The Antiviral Activity of Equine Mx1 against Thogoto Virus Is Determined by the Molecular Structure of Its Viral Specificity Region

ABSTRACT

Mammalian myxovirus resistance (Mx) proteins are interferon-induced, large dynamin-like GTPases with a broad antiviral spectrum. Here, we analyzed the antiviral activity of selected mammalian Mx1 proteins against Thogoto virus (THOV). Of those, equine Mx1 (eqMx1) showed antiviral activity comparable to that of the human MX1 gene product, designated huMxA, whereas most Mx1 proteins were antivirally inactive. We previously demonstrated that the flexible loop L4 protruding from the stalk domain of huMxA, and especially the phenylalanine at position 561 (F561), determines its antiviral specificity against THOV (P. S. Mitchell, C. Patzina, M. Emerman, O. Haller, et al., Cell Host Microbe 12:598–604, 2012, https://doi.org/10.1016/j.chom.2012.09.005). However, despite the similar antiviral activity against THOV, the loop L4 sequence of eqMx1 substantially differs from the one of huMxA. Mutational analysis of eqMx1 L4 identified a tryptophan (W562) and the adjacent glycine (G563) as critical antiviral determinants against THOV, whereas the neighboring residues could be exchanged for nonpolar alanines without affecting the antiviral activity. Further mutational analyses revealed that a single bulky residue at position 562 and the adjacent tiny residue G563 were sufficient for antiviral activity. Moreover, this minimal set of L4 amino acids transferred anti-THOV activity to the otherwise inactive bovine Mx1 (boMx1) protein. Taken together, our data suggest a fairly simple architecture of the antiviral loop L4 that could serve as a mutational hot spot in an evolutionary arms race between Mx-escaping viral variants and their hosts.

IMPORTANCE Most mammals encode two paralogs of the interferon-induced Mx proteins: Mx1, with antiviral activity largely against RNA viruses, like orthomyxoviruses and bunyaviruses; and Mx2, which is antivirally active against HIV-1 and herpesviruses. The human Mx1 protein, also called huMxA, is the best-characterized example of mammalian Mx1 proteins and was recently shown to prevent zoonotic virus transmissions. To evaluate the antiviral activity of other mammalian Mx1 proteins, we used Thogoto virus, a tick-transmitted orthomyxovirus, which is efficiently blocked by huMxA. Interestingly, we detected antiviral activity only with equine Mx1 (eqMx1) but not with other nonprimate Mx1 proteins. Detailed functional analysis of eqMx1 identified amino acid residues in the unstructured loop L4 of the stalk domain critical for antiviral activity. The structural insights of the present study explain the unique position of eqMx1 antiviral activity within the collection of nonhuman mammalian Mx1 proteins.

INTRODUCTION

Myxovirus resistance (Mx) proteins were initially discovered in mouse strains displaying resistance to lethal influenza A virus (IAV) infections (1). Later, two MX paralogs were discovered in most studied mammals (2). They encode the Mx1 and the Mx2 proteins. The Mx1 proteins, for humans called huMxA, confer antiviral activity mainly against RNA viruses, like orthomyxoviruses and bunyaviruses, whereas Mx2, for humans called huMxB, inhibits HIV-1 and herpesviruses (3, 4). An exception are rodents that lost their MX2 locus but by gene duplication gained two Mx proteins, nuclear Mx1 and cytoplasmic Mx2, which are both orthologs of mammalian Mx1 (5). Mouse strains lacking functional Mx1 expression are highly susceptible to IAV infection. However, transgenic expression of Mus musculus Mx1 (mmMx1) or huMxA renders the mice resistant against otherwise lethal IAV infections (6, 7).

Mx proteins belong to the family of dynamin-like large GTPases. Their overall structure is highly conserved, consisting of a globular G domain that binds and hydrolyzes GTP, a stalk region important for oligomerization and antiviral specificity, and a bundle signaling element (BSE) composed of three helices connecting the G-domain and stalk (see Fig. 2a below) (8). GTP binding and hydrolysis as well as homo-oligomerization are necessary for the antiviral activity of Mx proteins. Accordingly, mutations in the G-domain that abolish GTP hydrolysis, like huMxA(T103A), or mutations in the stalk that abolish oligomerization, like huMxA(M527D), result in the loss of antiviral activity (8). In the distal part of huMxA the unstructured loop L4 (amino acids 530 to 573), connecting stalk helices α3 and α4, protrudes from the compact rod-shaped structure of the stalk region (see Fig. 2a below). This unstructured loop is quite tolerant for mutations, can capture many conformations, and provides huMxA with the flexibility to evolutionarily adapt to and target multiple viral pathogens (9). Loop L4 can be divided into an N-terminal part (positions 530 to 557) and a C-terminal part (positions 558 to 573) joined by a tetralysine (KKKK) motif (8). The conserved N-terminal part and especially the KKKK motif were shown to be involved in membrane association of huMxA (10). In contrast, the C-terminal part of L4 is the most variable region between different Mx1 proteins (see Fig. S1 in the supplemental material) (11), suggesting that this region is under recurrent selective pressure to adapt to diverse, rapidly evolving viral pathogens. Accordingly, the C-terminal L4 part of primate Mx1 is a hot spot of positive selection, whereas the overall structure of Mx1 needs to be preserved for its complex architecture and enzymatic activity and shows purifying selection (12). By functionally analyzing the loop L4 of primate Mx1 proteins, the phenylalanine F561 in huMxA was identified to be critical for viral target recognition and antiviral specificity against IAV and Thogoto virus (THOV) (12, 13). However, the structural requirements of the loop L4 for the antiviral activity of other mammalian Mx1 proteins have not been studied in detail.

FIG 2.

