|Home | About | Journals | Submit | Contact Us | Français|
Detailed understanding of the signaling intermediates that confer the sensing of intracellular viral nucleic acids for induction of type I interferons is critical for strategies to curtail viral mechanisms that impede innate immune defenses. Here we show that the activation of the microtubule-associated guanine nucleotide exchange factor GEF-H1, encoded by Arhgef2, is essential for sensing of foreign RNA by RIG-I-like receptors. Activation of GEF-H1 controls RIG-I and Mda5-dependent phosphorylation of IRF3 and induction of interferon-β expression in macrophages. Generation of Arhgef2−/− mice revealed a pronounced signaling defect that prevented antiviral host responses to encephalomyocarditis virus and influenza A virus. Microtubule networks sequester GEF-H1 that upon activation is released to enable antiviral signaling by intracellular nucleic acid detection pathways.
The induction of type I interferon (IFN) and activation of IFN-inducible genes are central to innate immune defenses against viral infection1. Intracellular sensors for microbial nucleic acids initiate complex signaling cascades that lead to the induction of proinflammatory cytokines and type I IFN for antiviral innate immune responses and the development of adaptive immunity2.
Viral targeting of dynein-based transport mechanisms play an important role for intracellular movements and replication of viral pathogens3, although it is unresolved how microtubule-based trafficking of signaling components contributes to the induction of antiviral defenses. GEF-H1, also called lfc in mice, was originally identified as a member of the Dbl family that is sequestered on microtubules and directs spatio-temporal activation of Rho GTPases4. Inactive GEF-H1 binds to the dynein motor complex on microtubules5. GEF-H1 can be activated and released from microtubules upon cellular interactions with bacterial effectors6,7 and subsequently contributes to intracellular pathogen recognition7,8.
The innate immune system senses viral infection through cytosolic and transmembrane receptors leading to activation of cell type-specific regulatory networks that activate IFN regulatory factors (IRFs) and NF-κB for the induction of type I interferons and proinflammatory cytokines. IFN-β expression is initially induced after viral RNA binding to Toll-like receptors (TLRs) that signal through MyD88 and TRIF9. Single stranded viral RNA (ssRNA) can be detected by TLR7 in endosomes10. Double-stranded viral RNA (dsRNA) can be recognized by endosomal TLR311. In addition cell surface TLRs such as TLR4 and TLR2 are activated by viral glycoproteins12.
During viral replication several members of the DExD/H-box helicases (DDX) protein family comprised of RNA and DNA helicases function as viral RNA and DNA sensors13,14. The CARD domain-containing DDX proteins, retinoic acid-inducible gene I (RIG-I) and melanoma differentiation-associated gene 5 (Mda5), also recognized as RIG-I-like receptors (RLR), are important inducers of innate immunity and recognize complementary variants of viral RNA to distinguish between virus families15,16. Mda5 is required for type I IFN responses to long cytoplasmic viral and synthetic dsRNA and is activated by picornavirus15,17, whereas RIG-I recognizes short blunt or 5′-triphosphorylated ends of viral genomic RNA segments and is required for defense activation against influenza, paramyxovirus and rhabdovirus families16,17,18,19. RIG-I and Mda5 signal transduction requires mitochondrial antiviral signaling protein (MAVS), also called IFN-β promoter stimulator 1 (IPS-1), for the activation of TBK1 (TANK [TRAF (tumour necrosis factor receptor-associated factor) family member-associated nuclear factor κB activator]-binding kinase 1) and IKKε that mediate the phosphorylation of IRF3 for the induction of type I IFNs20. TBK1 and IRF3 activation also occurs downstream of stimulator of interferon genes (STING; also known as TMEM173, MPYS, ERIS or MITA), in the detection of viral nucleic acids and B-form DNA (B-DNA or poly(dA:dT)) by DDX proteins)21.
Here we demonstrate that GEF-H1 mediates the induction of antiviral host defenses by cytosolic receptors RIG-I and Mda5 in macrophages. The recognition of viral RNA and synthetic dsRNA in the MAVS pathway was dependent on microtubule networks that were required for the activation and interaction of GEF-H1 with TBK1-IKKε for the induction of IRF3 phosphorylation and subsequent induction of Ifnb1 gene expression. In contrast, deletion of GEF-H1 or disruption of microtubule function in macrophages still allowed NF-κB activation by cell surface and endosomal TLR activation. Consequently, GEF-H1 was required for the restriction of ssRNA virus replication and the induction of antiviral host defense against EMCV and influenza A.
To define the role of GEF-H1 in innate immune activation by foreign nucleic acids, we generated GEF-H1-deficient mice using C57BL/6 embryonic stem cells with a gene-trap insertion between exons 4 and 5 of Arhgef2 on mouse chromosome 3 that prevents GFH-H1 mRNA (Fig. 1a) and protein expression (Fig. 1b). These mice had normal T cell, B cell and mononuclear phagocyte numbers in spleen and lymph nodes (Supplementary Fig. 1).
IFN-β protein secretion and mRNA expression were determined in response to 1-8 kb (high molecular weight, HMW) polyriboinosinic:polyribocytidylic acid (poly(I:C)) as a ligand for Mda5 or 0.3-1.2 kb (low molecular weight, LMW) poly(I:C) and 5′-triphosphate (5′ppp)-double-stranded (ds)RNA were used as synthetic ligands for RIG-I18. In addition, cyclic diguanosine monophosphate (c-di-GMP) was used as a DDX41 ligand that induces STING-dependent IFN-β expression22. Expression of Ifnb1 mRNA was significantly reduced in bone marrow-derived macrophages derived from GEF-H1-deficient mice in response to MAVS and STING-mediated recognition of nucleic acids (Fig. 1c). In contrast, GEF-H1-deficient macrophages upregulated Ifnb1 mRNA expression in response to TLR1/2, TLR2, TLR4, TLR5, TLR2/6, TLR7 and TLR9 activation by specific ligands comparable to wild-type macrophages (Fig. 1d).
