Kane M, Mx2 HIV Nature 2013

8/13/2019 Kane M, Mx2 HIV Nature 2013

1/14

LETTER doi:10.1038/nature12653

MX2 is an interferon-induced inhibitor of

HIV-1 infection

Melissa Kane1,2, Shalini S. Yadav1,2,3{, Julia Bitzegeio1,2, Sebla B. Kutluay1,2, Trinity Zang1,2,3, Sam J. Wilson1,2,3{,John W. Schoggins4{, Charles M. Rice4, Masahiro Yamashita1, Theodora Hatziioannou1 & Paul D. Bieniasz1,2,3

HIV-1 replication can be inhibited by type I interferon (IFN),and the expression of a number of gene products with anti-HIV-1activity is induced by type I IFN1,2. However, none of the knownantiretroviral proteins can account for the ability of type I IFN toinhibitearly, preintegration phases of the HIV-1 replication cycle inhuman cells3,4. Here, by comparing gene expression profiles in celllines that differ in their ability to support the inhibitory action ofIFN-a at early steps of the HIV-1 replication cycle, we identifymyxovirus resistance 2 (MX2) as an interferon-induced inhibitor

of HIV-1 infection. Expression of MX2 reduces permissiveness to avariety of lentiviruses, whereas depletion of MX2 using RNA inter-ference reduces the anti-HIV-1 potency of IFN-a. HIV-1 reversetranscription proceedsnormally in MX2-expressing cells,but 2-longterminal repeat circular forms of HIV-1 DNA are less abundant,suggesting that MX2 inhibits HIV-1 nuclear import, or destabilizesnuclear HIV-1 DNA. Consistent with this notion, mutations inthe HIV-1 capsid protein that are known, or suspected, to alter thenuclear import pathways used by HIV-1 confer resistance to MX2,whereas preventing cell division increases MX2 potency. Overall,these findings indicate that MX2 is an effector of the anti-HIV-1activityof type-I IFN, andsuggest thatMX2 inhibitsHIV-1infectionby inhibitingcapsid-dependentnuclear import of subviral complexes.

We and others have previously identified proteins with antiretro-

viral activity on the basis of their differential expression in cells that arepermissive or non-permissive with respect to particular steps in theHIV-1 lifecycle5,6. We noticedthat monocytoid cell lines variedin theirability to support the anti-HIV-1 activity of type I IFN. Specifically,IFN-a treatmentof THP-1 cellscaused an,40-fold reductionin infec-tion by an HIV-1-based green fluorescent protein (GFP) reporter

vector, whereas treatment of K562 and U937 cells had little effect(Fig. 1a). When these cell lines were differentiated into a macrophage-like state by treatment with phorbol 12-myristate 13-acetate (PMA),theinhibitory effectof IFN-awasaccentuated in THP-1cells andaccen-tuated to a lesser extent in U937 cells, but remained nearly absent inK562 cells (Fig. 1a).

To identifycandidateeffectors of theantiviralaction of IFN-a,weusedmicroarrays to measuremessenger RNAlevels in theaforementionedcelllines. Twenty-twogenes whoseinduction,or non-induction, by IFN-acorrelated to varying degrees with the ability or inability of IFN-atoinhibit HIV-1GFP vector infection in the monocytoid cell lines wereselected for further study (Fig. 1b and Extended Data Figs 1 and 2).Among thesecandidates,MX2, a gene that wasnot previously thoughtto exhibit antiviral activity7, was of particular interest as we recentlyidentified it as a hit in an overexpression screen in a T-cell line duringwhich MX2 modestly inhibited infection by HIV-1 (ref. 8). Westernblot analyses confirmed that MX2 expression was strongly inducedby IFN-ain THP-1 cells but not K562 cells, and a basal level of MX2expression was slightly increased by IFN-a treatment in U937 cells

(Fig. 1c). MX2 was expressed at a basal level in primary CD41 T cellsandmacrophages,and wasinduced to varyingdegreesby IFN-a,depen-ding on the individual donor, and how cells were activated (ExtendedData Fig. 3).

Expression of the 22 candidate and control genes in K562 cellsrevealed that only MX2 and a control antiviral gene coding for rhesusmacaque TRIM5-a9 inhibited HIV-1 infection. (Fig. 2a). A rhesus ma-caque variant of MX2 also inhibitedHIV-1 infectionto a similar degreeas human MX2, whereas MX1 was inactive against HIV-1 (Fig. 2a),

even though it inhibits a variety of otherviruses7. Although MX2clearlyinhibited HIV-1 infection (Fig. 2ad), the fact that U937 cells (Fig. 1a),primary macrophages and anti-CD3/CD28-stimulated CD41 T cellsare readily infected by HIV-1, despite expressing appreciable levels ofMX2 (Fig. 1c and Extended Data Fig. 3), indicates that the blockimposed by MX2 is not absolute, or that MX2 potency is perhapsinfluenced by the cellular environment or cofactors.