Role of loop L4 in the antiviral effect of eqMx1 against THOV. (a) AlphaFold structure prediction of the huMxA (UniProt P20591) monomer in ribbon-type representation. The G domain is shown in orange, the bundle signaling element (BSE) in red, the stalk in green, and loop L4 in blue. The positions of critical amino acids, threonine 103, phenylalanine 561, and glycine 562, are indicated. The amino acid alignment depicts the C-terminal part of huL4(558–573) and eqL4(558–571). The aromatic residues F561 and W562 and the adjacent glycines in huMxA and eqMx1 are highlighted in boldface. The eqMx1(W562C) variant is shown in red. Gaps are indicated by hyphens. (b) Schematic representation of the complete L4 of huMxA(530–573) (black) and eqMx1(530–571) (white). Replaced parts of L4 of huMxA(558–569) or eqMx1(558–567) are indicated. (c) L4 loops of eqMx1 and huMxA are functional analog modules. Antiviral activities of Mx1 proteins were tested in 293T cells transfected with the components of the THOV polymerase system, as described in Fig. 1a, with 300 ng of the Mx1 expression plasmids for 24 h. Mx1(T103A) was used as inactive control. Firefly and Renilla luciferase activities were determined. Firefly luciferase activity was normalized to the Renilla luciferase activity, and the control without Mx1 (NP+) was set to 100% (mean ± SD; n = 3). Significance was calculated using one-way ANOVA (Tukey’s multiple-comparison test; ***, P < 0.001; ****, P < 0.0001; ns, not significant). The expression of Mx1, NP, and actin was controlled by Western blotting analyses. (d) Coprecipitation of THOV NP with Mx1. 293T cells in 6-wells were transfected with expression plasmids encoding the components of the polymerase reconstitution system of THOV, including 60 ng of PB1, PB2, and PA, 600 ng of NP, 300 ng of the pPol-I FF-Luc minigenome, and 900 ng of Flag-tagged Mx1 expression plasmids for 48 h. Cotransfected Flag-tagged chloramphenicol acetyltransferase (CAT) was used as a specificity control. At 48 h after transfection, the cells were lysed and the Flag-tagged Mx1 proteins were precipitated with Flag-specific antibodies. Flag-Mx1 and coprecipitated THOV NP as well as expression of the proteins in the whole-cell lysates (WCL) were detected by Western blotting analyses using anti-Flag- and anti-THOV NP-specific antibodies.

Seasonal and pandemic human IAVs showed reduced sensitivity to huMxA both in cell culture and in vivo (7, 14). Further analyses pointed to the viral nucleoprotein (NP) as the primary target structure of Mx1 action (15). Differences in Mx1 sensitivities of IAVs were then used for in-depth functional analyses and resulted in the identification of distinct patterns of surface-exposed amino acid residues in NP that reduce the sensitivity of seasonal IAVs and lead to the escape from huMxA restriction (16).

For the present study, we used THOV and Dhori virus (DHOV), two members of the Thogotovirus genus within the Orthomyxoviridae family. Similar to influenza viruses, thogotoviruses are enveloped and have a segmented single-stranded RNA genome of negative polarity. However, thogotoviruses have six genomic segments instead of eight and are transmitted by ticks (17). According to their phylogeny, thogotoviruses can be subdivided into two clades, the Thogoto-like viruses and the Dhori-like viruses (18). Similar to the host range of their reservoir tick species, thogotoviruses were found to infect a broad range of domestic and wild animals (19–25). In order to productively infect various mammalian species, these viruses have to overcome restriction by the host’s innate immune system, including the interferon (IFN)-induced, antiviral Mx proteins. Therefore, thogotoviruses represent a valuable tool to analyze the function and specificity of Mx1 proteins originating from different mammalian species. Interestingly, all thogotoviruses are sensitive to mmMx1 (26), but in contrast to THOV, DHOV is insensitive to huMxA (27–29). We recently discovered that also the THOV-like Jos virus (JOSV) is insensitive to huMxA. By comparing the NP structures of THOV and JOSV, we identified two surface-exposed amino acid residues that confer the reduced sensitivity to huMxA. This supports the idea that the viral NP is the primary viral target of huMxA (16, 29).

By analyzing the antiviral potential of different mammalian Mx1 proteins, only equine Mx1 (eqMx1) showed antiviral activity comparable to huMxA against THOV. A tryptophan (W562) and a glycine (G563) in the loop L4 of eqMx1 were identified as the critical residues determining this activity. Interestingly, other positions in the C-terminal part of equine loop L4 (eqL4) could be converted to alanines without functional repercussions. Finally, our mutational analysis allowed the design of an optimized eqL4 structure that conferred antiviral activity even against THOV escape variants.

RESULTS

Antiviral activity of equine eqMx1 against THOV.

To determine the effect of Mx1 from various species on viral replication of THOV and DHOV, we used polymerase reconstitution systems in the presence of various mammalian Mx1 proteins to compare the sensitivities of the two viruses (13). In this minireplicon system, expression plasmids encoding the three subunits of the viral polymerase, the viral NP, and an artificial minigenome encoding the firefly luciferase (FF-Luc) as a reporter are cotransfected. The resulting reporter activity directly correlates with the viral polymerase activity. Most of the Mx1 proteins were inactive against THOV except for huMxA, orangutan Mx1 (oruMx1), and, unexpectedly, eqMx1 (Fig. 1a). However, all tested mammalian Mx1 proteins were fairly inactive against the DHOV system (Fig. 1b).

FIG 1.

