Murine Cytomegalovirus Virion-Associated Protein M45 Mediates Rapid NF-κB Activation after Infection

ABSTRACT

Murine cytomegalovirus (MCMV) rapidly induces activation of nuclear factor κB (NF-κB) upon infection of host cells. After a transient phase of activation, the MCMV M45 protein blocks all canonical NF-κB-activating pathways by inducing the degradation of the gamma subunit of the inhibitor of κB kinase complex (IKKγ; commonly referred to as the NF-κB essential modulator [NEMO]). Here we show that the viral M45 protein also mediates rapid NF-κB activation immediately after infection. MCMV mutants lacking M45 or expressing C-terminally truncated M45 proteins induced neither NF-κB activation nor transcription of NF-κB-dependent genes within the first 3 h of infection. Rapid NF-κB activation was absent in MCMV-infected NEMO-deficient fibroblasts, indicating that activation occurs at or upstream of the IKK complex. NF-κB activation was strongly reduced in murine fibroblasts lacking receptor-interacting protein 1 (RIP1), a known M45-interacting protein, but was restored upon complementation with murine RIP1. However, the ability of M45 to interact with RIP1 and NEMO was not sufficient to induce NF-κB activation upon infection. In addition, incorporation of the M45 protein into virions was required. This was dependent on a C-terminal region of M45, which is not required for interaction with RIP1 and NEMO. We propose a model in which M45 delivered by viral particles activates NF-κB, presumably involving an interaction with RIP1 and NEMO. Later in infection, expression of M45 induces the degradation of NEMO and the shutdown of canonical NF-κB activation.

IMPORTANCE Transcription factor NF-κB is an important regulator of innate and adaptive immunity. Its activation can be beneficial or detrimental for viral pathogens. Therefore, many viruses interfere with NF-κB signaling by stimulating or inhibiting the activation of this transcription factor. Cytomegaloviruses, opportunistic pathogens that cause lifelong infections in their hosts, activate NF-κB rapidly and transiently upon infection but block NF-κB signaling soon thereafter. Here we report the surprising finding that the murine cytomegalovirus protein M45, a component of viral particles, plays a dual role in NF-κB signaling. It not only blocks NF-κB signaling later in infection but also triggers the rapid activation of NF-κB immediately following virus entry into host cells. Both activation and inhibition involve M45 interaction with the cellular signaling mediators RIP1 and NEMO. Similar dual functions in NF-κB signaling are likely to be found in other viral proteins.

INTRODUCTION

Transcription factor NF-κB functions as an important cellular regulator of the immediate response to infection by microbial pathogens. It induces the transcription of genes encoding inflammatory cytokines, chemokines, adhesion molecules, proinflammatory enzymes, and apoptosis-regulating proteins. These factors are essential components of the innate immune response against invading microbes and play an important role in shaping an effective adaptive immune response (1).

In the classical pathway, NF-κB is kept in an inactive form in the cytoplasm by inhibitor of NF-κB (IκB) proteins, of which IκBα is the most prominent member. The IκB proteins are, in turn, controlled by the IκB kinase (IKK) complex, which consists of two catalytic subunits (α and β) and a regulatory subunit (γ) commonly referred to as the NF-κB essential modulator (NEMO). Upon activation, the IKK complex phosphorylates IκBα, leading to rapid ubiquitination and proteasomal degradation of IκBα (2). Once released from its inhibitor, NF-κB translocates to the nucleus and activates the transcription of NF-κB-responsive genes (Fig. 1A).

FIG 1.

M45 induces IκBα degradation and nuclear translocation of NF-κB p65. (A) Schematic representation of NF-κB-activating signaling pathways. (B) NIH 3T3 cells were infected with wt MCMV, an M45 deletion mutant (ΔM45), or a revertant virus (RevM45) at an MOI of 10 TCID₅₀/cell. Cells were lysed at the indicated times postinfection, and protein levels were analyzed by immunoblotting. (C) NIH 3T3 cells were infected with wt MCMV or the ΔM45 mutant at an MOI of 10 TCID₅₀/cell. Cells were lysed at the indicated times postinfection; cytoplasmic and nuclear fractions were separated; and protein levels were analyzed by immunoblotting.

Several receptors initiate NF-κB-activating pathways that converge at the IKK complex. These include cytokine receptors, such as tumor necrosis factor (TNF) receptor 1 (TNFR1) and interleukin-1 receptor (IL-1R), and pattern recognition receptors (PRRs), such as the toll-like receptors (TLRs), which recognize pathogen-associated molecular patterns of bacterial and viral origin at the plasma membrane or within endosomes (2). Additional PRRs, such as the retinoic acid-inducible gene 1 (RIG-I)-like receptors and Z-DNA-binding protein 1 (ZBP1; also known as DNA-dependent activator of interferon regulatory factors [DAI]), recognize viral nucleic acids in the cytosol (3). All these receptors signal to the IKK complex through distinct sets of adaptor proteins. Most TLRs, the structurally related IL-1R, and the RIG-I-like receptors transmit signals to the IKK complex via the adaptor proteins interleukin receptor-associated kinase 1 (IRAK1) and IRAK4. In contrast, TNFR1, TLR3, TLR4, and DAI rely on receptor-interacting protein 1 (RIP1) for signal transduction to the IKK complex (4). These receptors can also recruit the adaptor kinase RIP3, a critical mediator of the pathway leading to programmed necrosis (5).