MX1 and MX2 are members of a family of dynamin-like GTPases7,but only MX2 is localized to the nucleus by virtue of a basic nuclearlocalizationsignal(NLS) contained within its amino-terminal25 aminoacids10,11. Notably, theN-terminal25 amino acids that encodethe MX2NLSwere strictly required forantiviral activity (Fig. 2b, c). Conversely,the mutations K131A and T151Awhich inhibit GTP binding andhydrolysis, respectively11did not block the anti-HIV-1 activity of

MX2(Fig.2b, c). This result is in contrast to findings with MX1, whoseantiviral activity is GTPase dependent7, but should be interpreted cau-tiously given the reported ability of these MX2 mutants to induce ageneralized perturbation of nucleocytoplasmic transport11. In additionto its activity against HIV-1 and HIV-2 (Fig. 2d), MX2 expression inHOS cells inhibited infection by GFP reporter viruses based on a

variety of primate lentiviruses, including simian immunodeficiencyviruses SIVMAC, SIVAGMTan and SIVAGMSab, with some variation inMX2 antiviral potency (Fig. 2e). The nonprimate lentivirusesequineinfectious anaemia virus and feline immunodeficiency viruswereless potentlyinhibited, whereasa gammaretrovirusmurine leukaemia

viruswas only marginally sensitive to MX2.The experiments described above all represented single-cycle in-

fection assays, using vesicular stomatitis virus glycoprotein (VSV-G)-pseudotypedreporterviruses.However, expression of MX2in GHOST-R5 cells also inhibited infection by two full-length primary HIV-1strains, suggesting that MX2 inhibition was independent of the routeof entry, and not counteracted by HIV-1 accessory genes (Fig. 3a).Moreover, MX2 expression in GHOST-X4 cells inhibited spreadinginfection by full-length replication-competent HIV-1NL4-3 (Fig. 3b),reducing the number of infected cells by,20-fold during the expo-nential phase of viral growth. Reduction of MX2 expression in THP-1cells (Fig. 3c, d) or in HOS cells (Extended Data Fig. 4a, b) using shorthairpin RNAs (shRNAs) reduced, but did not eliminate, the antiviraleffect of IFN-a. Thus, MX2 is required for the full potency of IFN-a,

1Aaron DiamondAIDS ResearchCenter,New York, NewYork10016,USA. 2Laboratoryof Retrovirology,TheRockefellerUniversity, NewYork,New York10065,USA.3HowardHughesMedicalInstitute,New

York, NewYork 10016,USA. 4Center forthe Studyof HepatitisC, The Rockefeller University,New York, NewYork 10065,USA. {Presentaddresses: Universityof Texas Southwestern Medical Center, 5323

Harry Hines Boulevard, Dallas, Texas 75390-9048, USA (J.W.S.); MRC Centre for Virus Research, Institute of Infection, Immunity and Inflammation, College of Medical, Veterinary and Life Sciences,

University of Glasgow, Glasgow G12 8QQ, UK (S.J.W.); Weill Cornell Medical College, 525 East 68th street, New York, New York 10025, USA (S.S.Y.).

2 4 O C T O B E R 2 0 1 3 | V O L 5 0 2 | N A T U R E | 5 6 3

Macmillan Publishers Limited. All rights reserved2013
http://www.nature.com/doifinder/10.1038/nature12653http://www.nature.com/doifinder/10.1038/nature12653


2/14

but is not solely responsible for the inhibitory action of IFN-aon theearly steps of the HIV-1 replication cycle.

Consistent with this conclusion, IFN-a treatment reduced the accu-mulation of HIV-1 reverse transcripts in HOS cells (Fig. 4a), as haspreviously been reported for other cell types12. Conversely, MX2 ex-pression did not inhibit reverse transcript accumulationin either HOSor K562 cells (Fig. 4a and Extended Data Fig. 5). However, MX2 didreduce the generationof 2-long terminal repeat (2-LTR)circles (Fig.4aand Extended Data Fig. 5), which are thought to form only after retro-

viral DNA has accessed the nucleus of infected cells. MX2 may, there-fore, inhibit the entry of HIV-1 into the nucleus, or perhaps cause

destabilization of viral DNA in the nucleus. Consistent with previous

reports10,11, we found thatthatN- or carboxy-terminally haemagglutinin-tagged forms of MX2 were particularly concentrated at nuclear poresmarked by the nucleoporin NUP98 (Extended Data Fig. 6). TheMX2(K131A) mutant is primarily cytoplasmic but nevertheless inhi-bits nucleocytoplasmic transport11 and also retains antiviral activity(Fig. 2c). Therefore, alteration of the fate of incoming HIV-1 DNA

with respect to the nucleus may underlie the antiviral activity of MX2,even though stable physical association with nuclear pores may not berequired for antiviral function.

