Potential conflict of interest: Nothing to report.
Human MxA, an interferon-inducible cytoplasmic dynamin-like GTPase, possesses antiviral activity against multiple RNA viruses. Recently, MxA has also been demonstrated to have activity against the hepatitis B virus (HBV), a well-known DNA virus responsible for acute and chronic liver disease in humans. We investigated the molecular mechanism for the anti-HBV activity of MxA. Our results demonstrated that in HepG2.2.15 cells, MxA GTPase independently suppressed the production of hepatitis B surface antigen and HBV DNA without changing the level of hepatitis B core antigen (HBcAg) and the distribution of HBV mRNA. MxA significantly reduced the level of the encapsidated pregenomic RNA. Through its central interactive domain, MxA interacted with HBcAg, causing accumulation of the proteins in perinuclear compartments. MxA-HBcAg interaction significantly affected the dynamics of HBcAg by immobilizing HBcAg in the perinuclear structures. Conclusion: MxA displays antiviral activity against HBV involving a mechanism of MxA-HBcAg interaction that may interfere with core particle formation. (HEPATOLOGY 2012;56:803–811)
Interferon (IFN)-inducible myxovirus resistance gene 1 (Mx1) is one of the best-studied genes of innate immunity to viral infection. Mx1 is expressed in almost all vertebrate species and exhibits wide antiviral activity. In humans, MxA, one of the two Mx proteins expressed in the cytoplasm in multiple cell types, has intrinsic antiviral properties,1 and serves as a major mediator of the antiviral action of type 1 (α/β) IFN.2 MxA belongs to a group of large GTP-binding proteins,3 and a common and notable feature of these proteins is their ability to self-assemble into a highly ordered oligomer that is associated with their function in the regulation of intracellular protein trafficking.4
To date, data from numerous studies have indicated a strong activity of MxA against RNA viruses.1, 5 Although the mechanisms by which MxA inhibits such a variety of viruses are yet to be precisely defined, observations from many groups appear to point to the conclusion that MxA obtains its antiviral effect by targeting the nucleoprotein components. As a consequence, these viral components may be trapped and sorted to locations where they become unavailable for either the transcription of the viral genome or the assembly of new virus particles.6, 7 The requirement of the oligomerization and guanosine triphosphatase (GTPase) activity of MxA for its antiviral function seems to be controversial, although functional analysis has suggested a critical role of the GTPase effector domain in its GTPase activity, oligomer formation, and antiviral activity.8, 9 Recently, a study based on the crystal structure of the stalk of MxA suggested that the oligomerization of MxA via the stalk region is not a prerequisite for its GTPase hydrolysis, but is essential for recognition of viral structure and antiviral function.10
In addition to RNA viruses, MxA has recently been found to provide resistance against DNA viruses, including hepatitis B virus (HBV).11, 12 Primary analysis indicates that the anti-HBV effect of MxA is mediated by inhibition of the nucleocytoplasmic transport of viral mRNA11 and is independent of GTPase activity.13 However, the events underlying how cytoplasmic MxA prevents the nuclear export of HBV mRNA is unclear. On the other hand, reports have demonstrated that HBV significantly down-regulates MxA expression, and this involves a role of hepatitis B core antigen (HBcAg) by interacting with the MxA promoter,14 making the interaction between HBV and MxA more complicated than had been predicted. Considering that HBV is one of the major causes of acute and chronic hepatitis (particularly in East Asia and central Africa, where some 10% of the population are HBV carriers, many of whom die from liver cirrhosis and hepatocellular carcinoma),15 it is therefore important to further elucidate the mechanisms underlying the anti-HBV activity of MxA, which may contribute to our understanding of the interaction between HBV and MxA, one of the major mediators of IFN function.
In this study, we verified the inhibitory effect of MxA on HBV replication in HepG2.2.15 cells. We provide evidence that the anti-HBV function of MxA is mediated by an interaction between MxA and HBcAg, the core protein of HBV. Through its central interactive domain (CID), MxA traps HBcAg in the perinuclear MxA-HBcAg complexes, and this interferes with HBV core particle formation.