During virus-host coevolution, viruses not only have acquired strategies to inhibit cellular response pathways that could be detrimental to their own survival but also have found ways to integrate activated cellular signaling pathways into their own life cycles. For instance, NF-κB activation can promote the survival of infected cells, prolonging the time span available for viral replication. Many oncogenic viruses activate NF-κB to induce the transformation of infected cells (e.g., Epstein-Barr virus, Kaposi's sarcoma-associated herpesvirus, Marek's disease virus, human T-lymphotropic virus). Moreover, a number of viruses harbor NF-κB binding sites in their promoters or enhancers, utilizing NF-κB for viral gene expression (e.g., human immunodeficiency virus, herpes simplex virus, cytomegalovirus) (6, 7).

Both human cytomegalovirus (HCMV) and murine cytomegalovirus (MCMV) regulate NF-κB activity in a time-dependent manner. They activate NF-κB in the first few hours of infection and inhibit its activation later on. The underlying molecular mechanisms remain incompletely understood. HCMV has been shown to induce an immediate and transient phase of NF-κB activation, which is triggered by virus attachment and entry into host cells (8, 9). The HCMV envelope glycoproteins gB and gH play a major role in infection-induced NF-κB activation (10, 11). Stimulation of TLR2 by these viral glycoproteins has been identified as an underlying mechanism (12–14). Direct phosphorylation of IκBα by casein kinase 2, which is delivered to cells by HCMV virions, also contributes to infection-induced NF-κB activation (15). At later times, HCMV inhibits TNF-α- and IL-1β-mediated NF-κB activation. Inhibition occurs upstream of the IKK complex and, in the case of TNF-α, is correlated with downregulation of TNFR1 (16, 17).

Like HCMV, MCMV uses several mechanisms to inhibit NF-κB activation later on in infection. Downregulation of TNFR1 has been observed in MCMV-infected macrophages and has been proposed as a mechanism by which MCMV inhibits TNF-α-induced NF-κB activation (18). More-recent work has revealed that the MCMV M45 protein is a very potent inhibitor of NF-κB activation. M45 interacts with the signaling mediator RIP1 and inhibits RIP1-dependent signaling, including activation of NF-κB after stimulation of TNFR1 and TLR3, as well as TNFR1-mediated activation of p38 mitogen-activated protein kinase (19). M45 also blocks RIP1-independent NF-κB activation and cytokine production after stimulation of IL-1R and TLRs. It binds to the IKK subunit NEMO and induces its lysosomal degradation by targeting NEMO to autophagosomes (20). The M45-induced degradation of NEMO results in an efficient block to all canonical NF-κB activation pathways and in the suppression of proinflammatory cytokine production. M45 also inhibits NF-κB activation mediated by the cytosolic DNA sensor DAI (21). In addition to its function as an inhibitor of NF-κB activation, M45 also acts as a powerful suppressor of infection-induced cell death (22). M45 inhibits a caspase-independent form of programmed cell death (19) called programmed necrosis, or necroptosis, which depends on the adaptor kinase RIP3 (23). RIP3 is activated by interaction with RIP1, DAI, or TRIF via a RIP homotypic interaction motif (RHIM) that is present in all three proteins (24–27). M45 itself contains a RHIM, which is required for interaction with RIP3 and DAI and for the inhibition of necrosis (21, 23, 25, 28).

Also like HCMV, MCMV induces a transient phase of NF-κB activation within the first few hours of infection (29, 30). However, the viral proteins and cellular signaling pathways involved have not been defined. Here we report the unexpected finding that M45—the same protein that inhibits NF-κB activation at early and late times of infection—is also responsible for the rapid activation of NF-κB immediately following infection. This activation depends, at least in part, on the interaction of M45 with RIP1 and possibly also on its interaction with NEMO. Finally, we demonstrate that virion-associated M45 protein, which is delivered to cells upon infection, mediates the rapid NF-κB activation.

MATERIALS AND METHODS

Cells and viruses.

Human embryonic kidney 293A cells were purchased from Invitrogen. NIH 3T3 fibroblasts (CRL-1658), IC-21 macrophages (TIB-186), and SVEC4-10 endothelial cells (CRL-2181) were obtained from the ATCC. Immortalized wild-type (wt) murine embryonic fibroblasts (MEFs) and RIP3^−/− MEFs (31) were provided by Edward Mocarski (Emory University, Atlanta, GA), RIP1^−/−, TNFR1^−/−, and TRAF2/5^−/− MEFs (32–34) by Michelle Kelliher (University of Massachusetts, Boston, MA), NEMO^−/− MEFs (35) by Michael Karin (University of California, San Diego, CA), MyD88^−/− and TLR2/4^−/− MEFs (36, 37) by Simon Fillatreau (German Rheumatism Research Centre, Berlin, Germany), TRIF^−/− MEFs (38) by Markus Heimesaat (Charité, Berlin, Germany), and TRAF6^−/− MEFs (39) by Arnd Kieser (Helmholtz Center, Munich, Germany). TLR3^−/− MEFs were isolated from TLR3^−/− mice (40) purchased from The Jackson Laboratory (Bar Harbor, ME).

