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

Protein Cell 2014, 5(12):954957DOI 10.1007/s13238-014-0113-5 Protein& Cell

LETTER

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

Dear Editor,

More than 50 years have passed since the myxovirus resistance (MX) genes were rst discovered and found to suppress infection with inuenza viruses in mice (Lindenmann, 1962). Like most mammals, mice carry two MX genes, MX1 and MX2, 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 MxA and MxB proteins, which are interferon-induced, dynamin-like large molecular weight guanosine triphosphatases (GTP-ases). The antiviral functions of MxA have been deeply explored: MxA can protect cells from infection by multiple groups of pathogenic DNA and RNA viruses, such as inuenza A virus and hepatitis B virus (Liu et al., 2013). In contrast, Mx2, although closely related to MxA (63% amino acid [aa] sequence identity), appears to have lost its antiviral function and has 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 immunodeciency virus type-1 (HIV-1) (Goujon et al., 2013; Kane et al., 2013; Liu et al., 2013). Mx2 restricts HIV-1 infection at a relatively late post-entry phase (Goujon et 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 identied as a critical determinant of Mx2s 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, Mx2 may bind to the HIV-1 core and 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 is also unknown. In this study, we have obtained a stable Mx2 protein containing the N-terminal 91-aa domain. Furthermore, we have observed that puried 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 difcult to obtain in E. coli. The expression levels of all the other deletion constructs (1387, 1413) were greatly reduced, but 183 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 (84387, 84413) without the N-terminal 83-aa domain showed high expression levels (Fig. S1B), but it exhibited poor solubility and stability (Fig. S1C), indicating that the completeness of BSE plays a signicant role in maintaining the structure of the Mx2 constructs. Thus, we were unable to obtain stable constructs with the N-terminal 83-aa domain.

To overcome this problem, on the basis of previous reports (Chappie et al., 2010)andthestructure of MxA (Gaoet al.,2011), we engineered a minimal GTPase-BSE fusion protein (GF) that connected residues 84413 and residues 683715 from human Mx2 via a exible 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 84387 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-terminal-GTPase-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

The Author(s) 2014. This article is published with open access at Springerlink.com and journal.hep.com.cn

Protein&Cell

Amino-terminal domain of Mx2/MxB-dependent interaction with HIV-1 capsid LETTER

A

B

Buffer+ His-Sumo-N-GF

CA+ His-Sumo-N-GF

Buffer+ N-GF

CA+ N-GF

t p s t p s t p s t p s

Buffer+ His-Sumo-GF

CA+His-Sumo-GFt p s t p s t p s t p s

Buffer+ (183)

CA+ (183)

t p s t p s

C

Buffer+ Stalk domain

CA+ Stalk domain

Protein&Cell

1 m 100 nm

Figure 1. Interactions of Mx2 variants with CA tubular assemblies. (A) Binding of His-sumo-N-GF and N-GF to pre-assembled wild-type CA tubes. Binding reactions were analyzed by SDS-PAGE using CA tubular assemblies at 80 mol/L or binding buffer alone, incubated with His-sumo-N-GF and N-GF (79 mol/L). Samples of the reaction mixture before centrifugation (t), of the supernatant (s), and of the pellet (p) are detected alone. (B) Binding of Mx2 constructs to preassembled wild-type CA tubes. Binding reactions were the same as in (A). No binding of any Mx2 construct without the N-terminal 83-aa domain was found. (C) Cryo-EM images of pre-assembled wild-type CA tubes used in the Mx2-CA binding experiments.

column (Table S1). Thus, we successfully generated a stable fusion construct of Mx2 for use in detecting further binding.

It had previously been demonstrated that assembled HIV-1 CA-NC tubular complexes can interact with full-length MX2 and its truncations containing the N-terminal 91-aa (isolated from HEK293T cell lysates) (Fricke et al., 2014). To further clarify whether Mx2 directly binds to HIV-1 CA assemblies (at 20 mol/L) in vitro, we tested puried His-Sumo-N-GF and N-GF (79 mol/L) in precipitation assays with preassembled CA tubes. The results clearly indicated that the N-GF proteins co-pellet with assembled CA, revealing that the binding process does not require other cellular co-factors (Fig. 1A); thus, established that Mx2 directly binds to HIV-1 CA tubes in vitro.

We performed a further dissection of the interactions between the Mx2 constructs and CA to address the

contribution of the different Mx2 domains to capsid binding. Puried GF, 183, and the stalk domain (79 mol/L) were each tested in the precipitation assay with preassembled CA tubes (20 mol/L). Essentially no binding was observed for GF, 183, or the stalk domain under the same assay conditions (Fig. 1B). All CA tubes were prepared as previously described (Yang et al., 20122012; Hung et al., 2013) and were conrmed by cryo-electron microscopy (cryo-EM) (Fig. 1C). The ndings described above indicate that Mx2-CA binding requires the N-terminal 83-aa of Mx2. The N-GF does not contain the stalk domain, but it is apparently sufcient to bind HIV-1 CA assembly. On the other hand, the stalk domain alone showed no binding to the assembled CA tubes (Fig. 1B).

