Abstract
Myxovirus resistance protein 1 (MX1) and MX2, are homologous, dynamin-like large GTPases, induced upon interferon (IFN) exposure. Human MX1 (HsMX1) is known to inhibit many viruses, including influenza A virus (IAV), by likely acting at various steps of their life cycles. Despite decades of studies, the mechanism(s) of action with which MX1 proteins manage to inhibit target viruses is not fully understood. MX1 proteins are mechano-enzymes and share a similar organization to dynamin, with an amino-terminal GTPase domain and a carboxy-terminal stalk domain, connected by a Bundle Signalling Element (BSE). These three elements are known to be essential for antiviral activity. HsMX1 has two unstructured regions, the L4 loop, also essential for antiviral activity, and a short amino (N)-terminal region, which greatly varies between MX1 proteins of different species. The role of this N-terminal domain in antiviral activity is not known. Herein, using mutagenesis, imaging and biochemical approaches, we demonstrate that the N-terminal domain of HsMX1 is essential for antiviral activity against IAV and Vesicular Stomatitis Virus (VSV), and for the ability to aggregate Orthobunyavirus nucleoproteins. Furthermore, we pinpoint a highly conserved leucine within this region, which is absolutely crucial for human, mouse and bat MX1 protein antiviral activity. Importantly, mutation of this leucine does not compromise GTPase activity or oligomerization capabilities, but does modify MX1 protein subcellular localisation. The discovery of this essential and highly conserved residue defines this region as key and may reveal insights as to the mechanism of action of MX1 proteins.
Introduction
Influenza A virus (IAV) is a member of the Orthomyxoviridae family and the causative agent of the disease commonly known as the flu. Upon infection of target epithelial cells within the respiratory tract, IAV is sensed by pattern recognition receptors, including retinoic acid-inducible gene I (RIG-I), which induce a signalling cascade leading to the production and secretion of type 1 and type 3 interferons (IFNs). The IFNs act in a paracrine and autocrine manner and, through binding to their cognate receptors and activation of the Janus Kinase (JAK) / Signal transducer and activator of transcription (STAT) pathway, leads to the regulation of hundreds of IFN stimulated genes (ISGs). Among these ISGs, many antiviral restriction factors have been described (1). These factors establish a so-called antiviral state, powerfully limiting IAV replication (2). The study of the IFN response against IAV in mice led to the discovery of the Myxovirus resistance (MX) proteins, which were later identified in humans (3–6) . There are two homologous MX genes in humans, MX1 and MX2. Human MX1, also called MxA (and hereafter referred to as HsMX1), inhibits a wide range of RNA and DNA viruses, replicating either in the cytoplasm, such as the bunyavirus La Crosse Virus (LACV) and the rhabdovirus VSV, or in the nucleus, such as IAV (7). Human MX2 (HsMX2), or MxB, has notably been shown to potently inhibit Human Immunodeficiency Virus 1 (HIV-1) and Herpes viruses (8–13). Interestingly, mice also possess two Mx genes: MmMx1 and MmMx2. The latter is more closely related to HsMX1 than the former, but, interestingly, MmMx1 is a more potent inhibitor of IAV than HsMX1 (14). However, to our knowledge, MmMx1 only restricts orthomyxoviruses, whereas HsMX1 is broadly antiviral and MmMx2 inhibits VSV and Hantaan River Virus (HTNV) (7). This could be due partly to the fact that MmMx1 is mainly localized inside the nucleus (15, 16), contrary to HsMX1 which is cytosolic.
Dynamin-like GTPases all share a very similar general organization and among members of this family of proteins, can present almost superimposable three-dimensional (3D) crystal structures (17, 18). The 3D structure of HsMX1 has been partially solved (19, 20) (Fig. 1A). HsMX1 sports a globular head, which contains the GTPase module, and possesses a stalk domain, attached to the head by a tripartite Bundle Signalling Element (BSE), composed of three separate alpha-helices (α-helices) that allow the fold back of the stalk towards the GTPase domain, which is akin to dynamin (21) (Fig. 1A). The major difference between HsMX1 and dynamin is the lack in the former of the Pleckstrin Homology (PH) domain, essential for the PI(4,5)P2 binding capacity of dynamin (22). In place of the PH domain, HsMX1 has a flexible loop, termed loop L4, of which the structure is unresolved (Fig. 1A). Another unstructured loop, the L2 loop, is also found at the extremity of the stalk and in the vicinity of the L4 loop (Fig. 1A). In addition to these domains, HsMX1 and all MX proteins possess an amino-terminal (N-terminal) extension of unknown structure, which highly differs in length and sequence (Fig. 1A, 1B, Sup. Fig. 1). HsMX1 has been shown to homodimerize through the stalk domain and further oligomerizes through numerous other interfaces on the stalk and the GTPase domains (19, 20, 23). This may suggest the possible cohabitation of different types of HsMX1 oligomers within the cell. Of note, MmMx1 possesses a nuclear localization signal (NLS) localized in the third helix of the BSE (24) allowing transport into the nucleus as opposed to HsMX1 and MmMx2, which are uniquely cytoplasmic.
The detailed antiviral mechanism(s) of action of MX1 proteins remain largely misunderstood, although certain intrinsic antiviral determinants have been well characterized. Four essential determinants have been identified to this day, the first being the binding/hydrolysis of GTP by the GTPase domain (25). Interestingly, HsMX2 does not require a functional GTPase domain for HIV-1 inhibition (8, 9, 26) and in the case of HsMX1, this might also be true for Hepatitis B Virus (HBV) inhibition (27). The second determinant is the presence of an intact BSE (28), the third being the possibility to oligomerize via the stalk domain (19), and the fourth requirement involves the extremity of the stalk, loops L2 (29–31) and L4 (32–34). Indeed, deletions or point mutations of the L4 loop of HsMX1 and MmMx1 abrogated antiviral activity against IAV and Thogoto virus (THOV) (32–34) and this loop was elegantly shown to have been under positive selection during evolution (35). Taken together, these studies showed that a number of intrinsic elements are needed for MX1 protein antiviral activity, however, to our knowledge, no study has so far addressed the importance of the N-terminal domain of MX1 proteins. The N-terminal domain of HsMX2, however, which is longer than that of HsMX1 (91 amino acids compared to 43 amino acids) has been shown to be a crucial determinant for HIV-1 restriction (26, 36, 37). Transferring the N-terminal domain of HsMX2 onto HsMX1 resulted in a chimeric protein able to inhibit HIV-1, without losing the anti-IAV activity (26). These data prompt that the N-terminal domain of MX proteins may generally be important for antiviral activity.