The lack of transcriptional activation of Ifnb1 upon RIG-I activation by 5′ppp-dsRNA resulted in significantly less IFN-β secretion in GEF-H1-deficient macrophages compared to wild-type macrophages (Fig. 1e). GEF-H1-deficient macrophages also secreted significantly less IFN-β after transfection of HMW and LMW poly(I:C) (Fig. 1f), and even demonstrated significantly attenuated IFN-β secretion when HMW poly(I:C) was directly added to the culture medium (Fig. 1g). Furthermore, GEF-H1 expression itself was upregulated by RIG-I signaling initiated by 5′ppp-dsRNA transfection into macrophages (Fig. 1h). Remarkably, two intact alleles of Arhgef2 were required to induce a full response to poly(I:C), since macrophages heterozygous for gene-trap insertion also demonstrated impaired Ifnb1 mRNA expression (Fig. 1i). GEF-H1-deficient macrophages also demonstrated reduced Il6 and Tnf mRNA expression in response to 5′ppp-dsRNA, indicating a profound innate signaling defect in the activation of MAVS-dependent RLR signaling (Fig. 1j). In contrast, TRIF- and MyD88-mediated induction of IFN-β secretion and Il6 and Tnf mRNA expression were not reduced in GEF-H1-deficient macrophages in response to the TLR4 ligand lipopolysaccharide (LPS) (Supplementary Fig. 2).
The RLR signaling deficiency in GEF-H1-deficient macrophages was not due to impaired poly(I:C) uptake. HMW rhodamine-labeled poly(I:C) was similarly absorbed from the medium in GEF-H1-deficient and wild-type macrophages and found in association with vesicular and tubular compartments in wild-type and GEF-H1-deficient macrophages (Supplementary Fig. 3a,b). Together these data indicated that GEF-H1 expression is induced by foreign intracellular dsRNA and required for the signaling of intracellular nucleotide sensors leading to IFN-β secretion and proinflammatory cytokine expression in macrophages.
RLRs-induced type I IFN gene transcription requires MAVS and TBK1-IKKε and is mediated primarily through IRF323. IRF3 is localized in the cytoplasm and, upon stimulation, becomes activated by serine/threonine phosphorylation leading to nuclear translocation and binding to recognition sequences in the promoters and enhancers of type I IFNs20. To determine whether GEF-H1-dependent type I interferon induction was mediated by IRF3 phosphorylation and nuclear translocation in response to RLR activation, we stimulated GEF-H1-deficient and wild-type macrophages with the RIG-I ligand 5′ppp-dsRNA and analyzed the resulting phosphorylation of IRF3 in cell lysates as well as nuclear translocation of IRF3. Phosphorylation of IRF3 in response to RIG-I activation was significantly reduced in GEF-H1-deficient macrophages when compared to wild-type macrophages (Fig. 2a). IRF3 remained undetectable 4 h after 5′ppp-dsRNA stimulation in the nuclei of GEF-H1-deficient macrophages, demonstrating a profound deficiency in IRF3 activation (Fig. 2b). In contrast, IRF3 phosphorylation in response to LPS occurred at much lower amounts in bone marrow-derived macrophages under same conditions but was detectable similarly in wild-type and GEF-H1-deficient macrophages (Fig. 2c).
We further found that the RIG-I ligand 5′ppp-dsRNA and Mda5 ligand HMW poly(I:C) induced p65 phosphorylation and IκBα degradation in GEF-H1-deficient and wild-type macrophages over a period of 4 h (Fig. 2d). Both, GEF-H1-deficient and wild-type macrophages also responded within 15 minutes with comparable NF-κB activation to TLR4 activation (Fig. 2d). This data indicated that GEF-H1 function was required for IRF3 activation in the RLR pathway but dispensable for p65 activation in the TLR4 and RLR pathways.
Furthermore, GEF-H1-deficient macrophages showed significantly less Ifnb1 promoter activation in response to MAVS expression compared to wild-type macrophages while GEF-H1 expression enhanced Ifnb1 promoter activation in wild-type macrophages and complemented MAVS-induced IFN-β responses in GEF-H1-deficient macrophages (Fig. 2e).
We further found that GEF-H1 enhanced MAVS signaling leading to the activation of a p561 interferon-stimulated response element (ISRE)-containing promoter that is activated by IRF3, but not NF-κB24 (Fig. 2f). GEF-H1 increased ISRE-induced transcriptional activity 10-fold compared to MAVS expression alone in HEK293T cells (Fig. 2f). Indeed, GEF-H1 augmented MAVS-dependent phosphorylation of endogenous IRF3 in HEK293T cells without significantly altering the baseline expression of IRF3 or TBK1 (Fig. 2g).
To further characterize the role of GEF-H1 in signaling of intracellular nucleotide receptors and induction of Ifnb1 promoter activation, we utilized HEK293T cells that lack TLR3, TLR4, TLR7/8 and TLR9. We assembled Mda5 or RIG-I signaling pathways in HEK293T cells in the absence or presence of GEF-H1 to assess the activation of a luciferase reporter containing the Ifnb1 promoter in response to RLR ligands. GEF-H1 expression significantly enhanced Ifnb1 promoter activation induced by the expression of Mda5 and RIG-I alone and further enhanced Mda5-mediated detection of HMW poly(I:C) and RIG-I-dependent responses to LMW poly(I:C) and 5′ppp-dsRNA (Fig. 2h). In contrast to RIG-I and Mda5, GEF-H1 expression in HEK293T cells did not induce Ifnb1 promoter activation by itself or rendered HEK293T cells responsive to RLR stimulation when expressed alone (Fig. 2h). Together these data demonstrated that GEF-H1 functions in conjunction with RLRs, enhancing the detection of intracellular poly(I:C) and 5′ppp-dsRNA and leading to the activation of the Ifnb1 promoter.