The MCMV Smith bacterial artificial chromosome (BAC) pSM3fr-MCK-2fl (41) was obtained from Barbara Adler (University of Munich, Munich, Germany). All mutant viruses were generated by en passant mutagenesis as described previously (42). The ΔM45 deletion mutant lacks the entire M45 open reading frame (ORF). The revertant, RevM45, was generated by inserting the full-length M45 ORF tagged at the 3′ end with a hemagglutinin (HA) epitope sequence into the ΔM45 BAC. All viruses expressing mutated versions of M45 were constructed in the same way as RevM45 and express C-terminally HA-tagged proteins. The M45 truncation constructs Nt2, Nt3, and Ct3 through Ct6 have been described elsewhere (19, 20). The Nt3b, Nt3c, and 1-277 truncation mutants were generated by PCR amplification of the corresponding DNA sequences. A plasmid encoding M45 with a mutated RHIM (mutRHIM) (28) was provided by Edward Mocarski.

Viruses were grown and titrated on NIH 3T3 cells according to standard procedures (43). Viral titers were determined using the median tissue culture infective dose (TCID₅₀) method (44). Infections were carried out with centrifugal enhancement (1,000 × g, 30 min). Viruses were inactivated by 254-nm-wavelength UV irradiation with 1 J/cm² for 30 s using a UV cross-linker (HL-2000 HybriLinker; UVP).

Antibodies.

The following antibodies were used: monoclonal antibodies against glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (14C10; Cell Signaling), gB (SN1.01; provided by Stipan Jonjic, University of Rijeka, Rijeka, Croatia), HA (16B12; Covance), IE1 (Chroma101; provided by Stipan Jonjic), NEMO (EA2-6; MBL), NF-κB p65 (F-6; Santa Cruz), phosphorylated IκBα (5A5; Cell Signaling), and RIP1 (clone 38; BD Transduction Laboratories) and polyclonal antibodies against HA (H6908; Sigma), IE3 (provided by Eva Borst, Hannover Medical School, Hannover, Germany), IκBα (C-21; Santa Cruz), LSD1 (Cell Signaling), M45 (45) (provided by David Lembo, University of Turin, Turin, Italy), and β-actin (AC-74; Sigma). Secondary antibodies coupled to horseradish peroxidase (HRP) were purchased from DakoCytomation or Jackson ImmunoResearch.

Immunoprecipitation and immunoblotting.

For immunoprecipitation, cells were grown in 6-well dishes and were infected at a multiplicity of infection (MOI) of 5 TCID₅₀/cell. At 15 h postinfection (hpi), cells were lysed (50 mM Tris, 150 mM NaCl, 1% [vol/vol] Nonidet P-40, and Complete Mini protease inhibitor cocktail [Roche]), and insoluble material was removed by centrifugation. After preclearing with protein A Sepharose (PAS; GE Healthcare), HA-tagged M45 protein was precipitated using an anti-HA antibody (H6908; Sigma) and PAS. Precipitates were washed six times, eluted by boiling in sample buffer, and subjected to SDS-PAGE and immunoblotting.

For infection kinetics, cells were grown in 12-well dishes, infected at an MOI of 10 TCID₅₀/cell, and lysed in SDS-PAGE sample buffer.

Immunofluorescence.

NIH 3T3 cells were grown on coverslips, infected at an MOI of 5 TCID₅₀/cell, and fixed for 20 min with 4% paraformaldehyde in phosphate-buffered saline (PBS). Cells were incubated with 50 mM ammonium chloride, permeabilized with 0.3% Triton X-100, and blocked with 0.2% porcine skin gelatin (Sigma). Cells were then incubated with primary antibodies for 2 h at room temperature (RT), washed three times with PBS, and incubated with secondary antibodies coupled to Alexa Fluor 488 or Alexa Fluor 555 (Life Technologies) for 2 h at RT. Nuclear DNA was stained with Draq5 (BioStatus). Samples were washed, mounted on slides with Aqua-Poly/Mount (Polysciences), and analyzed by confocal laser scanning microscopy using a Zeiss LSM510 META/FCS microscope.

4sU labeling and qRT-PCR.

RNA labeling was started by adding 200 μM 4-thiouridine (4sU; Carbosynth) to the cell culture medium for 1 h at different times of infection. At the end of labeling, total cellular RNA was isolated using TRIzol reagent (Invitrogen). Biotinylation and purification of 4sU-tagged RNA (newly transcribed RNA) were performed as described previously (46). Following cDNA synthesis using SuperScript III reverse transcriptase (Invitrogen), hexanucleotide random primers (Invitrogen), and the RNase inhibitor RNasin (Promega), reverse transcription-quantitative PCR (qRT-PCR) was performed using TaqMan PCR and the Universal Probe Library (Roche) or SYBR green PCR. The primers and probes used are listed in Table 1. Relative quantification was performed by normalizing to the housekeeping gene encoding the lamin B receptor.

TABLE 1.