ASFV, African swine fever virus; BFA, brefeldin A; ER, endoplasmic reticulum; GFP, green fluorescent protein; GTPase, guanosine triphosphatase; FLIP, fluorescence loss in photobleaching; FRAP, fluorescence recovery after photobleaching; FRET, fluorescence resonance energy transfer; HBcAg, hepatitis B core antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; IFN, interferon; PCR, polymerase chain reaction; pgRNA, pregenomic RNA; RC-DNA, relaxed circular DNA.
Materials and Methods
Cell Cultures, Reagents, and Antibodies.
HepG2.2.15 cells were grown in Roswell Park Memorial Institute 1640 at 37°C under an atmosphere of 5% CO2. HuH7 cells and Vero cells were maintained in Dulbecco's modified Eagle's medium. IFN-α2B was obtained from PeproTech (Rocky Hill, NJ), brefeldin A was obtained from Epicentre Technologies (Madison, WI), and nocodazole was obtained from Sigma (St. Louis, MO). The following antibodies were used: anti-Flag (Santa Cruz Biotechnology, Santa Cruz, CA), anti–green fluorescent protein (GFP) (Cell Signaling, Danvers, MA), monoclonal anti-HBcAg (Millipore, Billerica, MA), polyclonal anti-HBcAg (Dako, Carpinteria, CA), anti-MxA (Proteintech, Chicago, IL), anti-GM130 (BD Biosciences, San Jose, CA), anti-p58, and anti-α-tubulin (Sigma).
Vector Construction and Transfection.
All vectors used are described in the Supporting Materials and Methods. Transient transfections were performed using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according to an optimized protocol.
Intracellular HBV DNA Extraction and Southern Blot Analysis.
Intracellular HBV DNA was isolated as described16 with modifications. Briefly, cells were lysed and the nuclei were removed by centrifugation. The cytoplasmic DNA was then extracted from the supernatants with a Cell DNA Extraction Kit (Bioteke, Beijing, China) and analyzed via Southern blotting as described in the Supporting Materials and Methods.
Immunoprecipitation and Western Blotting.
Cells were lysed with Triton X-100 lysis buffer containing protease inhibitors. After centrifugation, the supernatants were incubated with antibody overnight and then Protein A agarose for 2 hours at 4°C. Immunocomplexes were washed and analyzed via western blotting as described.17
Confocal Imaging, Live-Cell Imaging, and Photobleaching.
Images were collected with a 63 × 1.4 NA objective using appropriate laser excitation on a confocal microscope. For quantification of fluorescence intensity, nonsaturated images were taken with a fully open pinhole, whereas nonquantitative images were obtained with a pinhole diameter equivalent to 1-2.5 Airy units. Live-cell imaging and photobleaching were performed as described in the Supporting Materials and Methods.
MxA Suppresses HBV Replication in HepG2.2.15 Cells.
To investigate the anti-HBV function of MxA, we used HepG2.2.15 cells, an HBV-replicating cell line carrying HBV DNA and stably secreting surface antigen particles, nucleocapsids, and virions. The cells were transfected with either wild-type MxA or MxAK83A, a mutant that lacks GTP-binding ability but exhibits extrinsic GTP hydrolytic activity,18 or MxAL612K, a mutant in which the intrinsic or extrinsic GTP hydrolytic activity is abolished but the GTP binding activity is retained.9 To elevate the transfection efficiency, we used suspended instead of dish-attached cells with double amounts of plasmid and Lipofectamine according to an optimized protocol. By this method, ≈70%-80% of the total cells were confirmed to be transfected (Supporting Fig. 1). We first determined the hepatitis B surface antigen (HBsAg) level in the extracellular culture medium 24 hours after transfection. Expression of MxA dramatically lowered the HBsAg level, by 90% of the control that was transfected with an empty pcDNA3.1-Flag vector (Fig. 1A). Likewise, both of the GTPase-defective mutants demonstrated a comparable inhibitory effect on HBsAg secretion (Fig. 1A). To determine that the reduction in HBsAg was associated with a change in HBV replication, the encapsulated viral DNA in the culture medium was measured by quantitative real-time polymerase chain reaction (PCR). In accord with the reduction in HBsAg, expression of wild-type MxA or each of the two mutants significantly decreased extracellular HBV DNA by equivalent levels (Fig. 1B).