The phosphorylation, dimerization and nuclear translocation of IRF3 for the activation of IFN-β transcription require IKKε and TBK123. Indeed, expression of either TBK1 or IKKε enhanced Ifnb1 promoter activation in HEK293T cells (Fig. 2i). GEF-H1 expression significantly enhanced Ifnb1 promoter activation by TBK1 alone or in conjunction with IKKε (Fig. 2i). In contrast, GEF-H1 did not augment TBK1 or IKKε mediated Nfkb1 promoter activation (Fig. 2i). Furthermore, the promotion of IRF3 and Ifnb1 gene promoter activation by GEF-H1 was dependent on functional TBK1 and therefore absent in the presences of a TBK1 kinase-inactive (K38A) mutant (Fig. 2j). Together these data demonstrate that GEF-H1 can function in the RLR pathway in conjunction with MAVS and TBK1/IKKε to enhance the phosphorylation of IRF3 and activation of the Ifnb1 promoter.
To better define the functional domains of GEF-H1, we created GEF-H1 variants by exchanging the tyrosine at position 393 with alanine to disable the GTP loading capacity of the Dbl homology domain (ΔDH)25, by replacing serine 885 with alanine (S885A) to prevent phosphorylation that inhibits GEF activity of GEF-H1 or by substituting cysteine 53 with an arginine (C53R) in the n-terminal zinc finger domain that is required for association of GEF-H1 with microtubules26,27. We expressed GFP-tagged GEF-H1 or its active mutants GEF-H1 (S885A) and GEF-H1 (C53R) or its GEF-deficient mutants to determine subcellular localization and association with the microtubule network. Confocal microscopy revealed that GEF-H1 or GEF-H1 (S885A) induced the formation of microtubules that formed aggregations and long curved filaments that contained α-tubulin (Fig. 3a). GEF-H1 ΔDH was found in the cytoplasm and co-localized with microtubules, but failed to induce microtubule aggregations (Fig. 3a). GEF-H1 (C53R) was unable to bind microtubules and therefore was found expressed in the cytoplasm and in intracellular aggregations (Fig. 3a).
Both the microtubule associating active GEF-H1 (S885A) and the cytoplasmic GEF-H1 (C53R) mutants significantly enhanced MAVS signaling when expressed in HEK-293T cells (Fig. 3b). In contrast, we found that the DH domain of GEF-H1 was required for the amplification of MAVS-induced Ifnb1 promoter activation (Fig. 3c). Furthermore, the ΔDH variant of GEF-H1 failed to enhance TBK1-mediated Ifnb1 promoter activation; together indicating that IRF3 activation in the presence of GEF-H1 was dependent on nucleotide exchange activity by GEF-H1 (Fig. 3d). Indeed, MAVS-induced Ifnb1 promoter activation was abrogated in the presence of a dominant negative RhoAT19N mutant (Fig. 3e).
Since GEF-H1 is sequestered on microtubules where its GEF function is inhibited by phosphorylation of S8855, we hypothesized that activation of GEF-H1 by dephosphorylation and release from microtubules may be required for RLR signaling. Immunostaining with antibodies directed against α-tubulin or phalloidin staining revealed intact microtubule and actin networks in unstimulated and poly(I:C)-stimulated wild-type and GEF-H1-deficient macrophages (Supplementary Fig. 3c,d). However, upon disruption of microtubules, macrophages failed to initiate Ifnb1 transcription after poly(I:C) and 5′ppp-dsRNA stimulation (Fig. 3f), although nocodazole treatment of macrophages did not prevent the uptake of poly(I:C) into macrophages (Supplementary Fig. 3e). Nocodazole treatment also reduced STING-mediated Ifnb1 mRNA expression in macrophages in response to c-di-GMP (Fig. 3g). The induction of TRIF-dependent IFN-β expression upon TLR4 activation in macrophages remained unchanged in the presence of nocodazole and thus occurred independent of the microtubule formation in macrophages (Fig. 3f).
We further found that a functional microtubule network was required for the interaction of GEF-H1 with TBK1 because GEF-H1-containing complexes lacked TBK1 in the presence of nocodazole, while neither TBK1 nor GEF-H1 expression was impaired under these conditions (Fig. 4a,b).
TBK1-containing signaling complexes preferably contained GEF-H1 that was dephosphorylated at S885 (Fig. 4c). Potential protein phosphatases that are associated with microtubule function, activated by foreign RNA and targeted by viral mediators include protein phosphatase 2A (PP2A)28. Inhibition of PP2A for 40 minutes using 1 nM okadaic acid, enhanced S885 phosphorylation of GEF-H1, but reduced its association with TBK1-containing signaling complexes (Fig. 4c,d). We also found reduced GEF-H1 in association with TBK1 after addition of forskolin (Fig. 4c), to stimulate the activation of adenylyl cyclase, increasing cellular concentrations of cAMP and subsequent S885 phosphorylation of GEF-H1 by protein kinase A (PKA)26 (Fig. 4d).