The HIV-1 capsid protein (CA) is a key determinant required for in-fection of non-dividing cells and nuclear entry of subviral complexes1315.Indeed, HIV-1 CA mutations have been shown to change the require-ment for specific nucleoporins (for example, NUP358 (also known asRANBP2), NUP85, NUP153, NUP155) during HIV-1 infection, andto alter the distribution of sites at which HIV-1 DNA integrates intohost chromosomes1618. Therefore, we tested whether a number of CAmutations that are known or suspected to affect the pathway used byHIV-1 DNA into the nucleus also affected sensitivity to inhibition byMX2 (Fig. 4b). Of these, a mutation (N57S) that confers cell cycledependence on HIV-1 infection19, and presumably restricts HIV-1

nuclear entry to the mitoticphase of the cell cycle, conferredresistance

None

Vecto

r

Lucif

erase

rhTR

IM5-

GBP1GB

P5

HERC

6IFI

44MX2

PARP

10RG

L1

SIGL

EC1

SP10

0TA

P1TN

K2ET

V7

EPST

I1HC

P5

HERC

5

HSPA

6IFI

35IF

IH1

LOC4

0075

9

RBM

43

RSAD

2

SERP

ING1

rhM

X2MX1

1

10

100

Infected(GFP+)cells(%)


m.o.i. = 0.6 m.o.i. = 0.2 m.o.i. = 0.07

0 1 2 3 40

20

40

60

80

100

Virus inoculum (m.o.i.)

MX2

T151A

K131A

delN25

Vector

MX2-

Tubulin-

a

b c d HIV-2

SIVMAC

SIVAGM

Sab EIAV

HIV-1

FIVSIVAGM

Tan MLV

Vecto

rMX2

0

Vecto

rMX2

0.0

Vecto

rMX2

0.0

Vecto

rMX2

0

Vecto

rMX2

0.0

Vecto

rMX2

0

Vecto

rMX

20.0

Vecto

rMX

20

6107

4107

2107

2.0107

1.5107

1.0107

1.5108

1.0108

5.0107

1.5104

1.0104

5.0103

1.5106

1.0106

5.0105

5105

4105

3105

2105

1105

1106

8105

6105

4105

21050.5107

6107

8107

4107

2107Titre(IUml1)

Titre(IUml1)

VectorMX2

T151AK131AdelN25

e

Figure 2| Inhibition of lentivirus infection by wild-type and mutant MX2,but not other differentially interferon-induced genes. a, Infection of K562cells, previously transduced with an HIV-1 vector (SCRPSY) expressingnegative (luciferase) or positive (rhesus macaque (rh)TRIM5-a-coding)control genes, or candidate antiviral genes, with GFP-expressing HIV-1 vector(CSGW) at the indicated multiplicity of infection (m.o.i.). b, Western blot

analysis of MX2 and tubulin expression in K562 cell clones transduced with anHIV-1 vector (CSIB) expressing wild-type and mutant MX2 proteins. delN25,MX2 mutant lacking theN-terminal NLS. c, Infection of thesame K562 cells asin b with an HIV-1GFP reporter virus.d, e, Infection of HOS cells, previouslytransduced with an MX2-expressing or empty HIV-1 vector (SCRPSY), withvarious GFP reporter viruses. Titres are mean1 s.d., n5 3 technical replicates,representative of four experiments. EIAV, equine infectious anaemia virus;FIV, feline immunodeficiency virus; MLV, murine leukaemia virus.

50 500 5,000 50,00050 500 5,000 50,00050

500

5,000

50,000

50 500 5,000 50,000

50 500 5,000 50,000 50 500 5,000 50,00050

500

5,000

50,000

50 500 5,000 50,000

MX2

MX2

MX2

MX2

MX2

MX2

Array signal in untreated cells (a.u.)

ArraysignalinIFN--treatedc

ells(a.u.)

0.1 1 10 100

0.1

1

10

100

No IFN + IFN

0.1 1 10 100

0.1

1

10

100

0.1 1 10 100

0.1

1

10

100

0.1 1 10 100

0.1

1

10

100

Inoculum (l) Inoculum (l) Inoculum (l)

GFPpositive(%)

GFPpositive(%)

0.1 1 10 100

0.1

1

10

100

0.1 1 10 100

0.1

1

10

100

THP-1 K562 U937

THP-1

(PMA)

K562

(PMA)

U937

(PMA)

THP-1 K562 U937

THP-1(PMA)

K562(PMA)

U937(PMA)

THP-1

0 10

100

1,000

0 10

100

1,000

0 10

100

1,000

K562 U937

IFN- (U ml1)

MX2-

Tubulin-

a

b

c

114

38

45

ND

ND

ND

ND 1

1.62.42.9

Figure 1| Differential effects of IFN-aon HIV-1 infection of monocytoidcell lines correlates with MX2 expression. a, Undifferentiated (top) orPMA-differentiated (bottom) THP-1, K562 and U937 cells with or withoutIFN-a treatment(1,000 U ml21) werechallenged witha GFP-expressingHIV-1vector (CSGW). b, RNA extracted from cells treated identically to those showninawas analysed on microarrays. The array signal is plotted in arbitrary units(a.u.), and the data points representingMX2are highlighted.c, Western blotanalysis of MX2and tubulin expression in monocytoid cell lines treated for24 h

with the indicated doses of IFN-a. Numbers below each lane indicated foldincrease in MX2 protein levels relative to untreated cells. ND, not detected.