In this study, we examined the role of the N-terminal domain of several MX1 proteins, showing that this domain is essential for antiviral activity. Indeed, deletion of the N-terminal region abrogates antiviral activity against IAV and this effect was mapped to a single essential residue, leucine 41 (L41) in HsMX1. This residue was also essential for inhibition of rhabdoviruses and bunyaviruses. We further show that this residue is highly conserved between MX1 proteins of different origins, and we show that the corresponding leucines in MmMx1 (leucine 7, L7) and little yellow-shouldered bat, Sturnira lilium, MX1 (SlMX1) (leucine 39, L39) are also essential for anti-IAV activity. Finally, we demonstrate that mutation of this highly conserved leucine did not seem to impact lower- or higher-order oligomerization status of HsMX1 or MmMx1 in cells, the propensity to hydrolyse GTP in vitro, or the natural structure of the first BSE helix in silico. However, we show that this residue is essential for correct subcellular localization of MX1 proteins. This study therefore confirms the complex multimodular nature of MX1 proteins and defines their N-terminal region as another important antiviral module governed by an essential and conserved leucine.
Materials and methods
Plasmid constructs
FLAG-tagged HsMX1 and MmMx1 pCAGGS constructs were previously described (26). The Sturnira lilium MX1 encoding plasmid was a gift from Prof Georg Kochs (38). All FLAG-tagged HsMX1, MmMx1 and SlMX1 mutants were generated by overlapping PCR. Coding DNA sequences (CDS) of interest were subcloned into pCAGGS using NotI and XhoI cloning sites. The pRRL.sin.cPPT.SFFV/IRES-puro.WPRE lentiviral vector system has been described previously (39). All lentiviral vectors used here were obtained by subcloning of the CDSs of interest using NotI and either XhoI or SalI cloning sites. The pcDNA3-RFP-LACV-N construct was a gift from Prof Georg Kochs and generated by inserting the cDNA for mRFP1 (Addgene #14435) into pcDNA3 flanked by BamHI/NheI sites and then the LACV-N cDNA in frame with NheI/NotI. Bunyamwera N coding sequence (sequence ID: D00353.1) was synthetized (Eurofins) and cloned into pcDNA3-RFP in place of LACV-N, between NheI and NotI restriction sites, to generate pcDNA3-RFP-BUNV-N. pCAGGS-Renilla, pHSPOM1-Firefly, and pRRL.sin.cPPT.SFFV/FLAG-E2Crimson-IRES-puro have previously been described (26).
The FLAG-HsMX1, FLAG-HsMX1T103A, FLAG-HsMX1L41A, FLAG-MmMx1, FLAG-MmMx1T69A and FLAG-MmMx1L7A CDS were PCR-amplified from the aforementioned pRRL.sin.cPPT.SFFV/IRES-puro.WPRE lentiviral vector plasmids and cloned into pET-30 Ek/LIC expression vectors (Novagen) using KpnI and XhoI cloning sites. The different inserts are in frame with the S- and His-tags and a sequence encoding the Tobacco Etch Virus (TEV) protease cleavage site was inserted upstream of the FLAG-HsMX1, FLAG-MmMx1 and FLAG-mutant coding regions to enable tag removal during the protein purification process.
Cells, cell culture, transduction and transfection
A549 (human lung adenocarcinoma; ATCC CCL-185), Human Embryonic Kidney 293T (HEK293T; ATCC CRL-3216) and Madin-Darby canine kidney (MDCK; ATCC CCL-34), were grown in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. Lentiviral vector (LV) stocks were produced as described previously (39). Transduction with LVs was performed by incubating cells for 8h prior to changing media for fresh media. Puromycin selection was performed 36h post-transduction and cells were used for experiments once confluency was attained. Transfection experiments were performed using Lipofectamine 2000 (Thermofisher Scientific) according to the manufacturer’s instructions.
Immunoblotting analysis
Cells were washed in phosphate buffered saline (PBS) 1X and frozen dry at -80°C or lysed directly in lysis buffer [10 mM Tris-HCl pH7.6, 150 mM NaCl,1% Triton X100, 1 mM EDTA, 0,1% deoxycholate, 2% SDS, 5% Glycerol, 100 mM DTT, 0,02% bromophenol blue] and boiled. The lysates were resolved by SDS-PAGE and analysed by immunoblotting. Incubation with a primary Flag (mouse monoclonal M2, Sigma-Aldrich) antibody coupled to horseradish peroxidase (HRP), or Actin (Sigma-Aldrich) antibody followed by an HRP-conjugated secondary antibody was performed. Bioluminescence was then measured (Clarity ECL Western Blotting Substrate, Bio-Rad) using a ChemiDoc system (Bio-Rad).
Immunofluorescence and Airyscan confocal microscopy
A549 cells stably expressing the FLAG-tagged MX1 proteins or mutants of interest were plated into 24-well plates on coverslips. The next day, cells were fixed using 2% paraformaldehyde (PFA) in PBS 1X for 10 minutes at room temperature (RT). Cells were permeabilized with 0.2% Triton X-100 in PBS 1X for 10 minutes followed by quenching and blocking in NGB buffer [50 mM NH4Cl, 2% Goat serum, 2% Bovine serum albumin, in PBS 1X] for 1h. Cells were incubated with primary antibodies for 1h, washed, then incubated for 1h with an Alexa Fluor-conjugated secondary antibody (Thermofisher). Cells were then incubated with Hoechst-33258 (Sigma-Aldrich) for 5 minutes and mounted on glass slides using ProLong™ Gold Antifade Mountant (Thermofisher). Slides were imaged on an LSM880 confocal microscope with an Airyscan module (Zeiss) using a 63x objective. Processing of the raw Airyscan images was performed on the ZEN Black software. Post-processing was performed using the FIJI software (40).
