by enhancing Gag stability, trafficking, and egress (20, 25). Moreover, Nishi et al. found that SOCS1 facilitated human immunodeficiency virus type 1 Gag ubiquitination and that a dominant-negative ubiquitin significantly inhibited Gag binding to microtubules (20). Interestingly, our preliminary findings suggest that SOCS1 also facilitates ubiquitination of Ebola virus VP40 (9) resulting in a modest (three- to fivefold) increase in VLP egress (Okumura and Harty, unpublished data). Ubiquitination and budding of VP40 are further enhanced in the presence of both SOCS1 and Nedd4 (Okumura and Harty, unpublished data). Additional studies will be needed to determine whether SOCS1 and/or TRAF6 might enhance VP40 VLP budding by coordinating ubiquitination complexes and/or by regulating trafficking of VP40 along microtubules to the site of budding. Further investigations into host innate immune interactions, ubiquitination, and virus budding should provide useful and fundamental knowledge necessary for the future development of novel antiviral therapies that target this late stage of virus replication.
This work was supported in part by grant AI-077014-02 and a grant from the University of Pennsylvania Research Foundation to R.N.H., by grant AI-19737-23 to P.M.P., and by the Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation (NIBIO) to A.Y.
We thank Yuliang Liu and Luis Cocka for helpful discussions.
Published ahead of print on 21 October 2009.
†Supplemental material for this article may be found at http://jvi.asm.org/.