Primers and probes used in this study

Gene	Gene name	Probe	RefSeq mRNA	Forward primer	Reverse primer
Lamin B receptor	Lbr	SYBR green	NM_133815	TGCAGAAGGAACACCTCTTG	CAGAATGAAGGCATACAATCCA
Nuclear factor of kappa light polypeptide gene enhancer in B cells inhibitor, alpha	Nfkbia	SYBR green	NM_010907	ACGAGCAAATGGTGAAGGAG	ATGATTGCCAAGTGCAGGA
Tumor necrosis factor alpha-induced protein 3	Tnfaip3	SYBR green	NM_001166402	TCATCGAATACAGAGAAAATAAGCAG	AGGCACGGGACATTGTTCT
Chemokine (C-X-C motif) ligand 10	Cxcl10	TaqMan probe 3	NM_021274	GCTGCCGTCATTTTCTGC	TCTCACTGGCCCGTCATC
MCMV immediate early gene 1	ie1	SYBR green	M11788	TCAGCCATCAACTCTGCTACCAAC	ATCTGAAACAGCCGTATATCATCTTG

NF-κB reporter assay.

pNiFty2-SEAP (InvivoGen) is an NF-κB-inducible reporter plasmid expressing secreted embryonic alkaline phosphatase (SEAP). NIH 3T3 cells grown in 24-well plates were cotransfected with pNiFty2-SEAP and plasmids encoding the proteins to be analyzed. SEAP activities in supernatants were determined photometrically 40 h after transfection by using the QUANTI-Blue detection reagent (InvivoGen) according to the manufacturer's protocol.

Cell fractionation.

Cytoplasmic and nuclear fractions were separated using the FractionPREP cell fractionation kit (BioVision).

Generation of stable cell lines.

RIP1^−/− MEFs stably transduced with an empty or murine RIP1-expressing murine stem cell virus (MSCV) retroviral vector have been described previously (19). RIP1^−/− MEFs stably expressing human RIP1 were generated in the same way. Briefly, human RIP1 was excised from pRK-mycRIP (47) (provided by Zheng-Gang Liu, NIH, Bethesda, MD) and was inserted into pMSCVpuro (Clontech). The retrovirus was produced using the Phoenix packaging cell line as described previously (48) and was used to transduce RIP1^−/− MEFs. Transduced cells were selected with 5 μg/ml puromycin (Sigma).

For the production of a ΔM45 virus that contains virion-associated M45, complementing NIH 3T3 cells were generated by stable transfection with the episomal plasmid pEpiNo-M45-ori. The plasmid was constructed by inserting the M45-HA sequence between the HindIII and XbaI sites of pEpiNo-luc-ori (49) (provided by Hermine Mohr, University of Munich, Munich, Germany), replacing the luciferase sequence. This plasmid contains the MCMV lytic origin of replication and allows for inducible expression of M45 upon MCMV infection. Transfected cells were selected with 2 mg/ml Geneticin (G418; Roth), and single cell clones were isolated and were tested by immunoblotting for inducible M45 expression.

Gradient purification of virions.

Purified virions for the analysis by immunoblotting were produced by glycerol tartrate gradient ultracentrifugation (50) according to a protocol provided by Bodo Plachter (University of Mainz, Mainz, Germany). Briefly, the supernatants of 10 145-mm dishes of infected cells were harvested, and cellular debris was removed by centrifugation (5,500 × g, 15 min). The virus was pelleted by centrifugation (3 h at 15,000 × g) and was resuspended in 1 ml PBS. A gradient of 35 to 15% tartrate and 0 to 30% glycerol was formed, overlaid with the virus suspension, and centrifuged overnight at 94,000 × g. The virus-containing band was extracted by needle aspiration, and virions were pelleted by centrifugation. The virus pellet was resuspended in 100 μl PBS.

RESULTS

M45 is required for rapid NF-κB activation immediately after infection.

MCMV infection causes a rapid but very transient activation of NF-κB (29, 30). To further investigate this phenomenon, we infected fibroblasts with MCMV at a high MOI and monitored the levels of the NF-κB inhibitor IκBα as an indicator of NF-κB activation. An M45 knockout virus (ΔM45) and the corresponding revertant (RevM45) were included, because the M45 gene encodes a potent inhibitor of the canonical NF-κB pathway (20). In cells infected with wt MCMV or RevM45, we observed a transient reduction of IκBα levels between 1 and 5 h postinfection (hpi), with maximum reductions at 2 and 3 hpi (Fig. 1B). IκBα levels recovered at 5 to 7 hpi. The recovery of IκBα levels coincided with the degradation of NEMO, the regulatory subunit of the IKK complex, and a decline in RIP1 levels (Fig. 1B). Surprisingly, this rapid and transient drop in IκBα levels was not seen in ΔM45 mutant-infected cells. Instead, we observed a delayed IκBα decrease starting around 5 hpi (Fig. 1B). NEMO was not degraded in ΔM45 mutant-infected cells, in agreement with previous results (20), and RIP1 levels did not decline.

In order to confirm that the IκBα degradation observed resulted in NF-κB activation, we determined the subcellular localization of the NF-κB p65 subunit by separating the nuclei from the cytoplasm of infected cells. In wt MCMV-infected cells, NF-κB p65 levels were strongly increased in the nuclear fraction between 2 and 7 hpi (Fig. 1C). In cells infected with the ΔM45 mutant, NF-κB p65 was detected at elevated levels in the nucleus starting at 5 hpi (Fig. 1C). In both cases, the appearance of NF-κB p65 in the nucleus closely followed the degradation of IκBα. Similar results were obtained by immunofluorescence analysis. At 2 to 5 hpi, NF-κB p65 was detected in the nuclei of wt MCMV-infected cells (Fig. 2). In contrast, p65 was found in the nuclei of ΔM45 mutant-infected cells predominantly between 5 and 9 hpi (Fig. 2). Thus, rapid NF-κB activation is dependent on M45. In the absence of M45, NF-κB is activated with delayed kinetics. It is noteworthy that expression of the viral immediate early 1 (IE1) protein was not delayed in ΔM45 mutant-infected fibroblasts (Fig. 1B and C). Therefore, the delay in NF-κB activation cannot be attributed to a delay in virus entry.