Antiviral activity of eqMx1 against THOV. (a and b) Antiviral activity of mammalian Mx1 proteins as determined by reconstituted viral polymerase activity assay. 293T cells in 12-well plates were transfected with expression plasmids encoding the components of the polymerase reconstitution system of THOV (a) or DHOV (b), including 10 ng of PB1, PB2, and PA, 50 ng of NP, 50 ng of the pPol-I FF-Luc minigenome, and 10 ng of Renilla luciferase (R-Luc) under the constitutive SV40 promoter. Three hundred nanograms of Flag-tagged MX1 expression plasmids was cotransfected. huMxA(T103A) is an inactive mutant of huMxA. At 24 h after transfection, cells were lysed and firefly and Renilla luciferase activities were determined. Firefly luciferase activity was normalized to Renilla luciferase activity, and the control without Mx1 (NP+) was set to 100% for the respective polymerase system (mean ± SD; n = 3). The expression of Mx1, NP, and actin was controlled by Western blotting analyses. Significance compared to the NP+ control was calculated using one-way analysis of variance (ANOVA) (Tukey’s multiple-comparison test; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant). The following different mammalian Mx1 proteins were tested: bat (Eidolon helvum), ehMx1; ferret, frMx1; canine, caMx1; porcine, poMx1; bovine, boMx1; equine, eqMx1; African green monkey, agmMx1; orangutan, oruMx1; and human, huMxA. (c and d) huMxA- and eqMx1-expressing cell cultures. Huh7 control cells, treated with IFN-α (500 U/mL for 24 h) or cells expressing Flag-tagged huMxA, eqMx1, or the inactive variants huMxA(T103A) and eqMx1(T103A) were analyzed for the expression of the recombinant Mx1 proteins by Western blotting (c) or immunofluorescence analyses (d) using an Mx-specific antibody (green). The nuclei were stained with DAPI (blue). Actin was used as a loading control. (e and f) Huh7 cells expressing recombinant Mx1 proteins were infected with THOV (e) or DHOV (f) (MOI of 0.001). Viral progeny in the supernatants were determined by plaque assay at the indicated time points (log-transformed values; mean ± SD; n = 3). Statistical analyses were performed using two-way ANOVA comparing Mx1(wt) to the mutant Mx1 variants (Tukey’s multiple-comparison test; ***, P < 0.001; ****, P < 0.0001; ns, not significant).

To confirm this antiviral effect of eqMx1, we generated Huh7 cell lines that constitutively express N-terminally Flag-tagged eqMx1 and huMxA as well as their corresponding GTPase-deficient and antivirally inactive (T103A) variants (30). Western blotting of the transduced cell lines using a pan-Mx1-specific antibody (31) showed uniform expressions of the respective Mx1 proteins similar to IFN-α induced endogenous huMxA (Fig. 1c). Immunofluorescence analysis of huMxA and eqMx1 showed comparable cytoplasmic distribution patterns in these cells (Fig. 1d). To determine the antiviral Mx1 effect during virus replication, the cells were infected with a low dose of THOV or DHOV, and progeny virus was detected in the supernatants by plaque assay. In cell lines transduced with empty vector or the T103A mutants, we observed a constant accumulation of infectious THOV over time, whereas THOV progeny virus could not be detected in the supernatants of wild-type (wt) huMxA- and eqMx1-expressing cells (Fig. 1e). In line with the minireplicon data, DHOV replication was not affected by the expression of any of the Mx1 proteins (Fig. 1f).

We previously showed that amino acids in the loop L4 of the stalk domain (Fig. 2a) determine the antiviral specificity of primate Mx1 proteins (12, 13). The critical region for this activity is the variable C-terminal part of loop L4 (amino acids 558 to 573 in huMxA) flanked by a conserved stretch of four lysines (K554 to 557) and the first methionine (M574) of the α4-helix in the huMxA stalk region (see Fig. S1 in the supplemental material). Comparison of this region with that of eqMx1 showed that tryptophan W562 of eqMx1 (NM_001082492) is aligned with the phenylalanine F561 in huMxA, which has been shown to be essential for its antiviral activity (12). However, a second sequence submission of eqMX1 (XM_005606071) has a cysteine instead of tryptophan at position 562 (Fig. 2a). To analyze the functional compatibility of eqL4 and huL4, we generated Mx1 chimeras by introducing eqL4(558–568) into the corresponding site of huMxA, creating huMxA(eqL4), and vice versa by inserting huL4(558–570) into eqMx1, creating eqMx1(huL4) (Fig. 2b). Furthermore, we introduced the W562C exchange into eqMx1 or huMxA(eqL4) to assess the effect of the cysteine at this position.

In the minireplicon assay, the exchange of huL4 for eqL4 in huMxA(eqL4) reduced the inhibitory effect of huMxA on the THOV polymerase activity from 90% to 50%. However, the introduction of huL4 into eqMx1, creating eqMx1(huL4), did not affect the antiviral potential of eqMx1 (Fig. 2c). Of note, the tryptophan-to-cysteine exchanges in eqMx1(W562C) and huMxA(eqL4/W562C) completely diminished their antiviral activity similar to the inactive T103A controls (Fig. 2c). The cells were transfected with equal amounts of Mx1 expression plasmids. Variations in the Western blot signals of viral NP seem to correlate with the actin signals, indicating comparable NP expression levels in the different assays.

Next, we analyzed the L4 chimeras for interaction with THOV NP, the viral target structure of huMxA. To this end, we precipitated Flag-tagged Mx1 proteins from cell lysates that were cotransfected with Mx1 expression constructs and the components of the minireplicon system or mixed with lysates of THOV-infected cells. THOV NP specifically coprecipitated with Flag-huMxA as described previously (32), whereas it only weakly coprecipitated with Flag-eqMx1 (Fig. 2d). Accordingly, huMxA(eqL4) showed reduced capacity and eqMx1(huL4) showed enhanced capacity to precipitate viral NP. Together, these experiments suggest that the strong antiviral activity of eqMx1 against the THOV polymerase complex is not reflected by a robust interaction with THOV NP as observed with huMxA.