Functionally, GEF-H1 phosphorylation by forskolin or okadaic acid prevented the induction of IFN-β in macrophages by poly(I:C) but failed to impede Ifnb1 expression in response to LPS (Fig. 4e). Furthermore, GEF-H1 was dephosphorylated when the RLR signaling pathway was activated by infecting COS-7 cells with NS1-deficient Influenza A (A/PR/8/34 ΔNS1) (Fig. 4f).
In aggregate, these data are consistent with a multistep activation and release of GEF-H1 from microtubules to make its GEF activity available for amplification of RLR-mediated activation of TBK1-IKKε-dependent Ifnb1 promoter activation.
Thus far our data indicated that GEF-H1 regulates MAVS-dependent utilization of TBK1 for IRF3 phosphorylation and nuclear translocation for type I IFN induction. Furthermore, macrophages derived from Arhgef2−/− mice were impaired in response to 5′ppp-dsRNA and poly(I:C) stimulation but responded to TLR activation with type I IFN secretion, indicating that GEF-H1 functions in the RIG-I and Mda5-dependent induction of type I IFNs for antiviral defense.
We next assessed susceptibility of GEF-H1-deficient macrophages to distinct RNA viruses that activate the innate immune system to varying degree through RLRs and TLRs and compared the innate immune responses to those elicited in MAVS-deficient macrophages. Encephalomyocarditis virus (EMCV) is a positive-sense ssRNA virus of the Picornaviridae that is primarily detected by Mda5-dependent host responses16. GEF-H1-deficient macrophages were severely impaired in their ability to respond to EMCV infection with IFN-β secretion compared to wild-type macrophages when infected at the multiplicity of infection (MOI) of 0.1 (Fig. 5a). The reduction in IFN-β secretion was similarly observed in MAVS-deficient macrophages upon infection with EMCV (Fig. 5a). This effect was likely due to reduced IRF3 phosphorylation in the absence of GEF-H1 (Fig. 5b), since NF-κB activation was similar in GEF-H1-deficient, MAVS-deficient and wild-type macrophages 16 h after EMCV infection (Fig. 5c). As a consequence of reduced IFN-β secretion, virus replication was enhanced in GEF-H1-deficient macrophages as indicated by the significantly increased expression of transcripts encoding for EMCV non-structural protein 2A and 2B (Fig. 5d). Together, these experiments indicated that GEF-H1, MAVS and Mda5 are similarly required for host defense activation to EMCV.
Influenza A (Puerto Rico 8/1934 strain, PR/8/1934), is recognized by RIG-I16 and DHX929. However, TLR7 contributes to type I interferon secretion by plasmacytoid dendritic cells to the virus Influenza A30. We assessed IFN-β expression in GEF-H1-deficient macrophages after infection with a non-structural protein (NS)1-deficient influenza A variant31, that is unable to inhibit host IFN-β responses during viral replication. GEF-H1-deficient macrophages lacked nuclear translocation of IRF3 in response to NS1-deficient influenza A infection (Fig. 5e), although the cytoplasmic amounts of IRF3 were comparable to wild-type macrophages (Fig. 5f). Consequently, GEF-H1-deficient macrophages secreted significantly less IFN-β at 8 and 12 h after infection compared to wild-type macrophages (Fig. 5g). The reduction in IFN-β secretion was similarly observed in MAVS-deficient macrophages upon infection with NS1-deficient influenza A (Fig. 5g). However, NF-κB activation in response to NS1-deficient influenza A infection was independent of GEF-H1 and MAVS in macrophages (Fig. 5h). This suggested that alternative pathways contribute to IFN-β secretion in response to Influenza A infection that are not impaired in either GEF-H1 or MAVS-deficient macrophages. Despite reduced IFN-β expression, GEF-H1-deficient macrophages demonstrated increased expression of Influenza A nucleoprotein (NP) indicating enhanced viral replication compared to wild-type macrophages (Fig. 5i). Even when we infected NS1-sufficient influenza A, Ifnb1 mRNA induction was significantly reduced in GEF-H1-deficient macrophages 12 and 24 h after infection (Fig. 5j). Moreover, viral replication was enhanced as measured by NS1 RNA expression and the replication of a recombinant influenza virus carrying a GFP reporter gene in the NS segment32 in GEF-H1-deficient macrophages (Fig. 5k and Supplementary Fig. 4a).
Vesicular stomatitis virus (VSV) is a negative single stranded rhabdovirus that activates IFN-α/β through RIG-I, but not protein kinase R (PKR), Mda5 or TLR316,33. However, glycoprotein G of VSV is a ligand for TLR4 and can trigger IFN-α/β production independent of RIG-I33. At a functional level TLR4 and MAVS or RIG-I-dependent type I IFN production appear non redundant since both MAVS-deficient mice and TLR4 mutant mice are highly susceptibility to VSV33,34. We found that GEF-H1-deficient macrophages were impaired in the ability to restrict VSV replication (Supplementary Fig. 4b-d). In these experiments, we infected macrophages from Arhgef2−/− mice and wild-type littermates with an MOI of 0.1 with VSV and followed virus production in the supernatants over two days by infecting BHK cells. Beginning 4 h after infection, GEF-H1-deficient BMDMs produced 0.8-1.3 log phase higher amounts of active virus for 24 h compared to wild-type macrophages (Supplementary Fig. 4b). However, VSV infection of GEF-H1 or MAVS-deficient macrophages induced IFN-β at similar amounts to infected macrophages with intact Arhgef2 loci 12 to 48 h after infection. (Supplementary Fig. 4c and d). This is consistent with the finding that TLR4 signaling leading to IFN-β expression remained intact in GEF-H1-deficient macrophages and indicated that GEF-H1 selectively controlled RLR-dependent antiviral defense.