FIG 2.

Immunofluorescence analysis of NF-κB p65 nuclear translocation. NIH 3T3 cells were infected with wt MCMV or the ΔM45 mutant at an MOI of 5 TCID₅₀/cell. Cells were fixed at the indicated times postinfection, and the subcellular localization of the NF-κB p65 subunit was analyzed by immunofluorescence. Costaining with IE3 served as an infection control. Nuclei were stained with Draq5. Arrowheads indicate cells with nuclear p65.

RIP1 and NEMO contribute to M45-induced NF-κB activation.

To gain insight into the mechanism of M45-induced NF-κB activation, we infected MEFs derived from different knockout mouse strains with wt MCMV and determined IκBα degradation during the course of infection by immunoblot analysis. NEMO proved to be essential for M45-induced NF-κB activation, since IκBα levels remained unchanged in MCMV-infected NEMO-deficient cells (Fig. 3A), indicating that NF-κB activation is triggered at or upstream of the IKK complex. Degradation of IκBα was impaired in RIP1 knockout cells (Fig. 3A), suggesting that RIP1 is involved in M45-induced NF-κB activation. In contrast, IκBα was degraded in TNFR1- and TLR2/4-deficient MEFs with kinetics similar to those observed previously in normal fibroblasts (Fig. 3A). IκBα degradation was also unaltered in MEFs deficient in other cellular proteins involved in NF-κB-activating pathways (TLR3, TRIF, MyD88, TRAF2, TRAF5, TRAF6, and RIP3 [data not shown]), suggesting that these proteins are not required for MCMV-induced NF-κB activation immediately after infection.

FIG 3.

MCMV-induced NF-κB activation in different cells. (A) MEFs deficient in TNFR1, NEMO, RIP1, or TLR2 and TLR4 were infected with wt MCMV at an MOI of 10 TCID₅₀/cell and were harvested at the indicated times postinfection. Protein levels were analyzed by immunoblotting. (B) Murine embryonic fibroblasts (MEFs), SVEC4-10 endothelial cells, IC-21 macrophages, and human embryonic kidney 293A cells were infected with wt MCMV at an MOI of 10 TCID₅₀/cell. Cells were harvested at the indicated times postinfection, and protein levels were analyzed by immunoblotting.

To test whether MCMV-induced NF-κB activation occurs in a cell type-specific manner, we infected murine fibroblasts, endothelial cells, and macrophages. In all murine cell types tested, immediate and transient IκBα degradation was observed (Fig. 3B). Interestingly, IκBα levels were only slightly reduced in MCMV-infected human embryonic kidney 293A cells (Fig. 3B), even though these cells can be infected and support MCMV replication at low levels (51).

To further investigate the lack of IκBα degradation in human 293A cells, we tested whether M45 interacts with RIP1 and NEMO in these cells. RIP1 and NEMO are known interaction partners of M45 (19, 20, 28) and are involved in MCMV-induced M45-dependent NF-κB activation (Fig. 3A). In coimmunoprecipitation experiments, M45 interacted with endogenous RIP1 and NEMO in MCMV-infected murine NIH 3T3 cells (Fig. 4A). In contrast, endogenous human NEMO, but not endogenous human RIP1, coprecipitated with M45 in lysates of infected 293A cells, suggesting that M45 interacts with human NEMO but not with human RIP1.

FIG 4.

Human RIP1 does not coprecipitate with M45 and does not restore M45-induced NF-κB activation in RIP1 knockout cells. (A) Human 293A cells and murine NIH 3T3 cells were infected with MCMV ΔM45 or the revertant virus RevM45 at an MOI of 5 TCID₅₀/cell. Cells were lysed at 15 hpi and were subjected to immunoprecipitation (IP) using an anti-HA antibody. Whole-cell lysates (WCL) and immunoprecipitates were analyzed by immunoblotting. (B) RIP1-deficient MEFs (RIP1^−/−) or RIP1^−/− MEFs complemented with either murine RIP1 (mRIP1) or human RIP1 (hRIP1) were infected with wt MCMV at an MOI of 10 TCID₅₀/cell. Cell lysates were harvested at the indicated times postinfection, and protein levels were analyzed by immunoblotting.

To further clarify the role of RIP1 in M45-induced NF-κB activation, we infected RIP1-deficient cells complemented with either murine or human RIP1 (mRIP1 or hRIP1) by retroviral transduction. After MCMV infection, IκBα levels were analyzed by immunoblotting. Infection of RIP1^−/− cells complemented with mRIP1 resulted in a pronounced drop in IκBα levels (Fig. 4B), similar to that seen in normal fibroblasts. In contrast, complementation with hRIP1 did not alter IκBα levels from those in the parental RIP1^−/− cells (Fig. 4B). It is noteworthy that hRIP1 protein levels did not decline during the course of infection, while mRIP1 levels did decline (Fig. 4B). This effect depends on M45 (Fig. 1B), and the failure of hRIP1 levels to drop (Fig. 4B) is likely related to the deficient interaction of M45 with hRIP1 (Fig. 4A). Taken together, the experiments for which results are depicted in Fig. 4 show that the interaction with RIP1 plays a major role in M45-mediated NF-κB activation.