To further analyze the effect of MxA on HBV replication, we measured HBV relaxed circular DNA (RC-DNA) in the cytoplasm via Southern blot analysis. We found that expression of MxA or each of the mutants significantly lowered the cytoplasmic RC-DNA together with the double-stranded DNA (Fig. 1C), indicating an inhibition of the replication intermediates by the proteins. Because the HBV DNA replication intermediates originate from reverse transcription of HBV pregenomic RNA (pgRNA), we then examined the HBV pgRNA in the nucleocapsids. In HepG2.2.15 cells 24 hours after transfection, the cytoplasmic capsids were precipitated and the encapsidated pgRNA was extracted and determined via real-time PCR. We found that expression of MxA or MxAK83A, or MxAL612K, significantly decreased the cytoplasmic encapsidated pgRNA level (Fig. 1D).
A previous study has shown an inhibitory effect of MxA on the nucleocytoplasmic export of HBV mRNA.11 We therefore checked the expression and the cytoplasmic/nuclear distribution of intracellular HBV RNAs in HepG2.2.15 cells. Results from real-time PCR demonstrated that neither the total RNAs nor its intracellular distribution was altered by MxA or each of the two mutants, as measured at 24 hours after transfection (Supporting Fig. 2A,B). Consistently, results of Western blot showed that the intracellular HBcAg protein level was not remarkably influenced (Supporting Fig. 2C).
Taken together, these results suggest that in HepG2.2.15 cells, MxA GTPase activity independently inhibits HBV replication without altering the cytoplasmic/nuclear HBV mRNA distribution and HBcAg level, at least in the early stage of MxA expression.
MxA Interacts with HBcAg.
To investigate the mechanism underlying the anti-HBV effect of MxA, we observed the location of HBcAg, the core protein of HBV, in hepatoma cells expressing HBV plasmid and CFP-tagged MxA. Immunofluorescence images revealed that in HBV-transfected Huh7 cells without CFP-MxA expression, HBcAg was spread throughout the cytoplasm and the nucleus in a small punctate pattern (data not shown). Strikingly, in CFP-MxA–overexpressing cells, HBcAg colocalized with CFP-MxA to generate large granular structures in the cytoplasm (Fig. 2A). To further verify the colocalization of MxA and HBcAg, we overexpressed the two proteins in a nonhepatoma cell type. In living Vero cells, YFP-HBcAg accumulated in the perinuclear region, overlapping with either the wild-type or the mutant CFP-MxA (Fig. 2B).
Colocalization of MxA with HBcAg prompted us to look for evidence of their possible interaction. First, we determined whether MxA and HBcAg could undergo coprecipitation. In Huh7 cells expressing Flag-MxA and YFP-HBcAg, immunoprecipitation of HBcAg using GFP antibody resulted in coprecipitation of Flag-MxA (Fig. 2C). The formation of the MxA-HBcAg complex was not dependent on the GTPase activity of MxA, because the Flag-L612K and Flag-K83A mutants of MxA were coimmunoprecipitated with YFP-HBcAg to a similar degree. We then tested whether exogenous HBcAg could interact with endogenous MxA, or exogenous MxA with endogenous HBcAg, because a previous study showed that in HBV-expressing HepG2.2.15 cells, IFN is unable to induce MxA expression.14 By treatment of Huh7 cells expressing Flag-HBcAg with IFNα, or transfection of HepG2.2.15 cells with Flag-MxA, we found that Flag-HBcAg coprecipitated IFNα-induced MxA (Fig. 2D), while Flag-MxA coprecipitated endogenous HBcAg (Fig. 2E), indicating a specific interaction between HBcAg and MxA.