Molecular determinants of loop L4 antiviral function in eqMx1.

In order to further analyze the structure of the C-terminal part of eqL4, we exchanged sets of amino acids to alanines (Fig. 3a). Most of the exchanges did not grossly influence the antiviral activity of eqMx1 except for amino acids in the vicinity of W562 (Fig. 3b). Single amino acid exchanges to alanine at positions 561 to 563 revealed the critical importance of tryptophan W562 and the neighboring glycine, G563. The Mx1 bands in the corresponding Western blots showed some variability that could have resulted due to differences in transfection efficacy or be attributed to differences in protein stability. Therefore, we compared the stability of a set of selected eqMx1 mutants with that of the wild-type protein by pulse-chase experiments, tracking ³⁵S-labeled Mx1 proteins over 22 h (Fig. S2a). The quantitative analysis of these experiments revealed a half-life of wild-type eqMx1 [eqMx1(wt)] of about 22 h, comparable to that of huMxA (33). The different mutations introduced into loop L4 of eqMx1 did not diminish their stability significantly (Fig. S2b and c), indicating that the mutations affected the antiviral function but not the overall protein stability of eqMx1.

FIG 3.

Alanine scanning of eqL4 points to positions required for antiviral activity. (a) Structure of the C-terminal moiety of eqL4(558–571). Examples of amino acid exchanges to alanines within the L4 region are indicated in red. (b) eqMx1 variants were tested in the THOV polymerase assay, as described in Fig. 2c, with 300 ng of the eqMx1 expression plasmids cotransfected. eqMx1(T103A) was used as an inactive control. After 24 h, firefly and Renilla luciferase activities were determined. Firefly luciferase activity was normalized to the Renilla luciferase activity, and the control without Mx1 (NP+) was set to 100% (mean ± SD; n = 3). Significance was calculated using one-way ANOVA (Tukey’s multiple-comparison test; *, P < 0.05; ***, P < 0.001; ****, P < 0.0001; ns, not significant). The expression of eqMx1, NP, and actin was controlled by Western blotting analyses. (c and d) Structural requirements for critical positions 562 and 563 in eqL4. Positions W562 and G563 were exchanged to different amino acids and tested in the THOV polymerase assay as described in panel b. (e) Coprecipitation of THOV NP with eqMx1(W562) variants. 293T cells in 6 wells were transfected with 1,000 ng of expression plasmids encoding Flag-tagged eqMx1 for 48 h. Separately, 293T cells were infected with THOV (MOI of 1) for 24 h. The transfected and the infected cell lysates were mixed and the Flag-tagged eqMx1 proteins were precipitated. Flag-eqMx1 variants and coprecipitated THOV NP as well as expression of the proteins in the whole-cell lysates (WCL) were detected by Western blotting analyses using anti-Flag- and anti-THOV NP-specific antibodies.

Amino acid exchanges of positions 562 and 563 to various acidic, basic, hydrophobic, or neutral amino acids demonstrated an absolute requirement of glycine at position 563 in close proximity to a bulky tryptophan, phenylalanine, or tyrosine at position 562 (Fig. 3c and d). Interestingly, the W562F or W562Y exchanges increased the capacity of eqMx1 to coprecipitate the viral NP (Fig. 3e). In contrast, wild-type eqMx1(W562) was very inefficient in coprecipitating THOV NP, as already shown in Fig. 2d, suggesting that the hydrophobic phenol ring stabilizes the interaction of eqMx1 with the viral NP without a significant increase in its antiviral activity compared to W562 (Fig. 3c).

To further underline the critical importance of W562 and G563 for the function of eqL4, we mutated the surrounding residues to alanines (Fig. 4a). Converting the three upstream and four downstream residues to alanines in eqMx1, creating eqMx1(3A/WG/4A), did not affect the antiviral activity. However, insertion of additional alanines led to a gradual reduction of the antiviral effect (Fig. 4b). Based on the eqMx1(3A/WG/4A) construct, we again analyzed the structural requirement of position 562. Converting W562 to phenylalanine or tryptophan slightly reduced the antiviral activity of the eqMx1 construct (Fig. 4c). However, both residues increased THOV NP coprecipitation (Fig. 4d) to a similar extent as observed for the introduction of phenylalanine or tyrosine into wild-type eqL4 (Fig. 3e). These data emphasize the central importance of positions 562 and 563 for the antiviral effect and the coprecipitation of viral NP with eqMx1.

FIG 4.

Structural analysis of eqL4. (a) Structure of the critical C-terminal part of eqL4(558–571). Examples of amino acid exchanges to alanines are indicated in red. (b and c) Stretches of amino acids in L4 of eqMx1 surrounding W562 and G563 were exchanged to alanines (b) and W562 in the 3A/WG/4A construct was exchanged with other hydrophobic amino acids (c). The THOV polymerase assay was performed, as described in Fig. 2c, with 100 ng of the eqMx1 expression plasmids cotransfected. eqMx1(T103A) was used as an inactive control. After 24 h, firefly and Renilla luciferase activities were determined. Firefly luciferase activity was normalized to the Renilla luciferase activity, and the control without Mx1 (NP+) was set to 100% (mean ± SD; n = 3). Significance was calculated using one-way ANOVA (Tukey’s multiple-comparison test; *, P < 0.05; ****, P < 0.0001; ns, not significant). The expression of eqMx1, THOV NP, and actin was controlled by Western blotting analyses. (d) Coprecipitation of THOV NP with eqMx1 variants. 293T cells in 6 wells were transfected with expression plasmids encoding the components of the polymerase reconstitution system and Flag-tagged eqMx1 as described in Fig. 2d. At 48 h after transfection, the cells were lysed and the Flag-eqMx1 proteins were precipitated. Flag-eqMx1 variants and coprecipitated THOV NP as well as expression of the proteins in the whole-cell lysates (WCL) were detected by Western blotting analyses using anti-Flag- and anti-THOV NP-specific antibodies.

