Characterization of the amino-terminal domain of Mx2/MxB-dependent interaction with the HIV-1 capsid

More than 50 years have passed since the myxovirus resistance (MX) genes were first discovered and found to suppress infection with influenza viruses in mice (Lindenmann, 1962). Likemost mammals, mice carry twoMX genes,MX1 andMX2, which have arisen by gene duplication; both of these genes exhibit antiviral activity against a wide range of viruses (Liu et al., 2013). Humans also have two MX genes, encoding the MxAandMxBproteins, which are interferon-induced, dynaminlike large molecular weight guanosine triphosphatases (GTPases). The antiviral functions of MxA have been deeply explored: MxA can protect cells from infection by multiple groups of pathogenic DNA andRNAviruses, such as influenza A virus and hepatitis B virus (Liu et al., 2013). In contrast, Mx2, althoughclosely related toMxA (63%aminoacid [aa] sequence identity), appears to have lost its antiviral function andhas been recognized as playing other cellular roles, since it does not suppress the viruses tested (Melen et al., 1996). Recently, Mx2 has been shown to serve as an inhibitor of human immunodeficiency virus type-1 (HIV-1) (Goujon et al., 2013; Kane et al., 2013; Liu et al., 2013). Mx2 restricts HIV-1 infectionat a relatively latepost-entryphase (Goujonet al., 2013; Kane et al., 2013) and leads to a reduced level of integrated viral DNA (Liu et al., 2013). The N-terminal 91-aa domain of Mx2 has been identified as a critical determinant of Mx2’s antiviral activity (Busnadiego et al., 2014). Interestingly, several mutations in the HIV-1 viral capsid (CA) region of Gag can overcome Mx2-mediated suppression (Goujon et al., 2013; Kane et al., 2013; Liu et al., 2013).Thus,Mx2maybind to theHIV-1coreand inhibit the early events in HIV-1 binding, thereby restricting viral infection. However, there is still no evidence showing that Mx2 directly binds to the HIV-1 capsid. Whether capsid binding of Mx2 requires cellular co-factors and/or higher-order assemblies of CA isalsounknown. In this study,wehaveobtainedastableMx2 protein containing the N-terminal 91-aa domain. Furthermore, we have observed that purified Mx2 recombinant proteins bind directly to HIV-1 CA assemblies in vitro. The N-terminal 83-aa domain of Mx2 is apparently critical for this interaction. Human MxA and Mx2 share a similar aa sequence and domain architecture (Liu et al., 2013; Melen et al., 1996). The crystal structure of MxA indicates that it includes a G domain that binds and hydrolyzes GTP; a hinge-like “bundle signaling element (BSE)” that connects the G domain to the elongated stalk domain; and the stalk domain, which is involved in self-assembly and oligomerization (Gao et al., 2010; Gao et al., 2011) (Fig. 1A). A unique feature of Mx2 is that it exhibits a longer N-terminal domain, including an NLS (N-terminal 25 aa). To identify the contribution of the diverse Mx2 domains to capsid binding, we generated and screened a series of deletion constructs of Mx2 with an N-terminal His-Sumo-tag (Fig. S1A) and characterized the expression, solubility, stability, and oligomerization behavior of these constructs (Table S1). We found that full-length Mx2 was difficult to obtain in E. coli. The expression levels of all the other deletion constructs (1–387, 1–413) were greatly reduced, but Δ1–83 exhibited a highly improved expression level (Fig. S1B), indicating that the N-terminal 83-aa domain of Mx2 may cause low expression and instability in solution (Table S1). Other deletion constructs (84–387, 84–413) without the N-terminal 83-aa domain showed high expression levels (Fig. S1B), but it exhibited poor solubility and stability (Fig. S1C), indicating that thecompletenessofBSEplaysasignificant role in maintaining the structure of theMx2 constructs. Thus, we wereunable toobtain stable constructswith theN-terminal 83-aa domain. To overcome this problem, on the basis of previous reports (Chappieetal., 2010)and thestructureofMxA(Gaoetal.,2011), we engineered aminimal GTPase-BSE fusion protein (GF) that connected residues 84–413 and residues 683–715 from human Mx2 via a flexible linker (Fig. S1A). GF eluted as a monodispersed peak from a size-exclusion column and had much better solubility (>20 mg/mL) than did residues 84–387 and 84– 413 (Fig. S1C). In the next phase, we added the N-terminal 83-aa domain of Mx2 to GF in order to generate an N-terminalGTPase-BSE fusion protein (N-GF). His-Sumo-N-GF showed a highly improved expression level (>20 mg/L in E. coli) and solubility when compared to other deletion constructs with the N-terminal 83-aa domain (Fig. S1C). After removing the Sumotag, we saw a reduction in the stability of the N-GF proteins, but N-GF eluted as a monodispersed peak from a size-exclusion