Finally, we performed fluorescence resonance energy transfer (FRET) experiments, which detect the proximity of interacting proteins. In living Vero cells expressing CFP-MxA and YFP-HBcAg, the proteins were first confirmed to colocalize to perinuclear structures. Photobleaching YFP-HBcAg, the energy acceptor, significantly enhanced the fluorescence intensity of CFP-MxA, the energy donor, in the perinuclear region (Fig. 2F). Quantification of the fluorescence of CFP-MxA in this region demonstrated a more than 1.22-fold increase after YFP-HBcAg photobleaching, whereas no increase was found in the two control groups YFP/CFP-MxA and YFP-HBcAg/CFP (data not shown), indicating an evident energy transfer between the two proteins in the perinuclear compartment. Taken together, our data suggest that MxA interacts with HBcAg in living animal cells.
MxA-HBcAg Interaction Is Mediated by the Central Interactive Region of MxA.
To further dissect the biochemical properties of MxA-HBcAg interaction and determine the relevance of the interaction to the anti-HBV activity of MxA, we created different truncated mutants of MxA (Fig. 3A) and tested their association with HBcAg. Huh7 cells were transfected with Flag-HBcAg and Myc-tagged full-length MxA or each of the truncated mutants, and the associations were checked by coimmunoprecipitation. We found that MxA deletion mutants either lacking the N-terminal GTP-binding domain, which contains the self-assembly sequence (MxAΔN, 359-662aa), or lacking the C-terminal leucine zipper region (MxAΔC, 1-574 aa), retained the ability to interact with HBcAg as demonstrated by coprecipitation with Flag-HBcAg (Fig. 3B). Interestingly, a MxA deletion mutant lacking the central interactive region (MxAΔCID) was not precipitated by Flag-HBcAg, indicating an essential role of this domain in mediating the MxA-HBcAg association (Fig. 3B).
We also coexpressed YFP-HBcAg and CFP-tagged each of the truncated mutants to observe the formation of the protein aggregates. We found that, well-correlated with the results of immunoprecipitation, CFP-MxAΔC and CFP-MxAΔN, but not the CFP-MxAΔCID, colocalized with YFP-HBcAg to form large perinuclear aggregates (Fig. 3C), indicating morphologically a requirement for the CID domain in the generation of MxA-HBcAg complexes.
Finally, we assessed the effects of the truncated MxA mutants on HBV replication by measuring the encapsulated viral DNA in the culture medium of HepG2.2.15 cells. Clearly, overexpression of either Myc-MxAΔC or Myc-MxAΔN dramatically decreased HBV DNA level, mimicking that of wild-type Myc-MxA. In contrast, no evident suppression was detected in cells expressing Myc-MxAΔCID (Fig. 3D). Therefore, our results suggest that the CID domain of MxA is the responsive region in mediating the interaction with HBcAg, and MxA-HBcAg interaction is essential to the anti-HBV function of MxA.
MxA Immobilizes HBcAg in the Perinuclear Compartment.
Given that MxA interacts with HBcAg to form a complex in the perinuclear compartment, and this interaction is required for the anti-HBV activity of MxA, we then aimed at investigating the effect of MxA on the intracellular kinetics of HBcAg. To address this, we performed fluorescence recovery after photobleaching (FRAP) in living cells. In Vero cells expressing YFP-HBcAg, a cytoplasmic region was photobleached and the extent of recovery of fluorescence in the region was monitored over time. Upon photobleaching, we observed rapid recovery from the nonbleached pool, with the original prebleached fraction of fluorescence clearly restored in less than 100 seconds (Fig. 4A,E). In contrast, in YFP-HBcAg and CFP-MxA cotransfected cells, the fluorescence recovery of YFP-HBcAg aggregated with CFP-MxA in the perinuclear region was dramatically slowed, with a great proportion of the initial YFP-HBcAg fluorescence in the region failing to recover, even a much longer time after photobleaching (Fig. 4B,E), suggesting this proportion of HBcAg was irreversibly bound to the compartment.