The interaction of M45 with RIP1 and NEMO is not sufficient for NF-κB activation.

To determine which part of the M45 protein is required for NF-κB activation and how this correlates with other known functions of M45, we constructed a set of virus mutants expressing different N- or C-terminally truncated versions of M45 (Fig. 5A). All mutant M45 proteins carry an HA epitope tag at their C terminus. We have reported previously that a region near the C terminus of M45 is required for interaction with RIP1 and NEMO and for the inhibition of NF-κB activation, whereas large parts of the N-terminal domain, including the RHIM, can be deleted without abolishing these functions (19, 20). However, most of these results were obtained in cells expressing plasmid- or retrovirus-encoded M45 mutants. To verify the results in the context of MCMV infection, we infected NIH 3T3 cells with different virus mutants, stimulated the cells with IL-1β at 15 hpi, and determined IL-1R-dependent IκBα degradation by immunoblot analysis. In accordance with previous results, IL-1β-induced IκBα degradation was blocked in MCMV-infected cells expressing full-length M45 or either of the N-terminally truncated M45 mutants Nt2 and Nt3, which lack the first 280 and 350 amino acids (aa), respectively (data not shown). The C-terminally truncated mutants Ct6, Ct5, and Ct4 were also capable of blocking IL-1β-induced IκBα degradation, but Ct3, which lacks the last 53 aa, was not (data not shown).

FIG 5.

Requirement of the M45 C terminus for interaction with RIP1 and NEMO and for infection-induced NF-κB activation. (A) Schematic representation of M45 mutants incorporated into the MCMV genome. M45 consists of a ribonucleotide reductase (RNR) R1 homology domain and a unique N-terminal domain containing a RHIM (61). (B) NIH 3T3 cells were infected with the indicated MCMV mutants at an MOI of 5 TCID₅₀/cell. Cells were harvested at 15 hpi and were subjected to immunoprecipitation (IP) using an anti-HA antibody. Whole-cell lysates (WCL) and immunoprecipitates were analyzed by immunoblotting. (C) NIH 3T3 cells were infected with mutant MCMVs at an MOI of 10 TCID₅₀/cell. Cells were harvested at the indicated times postinfection, and protein levels were analyzed by immunoblotting.

Next, we determined the abilities of the different M45 truncation mutants to interact with endogenous RIP1 and NEMO in infected NIH 3T3 cells by coimmunoprecipitation experiments. We found that all M45 mutants capable of inhibiting IL-1β-induced IκBα degradation were also capable of interacting with RIP1 and NEMO (Fig. 5B). The M45 Nt3b and Nt3c proteins were repeatedly found only at low levels in infected cells. Therefore, we cannot conclude whether or not they can interact with RIP1 and NEMO and inhibit IL-1β-induced IκBα degradation.

Finally, we tested which of the M45 mutant viruses were capable of inducing the initial NF-κB activation. To do this, we analyzed IκBα degradation during the course of infection. As shown in Fig. 5C, the amino acids up to aa 350 from the N terminus of M45 (Nt3) were dispensable for IκBα degradation at 1 to 5 hpi. Surprisingly, of all the C-terminally truncated M45 mutants, only Ct6 (lacking the last 7 aa) was capable of activating NF-κB immediately after infection. Ct5 and Ct4, which interacted with both RIP1 and NEMO and were also functional in all other assays tested, failed to activate NF-κB during infection (Fig. 5C). In contrast to the other M45 functions tested, which tolerated the removal of at least 37 aa of the C terminus, the ability of M45 to activate NF-κB was lost when 19 or more aa of the C terminus were deleted.

MCMVs expressing full-length M45 or the M45 truncation mutant Nt2 or Ct3 were also used to analyze the role of M45 in the regulation of NF-κB-induced genes. For highly sensitive detection of changes in gene expression, we employed 4-thiouridine (4sU) labeling of newly transcribed RNA (30). NIH 3T3 cells were infected with wt MCMV, MCMV-Nt2, or MCMV-Ct3. Newly transcribed RNA was labeled for 1 h at different times during the first 24 hpi, purified, and subjected to quantitative RT-PCR analysis as described previously (30). While wt MCMV and the Nt2 mutant induced the well-characterized NF-κB-inducible genes Nfkbia and Tnfaip3 to similar extents, induction of both genes was completely lost in cells infected with the Ct3 mutant virus (Fig. 6). For Nfkbia, infection with the Ct3 mutant even resulted in a transient drop in expression levels between 1 and 6 hpi. This was not seen for Tnfaip3. Of note, infection with these three viruses resulted in similar levels of IE1 expression, thus excluding gross differences in the multiplicity of infection. Interestingly, induction of the interferon-inducible gene Cxcl10 was also slightly impaired by the Ct3 mutant. This result indicated that NF-κB might also play a role in Cxcl10 expression, a suggestion consistent with a recent report by Brownell and colleagues (52).

FIG 6.