RESEARCH LETTER

5 6 4 | N A T U R E | V O L 5 0 2 | 2 4 O C T O B E R 2 0 1 3



3/14

to MX2 (Fig. 4b). Another mutation, G89V, which abolishes cyclophi-lin A binding by HIV-1 CA and the requirement for NUP358 duringHIV-1 infection17, also conferred apparently complete MX2 resistance(Fig. 4b). Another CA mutation, N74D, which abolishes CA inter-action with cleavage and polyadenylation specificity factor 6 (CPSF6)16,reduced but did not eliminate sensitivity to MX2, whereas the muta-tions G94D and A92E, which confer cyclophilin A sensitivity (cyclos-porin A dependence) during early replication steps20, slightly reducedMX2 sensitivity (Fig. 4b). These data demonstrate that the viral capsidgoverns the sensitivity of HIV-1 to MX2. In addition, they show thatthe antiviral activity of MX2 is specific, and unlikely to be the result ofsome generalized perturbation of cell physiology. Notably, the MX2-resistant CA mutant N57S exhibited a modest degree of resistanceto IFN-a, relative to wild-type HIV-1, in THP-1 cells and HOS cells(Fig. 4c and Extended Data Fig. 7), supporting the notion that MX2 isone, but not the only, effector of the antiviral activity of IFN-aduring

the early steps of the HIV-1 replication cycle.Because the cell-cycle-dependent HIV-1 CA mutant N57S was not

inhibited by MX2 (Fig. 4b), we reasoned that arresting the cell cycleand thereby restricting HIV-1 infection to non-mitotic cells mightpotentiate the antiviral activity of MX2. Growth arrest of HOS orK562 cells with aphidicolin blocked infection by a control cell-cycle-dependent retrovirus (murine leukaemia virus) irrespective of MX2expression, whereas HIV-1 was almost unaffected, as expected (Fig.4dandExtendedData Fig. 8a,b). However, the inhibitory activity of MX2wasincreased in non-dividing cells (Fig. 4d and Extended Data Fig. 8),in which it inhibited a single cycle of replication by,30-fold. In otherwords, MX2both inhibited and conferred a degree of cell cycle depen-dence on wild-type HIV-1 infection.

Type I IFN inhibits HIV-1 replication at multiple points in the life

cycle, both before and after the point at which MX2 seems to act2,6,12.

Thus MX2 is one of multiple effectors that contribute to the overallanti-HIV-1 activity of type I IFN. A few potential mechanisms mightunderlie the anti-HIV-1 activity of MX2. First, MX2 might directlytarget the incoming viral capsid, in a manner akin to the primateTRIM5-a and murine Fv1 antiretroviral proteins2,9,21, or mutant cyto-plasmic forms of CPSF6 (ref. 16). As with MX2, one consequence ofthe action of these capsid-targeting proteins is inhibition of the importof viral DNA into the nucleus, and in some cases their potency isenhanced in non-dividing cells16,22. A second possibility is that MX2inhibits particular nuclear import pathways, without regard to the pre-cise nature of the import cargo, as mutant forms of MX2 have beenshown to inhibit the nuclear accumulation of model cargos unrelatedto HIV-1 (ref. 11). A third possibility is that MX2 acts after nuclearentry to destabilize viral DNA and/or inhibit integration. In these sce-narios, CA mutations (G89V, N57S) could confer resistance by inhi-biting interaction with MX2, by modulating the timing or extent ofcapsid uncoating, or by directing HIV-1 to alternative nuclear entrypathways. We note that the MX2-resistant G89V and N57S mutantsexhibit reduced infectiousness in human cells, raising the possibilitythat the mutations abolish the use of pathways or processes duringinfection that are inhibitedby MX2. Finally, it is possible that MX2 actsindirectly, for example by affecting the nuclearcytoplasmic distri-bution of other cellular proteins that can interact with the viral capsid.

However, thepoor correlation in thedegree of MX2 (Fig. 4) andCPSF6

0 2 4 61

10

100

Days after infection

shRN

A(ctrl)

shRN

A(M

X2)

shRN

A(ctrl)

shRN

A(M

X2)

104

105

106

10717

7

a b

c shRNA(ctrl)

0 10

100

1,000

0 10

100

1,000

shRNA(MX2)

IFN-

U ml1

MX2-

Tubulin-

0.22.15.66.60.10.20.41.7

d

HIV-1TF2864 HIV-1TF2851

Vecto

rMX2

0.0

Vecto

rMX2

0.0

NoIFN-

1,000

Uml1IF

N-

1

2

3

4

5

Foldtitreincrease

inshRNA(MX2)cells

Vector

MX2MX2(T151A)

P= 0.27 P= 0.001P= 0.004

Titre(IUml1)

Titre(IUml1)

1.5105

1.0105

5.0104

1.5105

1.0105

5.0104


Figure 3| MX2 inhibits replication-competent HIV-1 and is required forthe full antiviral activity of IFN-a. a, Infection of empty vector (CSIB) orMX2-expressing GHOST-R5 cells with full-length primary HIV-1 strains.Titres are mean1 s.d.,n5 3 technical replicates, representative of twoexperiments.b, Growth of replication-competent HIV-1NL4-3in empty vector(CSIB) or MX2-expressing GHOST-X4 cells (containing an HIV-2-LTR-GFPgene). c, Western blot analysis of MX2and tubulin expression in IFN-a-treatedTHP-1cells expressing control orMX2-targeted shRNAs. Numbers beloweachlane indicate fluorescence intensity associated with the MX2 band.d, HIV-1GFP reporter virus infection of shRNA-expressing THP-1 cells from c, with(black) or without (white) IFN-atreatment. Titres are mean 1 s.d.,n5 3technical replicates,Pvalues calculated using unpairedt-test, representative ofthree experiments.