GEF-H1 functioned in mediating RLR recognition of viral RNA rather than by mediating type I interferon signaling, since IFN-α/β receptor activation leading to the phosphorylation of STAT1 was intact in GEF-H1-deficient macrophages (Supplementary Fig. 4e).
Finally, we determined innate host defense responses in GEF-H1-deficient macrophages to Salmonella typhimurium. The recognition of S. typhimurium is for the most part mediated by TLR2, TLR4 and TLR5 when cultured under conditions that favor Salmonella pathogenicity island 2 (SPI-2) expression35. GEF-H1-deficient macrophages expressed comparable Ifnb1 and Tnf mRNA levels to wild-type macrophages after infection with S. typhimurium (Supplementary Fig. 5a, b). Further, absence of GEF-H1 expression did not protect macrophages from invasion and intracellular replication of S. typhimurium (Supplementary Fig. 5c). Together these data demonstrate that GEF-H1 facilitates RIG-I and Mda5-dependent host defenses to viral pathogens without preventing the activation of IFN-α/β or TLR receptor signaling in macrophages.
We next assessed the susceptibility of Arhgef2−/− mice to influenza A infection to determine whether GEF-H1 was required for antiviral innate immune responses in vivo. We first determined whether alveolar macrophages of GEF-H1-deficient mice were impaired in recognizing poly(I:C). Bronchoalveolar fluid from GEF-H1-deficient mice lacked detectable IFN-β when challenged intranasally with poly (I:C), whereas wild-type mice secreted significant amounts of IFN-β in the airways in response to the same challenge (Fig. 6a). We also isolated alveolar macrophages and examined the concentration of IFN-β secretion after the initial intranasal challenge with poly(I:C) in vivo and assessed their responsiveness to additional challenges with poly(I:C). While alveolar macrophages from wild-type littermates significantly increased IFN-β secretion upon re-stimulation in vitro, no detectable IFN-β was released from GEF-H1-deficient alveolar macrophages after isolation or following re-stimulation in vitro suggesting a severe defect in the recognition of poly(I:C) by alveolar macrophages which we hypothesized would impair viral defense in these mice (Fig. 6b).
Indeed, GEF-H1-deficient mice were more susceptible to Influenza A infection compared to wild-type littermates. Four days after Influenza A infection, alveolar macrophages from GEF-H1-deficient mice expressed significantly lower levels of Ifnb1 and Il6 mRNA (Fig. 6c). Additionally, lungs of GEF-H1-deficient mice demonstrated significantly more signs of severe inflammation, with increased epithelial damage, mononuclear cell infiltrates, and alveolitis (Fig. 6d). This suggests that GEF-H1 is required for IFN-β induction for antiviral responses against influenza A infection.
GEF-H1-deficient macrophages have a profound defect in the induction of IFN-β following detection of synthetic dsRNAs including HMW and LMW poly(I:C) and 5′ppp-dsRNA. The inability to induce IFN-β in the absence of GEF-H1 was not due to impaired uptake of ligands or differential expression of signaling intermediates, but the requirement of the nucleotide exchange activity of GEF-H1 and polarized microtubules for RLR signaling. In macrophages GEF-H1 is dephosphorylated and released from microtubules during RLR activation and promotes IRF3 activation. Disruption of microtubule polarization prevents activation of GEF-H1 and consequently RLR signaling. Remarkably, TRIF and MyD88 dependent activation of IFN-β and proinflammatory cytokine expression was neither dependent on GEF-H1 nor nocodazole sensitive microtubule formation, suggesting that GEF-H1 has a distinct spatial function that is required for the activation of TBK1-IKKε in the RLR pathway (Supplementary Fig. 6).
Viral pathogens often target dynein machinery components with effectors to utilize the microtubule system for transport in host cells3. GEF-H1 may serve as gatekeeper on microtubules to locally modulate the activity of GTPases that in turn are responsible for the initial polarization of the microtubule cytoskeleton to facilitate antiviral responses36. Recently the microtubule network has been demonstrated to mediate the aggregation of mitochondria to facilitate the activity of the NLRP3 inflammasome37. Rho GTPase activation by GEF-H1 may stabilize actin association with mitochondria that occurs during RIG-I signaling during influenza A infection of macrophages38. In this context, GEF-H1 could regulate local Rho GTPase activation to promote stabilizing microtubules39 or membrane compartments as signaling platforms that allow GEF-H1 to interact with TBK1-IKKε containing signaling complexes. Furthermore, GEF-H1 may regulate the distribution of mitochondria within cells that requires crosstalk between microtubule and actin cytoskeleton through Rho GTPase activation40.
GEF-H1 and MAVS-deficient macrophages were impaired in RLR signaling leading to IRF3 activation. In contrast, NF-κB activation by surface and endosomal TLRs involved in the detection of viral RNA and viral glycoprotein remained intact in GEF-H1-as well as MAVS-deficient mice. Although TRIF-dependent TLR signaling can activate TBK141 and IRF3 activation42, we showed that in macrophages LPS stimulation induced primarily NF-κB activation. This is consistent with the finding that TLR-dependent NF-κB activation can occur independent of TBK141 but still can control IFN-β expression43 since the IFN-β promoter contains IRF3 and NF-κB sites that are utilized for induction by different signaling pathways44. Conversely, NF-κB is dispensable for IFN-β activation by RLRs that is mediated by IRF3 activation45,46.