M45 induces NF-κB-dependent transcription. NIH 3T3 cells were infected with wt MCMV (dark shaded bars) or with MCMV expressing the M45 mutant Nt2 (light shaded bars) or Ct3 (open bars) at an MOI of 10. Newly transcribed RNA was labeled for 1 h with 200 μM 4-thiouridine at the indicated times of infection. The expression levels of two NF-κB-inducible genes, Nfkbia and Tnfaip3, the interferon-inducible gene Cxcl10, and the viral gene ie1 were quantified in purified, newly transcribed RNA by qRT-PCR. Combined data from two independent experiments normalized to the values for the housekeeping gene Lbr are shown. uninf., uninfected; d.l., detection limit.

Virion-associated M45 mediates rapid NF-κB activation.

M45 is an MCMV gene expressed with early kinetics (45). Its expression can be detected by immunoblotting between 3 and 5 hpi after high-MOI infection (Fig. 1 and 3). However, M45 is also packaged into viral particles, presumably as a tegument protein (45), and is delivered to cells at the time of infection. Since M45-induced NF-κB activation starts as early as 1 hpi, we hypothesized that the virion-associated M45 might initiate NF-κB activation before de novo M45 expression starts. First, we wanted to test whether or not the incorporation of M45 mutants into virions correlates with the ability to induce NF-κB activation. To do this, we determined the presence of full-length and mutant M45 proteins in purified virion preparations. Virions were purified by glycerol tartrate gradient centrifugation, and the presence of M45 and control proteins was analyzed by immunoblotting. Full-length M45 and the Nt3 and Ct6 mutants were readily detectable in purified virus preparations. In contrast, the Ct5, Ct4, and Ct3 mutants were barely detectable in purified virions (Fig. 7A). Hence, the C terminus of M45 is required for efficient incorporation into virions.

FIG 7.

Virion-associated M45 mediates rapid NF-κB activation after infection. (A) Gradient-purified virions of the indicated MCMV mutants and whole-cell lysates (WCL) of infected cells were analyzed by immunoblotting. HA-tagged M45 proteins were detected using an anti-HA antibody. Note that WCL contain both the gB precursor and cleaved gB, whereas virions contain predominantly the mature (cleaved) form. (B) NIH 3T3 cells were infected with the revertant virus RevM45, UV-inactivated RevM45, the ΔM45 mutant, the ΔM45 mutant grown on M45-expressing NIH 3T3 cells (ΔM45comp), or the MCMV Ct3, Ct4, Ct5, or Ct6 mutant at an MOI of 10 TCID₅₀/cell. Cells were harvested at the indicated times postinfection, and levels of phosphorylated IκBα (p-IκBα), HA-tagged M45, and IE1 were analyzed by immunoblotting. (C) NIH 3T3 cells were infected with the indicated MCMV mutants at an MOI of 30 TCID₅₀/cell. Cells were harvested at 1 hpi and were subjected to immunoprecipitation (IP) using an anti-HA antibody. Whole-cell lysates (WCL) and immunoprecipitates were analyzed by immunoblotting.

Incorporation into MCMV virions coincided with the abilities of the respective M45 truncation mutants to trigger rapid NF-κB activation. In order to test whether virion-associated M45 is required for NF-κB activation immediately after infection, we wanted to use a virus that contains M45 as a virion component but does not express M45. Two approaches were used: (i) UV-inactivated MCMV and (ii) MCMV ΔM45 propagated on M45-expressing NIH 3T3 cells. Fibroblasts were infected at an MOI of 10, and phosphorylated IκBα was detected by immunoblotting as a sensitive measure of NF-κB activation. Successful UV inactivation was verified by testing for IE1 expression. While virtually no IκBα phosphorylation was detected in ΔM45 mutant-infected cells within the first 3 hpi, infection with UV-inactivated RevM45 still resulted in robust phosphorylation of IκBα. Similarly, the M45-complemented ΔM45 virus also induced IκBα phosphorylation, although to a lesser degree than RevM45, probably because the complemented virions delivered smaller quantities of M45 to infected cells (Fig. 7B). We also tested the virus mutants expressing Ct6, Ct5, Ct4, or Ct3 for their abilities to induce the phosphorylation of IκBα within the first 3 hpi. As expected, only Ct6 was able to induce IκBα phosphorylation. Ct6 was also the only C-terminally truncated M45 protein delivered in substantial amounts to infected cells (Fig. 7B). Taken together, these results demonstrated that virion-associated M45 mediates the rapid NF-κB activation following MCMV infection.

We wondered whether virion-associated M45 differs from de novo-expressed M45 in its ability to interact with RIP1 and NEMO. We thus performed coimmunoprecipitation in infected NIH 3T3 cells at 1 hpi, when only virion-associated M45 is present. All virion-associated M45 mutants tested were capable of interacting with RIP1 and NEMO (Fig. 7C).

Last, we tested whether M45 can activate NF-κB independently of an MCMV infection, i.e., in the absence of other viral proteins. Plasmid expression vectors encoding M45 mutants were transfected into NIH 3T3 cells together with an NF-κB reporter plasmid. As shown in Fig. 8, full-length M45 induced moderate reporter gene expression similar to the level obtained by overexpression of MyD88 or IKKβ, which served as positive controls. NF-κB was also activated by all M45 mutants capable of interacting with RIP1 and NEMO, including Ct5 and Ct4, which did not activate NF-κB upon MCMV infection. Thus, aa 1138 through 1167 of M45 (present in Ct6 but not in Ct4) are not necessary for NF-κB activation but are required for the incorporation of M45 into virions, the delivery of M45 to infected cells, and rapid NF-κB activation immediately after infection.