The irreversible binding of HBcAg to the perinuclear pool was further confirmed by fluorescence loss in photobleaching (FLIP). YFP-HBcAg fluorescence in a small cytoplasmic region of the cell expressing only YFP-HBcAg was bleached repetitively. After about 200 seconds, the fluorescence signals were completely lost in the areas outside the region, indicating that YFP-HBcAg diffused between the bleached and unbleached areas (Fig. 4C,F). In cells coexpressing YFP-HBcAg and CFP-MxA, repetitive photobleaching of a similar cytoplasmic pool failed to cause the loss of YFP-HBcAg from the perinuclear area in which it colocalized with CFP-MxA, indicating fixation of YFP-HBcAg in that area (Fig. 4D,F). These results indicate that the formation of MxA-HBcAg complexes is able to immobilize HBcAg in the perinuclear area.
Characterization of the Perinuclear Compartment.
Interaction of MxA and viral nucleocapsid protein to generate complexes in the perinuclear compartment has been reported in many types of viral infection.7, 19 However, so far, the compartment has not been well-characterized. In an attempt to determine the essence of the perinuclear compartment where the MxA-HBcAg complex forms, we tested whether the MxA-HBcAg aggregates overlapped with the Golgi apparatus, the endoplasmic reticulum (ER), or the ER-Golgi intermediates, using distinct organelle markers. Vero cells expressing CFP-MxA and YFP-HBcAg were either transfected with RFP-tagged ssKDEL, or stained by GM130 or p58 antibody. We found that the perinuclear MxA-HBcAg complexes colocalized with neither the ssKDEL-RFP used as an ER marker, nor the GM130, a Golgi matrix protein, nor the p58 as a marker of ER-Golgi intermediates (Fig. 5A).
We also treated the cells with brefeldin A (BFA), a drug that disassembles the Golgi structure by inhibition of early steps in ER-Golgi transport.20 In the presence of BFA, when RFP-tagged galactosyltransferase, a Golgi-resident enzyme, redistributed to the ER, and the endogenous GM130 disassociated from the Golgi membranes (Fig. 5B), the MxA-HBcAg aggregation remained in the perinuclear region without notable changes (Fig. 5B), suggesting that the Golgi and the ER-Golgi intermediates are not involved in the formation of the aggregation.
To further characterize the perinuclear compartment, we disrupted microtubules by treating the cells with nocodazole21 or by putting the cells on ice.22 Destruction of microtubules by these methods caused dramatic disassembly of the perinuclear compartment into spot-like structures throughout the cell (Fig. 5C). This observation suggests that the generation and maintenance of the compartment are microtubule-dependent.
As an IFN-inducible cytoplasmic protein, the effect of MxA on DNA virus replication has just recently been recognized, and the underlying mechanisms have not been fully elucidated. In this study, we verified the anti-HBV effect of MxA in HepG2.2.15 cells. Our results suggest that MxA inhibits HBV replication by a direct interaction with the HBV core protein HBcAg via its CID domain, causing the immobilization of HBcAg and subsequently the loss of capsid assembly.