Transferring antiviral activity to inactive bovine Mx1.

In contrast to eqMx1, bovine Mx1 (boMx1) showed no antiviral activity in the THOV minireplicon assay (Fig. 1a). A comparison of the amino acid sequences of the C-terminal parts of the loops L4 of boMx1, eqMx1, and huMxA revealed the lack of the glycine downstream of the phenylalanine (F557) in boMx1 L4 (Fig. 5a). In an attempt to reconstitute the antiviral activity of boMx1, we inserted an additional glycine after F557. Indeed, this modified boMx1, boMx1(F557+G), showed antiviral activity in the THOV minireplicon system comparable to that of eqMx1 (Fig. 5b). In order to test the specificity of this insertion, the additional glycine was converted to alanine in boMx1(F557+A) or the F557 was exchanged to alanine in boMx1(F557A+G) (Fig. 5a). Both constructs were antivirally inactive (Fig. 5b). Based on these observations, we suggest that both a bulky residue and the glycine that follows it in the C-terminal part of loop L4 are critical requirements for anti-THOV activity of mammalian Mx1 proteins.

FIG 5.

Restoration of boMx1 activity against THOV. (a) Amino acid alignment of loop L4 of boMx1 compared to eqL4 and huL4. Single amino acid changes introduced in boL4 are indicated in red. Gaps are indicated by hyphens. (b) Antiviral activity of boMx1 against THOV in the polymerase assay, as described in Fig. 2c, with 300 ng of the boMx1 expression plasmids cotransfected. eqMx1(delL4), with deletion of residues 533 to 562 in eqL4, was used as an inactive control. After 24 h, firefly and Renilla luciferase activities were determined. Firefly luciferase activity was normalized to the Renilla luciferase activity, and the control without Mx1 (NP+) was set to 100% (mean ± SD; n = 3). Significance was calculated using one-way ANOVA (Tukey’s multiple-comparison test; *, P < 0.05; ****, P < 0.0001; ns, not significant). The expression of Mx1 and THOV NP was controlled by Western blotting analyses.

THOV NP variant escapes the antiviral effect of eqMx1.

A recent analysis of THOV NP variants identified two surface-exposed mutations, G327R and R328V (Fig. 6a), that led to the escape of THOV polymerase activity from the antiviral effect of huMxA and to the loss of huMxA-NP coprecipitation (Fig. 6b and c) (29). Similar as huMxA, eqMx1 failed to block THOV encoding either NP(G327R) or NP(G327R/R328V) (Fig. 6b). Surprisingly, NP(R328V) was still as sensitive as wild-type NP to eqMx1 (Fig. 6b), suggesting that eqMx1 targets a slightly different interface on the viral NP compared to huMxA.

FIG 6.

Antiviral activity of eqMx1 against THOV escape mutants. (a) AlphaFold protein structure prediction of THOV strain SiAr126 nucleoprotein colored according to the electrostatic potential. The positions of the MxA escape mutations, as described by Fuchs et al. (29), are indicated in green (G327 and R328). The three THOV NP mutants used in this study are listed to the left. (b) Activity of huMxA and eqMx1 against the NP mutants was determined in the polymerase assay, as described in Fig. 2c, with 100 ng of the Mx1 expression plasmids cotransfected. huMxA(delL4) and eqMx1(delL4) were used as inactive controls. After 24 h, firefly and Renilla luciferase activities were determined. Firefly luciferase activity was normalized to the Renilla luciferase activity, and the control of huMxA(delL4) was set to 100% (mean ± SD; n = 3). Significance was calculated using one-way ANOVA (Tukey’s multiple-comparison test; ****, P < 0.0001; ns, not significant). The expression of Mx1 and THOV NP was controlled by Western blotting analyses. (c) Coprecipitation of the THOV NP mutants with huMxA. 293T cells in 6 wells were transfected with expression plasmids encoding the components of the polymerase reconstitution system and Flag-tagged huMxA as described in Fig. 2d. Flag-huMxA(delL4) was used as an inactive control. At 48 h after transfection, the cells were lysed and the Flag-tagged huMxA proteins were precipitated. Flag-huMxA and coprecipitated THOV NP as well as expression of the proteins in the whole-cell lysates (WCL) were detected by Western blotting analyses using anti-Flag- and anti-THOV NP-specific antibodies.

Next, we analyzed the effects of tryptophan, phenylalanine and tyrosine at position 562 in eqMx1 and eqMx1(3A/WG/4A) for their antiviral activity on the THOV NP mutants in the minireplicon system. As a control for an inactive Mx protein, we mutated position 562 to alanine. NP(wt) showed the same sensitivity against all active eqMx1 constructs (Fig. 7a and b). Interestingly, the NP(G327R) escape mutant resistant to wild-type eqMx1(W562) was inhibited by the eqMx1 variants with either W562F or W562Y substitutions in eqL4 (Fig. 7a and b). These eqMx1 variants also showed enhanced antiviral activity in the NP(R328V)-based minireplicon system (Fig. 7b). Of note, residual antiviral activity by eqMx1(W562Y) was even observed with the highly resistant NP(G327R/R328V) protein (Fig. 7a). Accordingly, NP(R328V) and to a lesser extent NP(G327R) but not NP(G327R/R328V) coprecipitated with the W562Y-mutated Mx1 variants (Fig. 7c and d). Thus, our analyses unveiled the potential of the eqMx1 loop L4 to increase its antiviral activity by the exchange of tryptophan W562 to phenylalanine or tyrosine against antiviral escape variants of THOV NP.