Vecto

rMX

20

Vecto

rMX2

0.0

Vecto

rMX

20.0

Vecto

rMX2

0.0

Vecto

rMX2

0.0

Vecto

rMX

20

N57S G89V G94DA92EWT N74D

0

5

10

15

20

25

Foldinhibition

HIV-1 (WT)HIV-1 (N57S)

NoIFN 1010

01,0

00

10,00

0

NoIFN 1010

01,0

00

10,00

0103

104

105

106

0.01 0.10 10.1

1

10

100

0.01 0.10 10.1

1

10

100

Virus inoculum (l) Virus inoculum (l)

VectorMX2

0 5 10 15 20 250

10,000

20,000

30,000

40,000

Time after infect ion (h) Time after infection (h)

RTproducts(GFP,

copiesperl)

a

c

b

dDividing Non-dividing

0 5 10 15 20 250

50

100

150

2-LTRcircles

(copiesperl)

NoneIFN-

NevirapineMX2

Titre(IUml1)

Titre(IUml1)


3107

2107

1107

3107

2107

1107

1.2106

8.0105

4.0105

1.2107

8.0106

4.0106

1.5106

1.0106

5.0105

1.5107

1.0107

5.0106

Figure 4| MX2 activity reduces levels of nuclear HIV-1 DNA, is capsiddependent and is more potent in non-dividing cells. a, Quantitative PCRanalysis of reverse transcript (RT, left) and 2-LTR circle (right) abundancein inhibitor-treated or MX2-expressing HOS cells. b, Wild-type (WT) orCA-mutant HIV-1GFP reporter virus infection of vector or MX2-expressingHOS cells. Titres are mean 1 s.d.,n5 3 technical replicates, representative offour experiments. c, Infectivity of wild-typeand N57S CA-mutant HIV-1GFP

reporter viruses in untreated and IFN-a-treated THP-1 cells. Titres aremean1 s.d.,n5 3 technicalreplicates, representativeof fourexperiments.Foldinhibition is the ratio of the mean titres on untreated and IFN-a-treated cells.d, HIV-1GFP reporter virus infection of dividing and non-dividing(aphidicolin-treated) vector- or MX2-expressing HOS cell clones.

LETTER RESEARCH

2 4 O C T O B E R 2 0 1 3 | V O L 5 0 2 | N A T U R E | 5 6 5



4/14

(ref. 23) resistance/sensitivity exhibited by HIV-1 CA mutants sug-gests that redistribution of CPSF6 is unlikely to underlie the antiviralaction of MX2. Although further work will be required to preciselydefine the molecular mechanisms involved, our findings demonstratethat MX2 is an effector in the anti-HIV-1 activity of type I IFN andunderscore the remarkable diversity of proteins that cells can mobilizeas antiretroviral defences.

METHODS SUMMARY

Gene expression in monocytoid cell lines was measured using human HT12Expression Beadchip (Illumina) containing,48,000 transcript probes, accordingto the manufacturers instructions. Candidate antiviral genes, MX2 and MX2mutants were expressed in K562, HOS or GHOST cells using the HIV-1-based

vectors SCRPSY (which encodes TagRFP and puromycin resistance) or CSIB(which confers blasticidin resistance). MX2- and control-vector-expressing cellswere used as populations or as single-cell clones in infection assays to evaluateMX2 antiviral activity.

All single-cycle GFP reporter viruses were pseudotyped with VSV-G. Virusstocks were generated by transfecting 293T cells with Env-defective proviralDNA that encoded GFP in place of the nefgene, or in the case of primary HIV-1strains, full-length proviral plasmids. Alternatively, packageable GFP-expressingretroviral vector and Gag-Pol packaging plasmids were cotransfected. Target cellsin microwell plates were challenged with various doses of virus and single-cyclereplication evaluated after 2 days. The proportion of cells infected with GFP

reporter viruses, or replication-competent virus infection in GHOST cells (whichcontain an LTR-GFP indicatorgene) in single cycleor spreadingreplicationassayswas measuredby flowcytometry. MX2expression was reduced in target cellsusinga modifiedlentiviral shRNAexpressionvector(Origene). Non-dividingtarget cellswere generated by aphidicolin treatment for 24 h before and during infection.

The abundanceof viralDNA species was measuredusing quantitative PCRwithprimers directed to the GFP reporter gene, or to viral LTR sequences that areproximate only in 2-LTR circles. Western blotting was done using fluorescentantibodies and signals quantitated with a LI-COR Odyssey scanner. Deconvolu-tionmicroscopy and imageanalysiswas doneusing a Deltavision microcopysuite.

OnlineContentAny additionalMethods, ExtendedData display items and SourceData are available in theonline version of the paper; references unique to thesesections appear only in the online paper.

Received 23 August; accepted 12 September 2013.