Quantification of RFP-LACV/BUNV-N HsMX1 aggregates
HEK293T cells were plated in 96-well plates and co-transfected with of constructs of interest. 24h later, images of the transfected cells were acquired using an EVOS XL Core microscope (Thermofisher) and quantification of aggregated versus cytoplasmic RFP-LACV/BUNV-N phenotypes was performed manually on FIJI. Results were analysed using GraphPad PRISM. For representative images, HEK293T cells were plated in 24-well plates on coverslips pre-coated with Poly-L-lysine (Sigma-Aldrich) and co-transfected with constructs of interest. Cells were fixed with 2% PFA for 10 minutes, incubated with Hoechst-33258 and mounted as above, and images were acquired on an LSM880 confocal microscope with Airyscan module.
VSV production and infection
G-pseudotyped VSVΔG-GFP-Firefly Luciferase virus was a gift from Dr Gert Zimmer and was produced as described in (41). 3.104 HEK293T cells were plated in 96-well plates and co-transfected with 0.04 µg of constructs of interest together with 0.015 µg of a Renilla luciferase-encoding pCAGGS plasmid (pCAGGS-Renilla). The following day, G-pseudotyped VSVΔG-GFP-Firefly Luciferase virus stock was diluted in complete DMEM and cells were infected at a multiplicity of infection (MOI) of 0,5. 16h post-infection, cells were frozen dry at -80°C for 30 minutes and then lysed in Passive Lysis Buffer (Promega) for 30 minutes. Bioluminescence was measured using the Dual-Luciferase® Reporter Assay System (Promega) on a microplate reader (Tecan Infinite Lumi). Firefly signals were normalized by Renilla luciferase levels and results were analysed using GraphPad PRISM.
Influenza virus production and infection
A/Victoria/3/75 (H3N2) containing the Nanoluciferase coding sequence in the PA segment (IAV-NLuc reporter virus) was described before (39). A/Victoria/3/75 (H3N2) wild-type (IAV-WT) and IAV-NLuc were produced as described previously (8). Briefly, the 8 Pol I plasmids (0.5 μg each) and 4 rescue plasmids (PB1, PB2, PA (0.32 μg each) and NP (0.64 μg)) were co-transfected into HEK293T cells in 6-well plates using Lipofectamine 3000 (Thermo Scientific). After 24h, the cells were detached and co-cultured with MDCK cells in 25 mL flasks. After 8h of co-culture in 10% serum, the medium was replaced with serum-free medium containing 0.5 μg/mL of TPCK-treated trypsin. Supernatants from day 5 post-transfection were used for virus amplification on MDCK cells. For IAV-NLuc reporter virus assays, A549 or HEK293T cells stably expressing the constructs of interest were infected at MOI 0.1 for 16h. Levels of infection were measured as described above using the Nano-Glo® Luciferase Assay System (Promega).
For the influenza minigenome infection reporter assay, 3.104 HEK293T cells were co-transfected in 96-well plates with 0.015 µg of pCAGGS-Renilla and 0.04 µg of pCAGGS expressing FLAG-tagged MX1, mutants or controls and 0.03 µg of the minigenome plasmid pHSPOM1-Firefly. The negative sense, Firefly-coding minigenome transcribed from pHSPOM1-Firefly by the cellular RNA Pol I contains mutations 3-5-8 in the 3’ end of the promoter to increase replication and transcription efficiency following IAV infection (42). 24h later, the cells were infected with wild-type A/Victoria/3/75 (H3N2) at MOI 0.1. 16h post infection, cells were frozen dry at -80°C for 30 minutes and lysed in Passive Lysis Buffer (Promega). Firefly and Renilla activities were then measured using the Dual-Luciferase® Reporter Assay System (Promega). Firefly signals were normalized by Renilla luciferase levels. Results were analysed using GraphPad PRISM.
Cross-link experiments
Crosslinking experiments were performed as described in (43). Briefly, cells expressing the FLAG-tagged MX1 constructs of interest were lysed in 0.5% Triton X-100-PBS 1X buffer in the presence of complete protease inhibitor cocktail (Roche) for 20 minutes on ice. Cells were water bath sonicated and centrifuged at 1,500 x g for 10 minutes at 4°C. Disuccinimidyl suberate (DSS, ThermoFisher) was added to the cells at a final concentration of 100 µg/ml (or an equivalent volume of Dimethyl sulfoxide (DMSO) was added to the control conditions) and incubated for 1h at RT. Laemmli was added at 1X final and samples were resolved by SDS-PAGE without boiling on a 5% acrylamide gel. Immunoblotting was performed as mentioned above.
Protein expression and purification
The recombinant plasmids pET-30 Ek/LIC expressing FLAG-HsMX1, FLAG-MmMx1 and FLAG-mutants were transformed in an Escherichia coli BL21 (DE3) strain resistant to Phage T1 (New England Biolabs) carrying pRARE2. One colony was used to inoculate an overnight culture of 125mL Lysogeny broth (LB) medium supplemented with kanamycin (50 μg/mL) and chloramphenicol (30 μg/mL). This culture was diluted in 2.5L of LB medium supplemented with the two antibiotics. The cells were grown at 16°C to an optical density at 600 nm of 0.8, then protein expression was induced with 1 mM Isopropyl-β-D-thiogalactoside (IPTG) and the culture was grown overnight at 16°C. The cells were harvested by centrifugation at 8 200g for 20 minutes and resuspended in 30 mL of buffer A (50 mM Tris–HCl pH 8, 400 mM NaCl, 5 mM MgCl2, 7 mM β-mercaptoethanol, 40 mM imidazole, 10% glycerol and 1 mM benzamidine). The cells were disrupted by sonication and cell debris were removed by centrifugation at 28 000g for 60 minutes. The supernatant was loaded at 4°C on Ni–NTA agarose beads previously equilibrated with buffer A. The beads were washed once with buffer A and twice with buffer B (50 mM Tris–HCl pH 8, 1 M NaCl, 5 mM MgCl2, 7 mM β-mercaptoethanol, 40 mM imidazole and 10% glycerol) and elution was performed with buffer E (50 mM Tris–HCl pH 8, 200 mM NaCl, 5 mM MgCl2, 7 mM β-mercaptoethanol, 500 mM imidazole, 10% glycerol). The eluted protein was incubated with His-tagged TEV protease purified in our laboratory in a 1:100 (w:w) ratio; the cleavage reaction was performed during dialysis (dialysis-bag cutoff 12–15 kDa) against 1L dialysis buffer D (50 mM Tris–HCl pH 8, 200 mM NaCl, 5 mM MgCl2, 5 mM β-mercaptoethanol) overnight at 4°C. After dialysis, the proteins were centrifuged for 20 minutes at 28 000g and the supernatant was loaded again at 4°C on Ni–NTA agarose beads equilibrated with buffer D. The proteins without the His-tag were collected in the flow-through, concentrated to 5 mg/mL using a Vivaspin® column (50 kDa cutoff, Sartorius), loaded onto a size-exclusion chromatography column (Superose 6 Increase 10/300 GL, GE Healthcare) and eluted with buffer F (50 mM Tris–HCl pH 8, 500 mM NaCl, 5 mM MgCl2, 0,5 mM β-mercaptoethanol, 10% glycerol). Aliquots of purified FLAG-tagged proteins were snap frozen in liquid nitrogen and stored at -80 °C.