GEF-H1 was not required for IFN-β secretion initiated by ligand binding to TLR1/2, TLR2, TLR2/6, TLR4, TLR5, TLR7 and TLR9 activation in macrophages. In addition, the induction of proinflammatory cytokines by TLR4 activation, which requires both MyD88- and TRIF-dependent signals47, was intact in GEF-H1-deficient macrophages. We further demonstrate that innate immune responses to viruses whose RNA or glycoptroteins also activates TLRs or invasive bacteria that activate TLRs were able to induce IFN-β expression even in the absence of GEF-H1 or MAVS. However the current experiments do not exclude a further function of GEF-H1 in TLR signaling in different cell types or in response to distinct pathogens that have different effector functions to evade host detection.
Our experiments demonstrate that GEF-H1-deficient macrophages fail to activate IRF3 but not NF-κB in response to ssRNA viruses. Because both IRF3 and NF-κB activate IFN-β, the role of GEF-H1 may depend on the degree to which antiviral host response to a particular pathogen includes the activation of RLR and TLR pathways. GEF-H1- and MAVS-deficient macrophages responded similarly to infection with EMCV with a profound lack of IFN-β secretion. Thus GEF-H1 is required for the recognition of EMCV infection that primarily occurs through Mda515,16. Mda5 is required and dominant over TLR3 for type I IFN induction by uncomplexed poly(I:C) in bone marrow-derived macrophages and DCs in vitro15 and we therefore cannot exclude a role of GEF-H1 in TLR3 signaling in different cell types.
GEF-H1- as well as MAVS-deficient macrophages demonstrate a significant reduction in IFN-β secretion in response to influenza A infection. In conventional dendritic cells and macrophages, RIG-I is required for the detection of influenza virus and induction of type I IFN via recognition of 5′-triphosphates on genomic ssRNA, which are generated after viral fusion and replication19. Our data demonstrate RIG-I activation in response to 5′ppp-dsRNA was impaired in GEF-H1-deficient macrophages resulting in attenuation of IRF3 phosphorylation, nuclear translocation and Ifnb1 gene transcription. Furthermore, GEF-H1-deficient alveolar macrophages failed to respond directly to stimulation or re-stimulation with poly(I:C) with IFN-β secretion. GEF-H1 was essential for host response to Influenza A, which was pronounced during infection with an NS1-deficient influenza A variant that lacks the ability of the wild-type virus to inhibit type I IFN secretion. It will be important to determine whether NS1 targets GEF-H1 function in host cells for immune evasion since variants of NS1 can regulate viral RNA load and RIG-I-mediated innate immune activation through mechanisms that may include targeting Rho GTPase function but have not been fully established48.
Our data indicate that GEF-H1 and polarization of microtubules are also required for the recognition of c-di-GMP that induces STING-mediated activation of TBK1 and IRF322. c-di-GMP serves as an important non-coding RNA binding second messenger in bacteria regulating the expression of many virulence genes49. Through binding to DDX41 and STING, c-di-GMP is recognized as a pathogen-associated molecular pattern that triggers the host type I interferon innate immune response22. Thus the role of GEF-H1 in the activation of IRF3 may also mediate DDX protein functions for the induction of type I interferon responses to bacterial infections. It will be important to define the role of GEF-H1 in other STING-dependent recognition pathways that include a large number of proposed sensors for cytosolic DNA sensing whose role in antimicrobial immunity and viral defense activation is currently being investigated50
In conclusion, our findings identify GEF-H1 as an antiviral signaling component that directs utilization of TBK1-IKKε in the MAVS-dependent nucleic acid detection pathways for the sensing of ssRNA virus infection and induction of IFN-β expression and secretion. GEF-H1 therefore possesses a pivotal role in mounting defenses against non-self RNA through RLRs, and thus understanding of the underlying molecular mechanisms of GEF-H1 activation and release from microtubules could lead to new therapeutic strategies against viral infection.
Arhgef2−/− mice were generated using the gene-trap ES cell clone IST13976A8 for Arhgef2 from the Texas Institute of Genomic Medicine (TIGM). ES cell clone were microinjected into C57BL/6 blastocysts. Chimeric offspring were used for the generation of homozygous mice for the targeted null allele (Arhgef2−/−). Genotyping was performed by PCR of tail genomic DNA using two sets of primers: F139_1, 5′-agtcccctgtccagtggtttacc-3′ and R_V76, 5′-ccaataaaccctcttgcagttgc-3′; F_V76, 5′-cttgcaaaatggcgttacttaagc-3′ and R139_4, 5′-agactcagggtcactggttgtga-3′ to produce amplicons that span the gene-trap vector junction to distinguish the Arhgef2− allele from the Arhgef2+ allele. PCR of tail genomic DNA was performed to distinguish the Arhgef2+ allele from the Arhgef2− allele using the F139_1 and R139_4 primers to produce an amplicon that spans the gene-trap vector insertion site. GEF-H1 mRNA was confirmed by RT-PCR using a forward primer spanning from exon 4~5 junction, 5′-aggcaaccaagacccgggaaaa-3′ and a reverse primer located in exon 12, 5′-taaggccttggtgtggcggc-3′. GEF-H1 protein expression was confirmed by immunoblotting. Mice were housed in pathogen-free barrier facilities. Arhgef2−/− and littermate control mice were used throughout the experiments. All animal experiments were performed according to animal protocols approved by the Subcommittee on Research Animal Care at Massachusetts General Hospital. Unless otherwise indicated, experiments used sex-matched littermates at 8-12 weeks of age.