Published online 13 October 2013.

1. Ho, D. D. et al.Recombinant human interferon alfa-A suppresses HTLV-IIIreplication in vitro. Lancet325, 602604 (1985).

2. Neil,S. & Bieniasz, P. Human immunodeficiency virus, restriction factors, andinterferon.J. InterferonCytokine Res. 29, 569580 (2009).

3. Bitzegeio, J., Sampias, M., Bieniasz, P. D. & Hatziioannou, T. Adaptation to theinterferon-induced antiviral state by human and simian immunodeficiencyviruses. J. Virol. 87,35493560 (2013).

4. Goujon, C.et al. Evidence for IFNa-induced, SAMHD1-independent inhibitors ofearly HIV-1 infection. Retrovirology10,23 (2013).

5. Sheehy, A.M.,Gaddis,N.C.,Choi,J. D.& Malim,M.H. Isolationof a humangenethatinhibits HIV-1 infection and is suppressed by the viral Vif protein. Nature 418,646650 (2002).

6. Neil, S. J., Zang, T. & Bieniasz, P. D. Tetherininhibits retrovirus release and isantagonized by HIV-1 Vpu. Nature 451,425430 (2008).

7. Haller, O., Staeheli, P. & Kochs, G. Interferon-induced Mx proteinsin antiviral hostdefense.Biochimie 89,812818 (2007).

8. Schoggins, J. W. et al.A diverse range of gene products are effectors of the type Iinterferon antiviral response.Nature 472, 481485 (2011).

9. Stremlau, M.et al. The cytoplasmic body component TRIM5arestricts HIV-1infection in Old World monkeys.Nature 427,848853 (2004).

10. Melen,K. et al. Human MxB protein, an interferon-a-inducible GTPase, contains anuclear targeting signal and is localized in the heterochromatin region beneaththe nuclear envelope.J. Biol. Chem. 271, 2347823486 (1996).

11. King, M. C.,Raposo,G. & Lemmon,M. A. Inhibitionof nuclearimportand cell-cycleprogression by mutated forms of the dynamin-like GTPase MxB.Proc. Natl Acad.Sci. USA101, 89578962 (2004).

12. Goujon, C. & Malim, M. H. Characterization of the alpha interferon-inducedpostentry block to HIV-1 infection in primary human macrophages and T cells.J. Virol. 84, 92549266 (2010).

13. Yamashita,M., Perez,O., Hope,T. J.& Emerman,M. Evidence fordirectinvolvementof the capsid protein in HIV infection of nondividing cells.PLoS Pathog.3, e156(2007).

14. Yamashita, M. & Emerman, M. Capsid is a dominant determinant of retrovirusinfectivity in nondividing cells.J. Virol.78,56705678 (2004).

15. Dismuke,D. J. & Aiken, C. Evidence fora functionallink between uncoating of thehuman immunodeficiency virus type1 core and nuclear import of the viralpreintegration complex.J. Virol. 80, 37123720 (2006).

16. Lee, K. etal. Flexibleuse of nuclearimportpathways by HIV-1. CellHost Microbe7,221233 (2010).

17. Schaller, T.et al.HIV-1 capsid-cyclophilin interactions determine nuclear importpathway, integration targeting and replication efficiency. PLoS Pathog. 7,e1002439 (2011).

18. Koh, Y.et al. Differential effects of human immunodeficiency virus type 1 capsidand cellular factors nucleoporin 153 and LEDGF/p75 on the efficiency andspecificity of viral DNA integration.J. Virol.87,648658 (2013).

19. Rihn, S. J. et al. Extreme genetic fragility of the HIV-1 capsid. PLoS Pathog. 9,e1003461 (2013).

20. Sokolskaja, E.,Sayah,D. M. & Luban, J. Target cellcyclophilinA modulates humanimmunodeficiency virus type 1 infectivity.J. Virol. 78, 1280012808 (2004).

21. Stoye, J. P. Fv1, themouse retrovirus resistance gene. Rev.Sci. Tech. 17, 269277(1998).

22. Yamashita, M. & Emerman, M. Cellularrestriction targeting viral capsids perturbshuman immunodeficiency virus type 1 infection of nondividing cells.J. Virol.83,98359843 (2009).

23. DeIaco, A.et al. TNPO3 protects HIV-1 replication from CPSF6-mediated capsidstabilization in the host cell cytoplasm. Retrovirology10, 20 (2013).

AcknowledgementsWe thank members of The Rockefeller University GenomicsResource Center for assistance with the microarray experiments and members of theBieniaszlaboratoryfor discussion andadvice.This workwas supported by grants fromthe National Institutes of Health; R37AI64003 (to P.D.B.), R01AI078788 (to T.H.)

R01AI100720 (to M.Y.), AI091707 to C.M.R., AI057158 (to I. Lipkin, NortheastBiodefense Center, subcontracted to C.M.R.) and DK095031 to J.W.S., the GreenbergMedical Research Institute and the Starr Foundation (C.M.R.) and by the HowardHughes Medical Institute.