FIG 8.

M45 induces NF-κB-dependent reporter gene expression. NIH 3T3 cells were cotransfected with an NF-κB-inducible SEAP reporter plasmid (pNiFty2-SEAP) and expression plasmids encoding the indicated proteins. SEAP activity in the supernatants of transfected cells was quantified by a colorimetric assay. Bars represent the means of results from three parallel experiments; error bars, standard errors of the means. Values were normalized to that for green fluorescent protein (GFP), which was used as a negative control.

DISCUSSION

Previous studies have shown that the HCMV envelope glycoproteins gB and gH are involved in NF-κB activation immediately after HCMV infection of fibroblasts (10, 11). HCMV gB and gH can interact with TLR2 and activate TLR2-dependent NF-κB activation (12, 14). Based on these results, it would be reasonable to assume that MCMV also activates NF-κB by gB and/or gH in a TLR2-dependent manner. However, our results indicate that TLR2 is not required for MCMV-induced NF-κB activation in fibroblasts (Fig. 3A) and that M45 (rather than gB or gH) is the decisive virion component. However, we cannot exclude the possibility that the rapid NF-κB activation requires both M45 and gB or gH. This possibility is difficult to exclude because both gB and gH are proteins essential for MCMV infectivity. However, the fact that M45 can activate NF-κB in the absence of other MCMV proteins (Fig. 8) suggests that M45 is sufficient for infection-induced NF-κB activation.

The finding that the same protein that blocks all canonical NF-κB pathways at later times is also responsible for early NF-κB activation came as a great surprise. How can the same protein first activate and then inhibit NF-κB? One could speculate that virion-associated M45 might differ from de novo-synthesized M45, for instance, by different posttranslational modifications. So far, little is known about posttranslational modifications of M45 apart from the fact that M45 is subject to proteolytic cleavage between aa 277 and 278 (45) and to ubiquitination (19). However, the observation that M45 also activates NF-κB when expressed by a plasmid vector (Fig. 8) suggests that modifications specific for virion-associated M45 are not required. Alternatively, the amount of M45 present in the infected cell could determine whether M45 activates or inhibits NF-κB. A similar quantity-dependent mechanism of NF-κB activation has been reported for the molluscum contagiosum virus protein MC159 (53), even though this protein is generally thought to function as an inhibitor of NF-κB signaling (54). By interacting with RIP1 and NEMO, the low levels of incoming M45 might enforce an interaction between RIP1 and NEMO and/or induce NEMO oligomerization. Both processes have been reported to result in NF-κB activation (55, 56). Later on in infection, when M45 is expressed at higher levels, it targets NEMO for degradation via the autophagosome-lysosome pathway, and this process appears to involve the formation of aggregates (20).

The central region of M45 between aa 351 and 1137 is sufficient for interaction with RIP1 and NEMO (Fig. 5B). This region is also necessary, but not sufficient, for NF-κB activation immediately after infection: M45 mutants Ct4 and Ct5 contain this region but do not induce IκBα phosphorylation and degradation upon infection (Fig. 5C and 7B). Thus, the region between aa 1138 and 1167 mediates an additional function required for rapid NF-κB activation after infection, namely, the incorporation of M45 protein into viral particles.

Apart from RIP1 and NEMO, M45 is also known to interact with RIP3, DAI, and possibly TRIF (21, 28). By infecting RIP3^−/− and TRIF^−/− MEFs, we could exclude the possibility that these molecules are necessary for rapid NF-κB activation (data not shown). We have not been able to formally rule out a role for DAI, since MEFs derived from DAI knockout mice were not available to us. However, it has been shown that NIH 3T3 cells do not express DAI at detectable levels (25), but these cells activate NF-κB in response to MCMV infection (Fig. 1B). Moreover, the M45–DAI interaction was reported to be RHIM dependent (21, 57), and MCMVs expressing RHIM-deficient M45 can activate NF-κB (Fig. 5C and 6). Thus, DAI is unlikely to play an important role in M45-mediated NF-κB activation.

While it is well established that HCMV and MCMV induce a transient phase of NF-κB activation immediately after infection, the importance of this rapid NF-κB activation for CMV replication and pathogenesis remains to be investigated. Because the major immediate early enhancers of HCMV and MCMV contain several NF-κB binding sites, it has been proposed that NF-κB can promote IE gene expression and stimulate the lytic replication cycle. A few experimental studies with HCMV do, in fact, support this hypothesis (58, 59). On the other hand, an MCMV “enhancer swap” mutant harboring the HCMV major immediate early enhancer with all NF-κB response elements mutated replicated as well as its control virus in NIH 3T3 fibroblasts (60). This result suggested that NF-κB is not important for MCMV gene expression and replication in cultured fibroblasts. However, it remains possible that NF-κB is important for full activation of the major immediate early promoter in other cell types. Moreover, limited NF-κB activation might also have other benefits for the virus: for instance, NF-κB-induced expression of cellular antiapoptotic proteins could increase cell survival, and expression of chemokines could attract cells important for virus dissemination. Now that a small region necessary for the packaging of M45 into virions has been identified, it might be possible to construct a mutant virus that selectively lacks the ability to package M45 into viral particles while leaving all other M45 functions intact. With such a virus, the biological importance of the rapid NF-κB activation could be analyzed in more-complex cell culture systems and in the mouse.