GTPase activity assays
Recombinant FLAG-tagged, wild-type and mutant HsMX1 or MmMx1 proteins (0.2 mg/mL) were incubated at 37°C in GTPase assay buffer [50 mM Tris–HCl pH 8, 500 mM NaCl, 5 mM MgCl2, 0,5 mM β-mercaptoethanol, 10% glycerol, 50 µM GTP, 26 nM [α-32P]-GTP], as described previously (26). The reactions were stopped at the indicated time by addition of [2 mM EDTA, 0,5% SDS]. The reaction products were resolved by thin-layer chromatography (TLC, Merck Millipore) using TLC buffer [1 M LiCl, 1 M formic acid] and detected using phosphor screen autoradiography (Amersham Typhoon apparatus).
Molecular modelling and dynamic simulations of wild-type HsMX1 and HsMX1L41A
HsMX1 WT (including the complete N-terminal domain and L4 loop) as well as HsMX1L41A were modelled using AlphaFold (44) through the ColabFold (45) platform. Predicted 3D models were compared to crystal (PDB: 3SZR) or low-resolution EM (PDB: 3ZYS) structures for examination of the overall folding of the three main domains (GTPase, BSE and stalk) and also compared to 3D models based on these templates and built with Modeller v9.19 (46). Structural alignments of AlphaFold models and X-ray structure gave RMSD values (for all backbone atoms) of 0.4, 0.2 and 1.4 Å for GTPase, BSE and Stalk domains, respectively. Molecular dynamics (MD) simulations were performed using NAMD3 (47) and CHARMM36m force field (48). Briefly, each protein was immersed in an explicit solvent box (TIP3P water model) using 10 Å of edge in each direction (in respect to protein dimensions), then neutralized with NaCl ions at a physiological concentration (0.154 M) and energy minimized for 50 ps using the conjugate gradients method. After a gradual heating from -273°C to 37°C, each system was further equilibrated for 300 ps using periodic boundary conditions to replicate the system in each direction. A trajectory of 500 ns was then produced (1 frame saved every 20 ps) for analysis in the isobaric-isothermal ensemble to keep constant temperature (37°C) and pression (1 atm) using Langevin dynamics and Langevin piston methods. The Newton’s equation of motions was integrated using a timestep of 2 fs using the r-RESPA algorithm (49) with the short-range Lennard-Jones potential smoothly truncated from 10 to 12 Å and the PME (Particle Mesh Ewald) approach (50) used for calculating long-range electrostatics with a grid spacing of 1 Å. Trajectories analysis was performed using VMD (51), its Timeline plugin for per-residue Phi and Psi dihedral angles calculation, 3D-graphs (dihedral angles plots) were made with GnuPlot-5.4 and MD snapshots were illustrated with the PyMol Molecular Graphics System (v2.5, Schrödinger, LLC).
Data availability statement
Datasets are currently being deposited on a public repository and the URL will be provided.
Statistical analyses
Statistical analyses were performed with the GraphPad Prism software. The analysis types performed are indicated in the figure legends. Comparisons are relative to the indicated condition. P-values are indicated as the following: ns = not significant, p<0.05 = *, p<0.01 = **, p<0.001 = *** and p<0.0001 = ****.
Results
Leucine 41 from the N-terminal region of human MX1 is essential for antiviral activity against IAV
The loop L4 and all 43 amino acids of the N-terminal domain are absent from the published crystal structures of HsMX1 (19, 20) (Fig. 1A). While the L4 loop has been extensively studied in the past (33, 35), to date, the importance of the N-terminal region for the antiviral activity of HsMX1 has not been evaluated. Therefore, to address this, we generated a series of N-terminal truncation mutants for HsMX1, with mutants missing either the first 14 (HsMX115-662) or 28 (HsMX129-662) amino acids, or the entire N-terminal region (HsMX144-662, named hereafter HsMX1ΔNter) (Fig. 1B). In parallel to two negative controls (E2 Crimson fluorescent protein, termed CTRL, and HsMX1 inactive GTPase mutant, HsMX1T103A), the wild-type (WT) protein and mutants were ectopically expressed in HEK293T cells, and an IAV minigenome infection reporter assay was performed, as reported previously (8). In this assay, a negative sense minigenome coding for the Firefly luciferase is recognized, replicated and transcribed by IAV polymerase in IAV infected cells. Thus, Firefly activity is used to monitor replication efficiency. Cells were also co-transfected with a Renilla luciferase coding plasmid for normalization. We observed that the deletion of the first 14 or 28 amino acids had no effect on HsMX1 antiviral activity (Fig. 2A). In contrast, the deletion of the entire N-terminal region totally abrogated antiviral activity in a comparable manner to the control inactive GTPase mutant HsMX1T103A (Fig. 2A). This suggested the presence of essential residues located between positions 29 and 43 of the N-terminal domain of HsMX1. Mus musculus Mx1 (MmMx1) is another well studied IAV-inhibiting MX1 protein. Alignment of MmMx1 and HsMX1 proteins revealed a conserved motif of four amino acids located between positions 29 and 43 of HsMX1: 39-NNLC-42 (or 5-NNLC-8 in the case of MmMx1) (Fig. 1B). Upon replacement in HsMX1 of these four amino acids with four alanines (HsMX1NNLC39-42A), we saw a complete loss of antiviral activity comparable to that of HsMX1ΔNter and HsMX1T103A (Fig. 2A, top panel). Alanine point mutations of these four residues (HsMX1N39A, HsMX1N40A, HsMX1L41A and HsMX1C42A), revealed that leucine 41 (L41) was responsible for the loss of function phenotype observed with the MX1NNLC39-42A mutant, whereas the other three mutants completely retained their anti-IAV activity (Fig. 2A, top panel). While the HsMX129-662 and HsMX1ΔNter truncation mutants showed decreased expression levels compared to the WT protein, as assessed by immunoblotting, the loss of antiviral activity of the HsMX1NNLC39-42A and HsMX1L41A mutants could not be attributed to a decrease in expression levels (Fig. 2A, bottom panel).