Mx2 constructs
a Expression was measured as the percentage of the expressed His-Sumo-tag-Mx2 construct in the total cell lysate. The percentage expression was determined by SDS-PAGE. "+++": indicates 100%, which corresponds to the amount of GF expression; "+": indicates ~10% of the GF expression level; "-": indicates no detectable expression. b Solubility was determined by the amount of each His-sumo-Mx2 variant purified from Ni-NTA-affinity chromatography. "+++": indicates 100% of the amount of purified GF; "+": indicates ~10% of the amount of purified GF; "-": indicates no target protein was obtained. c Stability was measured by the amount of target Mx2 variant in solution after removal of the Sumo-tag/GST-tag. "+++": indicates 100% of the amount of GF; "+": indicates ~10% of the amount of GF; "-": indicates no target protein was detected. d Oligomerization was analyzed by size chromatography. ,"+": indicates most of the protein oligomerized, "-":indicates most of the protein was monomeric.

Expression and purification of Mx2 constructs
The fragments encoding Mx2 deletion constructs were PCR-amplified from full-length Mx2 cDNA (XM_005260983) and cloned into the modified pET-30a plasmid with a His-Sumo-tag at the N-terminus. The N-GF and GF fusion proteins were created by PCR amplification from His-sumo-Mx2 (full-length) and His-Sumo-∆1-83 using the following 5'-phosphate primers: 5'-ggatcccaagagcagagtgagaccgctacc-3'; 5'-gcttccgatgtcagccccgcaacgc-3'. All proteins were over-expressed overnight at 18°C in Escherichia coli BL21 (DE3) cells. The cells were lysed using a French press in buffer containing 50 mM Tris-HCl, pH 8.0, with 300 mM NaCl, 20 mM imidazole, 1 mM PMSF, and 0.5% Triton X-100. The eluted fractions from Ni-NTA affinity chromatography (GE Healthcare) were collected and analyzed by SDS-PAGE to assess expression and solubility. The His-Sumo tag was cleaved by ULP in buffer (20 mM Tris-HCl, pH 8.0, with 300 mM NaCl and 5 mM DTT) and removed by Ni-NTA affinity chromatography. The target proteins were sequentially detected on a Superdex200 sizing column (GE Healthcare) in buffer (20 mM Tris-HCl, pH 8.0, with 150 mM NaCl and 0.5 mM TCEP).

In vitro assembly of HIV-1 CA
The expression and purification of the monomeric HIV-1 NL4-3 CA protein have been described in detail (Hung et al., 2013). Wild-type CA proteins were assembled in vitro by overnight dialysis into assembly buffer (50 mM Tris-HCl, pH 8.0, with 1 M NaCl) containing 20 mM βME at 4°C. Final protein concentrations were about 10 mg/mL. Assembled particles were visualized by cryo-EM. The CA mutations A14C/E45C/W184A/M185A and A21C/E22C/W184A/M185A were generated by site-directed mutagenesis and confirmed by custom sequencing analysis. Cross-linked CA A14C/E45C/W184A/M185A hexamers and CA A21C/E22C/W184A/M185A pentamers were prepared separately as previously described (Pornillos et al., 2009;Pornillos et al., 2011). In brief, the monomers of CA mutants were purified like wild-type CA, with only the minor difference being that the buffer contained 200 mM βME. Purified CA mutants were sequentially dialyzed into assembly buffer containing 200 mM βME, assembly buffer with 0.2 mM βME, and, finally, 50 mM Tris-HCl, pH 8.0. The CA hexamers and pentamers were then detected on a Superdex200 sizing column.

Binding of Mx2 variants to assemblies of CA
To verify binding, 20 µl of assembled CA tubes (20 µM) were incubated in vitro with 80 µl of PBS supplemented with 1M NaCl, pH 7.5, and each Mx2 variant (79 µM) at 37°C for 1 h. A fraction of this mixture was stored (total). The mixture was then centrifuged at 100,000g for 1 h at 4 °C. After centrifugation, the supernatant (soluble) was carefully removed, and the pellet (pellet) was resuspended in 1× SDS-PAGE loading buffer (pellet). The protein concentration was determined by SDS-PAGE.

In vitro GST pull-down assays
The in vitro binding assays were performed according to standard procedure: GST-N-GF-bound glutathione-sepharose beads were incubated with purified CA hexamers, pentamers, and monomers in binding buffer (20 mM HEPES, pH 7.5, with 150 mM NaCl, 1