FIG 7.

Antiviral activity of eqMx1 L4 variants against THOV escape mutants. (a and b) Activity of eqMx1 variants against the NP mutants was determined in the THOV polymerase assay, as described in Fig. 2c, with 100 ng of the eqMx1 expression plasmids cotransfected. eqMx1(T103A) (a) and eqMx1(YWG-AAA) (b) were used as inactive controls, respectively. After 24 h, firefly and Renilla luciferase activities were determined. Firefly luciferase activity was normalized to the Renilla luciferase activity, and the controls eqMx1(T103A) (a) and eqMx1(YWG-AAA) (b) were set to 100% (mean ± SD; n = 3). Significance was calculated using one-way ANOVA (Tukey’s multiple-comparison test; *, P < 0.05; **, P < 0.01; ****, P < 0.0001; ns, not significant). The expression of eqMx1 and THOV NP was controlled by Western blotting. (c and d) Coprecipitation of the THOV NP mutants with the eqMx1 variants. 293T cells in 6-well plates were transfected with expression plasmids encoding the components of the polymerase reconstitution system and Flag-tagged eqMx1 as described in Fig. 2d. Flag-eqMx1(W562A) and Flag-eqMx1(3A/AA/4A) were used as inactive controls. At 48 h after transfection, the cells were lysed and the Flag-tagged eqMx1 proteins were precipitated. Flag-eqMx1 variants and coprecipitated THOV NP as well as expression of the proteins in the whole-cell lysates (WCL) were detected by Western blotting analyses using anti-Flag- and anti-THOV NP-specific antibodies.

DISCUSSION

We recently demonstrated antiviral activity of eqMx1 against THOV (29), while two recent publications showed an antiviral effect of eqMx1 against IAV (34, 35). By analysis of a collection of mammalian Mx1 proteins, we here confirmed the unforeseen antiviral specificity of eqMx1 against THOV, comparable to that of huMxA and oruMx1, while all other tested mammalian Mx1 proteins were inactive. However, DHOV, another member of the Thogotovirus genus, was blocked neither by eqMx1 nor huMxA as previously reported (27–29). Thus, the surprisingly narrow spectrum of antivirally active Mx1 proteins against THOV differs from the broader spectrum against IAV (34).

By detailed molecular analyses of eqMx1, we found that the loop L4 and especially W562 and G563 are crucial for this antiviral effect. For primate Mx1 proteins, the loop L4 that protrudes from the compact structure of the stalk was shown to be an important module determining antiviral specificity (12). An amino acid sequence alignment of various mammalian L4 loops (see Fig. S1 in the supplemental material) shows that only eqMx1 has some similarity to hominoid Mx1 in the variable C-terminal part (from positions 558 to 573). We propose that this fulfills the basic structural requirement for an antiviral function against THOV: a bulky residue at position 562 and an adjacent tiny residue, G563, playing similar roles to F561 and G562 in huL4 (see alignment in Fig. 2a) as speculated recently (34).

In the NCBI GenBank database, two cDNA sequences of eqMx1 are available. The most remarkable difference between these two sequences is an amino acid exchange in eqL4 at position 562 with either tryptophan or cysteine. By introducing this cysteine, we completely lost the antiviral activity of eqMx1 against THOV, emphasizing the importance of W562 in eqL4. Our mutational analysis of this L4 area confirmed that a bulky, hydrophobic, and aromatic residue (tryptophan, phenylalanine, or tyrosine) at position 562 and a glycine at position 563 are indispensable for anti-THOV activity. The importance of these two residues has been previously demonstrated in several publications: by reconstitution of the antiviral activity of the inactive African green monkey Mx1 (agmMx1) (Fig. S1) through the exchange of the valine 561 to phenylalanine in agmL4 (12), by an alanine scanning analysis of huL4 (13), and by a random mutagenesis approach in huL4 (36). The requirement of the glycine downstream of the bulky residue was confirmed in the present study for boMx1, which gained anti-THOV activity after the introduction of a missing glycine adjacent to the critical bulky residue F557. This arrangement of a bulky residue at position 562 combined with a tiny glycine in eqL4 suggests the formation of a platform at the tip of the loop L4 that might represent an interface for direct viral target recognition. Accordingly, we found that most of the C-terminal part of eqL4 could be converted to alanines except for W562 and G563. However, increasing numbers of alanine residues gradually reduced the antiviral activity, pointing to a modulatory role of the amino acids surrounding W562 and G563 amino acids, which might stabilize the molecular interactions within Mx1 oligomers or interactions between Mx1 and the viral target.

In previous studies, we showed that L4 determines the antiviral activity as well as the interaction of huMxA with THOV NP in coprecipitation experiments (13, 32). Therefore, it was rather unexpected that eqMx1 had only a minimal capacity to precipitate the viral NP, although it showed a strong antiviral activity against THOV comparable to that of huMxA. The swapping experiments demonstrated that the reduced Mx1-NP coprecipitation was due to the eqL4 and especially W562, suggesting that the primary structure of L4 determines the Mx1-NP interaction.