The observed phenotypes were then confirmed using a Nanoluciferase reporter-expressing version of A/Victoria/3/75 (IAV-NLuc) (39) as a second infection readout, both in HEK293T (Fig. 2B) and in lung-derived A549 cells (Fig. 2C), which stably expressed the control and mutant proteins. Similarly to what was observed in the minigenome infection reporter assay (Fig. 2A), HsMX1ΔNter, HsMX1NNLC39-42A and HsMX1L41A were completely inactive against IAV, comparable to the inactive HsMX1T103A mutant, whereas the other mutants retained antiviral activity (Fig. 2B-C).
Next, we investigated the subcellular localization of the N-terminal HsMX1 mutants using super-resolution Airyscan microscopy. As reported previously, HsMX1 showed a honeycomb-like, punctate cytoplasmic staining and HsMX1T103A presented a juxtanuclear accumulation (Fig. 2D) (26, 52, 53). In contrast, the HsMX1ΔNter mutant formed cytoplasmic aggregate-like structures that varied in size and were dispersed throughout the cytoplasm (Fig. 2D). Unlike this mutant, HsMX1NNLC39-42A and HsMX1L41A accumulated at the perinuclear region into small spherical structures that were phenotypically different from those of HsMX1T103A (Fig. 2D). This attests to a potential difference between the L41 point and N-terminal truncation mutants, which might be linked to differences in protein synthesis or stability as seen by immunoblot (Fig. 2A).
Taken together, this data showed that the N-terminal region, and more precisely, leucine 41, was essential for anti-IAV activity and correct subcellular localization of HsMX1.
L41 is essential for HsMX1 restriction of other families of RNA viruses
To further understand the importance of this newly discovered essential residue for the antiviral activity of HsMX1, we tested the restriction abilities of aforementioned HsMX1 mutants against other RNA viruses known to be inhibited by HsMX1. HEK293T cells were co-transfected with either a control or HsMX1 WT or mutant expression constructs along with a Renilla luciferase-coding plasmid and the infection levels of Firefly-expressing, Vesicular Stomatitis Virus (VSV) rhabdovirus were measured (Fig. 3A). Reminiscent of what observed for IAV (Fig. 2A), the HsMX1ΔNter, HsMX1NNLC39-42A and HsMX1L41A mutants totally lost their ability to restrict VSV (Fig. 3A). This therefore demonstrates the essential role of L41 for the restriction of VSV by HsMX1.
As HsMX1 is well-known to inhibit the Orthobunyaviruses La Crosse Virus (LACV) and Bunyamwera Virus (BUNV) (54), we tested the importance of the N-terminal domain against these viruses. The inhibition phenotype of HsMX1 towards LACV and BUNV is characterized by the aggregation of their Nucleoprotein (N) into perinuclear aggregates (54), a phenotype that can be recapitulated by co-expressing HsMX1 and the N protein in cells and easily visualised by using a fusion of N with the red fluorescent protein (RFP) (54). Hence, HEK293T cells were co-transfected with plasmids expressing HsMX1 or the mutants of interest, together with an RFP-LACV-N expression plasmid. As reported previously (54), HsMX1 induced LACV-N aggregates and this ability was lost by HsMX1T103A, as shown on the representative images (Fig. 3B). To quantify HsMX1 protein activity here, the percentage of cells showing a diffuse or aggregated RFP-LACV-N staining was determined. In control conditions (RFP-LACV-N alone), around 70% of the cells displayed a diffuse cytoplasmic pattern (Fig. 3C), with 30% of cells showing limited aggregation, possibly due to overexpression levels. In the presence of HsMX1, only around 10% of cells showed a diffuse cytoplasmic staining, with around 90% of cells showing an RFP-LACV-N aggregation at the perinuclear region (Fig. 3C, 3B). In the presence of HsMX1ΔNter, HsMX1NNLC39-42A and HsMX1L41A, there was a complete loss of RFP-LACV-N aggregation phenotype, with distributions comparable to the RFP-LACV-N alone or HsMX1T103A conditions (Fig. 3C, 3B). The same experiment was performed for the RFP-tagged Nucleoprotein of BUNV (RFP-BUNV-N) (Fig. 3D). Similar results to the ones obtained with LACV-N were observed, although with a somewhat reduced, but nevertheless significant, impact of HsMX1L41A and HsMX1NNLC39-42A mutations on the ability of HsMX1 to induce RFP-BUNV-N aggregation. These data show that the N-terminal domain, and in particular L41, is important for LACV and BUNV restriction as well as for VSV and IAV.
Conserved importance of the N-terminal domain for the anti-IAV activity of MX1 proteins of other mammals
MX1 proteins from various mammalian species are known to harbour anti-IAV activity (7, 55). Therefore, we sought to explore the importance of the conserved leucine we identified in the N-terminal region in the anti-IAV activity of MX1 proteins from other mammals. Mouse Mx1 (MmMx1) strongly inhibits IAV, and, interestingly, the N-terminal domain of MmMx1 is highly different to that of HsMX1, consisting of only 9 amino acids versus 43 for HsMX1 (Fig. 1B). Nevertheless, MmMx1 contains a leucine at position 7, placed exactly 3 amino acids before the start of the first α-helix of the BSE domain, similarly to leucine 41 of HsMX1 (Fig. 1B). To determine whether the N-terminal domain, and notably L7, were also essential for MmMx1 anti-IAV activity, we performed a truncation of the N-terminal domain (MmMx1ΔNter) and an alanine point mutation of L7 (MmMx1L7A). We stably expressed these constructs, as well as a GTPase inactive mutant (MmMx1T69A), in A549 cells and infected them with the IAV-NLuc reporter virus (Fig. 4A, top panel). Whereas MmMx1 potently inhibited IAV replication, MmMx1ΔNter and MmMx1L7A showed no antiviral activity, similar to the GTPase inactive mutant MmMx1T69A. It is noteworthy that, similarly to HsMX1ΔNter (Fig 2A), MmMx1ΔNter had reduced expression levels (Fig. 4A, bottom panel). However, MmMx1L7A showed similar expression levels as compared to WT MmMx1 (Fig. 4A, bottom panel). Using super-resolution Airyscan microscopy, WT MmMx1 was located into classical nuclear bodies as well as being present to a lesser extent in the cytoplasm (Fig. 4B). MmMx1ΔNter was localized in nuclear rod-like structures, and MmMx1L7A was found as nuclear rod-like structures that associated into star-shaped superstructures, as well as in perinuclear, aggresome-like structures (Fig. 4B). Taking all these data together, we show that the N-terminal domain, and more specifically leucine 7, is essential for MmMx1 anti-IAV activity and subcellular localization.