HEK293T and COS-7 cells were purchased from American Type Culture Collection and maintained in DMEM supplemented with 10% fetal bovine serum and 0.5% penicillin/streptomycin mixture and tested to be free of mycoplasma. Bone marrow-derived macrophages were generated in DMEM containing 10% FBS, 0.5% penicillin/streptomycin mixture, and 20 ng/ml M-CSF for 5-7 days. Control dsRNA, 5′ppp-dsRNA, LMW poly(I:C), HMW poly(I:C), LPS,c-di-GMP, Pam3CSK4, HKLM, Flagellin, FSL-1, ssRNA, and CpG ODN1826 were purchased from Invivogen. Nocodazole and gentamicin solution was obtained from Sigma. Forskolin and okadaic acid were purchased from Abcam.
Plasmids encoding VSV-tagged GEF-H1, RhoA, RhoAT19N, and RhoAG14V have been described previously7. Plasmid encoding FLAG-tagged GEF-H1 (pCMV6-Entry-hGEF-H1) and GFP-tagged GEF-H1 (pCMV6-AC-GFP-hGEF-H1) were purchased from OriGene. FLAG-GEF-H1 (Y393A, ΔDH), GFP-GEF-H1 (Y393A, ΔDH), GFP-GEF-H1 (S885A), and GFP-GEF-H1 (C53R) mutations were generated using the Quikchange Site-Directed Mutagenesis Kit (Stratagene). pEF-BOS-huTBK1, pEF-BOS-huTBK1K38A, pcDNA3-IKKε-flag, pEF-BOS-huMAVS, pEF-BOS-huRIG-I and pEF-BOS-huMDA5 plasmids were obtained from Addgene. P561-luc-IRF3 firefly reporter was kindly provided by Xiaoxia Li (Cleveland Clinic, Cleveland, OH). The pGL4.20(luc2/puro)-IFNβ firefly luciferase promoter reporter was a gift from Tilmann Bürckstümmer (Austrian Academy of Sciences, Vienna, Austria). pNF-κβ firefly luciferase and pRL-TK Renilla plasmids were obtained from ClonTech. Horseradish peroxidase-anti-FLAG (M2; F1804) and anti-VSV (P5D4; V5507) antibodies were obtained from Sigma. Anti-GEF-H1 phosphorylated at S885 (ab74156), anti-IRF3 (D83B9; 4302), anti-IRF3 phosphorylated at Ser396 (4D4G; 4947), anti-TBK1 (D1B4; 3504), anti-MAVS (3993), anti-STAT1 (9172), anti-STAT1 phosphorylated at Tyr701 (58D6; 9167), anti-IKKε (D61F9; 3416), anti-IκBα (L35A5; 4814), anti-p65 phosphorylated at S536 (93H1; 3033), anti-p65 (D14E12; 8242), anti-β-actin (8H10D10; 3700), and anti-PCNA (D3H8P; 13110) antibodies were purchased from Cell Signaling Technology. Anti-α-tubulin antibody (ab4074) was obtained from Abcam. Anti-GEF-H1 antibody (x1089p) was purchased from Exalpha Biologicals. Alexa Fluor 488 Phallodin was obtained from Invitrogen. Anti-influenza A/PR/8/34 nucleoprotein antibody was a gift from Adolfo García-Sastre (Mount Sinai School of Medicine, New York, NY). FITC-conjugated donkey anti-sheep IgG (H+L) (713-095-147) and Cy3-comjugated donkey anti-rabbit IgG (H+L) (711-165-152) antibodies were purchased from Jackson ImmunoResearch Laboratories.
HEK293T cells were transfected with indicated plasmids using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol. For the experiment of GEF-H1 and TBK1 interaction, HEK293T cells were first transfected with GEF-H1-VSV and TBK1-Flag plasmid by Lipofectamine 2000 for 24 h, then were treated with nocodazole (1 or 10 μM), forskolin (10 μM), or okadaic acid (1 nM) for 40 minutes followed by immunoprecipitation and immunoblotting. The preparation of cell total lysates for immunoprecipitation and immunoblotting has been described previously7. Nuclear and cytoplasmic extracts from bone marrow-derived macrophages were prepared according to published methods41 with modifications: buffer A (10 mM HEPES, 1.5 mM MgCl2, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM DTT, 0.3 mM Na3VO4) and buffer C (20 mM HEPES, 1.5 mM MgCl2, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 0.3 mM Na3VO4, 0.4 M NaCl, 1 mM PMSF). Densitometry analysis of Immunoblots was done by ImageJ (NIH).
Influenza A virus (PR/8/1934), A/PR/8/1934 ΔNS1, A/PR/8/1934 NS1-GFP and anti-NP antibody were kindly provided by Adolfo García-Sastre (Mount Sinai School of Medicine, New York, NY). EMCV was purchased from ATCC. VSV was a gift from Charles Rice and Margaret MacDonald at The Rockefeller University. Bone marrow-derived macrophages from WT and Arhgef2−/− mice were infected with virus in DPBS with 1% FBS at indicated MOI for 1 h. Infection was continued for various times in the presence of serum-containing DMEM. For the in vivo influenza A/PR/8/1934 infection, Arhgef2−/− mice and their littermate controls were anesthetized and challenged with 20 μl (10 μl per nostril) of influenza A/PR/8/1934 suspension (103 pfu) intranasally.
Supernatants from infected cells were used to measure viral titers by plaque assays. Monolayers of BHK-21 cells were used for VSV plaque assays. After 1 h of viral infection, BHK-21 cells were overlaid with 1.5% LE agarose for 1 day. The cells were then fixed with 7% formaldehyde followed by crystal violet staining. Plaques were counted to determine the titers.