The missing correlation between the antiviral activity and NP coprecipitation suggests two alternative strategies of how Mx1 proteins block THOV replication as also discussed for IAV previously (37, 38). It has been shown that huMxA inhibits THOV by interacting with the viral ribonucleoprotein complexes (vRNPs) (13, 32) and interferes with their nuclear import, resulting in an early block of viral replication (39). Alternatively, it was speculated that huMxA can affect a later step of the viral replication cycle (40) by preventing the nuclear import of newly synthesized viral NP (38). Our current study suggests that the amino acid composition of the central L4 core around F561 and G562 in huMxA (12, 13) and W562 and G563 in eqMx1 (this study) in concert with surrounding amino acid residues (36) might decide the mode of antiviral action. Whereas huMxA interacts with incoming vRNPs and prevents their transport into the nucleus, eqMx1 might affect a later step in THOV replication independent of a direct NP interaction. Moreover, additional cellular components might be involved in these processes. For example, the UAP56 RNA helicase that interacts with IAV NP (41) supports viral mRNA transport (42) and was shown to interact with huMxA (43). Likewise, components of the cytoplasmic transport pathways, including actin, tubulin, or kinesin (44, 45), might play important roles for the proper transport and localization of the viral proteins.

In recent years, several studies pointed to the viral NP as the target of Mx1. Sequence and functional analyses of IAV identified amino acid motifs on the surface of the viral NP that determine sensitivity to mammalian Mx1 proteins (16, 35, 46, 47). In a similar region of the body domain of THOV NP, we recently identified two amino acid exchanges, NP(G327R/R328V), that abolished the sensitivity of the THOV strain SiAr126 to huMxA (29). Interestingly, this mutant NP(G327R/R328V) also escaped inhibition by eqMx1. Unexpectedly, however, the single exchange in NP(R328V) was not sufficient to escape from wild-type eqMx1 activity, although it conferred resistance to huMxA, supporting our hypothesis of two alternative mechanisms of antiviral action. Whereas huMxA inhibits an early step of THOV replication that the virus can escape by the NP(R328V) mutation, eqMx1 can inhibit a later step that the mutant virus is not able to escape. These modes of Mx1 action might have been shaped during the evolutionary virus-host arms race of thogotovirus-like viruses in their mammalian hosts.

Thogotoviruses are transmitted to a wide range of mammals by their tick vectors (19–25); therefore, horses are likely potential hosts for thogotoviruses. Our data suggest that position 562 in eqMx1 marks one side of the host-virus interface that is under selection pressure during an evolutionary virus-host arms race between horses and thogotoviruses. Prolonged prevalence of thogotoviruses in horse populations could result in changes at position 562, representing a possible strategy to enhance the antiviral capacity of eqMx1. However, it remains unclear why MX1 genes from other species infected with thogotoviruses, such as cattle, were not selected for more optimized antiviral L4 structures.

HuMxA and other mammalian Mx1 proteins have maintained a broad antiviral activity against a wide range of RNA and DNA viruses, among others, orthomyxoviruses, bunyaviruses, and vesicular stomatitis virus (3, 9, 48). Whether the mutations we describe in the present eqMx1 loop L4 study might contribute to the antiviral breadth of Mx1 proteins against other viruses, as discussed recently (36), remains to be investigated. Likewise, future studies are required to elucidate whether the enhanced activity of eqMx1(W562Y) against THOV escape variants might impact the antiviral control of equine IAVs that are of epidemiological and economic importance in horse breeding and keeping facilities (49, 50).

MATERIALS AND METHODS

Cell lines, culture, and interferon treatment.

Human hepatoma Huh7 cells (51), African green monkey kidney Vero cells (ATCC CCL-81), and human embryonic kidney 293T cells (ATCC CRL-3216) were cultivated in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (FCS) at 37°C and 5% CO₂. Huh7 cells were treated with recombinant human IFN-αB/D (52) for 24 h before Western blotting.

For the generation of stable Huh7 cell lines expressing eqMx1 or huMxA, lentiviral transduction was used. The MX1 cDNAs were cloned into the lentiviral pLVX-Puro expression vector (Clontech). Lentiviruses were produced in 293T cells by cotransfecting pLVX empty vector, pLVX-eqMx1(wt), pLVX-eqMx1(T103A), pLVX-huMxA(wt), or pLVX-huMxA(T103A), as well as pCMVR8.91 and pMDG(VSV-G) for 72 h as previously described (53). Huh7 cells were infected with the lentivirus, and selection with 2 μg/mL puromycin was started at 48 h postransduction.

Virus infections.

For the present study, we used thogotovirus (THOV) strain SiAr/126/72 (54) and Dhori virus (DHOV) strain India/1313/61 (55). All infection experiments were performed under biosafety level 2 (BSL2) conditions.

For growth kinetics, Huh7 cells in a 6-well format (9 × 10⁵ cells) were infected at a multiplicity of infection (MOI) of 0.001 in DMEM without FCS for 1.5 h at room temperature. Then, the cells were washed three times with phosphate-buffered saline (PBS) and incubated with DMEM containing 2% FCS and 20 mM HEPES. Virus-containing supernatants were stored at –80°C, and viral titers were determined by plaque assay on Vero cells.

Bioinformatics.

The GenBank accession numbers for the proteins studied are as follows: for huMxA, NM_001178046; oruMx1, NM_001134146; agmMx1, KJ650325 (12); eqMx1, NM_001082492 and XM_005606071 (34, 35); bat (Eidolon helvum) Mx1 (ehMx1), KR362562 (56); ferret Mx1 (frMx1), JF906988 (57); canine Mx1 (caMx1), NM_001003134 (58); porcine Mx1 (poMx1), NM_214061 (16); and boMx1, NM_173940 (repaired by an 18-nucleotide insertion between nucleotides 73 and 74 of the open reading frame [ORF]).

The protein structures of huMxA (UniProt no. P20591) and THOV(SiAr126) NP (UniProt no. P89216) were modeled using AlphaFold (59) within the ColabFold (60) framework. Processing and annotation of protein structures were performed with UCSF ChimeraX v.1.2 (61) software.

Cloning of eqMx1 expression plasmids.