We next decided to look at a more distant relative of humans and therefore chose to study a bat MX1 protein, Sturnira lilium (little yellow-shouldered bat) MX1 (SlMX1), that also possesses anti-IAV activity (38). SlMX1 has a similar sized N-terminal domain (41 amino acids in length) to that of HsMX1, with, again, a leucine located 3 amino acids before the start of the BSE, leucine 39 (L39) (Sup. Fig. 1). As previously, we generated a truncation of the N-terminal domain (SlMX1ΔNter) and an alanine point mutation of L39 (SlMX1L39A). We produced A549 cells stably expressing WT SlMX1 and mutants and infected them with the IAV-NLuc reporter virus (Fig. 4C). As for HsMX1 and MmMx1, SlMX1ΔNter and SlMX1L39A mutants lost their antiviral activity against IAV, similar to the GTPase inactive mutant SlMX1T101A (Fig. 4C, top panel). In opposition to HsMX1ΔNter and MmMx1ΔNter, however, SlMX1ΔNter did not show profoundly decreased expression levels and the SlMX1L39A mutant seemed to show slightly higher expression levels than the WT SlMX1 (Fig. 4C, bottom panel). In terms of subcellular localization, SlMX1 was found in cytoplasmic puncta forming honeycomb-like subcellular structures, reminiscent of HsMX1 (Fig. 4D, 2D). In contrast, SlMX1ΔNter and SlMX1L39A were present in small puncta in the cytoplasm but lost their honeycomb-like network (Fig. 4D). This confirmed that the conserved leucine present in the N-terminal domain of mammalian MX1 proteins is a crucial residue for their antiviral activity and subcellular localization.
Considering that this leucine is conserved for human, mouse and bat MX1 proteins, we aligned the amino acid sequence of MX1 from 76 different species, comprising mammals, birds, amphibians and fish (Sup. Fig. 1). An analysis using the WebLogo3 tool (weblogo.threeplusone.com) (56, 57) of amino acids corresponding to positions 34-52 of HsMX1 from all 76 MX1 protein sequences shows the extreme conservation of the leucine corresponding to L41 throughout all MX1 proteins aligned (Fig. 4E), with the rest of the N-terminal domain before and after this residue being much less conserved than the leucine residue. Indeed, the leucine residue is conserved to a level similar to amino-acids from the BSE domain (Fig. 4E). Of note, only a few species of fish, including zebrafish (Danio rerio) and goldfish (Carassius auratus) did not show conservation of this leucine, with a phenylalanine (F) found in place of this leucine (Fig. 4E and Sup. Fig. 1).
Altogether these results show that similarly to HsMX1 L41, L7 and L39 from MmMx1 and SlMX1 respectively, are essential for antiviral activity against IAV and that this leucine is the most conserved residue of the N-terminal domain across many species.
Loss of antiviral activity of N-terminal Leucine mutants is neither due to an oligomerization defect nor the inability to hydrolyse GTP
Correct oligomerization and GTPase activity being both essential for HsMX1 antiviral activity (19, 25), we wondered whether mutating L41 in HsMX1 could impair these properties. First, we performed crosslinking experiments on WT HsMX1, HsMX1T103A, HsMX1M527D (a monomeric mutant (19)) and HsMX1L41A. HEK293T cells stably expressing these proteins showed the expected phenotypes with respect to IAV-NLuc infection, with a loss of antiviral activity for HsMX1M527D comparable to that of HsMX1T103A and HsMX1L41A (Fig. 5A). Disuccinimidyl suberate (DSS) crosslinking followed by immunoblotting experiments showed that, contrary to WT HsMX1, HsMX1M527D did not assemble into lower- or higher-order oligomers and was only detected as a monomer (Fig. 5B). In contrast, HsMX1L41A showed a similar capacity to oligomerize into lower- and higher-order oligomers than WT HsMX1 or the GTPase inactive mutant HsMX1T103A (Fig. 5B). We can therefore conclude that the absence of antiviral activity of this mutant is not due to an oligomerization defect. Next, the same experiments were performed with MmMx1 and mutants stably expressed in HEK293T cells (Fig. 5C, 5D). Of note, MmMx1M493D, the potential monomeric mutant corresponding to HsMX1M527D, could not be used as expression levels were undetectable (data not shown). MmMx1 stable overexpression in HEK293T cells strongly inhibited IAV replication, although somewhat less efficiently compared to stable overexpression in A549 cells (compare Fig. 5C with Fig. 4A). Both MmMx1T69A and MmMx1L7A mutants totally lost antiviral activity (Fig. 5C). In a similar fashion to HsMX1L41A, MmMx1L7A was able to oligomerize into lower- and higher-order oligomers (Fig. 5D), showing that the inability of MmMx1L7A to inhibit IAV was not due to an oligomerization defect.
To assess the potential impact of leucine mutations on GTPase activity, recombinant WT HsMX1, MmMx1 and mutant proteins HsMX1T103A, HsMX1L41A, MmMx1T69A and MmMx1L7A were produced in E. coli and purified (Fig. 6A). A GTPase activity assay was performed to assess the ability of these proteins to convert GTP into GDP (Fig. 6B-D). While the WT HsMX1 and the GTPase inactive mutant HsMX1T103A were respectively able and unable to hydrolyse GTP, as expected (58) (Fig. 6B), the HsMX1L41A mutant showed comparable GTPase activity to WT HsMX1, indicating no defect in enzymatic activity. In the case of MmMx1, WT MmMx1 could hydrolyse GTP and MmMx1T69A was unable to hydrolyse it (Fig. 6D), as expected (34). However, while MmMx1L7A was still able to hydrolyse GTP to some extent (Fig. 6D), in comparison to WT MmMx1, MmMx1L7A showed ∼60% decrease in enzymatic activity (Fig. 6D). Time course experiments showed that GTPase activity efficiency of HsMX1L41A was identical to that of WT HsMX1 regardless of the duration of the assay (Fig. 6C). In contrast, MmMx1L7A appeared less active than WT MmMx1 and showed a 30% to 50% decrease in GTPase activity at the different time points (Fig. 6E). Nevertheless, these experiments showed that the loss of antiviral activity of HsMX1L41A and MmMx1L7A mutants could not be attributed to a complete inability to hydrolyse GTP.