Salmonella enterica serovar typhimurium strain SL1344 was used. Bacterial cultures were prepared by inoculating 10 ml of LB with 0.01 ml of a stationary phase culture, followed by a 16 h incubation at 37°C. Bone marrow derived macrophages were spininfected at moi=10 for 15 minutes at 750 rpm followed by incubation at 37°C for 45 minutes. Cells were washed twice before the addition of 100 μg/ml gentamicin in DMEM with 10 % FBS. 1 h after infection, macrophages were given fresh DMEM medium containing 10% FBS and 10 μg/ml gentamicin for the reminder of the experiment. For the determination of intracellular replication of S. typhimurium, cells were lysed in 1% Triton X-100 in PBS. Lysates were serially diluted in DMEM and plated on LB agar plates containing 100 μg/ml streptomycin for colony enumeration.
Mouse bone marrow-derived macrophages and alveolar macrophages were isolated and treated as described above. Qiagen RNeasy kit was used for the extraction of RNA from all cell types examined and, after synthesis of cDNA with iScript cDNA synthesis kit (Bio-Rad), iQ SYBR Green Supermix kit (Bio-Rad) was used for real-time PCR (Bio-Rad CFX96 Real-Time PCR Detection System) according to the manufacturer’s specifications. The value obtained for each gene was normalized to that of the GAPDH gene. Primers were as follows: IFN-β forward, 5′-ccctatggagatgacggaga-3′, and reverse, 5′-ctgtctgctggtggagttca-3′; IL-6 forward, 5′-ctgatgctggtgacaaccac-3′, and reverse, 5′-tccacgatttcccagagaac-3′; TNFα forward, 5′-tagccaggagggagaacaga-3′, and reverse, 5′-ttttctggagggagatgtgg-3′; GAPDH forward, 5′-aactttggcattgtggaagg-3′, and reverse, 5′-ggatgcagggatgatgttct-3′; Influenza NS1 forward, 5′-tcgagacagccacacgtgctggaaa-3′; Influenza NS1 reverse, 5′-aagagggcctgccactttctgcttg-3′; EMCV 2A-2B forward, 5′-aatgcccactacgctggt-3′; EMCV 2A-2B reverse, 5′-gtcgttcggcagtagggt-3′.
Concentration of IFN-β in cell culture supernatants was measured using commercial ELISA kits (PBL interferon source) or Luminex assays (Affymetrix) according the manufacturers’ instructions.
HEK293T cells were transfected with 50 ng of IFNβ, p561, or NF-κB firefly luciferase reporter and 0.5 ng of pRL-Renilla reporter together with expression plasmids by Lipofectamine 2000. For control experiments, empty pcDNA3.1 were utilized. Transfection of WT and Arhgef2−/− macrophages was carried out using Amaxa Mouse Macrophage Nucleofector kits. Luciferase assays were performed 24 h post transfection using the Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s instructions.
Isolated cells from spleen and mesenteric lymph nodes were incubated in 10% donkey serum and Fc block for 20 minutes at 4°C and then stained with the following fluorescent-conjugated antibodies: APC-conjugated anti-CD11c (HL3; 550261), PE-conjugated anti-F4/80 (6F12; 552958), PE-conjugated anti-CD103 (M290; 557495), APC-Cy7-conjugated CD4 (GK1.5; 552051), PerCP-conjugated CD8 (53-6.7; 553036), FITC-conjugated anti-NK1.1 (PK136; 553164), and APC-conjugated a anti-CD19 (1D3; 550992). All antibodies were obtained from BD Pharmingen. Cells were analyzed on a FACSCalibur flow cytometer (BD Bioscience) and analyzed by FlowJo (Tree Star).
WT or Arhgef2−/− bone marrow-derived macrophages were plated on 4-well chamber coverglass (Lab-Tek) and stimulated with poly(I:C)-Rhodamine or infected with influenza A/PR/8 NS1-GFP at indicated time periods. Live cells were imaged with a Nikon A1R-A1 confocal microscope. Image acquisition was carried out with NIS-Elements imaging software (Nikon) followed by analyses by Volocity (PerkinElmer). Immunofluorescence staining and imaging were performed as previously described7.
Bronchoalveolar lavages were recovered by cannulation with 1000 μl of PBS after terminal exsanguination. To obtain alveolar macrophages, cells were washed and then enriched by centrifugation.
To assess histological changes of lung following infection with influenza A/PR/8/1934 infection, anesthetized mice were exsanguinated via the abdominal aorta, and their lung tissues were fixed in 4% formaldehyde overnight at 4°C. The tissues were dehydrated by gradually soaking in alcohol and xylene and then embedded in paraffin. Specimens were cut into 5 μm sections and stained with Hamp;E. Lung sections were evaluated ‘blinded’ to sample identity and scored based on assessments of lung tissue destruction, epithelial cell layer damage, polymorphonuclear cell infiltration into the inflammation site, and alveolitis on a scale of 0-5 each parameters (0, none; 5 severe).
GraphPad Prism was used for all statistical analysis. Total sample size was determined based on the previous studies with similar genetically modified mouse with comparable functional defects. All experiments were repeated at least three times. Statistical analysis was carried out by two-way analysis of variance (ANOVA) followed by Student’s t-test. A P value < 0.05 was considered statistically significant.
This work was supported by grants AI093588 (HCR), DK-068181 (HCR), DK-033506 (HCR), DK-043351 (HCR, CT) and DK-52510 (CT) from the National Institutes of Health
Note: Supplementary Information is available in the online version of the paper.
HSC, YZ, JHS, SL, NW, and MB carried out experiments; KJ, AHS and CT supported the development of research tools and mice; KJ provided virus and advised on virus infection experiments; HCR conceived of, directed all research and along with HSC, prepared the manuscript.
COMPETING FINICIAL INTERESTS
The authors declare no competing financial interests.