Equine MX1 (accession no. NM_001082492) and human MX1 (accession no. NM_001178046) were cloned into pCAGGS expression plasmids using EcoRI and XhoI restriction sites (62). All MX1 cDNAs carried a Flag epitope at the N terminus after the ATG start codon and were described previously (29). Mutations and chimeric exchanges were introduced by PCR with overlapping primers as described previously (13).

Viral polymerase reconstitution system.

To reconstitute the polymerase activity of THOV(SiAr126) or DHOV(India/61), 293T cells in 12-well plates (4 × 10⁵ cells) were cotransfected (JetPEI; Polyplus) with 10 ng of pCAGGS expression plasmids encoding the polymerase subunits PB2, PB1, PA and 50 ng of NP plasmids as previously described (13). In addition, 50 ng of an artificial viral minigenome encoding firefly luciferase in negative-sense orientation flanked by viral noncoding regions (Pol-I-FF-Luc) and 10 ng of a plasmid coding for a Renilla luciferase under the constitutive SV40 promoter (SV40-RL) were added. To determine the effect of Mx1 proteins on the activity of the viral polymerase, 100 to 300 ng each of various Mx1 expression constructs were cotransfected. At 24 h posttransfection, firefly and Renilla luciferase activities were measured (Dual-luciferase reporter kit; Promega). Firefly luciferase activity was normalized to Renilla luciferase activity. Western blotting of the cell lysates was performed with the following specific antibodies: FLAG-M2 (Sigma-Aldrich) or M143 against Mx1 (31), SiAr126 NP (32), and β-actin (Sigma-Aldrich). DHOV NP was detected using a polyclonal rabbit antiserum raised against the purified His-tagged NP produced in Escherichia coli as described previously (32).

Fluorescence microscopy.

For immunofluorescence microscopy analysis, the cells were seeded onto coverslips. The cells were fixed in paraformaldehyde (4% in PBS) for 15 min at room temperature and washed with PBS. Afterwards, the cells were permeabilized with 0.5% Triton X-100 in PBS. After undergoing blocking for 1 h with blocking buffer (PBS with 1% bovine serum albumin [BSA] and 0.1% Tween 20), the coverslips were incubated with the primary antibody for 1 h at room temperature. Unspecific binding was reduced by washing with PBS. The secondary antibody was incubated in the dark at room temperature for 1 h. After washing, the cells were stained with DAPI (4′,6-diamidino-2-phenylindole) (0.3 μM in PBS) for 10 min, washed with PBS, and mounted onto microscope slides using FluorSave reagent (Millipore). Pictures were taken with an Axioplan-2 (Carl Zeiss AG). The primary antibodies that were used for immunofluorescence were rabbit polyclonal anti-huMxA (32).

Metabolic labeling and immunoprecipitation.

To determine the stability of eqMx1 variants, 293T cells in 6-well plates (9 × 10⁵ cells) were transfected with 2 μg of pCAGGS expression plasmids encoding eqMx1 proteins. At 14 h posttransfection, newly synthesized proteins were metabolically labeled with [³⁵S]methionine (6.3 MBq/mL) (Perkin Elmer) for 2 h and then chased for 5, 10, and 22 h as described previously (33). Cells were lysed in 800 μL immunoprecipitation (IP) buffer (50 mM Tris [pH 8.0], 150 mM NaCl, 1.0% Triton X-100, 0.05% SDS, 1 mM dithiothreitol, and protease inhibitors [Roche]) and immunoprecipitated using protein A Sepharose (GE Healthcare) and the M143 monoclonal antibody against Mx1 (31). The Sepharose pellets were washed three times with IP buffer. Then, the precipitated proteins were dissociated in SDS sample buffer at 95°C and separated by SDS-PAGE, followed by PhosphorImager analysis of the fixed and dried gels (Typhoon FLA 7000; GE Healthcare). The PhosphorImager signals were quantified using the ImageJ v.1.53 software and further processed using GraphPad Prism9, and the half-life was calculated using the GraphPad Prism9 one-phase decay nonlinear regression model. Expression of transfected recombinant eqMx1 was analyzed by Western blotting of the whole-cell lysates using the M143 antibody.

Coimmunoprecipitation.

Coimmunoprecipitation (CoIP) studies were performed as described previously (13), using HEK-293T cells seeded in 6-well plates (9 × 10⁵ cells per well) and transfected with pCAGGS expression plasmids: 60 ng of each PB1, PB2, and PA, 600 ng of NP, 300 ng of pPol-I FF-Luc, and 900 ng of FLAG-huMxA or FLAG-eqMx1. Alternatively, the cells were transfected only with 1,000 ng FLAG-huMxA or FLAG-eqMx1 for 48 h and another 6-well plate of confluent 293T cells was infected with THOV at an MOI of 1 for 24 h. The cells were lysed on ice in 800 μL CoIP buffer (50 mM Tris [pH 8.0], 150 mM NaCl, 2 mM Mg acetate, 0.2% NP-40, 1 mM dithiothreitol, and 50 μM GTP-γS) containing protease inhibitors (Roche). The lysates of the transfected cells or the mixed lysates of the transfected and the infected cells were incubated with 25 μL anti-FLAG M1 agarose affinity gel (Sigma) at 4°C for 2 h. The beads were washed five times with 900 μL CoIP buffer at 4°C, and the precipitated proteins were subjected to Western blot analysis as described above.

Statistical analyses.

Data were visualized and statistically evaluated with GraphPad Prism 9.4.1. Viral titers in cell culture supernatants are displayed as log-transformed values on a linear scale (mean with standard deviation). The polymerase reconstitution assay data were visualized as mean ± standard deviation (SD) (n = 3).

Data availability.

The GenBank accession numbers for equine MX1 are NM_001082492 and XM_005606071.