Mutating L41 of HsMX1 does not seem to impact BSE structural properties
As mentioned above, the structure of the N-terminal domain of HsMX1 has not been solved (19, 20) (Fig. 1A). The AlphaFold tool (44, 59) may accurately predict structures of proteins (or regions of proteins) that have not been experimentally determined by structural approaches. Concerning HsMX1, the N-terminal domain can be modelled starting from serine 35 (S35) (alphafold.ebi.ac.uk/entry/P20591), although the confidence levels are not high and even lower for the long N-terminal unstructured tail. Nevertheless, AlphaFold predicts that the first α-helix of the BSE starts at S35, which would entail a redefinition of BSE region and would make L41 actually part of the BSE (Fig. 7A). Hence, mutation of L41 into an alanine could entail a restructuration of the BSE helix, explaining the loss of antiviral activity of the HsMX1L41A mutant. However, performing molecular dynamics (MD) simulations with WT HsMX1 or the L41A mutant (Fig. 7B, 7C, Sup. Movie 1, Sup. Movie 2) showed that the structural behaviour of this α-helix did not change when comparing both proteins. Indeed, except of a hinge or a turn-like structure occurring quite rapidly around residues E46-E47 but observed in both cases, the α-helicoidal secondary structure is maintained all throughout the simulations (0.5 µs) (Fig. 7B, 7C, Sup. Movie 1, Sup. Movie 2). To confirm this point, two dihedral angles (Phi and Psi) from part of the N-terminal helix (residues ranging from 33 to 60) were measured over time leading to very similar values for both angles between both WT and the L41A mutant (Sup. Fig. 2). Angle values were characteristic of α-helix structure with average values centred at -60° and - 50° for Phi and Psi, respectively. This hinge or partial helix brake seems slightly more pronounced for WT HsMX1 than for the mutant when examining thoroughly angles values, however, this unique exception may not be sufficient to explain the difference observed in antiviral activity between the two constructs.
Discussion
Several HsMX1 intrinsic antiviral determinants have been described in the past, namely an active GTPase domain, the capability to oligomerize via the stalk, and intact BSE, L2 and L4 loops, but the importance of the N-terminal domain had never been assessed. In the case of HsMX2, the N-terminal domain was found to be one of the two main determinants, together with the oligomerization capacity, for anti-HIV activity (36, 37). Therefore, it was plausible that HsMX1 might also need this domain for antiviral activity despite a high variability between orthologous MX proteins (Fig. 1B, 4E, Sup. Fig. 1). Herein, we show that the N-terminal domain, more specifically the highly conserved leucine situated 3 amino acids before the start of the first BSE helix, from human, mouse and bat (Sturnira lilium) MX1 proteins, are essential for their anti-IAV activity. The importance of this leucine is highlighted by the fact that its mutation induces a complete loss of anti-IAV activity for all three proteins without being correlated with a decrease in protein expression levels, or, for HsMX1 and MmMx1, without being correlated to a defect of oligomerization or of total inability to hydrolyse GTP. In addition, according to MD simulations based on the AlphaFold prediction, the mutation of L41 in HsMX1 into an alanine does not seem to affect the structure of the first BSE α-helix. Nevertheless, this residue was shown to be essential for subcellular localization of these three MX1 proteins. Finally, we show that this leucine is not only essential for HsMX1 restriction of IAV but also VSV, as well as LACV or BUNV nucleoprotein aggregation capacity.
The N-terminal domain is not present in the crystal structure of HsMX1 (Fig. 1A), the non-structured and highly flexible nature of this extension likely explains the lack of structural information (Fig. 1A, 7A, 7B, Sup. Movie. 1). Furthermore, MmMx1 and SlMX1 3D structures have not yet been determined. As we now show that this conserved leucine is essential for the antiviral activity of these proteins, it would therefore be interesting to be able to decipher experimentally the structure of an MX1 protein with an intact N-terminal domain. This would allow us to understand the three-dimensional architecture of this part of the protein and confirm the AlphaFold predictions. In terms of the role of this residue for MX1 function, we show above that it is not essential for the oligomerization of both HsMX1 and MmMx1. Of note, the elution profiles of the recombinant proteins are very similar for the WT and mutant proteins, confirming our crosslink data and the fact that the leucine point mutants do not seem to affect oligomerization status (Fig. 6A, 5B, 5D). AlphaFold predictions nevertheless suggest that L41 is part of the α-helix of the first BSE (Fig. 7A), which, if confirmed experimentally, should lead to a redefinition of the boundaries of MX1 N-terminal and BSE domains. A plausible effect of these leucine mutants could therefore be a partial hindrance to the correct folding or function of the first helix of the BSE, but MD simulations in this region did not support this hypothesis as similar structural behaviour was observed in both cases (Fig. 7B, 7C, Sup. Movie 1-2). The AlphaFold prediction models the conserved leucine as pointing away from the core of the protein (Fig. 7A). This could potentially suggest that L41 is a critical residue important for interaction with a cellular cofactor or a viral element, although the latter possibility is more difficult to imagine due to the wide breadth of viral families inhibited by MX1 proteins (7, 55). In support of the former idea, the subcellular localization of HsMX1L41A, MmMx1L7A, and SlMX1L39A were all displaced compared to their WT counterparts (Fig. 1D, 4B, 4D), which might be explained by the loss of interaction with such a cellular cofactor, which would be essential for correct subcellular localization. Because of this, these single point leucine mutants therefore present themselves as ideal candidates for differential interactomic approaches to identify such putative cellular cofactors. To date, no cofactor that would be essential for MX1 effector antiviral activity has been identified. The discovery of such cofactor(s), if they exist, would probably be a major turning point to understand the mechanism of action of MX1 proteins.