Author ContributionsM.K., S.S.Y.,J.B.,S.B.K.,T.Z. andS.J.W.designedand executed theexperiments and analysed the data. J.W.S. and C.M.R. provided aninterferon-stimulated genelibrary and advice. M.Y.providedreagents and advice. T.H.provided reagents and advice and supervised the work. P.D.B. conceived the study,supervised the work and wrote the paper, with additional input from all authors.

Author Information Reprints and permissions information is available atwww.nature.com/reprints. The authors declare no competing financial interests.Readersare welcome to commenton the online version of thepaper. Correspondenceand requests for materials should be addressed to P.D.B. ([email protected]).

RESEARCH LETTER

5 6 6 | N A T U R E | V O L 5 0 2 | 2 4 O C T O B E R 2 0 1 3

http://www.nature.com/doifinder/10.1038/nature12653http://www.nature.com/reprintshttp://www.nature.com/doifinder/10.1038/nature12653mailto:[email protected]:[email protected]://www.nature.com/doifinder/10.1038/nature12653http://www.nature.com/reprintshttp://www.nature.com/doifinder/10.1038/nature12653


5/14

METHODSMicroarray analyses.Total RNA was extracted, using the RNeasy Plus Mini kit(Qiagen), from THP-1, K562 and U937 cells that were either undifferentiated, ordifferentiated using PMA, and untreated or treated with 1,000U ml21 IFN-afor24 h before collection. Complementary RNA was prepared and probed usingHuman HT12 Expression Beadchip (Illumina), containing ,48,000 transcriptprobes, according to the manufacturers instructions.Plasmid construction.The HIV-1-based expression vector SCRPSY expresses ared fluorescent protein (TagRFP) and puromycin resistance from a TagRFP-

FMDV2A-Puro cassette in the unspliced viral transcript (in place of Gag codingsequence). SCRPSY also expresses spliced transcripts that encode HIV-1 Tat andRev proteins, as well as genes inserted in Gateway-compatible sequences in placeof Nef-coding sequences. The HIV-1-based expression vectors CSIB, HA-CSIBand CSIB-HA are derived from CSGW24 by replacing GFP coding sequences witha multi-cloning site (with or without sequencesencodinga HAtag) followedby aninternal ribosome entry site sequence and a blasticidin-resistance cassette.

Candidate antiviral genes, includingMX2, were also transferred into theSCRPSY vector from a library of IFN-stimulated genes8 in Gateway-compatibleentry vectors using the clonase reaction (Life Technologies). Alternatively, openreading frames were amplified from complementary DNA prepared from THP-1cells following treatment with 1,000U ml21 IFN-a for 24 h. The GTP-hydrolysis-defective MX2 mutant (T151A), the GTP-binding-defective MX2 mutant (K131A)andMX2 lacking the N-terminal NLS(delN25)10,11weregenerated usingPCRbasedsite-directed mutagenesis. For insertion into CSIB-based vectors,MX1and wild-type and mutantMX2 genes were amplified by PCR and inserted following SfiIrestriction digestion.Cells. The adherent cell lines 293T and HOS were maintained in DMEM. Thesuspension cell lines U937 and K562 were maintained in RPMI. THP-1 cells weremaintained in RPMI supplemented with 0.05 mM beta-mercaptoethanol. Thesecell lines wereobtained fromthe ATCC. Single-cell clones of GHOSTcells expressingCXCR4 or CCR5obtainedthroughthe AIDSReagent Program (Division of AIDS,NIAID,NIHfromV. N.KewalRamaniandD. R.Littman)werederivedby limitingdilution and maintained in DMEM supplemented with 2.5mg ml21 puromycin,50mg ml21 hygromycin and 500mg ml21 G418. All growth media were supple-mented with 10% FCS and gentamicin.

Derivatives of K562 and GHOST cells expressing wild-type and mutant MX2proteins were generated by transduction with CSIB-based vectors followed by se-lection in 5mg ml21 blasticidin. HOS cells expressing wild-type or mutant MX2proteins were prepared by transduction with SCRPSY-based vectors followed byselection in 2.5mg ml21 puromycin. HOS cells expressing wild-type or mutant

HA-tagged MX1 or MX2 proteins were prepared by transduction with HA-CSIBbased vectors followed by selection in 5 mg ml21 blasticidin. In general, the MX2-expressing cells were used as populations of blasticidin- or puromycin-resistantcells, but some experiments used single-cell clones of MX2-expressing HOS andK562 cells, which were derived from puromycin- or blasticidin-resistant popula-tions by limiting dilution. For experiments in which the panels of candidate anti-

viral genes were screened (Fig.2a), K562 cells were transducedwith SCRPSY-basedvectors, but were not selected in puromycin; infection was measured in theTagRFP-positive population. To generate the MX2-expressing or candidate genecell lines, SCRPSY or CSIB vector stocks for transduction were generated byco-transfection of 293T cells with a VSV-G expression plasmid, an HIV-1NL4-3Gag-Pol expression plasmid, and an MX2-expressing vector using polyethylenei-mine (PolySciences).