HIV-1 capsid (CA) proteins can assemble into closed fullerene cones or helical tubes in vivo and in vitro. It is widely known that tripartite motif protein isoform 5 alpha

The Author(s) 2014. This article is published with open access at Springerlink.com and journal.hep.com.cn 955

LETTER Jia Kong et al.

A

B

CA monomers CA pentamers CA hexamers

Elution volume (Superdex 200)

OD 280(mAU)

150

100

50

0 5 10 15 20

GST pull down Input

GST-N-GF

GST-N-GF

CA

Protein&Cell

CA monomers

CA pentamers CA hexamers

Figure 2. GST Pull-down assays of GST-N-GF with CA hexamers, pentamers, and monomers. (A) Size-exclusion chromatographic proles of individual CA hexamers, pentamers, and monomers. (B) Pull-down assays of GST-N-GF with CA hexamers, pentamers, and monomers. CA hexamers, pentamers, and monomers, incubated with GST-N-GF-bound glutathionesepharose beads, showed no more interaction than those incubated with glutathione-sepharose beads alone.

(TRIM5) is an antiviral protein that restricts infection of HIV-1 by binding to the viral capsid. The TRIM5-CA binding interaction requires an assembled capsid lattice, since individual CA monomers do not have an appreciable afnity for TRIM5 (Stremlau et al., 2006). It is the shape of the HIV-1 capsid that is recognized by the TRIM5 protein (Yang et al., 2012). Having conrmed the binding of Mx2-CA assemblies, we wanted to investigate whether, like TRIM5, Mx2 binds to the viral capsid by recognizing the shape of the higher-order assembled HIV-1 capsid lattice. Therefore, we over-expressed and puried individual hexamers (A14C/E45C/W184A/ M185A), pentamers (A21C/E22C/W184A/M185A), and monomers of CA as previously described (Fig. 2A) (Pornillos et al., 2011), and separately examined the binding afnity between each of them and GST-N-GF by GST pull-down assays, detecting the results by SDS-PAGE. No complex formation was detected, even at very high concentrations of both proteins (Fig. 2B). This result demonstrates that Mx2 shows no detectable binding to CA hexamers, pentamers, or monomers, implying that a higher-order lattice of CA tubes is required for an efcient Mx2-CA interaction.

In the current study, we have demonstrated for the rst time that a direct interaction occurs between Mx2 and the HIV-1 capsid. The N-terminal region of Mx2 is critical for this interaction. We identied a fusion protein of Mx2 containing the N-terminal 83-aa and the complete BSE (N-GF), which was fundamental for the detection of Mx2-CA binding in vitro. The N-GF construct may also facilitate further biochemical and structural studies of Mx2. The co-pelleting of puried N-GF and CA tubes, as well as domain mapping studies, showed that Mx2 directly recognizes CA assembles through its N-terminal 83-aa domain. Based on the results of pull-down assays, Mx2-CA binding requires the higher-order assembled capsid lattice, just as TRIM5-CA binding does.

While this work was in progress, it was reported that assembled HIV-1 CA-NC tubular complexes can co-pellet Mx2 from whole-cell lysates (Fricke et al., 2014). However, there is no evidence to show that Mx2 directly binds to CA in vitro (i.e., whether capsid binding of Mx2 requires co-factors in cells). The contribution of the NC domain to the binding of Mx2 to higher-order assemblies of CA-NC was also not clear. Our data argue that there is direct interaction between Mx2 and HIV-1 CA.

956 The Author(s) 2014. This article is published with open access at Springerlink.com and journal.hep.com.cn

Amino-terminal domain of Mx2/MxB-dependent interaction with HIV-1 capsid LETTER

Interestingly, several mutations in the HIV-1 viral capsid (CA) region of Gag can overcome MX2-mediated suppression (Fricke et al., 2014; Liu et al., 2013). The CA-NC tubular complexes that are formed with mutations in CA that help HIV-1 escape from Mx2 restriction exhibit weakened interactions with Mx2 (Fricke et al., 2014). Thus, Mx2 may bind to the HIV-1 core and inhibit the process of HIV-1 infection.

It has been reported that Mx2 variants 572715 and 623715 have lost their ability to oligomerize, and they fail to bind to CA-NC assemblies in vitro (Fricke et al., 2014). Our results show that although N-GF lacks the stalk domain involved in oligomerization, it still has the ability to bind CA tubes. This binding may occur because N-GF has the complete BSE, which is of great importance in maintaining normal Mx2 structure. These results suggest that although it is possible that the oligomerization of Mx2 produces a higher binding afnity for the viral capsid, oligomerization is not necessary for Mx2-CA binding. Future studies will be required to explore the oligomerization functions that are required for Mx2-mediated HIV-1 restriction (Haller et al., 2010).

ACKNOWLEDGEMENT

This work was supported in part by funding from the Chinese Ministry of Science and Technology (No. 2012CB911100 and No. 2013ZX0001-005) and Tianjin Research Program of Application Foundation and Advanced Technology (No. 14JCQNJC09400).