Allelic variation studies have been performed for HsMX1 (60, 61), with only one variant being discovered in the N-terminal domain where a premature stop codon was found at amino acid position 30 (Q30*) causing a truncated, antivirally inactive protein (61). Using the gnomAD browser (https://gnomad.broadinstitute.org), a single variant can be found in a single individual implicating L41 of HsMX1 where a missense deletion of 1 base occurred (21-42807777-AC-A) inducing a frameshift. With no reported naturally occurring allelic variation occurring at this position so far, it would be interesting to find such allelic variants to see if these individuals would be more susceptible to viral infections due to a non-functional HsMX1 protein.
The amino acid placed exactly three residues before the (previously defined) beginning of the first BSE α-helix is consistently a leucine for almost all species, with the exception of some fish species where it is replaced by a phenylalanine (Fig. 4E, Sup. Fig. 1). The difference of the genetic code between leucine and phenylalanine is minimal (UUA or UUG for the former and UUU or UUC for the latter). A single point mutation would have therefore been necessary to induce this exchange of amino acids. Another interesting observation is the fact that when the N-terminal domain of HsMX2 was transferred onto HsMX1, this chimeric protein retained its anti-IAV activity while gaining anti-HIV activity (26). Despite being much longer than the N-terminal domain of HsMX1, that of HsMX2 also contains a leucine, 3 amino acids before start of the BSE, which was also the case for the chimeric protein (26). This information, coupled with the fact that the rest of the N-terminal domain is extremely divergent between species, attests to the high evolutionary selective pressure placed on this residue, further reinforcing the importance of this leucine for MX1 protein antiviral activity.
In conclusion, we show herein that the N-terminal domain of MX1 proteins, more specifically a highly conserved leucine residue, is important for their antiviral activity against multiple RNA viruses and governs correct subcellular localization. Further characterizing the importance of this residue could pave the way to a better understanding of the molecular mechanisms at play in the antiviral activity of these long-studied, yet still not well understood restriction factors.
Author contributions
J.M., O.M. and C.G. conceived the study, designed the experiments and interpreted the data; J.M. performed the experiments; M.A-A performed immunoblots, produced the recombinant proteins and did the GTPase assays; M.B. provided expertise and supervision on recombinant protein production and biochemical assays; M.T., C.A-A., and O.P. provided technical help; L.C. performed and analysed the MD studies. J.M., O.M. and C.G. wrote the manuscript with input from all authors.
Competing interests
The authors declare no competing interests.
Footnotes
3D: Three-dimensional
BSE: Bundle Signalling Element
BUNV: Bunyamwera Virus
CDS: Coding DNA Sequence
DMEM: Dulbecco’s Modified Eagle Medium
DMSO: DiMethyl SulfOxide
DNA: DeoxyriboNucleic Acid
DSS: DiSuccinimidyl Suberate
EDTA: EthyleneDiamineTetraacetic Acid
EM: Electron Microscopy
FBS: Fetal Bovine Serum
GDP: Guanosine DiPhosphate
GFP: Green Fluorescent Protein
GTP: Guanosine TriPhosphate
GTPase: Guanosine TriPhosphatase
HBV: Hepatitis B Virus
HEK293T: Human Embryonic Kidney 293T
HIV-1: Human Immunodeficiency Virus 1
HRP: HorseRadish Peroxidase
HsMX1/2: Human Myxovirus resistance protein 1/2
HTNV: Hantaan River Virus
IAV: Influenza A virus
IFN: Interferon
ISG: Interferon Stimulated Genes
IPTG: IsoPropyl-β-D-ThioGalactoside
JAK: Janus Kinase
L7/39/41: Leucine 7/39/41
LACV: La Crosse Virus
LACV/BUNV-N: LACV/BUNV Nucleoprotein
LB: Lysogeny Broth
LV: Lentiviral Vector
MD: Molecular Dynamics
MDCK: Madin-Darby Canine Kidney
MmMx1/2: Mouse Myxovirus resistance protein 1/2
MOI: Multiplicity of Infection
MX1/2: Myxovirus resistance protein 1/2
MxA/B: Myxovirus resistance protein
A/B N-terminal: Amino-terminal
NLS: Nuclear Localization Signal
NLuc: Nanoluciferase
NP: NucleoProtein
PA: Polymerase Acidic
PB1/2: Polymerase Basic 1/2
PBS: Phosphate Buffered Saline
PFA: ParaFormAldehyde
PH: Pleckstrin Homology
PI(4,5)P2: PhosphatidylInositol 4,5-bisPhosphate
RIG-I: Retinoic acid-Inducible Gene I
RFP: Red Fluorescent Protein
RMSD: Root Mean Square Deviation
RNA: RiboNucleic Acid
RT: Room Temperature S35: Serine 35
SDS-PAGE: Sodium Dodecyl Sulfate–Polyacrylamide Gel Electrophoresis
SlMX1: Sturnira lilium Myxovirus resistance protein 1
STAT: Signal Transducer and Activator of Transcription
TEV: Tobacco Etch Virus
THOV: Thogoto Virus
TPCK: N-Tosyl-L-Phenylalanine Chloromethyl Ketone
VSV: Vesicular Stomatitis Virus
WT: Wild-Type
Acknowledgements
We are thankful to Prof. Georg Kochs (Freiburg University, Germany) for the pcDNA3-RFP-LACV-N and SlMX1 plasmids; to Prof. Gert Zimmer (Institute of Virology and Immunology, Switzerland) for the G-pseudotyped VSVΔG-GFP-Firefly Luciferase virus. This work was supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement 759226, ANTIViR, to CG), the ATIP-Avenir program (to CG), the Institut National de la Santé et de la Recherche Médicale (INSERM) (to CG), institutional funds from the Centre National de la Recherche Scientifique (CNRS) and Montpellier University, a 3-year PhD studentship from the Ministry of Higher Education and Research (to JM) and a 4th year PhD funding from by the Fondation pour la Recherche Médicale, grant number [FDT202106013175] (to JM), and institutional funds from IRIM (inter-team project, to MAA and MB). We acknowledge the imaging facility MRI, member of the national infrastructure France-BioImaging supported by the French National Research Agency (ANR-10-INBS-04). JM, OM and CG acknowledge support from the French research network on influenza viruses (ResaFlu; GDR2073) financed by the CNRS.
Footnotes
This article contains supporting information