Interaction with viral nucleoprotein is the most likely common pathway for MxA to perform its antiviral function against RNA viruses. Nevertheless, in the case of HBV, it has been shown that MxA suppression of HBV replication involves inhibition of the export of viral mRNA from the nucleus to the cytoplasm via the PRE sequence.11 However, results from recent studies indicate that this might not be the case. Expression of two nuclear forms of the wild-type only slightly decreases the expression of extra- and intracellular HBV DNA in HepG2 cells, indicating that MxA has only a minimal effect on the replicative cycle of HBV in the nucleus.13 In HBV and HBV/MxA transgenic mice lacking functional IFN receptors, while MxA evidently inhibits HBV, the cytoplasmic HBV RNA level is not dramatically changed.12 In Vero cells, MxA inhibition of the replication of African swine fever virus (ASFV), a large double-stranded DNA virus, involves recruitment of MxA to perinuclear viral assembly sites,19 implying an interaction between ASFV and MxA. Using biochemical and fluorescence imaging techniques, we here identified an MxA-HBcAg interaction and its necessity for the anti-HBV activity of MxA, suggesting a mechanism common to that in RNA viruses and ASFV. Our results contrast with the results of Kremsdorf and colleagues in which a lack of MxA-HBcAg interaction was indicated.11 The major cause for the differences in results and interpretation could be the experimental conditions. Instead of a cosedimentation assay using purified HBcAg, we performed immunoprecipitation in cells coexpressing the proteins. This may facilitate the encounter efficiency of the proteins by positioning them in a relatively physiological condition without losing possible unknown modifications required for their interaction. Identification of the interaction domain together with the colocalization of the proteins and FRET in living cells further support our conclusion.
Our results showing an MxA interaction with transfected HBcAg suggest that this interaction is independent of additional HBV viral components, further supporting a direct association between MxA and HBcAg. In addition to revealing the interaction, we identified here the region in MxA responsible for the interaction. Our results suggest that the CID domain, which is required for MxA self-assembly,23 is also essential for MxA to interact with HBcAg. The MxAL612K and the MxAΔC mutants, which are unable to self-assemble, retain the ability to interact with HBcAg, suggesting that the self-assembly of MxA is not required for the recruitment of HBcAg. This supports the current model in which high molecular weight MxA oligomers are a storage form, whereas MxA monomers are the active form of MxA,24 at least in terms of its anti-HBV action.
Using distinct intracellular membrane structural markers, we identified the large perinuclear complexes in which MxA and HBcAg aggregate. It is reasonable to speculate that the MxA sequesters the viral nucleocapsid protein to form complexes at sites where either MxA assembles or the viral particles form. Recently, it has been found that MxA self-assembles into rings and associates with the smooth ER.25 On the other hand, the envelopment and budding of the mature capsids of HBV enclosed with HBcAg also occurs in the ER.26 Nevertheless, our colocalization and BFA experiments clearly excluded association of the large MxA-HBcAg complexes with either the ER or the Golgi apparatus. Rather, our results showed that the perinuclear location of the complexes is dependent on the stability of microtubules. The dependence on microtubules supports the recently proposed concept of aggresomes,27 implying that either HBV capsids assemble in aggresomes or MxA takes the HBcAg to the aggresomes for degradation.
It has been proposed that association of MxA with viral nucleoproteins may hijack the nucleoproteins, preventing them from transcription of the viral genome or the assembly of new viral particles; however, so far, no direct evidence has been provided. Real-time imaging and photobleaching techniques allowed us to investigate the mobility of nucleoprotein in living cells. Our data indicate that the formation of MxA-HBcAg complexes immobilizes the HBcAg. Although this mechanism may be involved in both the inhibition of nucleocapsid assembly and the enveloping of viral nucleocapsids, our data suggest that MxA-HBcAg interaction interferes in the early stage of core particle formation, based on the decrease in cytoplasmic encapsidated pgRNA and the RC-DNA. Our data is consistent with previous studies showing that IFN prevents the formation of replication-competent HBV capsids.28, 29
Although the anti-HBV activity of MxA has been defined, in view of the antagonistic effects of HBcAg on the antiviral activity of MxA and the lack of global efficiency of IFNα in clinical treatment, the discovery of methods that either strengthen the trapping of HBcAg or disrupt the binding of HBcAg to the MxA promoter is a practical strategy. In this context, the findings of our present study suggest that small molecules based on the MxA CID domain may be a promising choice.