Jia Kong, Bo Xu, Wei Wei, Xin Wang, Wei Xie, and Xiao-Fang Yu declare that they have no conict of interest. This article does not contain any studies with human or animal subjects performed by any of the authors.

Jia Kong1,2, Bo Xu1,2, Wei Wei2,3,4, Xin Wang1,2, Wei Xie1,2&, Xiao-Fang Yu1,2,3,4&

1 School of Life Science, Tianjin University, Tianjin 300072, China

2 Collaborative Innovation Center of Chemical Science and Engineering, Tianjin 300072, China

3 Department of Molecular Microbiology and Immunology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD 21205, USA

4 Institute of Virology and AIDS Research, First Hospital of Jilin University, Changchun 130061, China

& Correspondence: [email protected] (W. Xie), xfyu@tju. edu.cn (X.-F. Yu)

OPEN ACCESS

This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and

reproduction in any medium, provided the original author(s) and the source are credited.

REFERENCES

Busnadiego I, Kane M, Rihn SJ, Preugschas HF, Hughes J, Blanco-

Melo D, Strouvelle VP, Zang TM, Willett BJ, Boutell C et al (2014) Host and Viral Determinants of Mx2 Antiretroviral Activity. J Virol 88:77387752Chappie JS, Acharya S, Leonard M, Schmid SL, Dyda F (2010) G domain dimerization controls dynamins assembly-stimulated GTPase activity. Nature 465:435440Fricke T, White T, Schulte B, de Souza Aranha Vieira D, Dharan A,

Campbell E, Brandariz-Nunez A, Diaz-Griffero F (2014) Mx2 binds to the HIV-1 core and prevents the uncoating process of HIV-1. Retrovirology 11:68Gao S, von der Malsburg A, Paeschke S, Behlke J, Haller O, Kochs

G, Daumke O (2010) Structural basis of oligomerization in the stalk region of dynamin-like MxA. Nature 465:502506Gao S, von der Malsburg A, Dick A, Faelber K, Schrder, Gunnar F,

Haller O, Kochs G, Daumke O (2011) Structure of myxovirus resistance protein A reveals intra- and intermolecular domain interactions required for the antiviral function. Immunity 35:514525 Goujon C, Moncorge O, Bauby H, Doyle T, Ward C, Schaller T, Hue

S, Barclay W, Schulz R, Malim M (2013) Human MX2 is an interferon-induced post-entry inhibitor of HIV-1 infection. Nature 502:559562Haller O, Gao S, von der Malsburg A, Daumke O, Kochs G (2010)

Dynamin-like MxA GTPase: structural insights into oligomerization and implications for antiviral activity. J Biol Chem 285:2841928424 Hung M, Niedziela-Majka A, Jin D, Wong M, Leavitt S, Brendza KM,

Liu X, Sakowicz R (2013) Large-scale functional purication of recombinant HIV-1 capsid. PLoS One 8:e58035Kane M, Yadav S, Bitzegeio J, Kutluay S, Zang T, Wilson S,

Schoggins J, Rice C, Yamashita M, Hatziioannou T et al (2013) MX2 is an interferon-induced inhibitor of HIV-1 infection. Nature 502:563566Lindenmann J (1962) Resistance of mice to mouse-adapted inuenza A virus. Virology 16:203204Liu Z, Pan Q, Ding S, Qian J, Xu F, Zhou J, Cen S, Guo F, Liang C

(2013) The interferon-inducible Mx2 protein inhibits HIV-1 infection. Cell Host Microbe 14:398410Melen K, Keskinen P, Ronni T, Sareneva T, Lounatmaa K, Julkunen I

(1996) Human Mx2 protein, an interferon-alpha-inducible GTPase, contains a nuclear targeting signal and is localized in the heterochromatin region beneath the nuclear envelope. J Biol Chem 271:2347823486Pornillos O, Ganser-Pornillos BK, Yeager M (2011) Atomic-level modelling of the HIV capsid. Nature 469:424427Stremlau M, Perron M, Lee M, Li Y, Song B, Javanbakht H, Diaz-

Griffero F, Anderson D, Sundquist W, Sodroski J (2006) Specic recognition and accelerated uncoating of retroviral capsids by the TRIM5alpha restriction factor. Proc Natl Acad Sci USA 103:5514 5519Yang H, Ji X, Zhao G, Ning J, Zhao Q, Aiken C, Gronenborn AM,

Zhang P, Xiong Y (2012) Structural insight into HIV-1 capsid recognition by rhesus TRIM5. Proc Natl Acad Sci USA 109:1837218377

Protein&Cell

Jia Kong and Bo Xu have contributed equally to this work.

Electronic supplementary material The online version of this article (doi:http://dx.doi.org/10.1007/s13238-014-0113-5

Web End =10.1007/s13238-014-0113-5 ) contains supplementary material, which is available to authorized users.

The Author(s) 2014. This article is published with open access at Springerlink.com and journal.hep.com.cn 957