Primary CD41 T cells were isolated from human blood by Ficoll-Paque gra-dient centrifugation and negative selection (RosetteSep Human CD41 T CellsEnrichment Cocktail, StemCell Technologies). Cells were activated with phyto-

haemagglutinin(Sigma,5 mg ml21

) or anti-CD3/CD28beads (Dynabeads HumanT-Activator CD3/CD28, Gibco) for 48 h, cultured in the presence of interleukin-2(50Uml21, PeproTech) and treated with or without IFN-afor 24h.

Primary macrophages were differentiated from fresh human peripheral bloodmononuclear cells. Cells were plated in serum-free medium for 3 h at 37 uC, thesupernatant with non-adherent cells was discarded and adherent monocytes werecultured in RPMI with 10% FCS, 1% L-glutamine and granulocytemacrophagecolony-stimulating factor (GM-CSF) (100 ng ml21, PeproTech) for 6 days andtreated with or without IFN-afor 24h.Viruses.All viruses were generated by transfection in 293T cells using polyethy-leneimine (PolySciences). For the GFP reporter proviral plasmidsHIV-1NL4-3DEnv-GFP(wild type,G89V, A92Eand G94DCA mutants)25, NHGCapNM(wildtype andN57SCA mutant)19, pLai3DEnvGFP(wildtype andN74DCA mutant)14,SIVMACDEnv-GFP, SIVAGMTanDEnv-GFP, SIVAGMSabDEnv-GFP, HIV-2RODDEnv-GFP2610mg of proviral plasmid was co-transfected with 1 mg of VSV-Gexpression plasmid. For HIV-1 (in Figs 1 and 2a), MLV, EIAV and FIV, three-

plasmid vector systems were also used to generate GFP reporter viruses, whereby

5mg of Gag-Pol, 5 mg of packageable genome and 1mg of VSV-G expressionplasmids27,28 were co-transfected. Plasmids containing full-length replication-competent transmitted founder HIV-1 proviruses (pTRJO.c/2851 and pREJO.c/2864)29 were obtained through the AIDS Reagent Program (Division of AIDS,NIAID, NIH from J. Kappes and C. Ochsenbauer). At 48 h after transfection, viralsupernatants were collected and their infectivity was determined using MT4 orGHOST target cells.

Infection assays. Single-cycle infectivity in HOS,K562, THP-1and U937cell lineswas measured in cells seeded in 96-well plates at 53 103 cells per well and inocu-

lated with serial-dilutions of VSV-G-pseudotyped GFP reporter viruses in thepresence of 5mg ml21 polybrene. For replication-competent primary HIV-1strains, GHOST-R5 cells, which contain an HIV-2-LTR-GFP reporter construct,were used. Two days post-infection, cells were trypsinized (adherent cells only)and fixed in 2% paraformaldehyde. In some experiments, cells were pre-treatedwith IFN-a (Sigma-Aldrich) for 24 h, re-plated, followedby inoculation with GFPreporter viruses.

For experiments in which infection of dividing and non-dividing cells wascompared,single-cell clones of K562and HOScellstransduced withMX2-expressingor empty vector (CSIB in the case of K562, SCRPSY in the case of HOS) wereseededat33 104 cellsper wellin 48-well plates. Cellswere treated withaphidicolin(1mg ml21, Sigma-Aldrich) or an equivalent volume of dimethylsulphoxide alonefor 24 h before infection with HIV-1 or MLV GFP reporter viruses.

For spreading replication assays, empty vector or MX2-expressing HIV-1(CSIB)-transduced GHOST-X4 cell lines (which contain an HIV-2-LTR-GFPindicator gene) were seeded at 13 105 cells in a six-well plate. Thereafter they

wereinoculated withHIV-1NL4-3at a m.o.i.of 0.01.Cellsweresplit at a 1:4dilution,and the percentage of infected (GFP-expressing) cellsmeasured, every2 daysuntilthe empty vector cultures died owing to HIV-1-induced cytopathic effects.

In all infection assays, infected cells (% GFP positive of viable cells) wereenumerated by FACS analysis using a CyFlow cytometer (Partec) coupled to aHypercyteAutosampler (Intellicyt). For cells transducedwith the SCRPSYvector,the percentage of infected cells was determined as the percentage of RFP/GFPdouble-positive cells in the total RFP-positive population.

RNA interference. Lentiviral MX2-specific shRNA (target sequence: 59-TCGCTATTCCTGGCTGCTTCAAGAGCAGA-3 9) or scrambled negative controlshRNA pGFP-C-shLenti expression plasmids (OriGene) were digested withXba1 and Bsu361 to excise the cytomegalovirus (CMV) promotor and GFP.Virus stocks for transducing the modified shRNA expression vectors were gener-ated by co-transfection of 293T cells with VSV-G and HIV-1NL4-3 Gag-Polexpression plasmids. THP-1 and HOS cells were transduced and selected in 1m

g ml

21

and 2.5m

g ml

21

puromycin, respectively. Puromycin-selected cells weretreated with IFN-a for 24 h before assessment of MX2 expression by westernblotting and inoculation with GFP reporter viruses.

Measurement of HIV-1 DNA species in infected cells. For analysis of HIV-1reverse transcription products and 2-LTR circles in infected cells, 13 105 HOS orK562 cells were seeded in 24-well culture plates and infected with a VSV-G-pseudotyped HIV-1 reporter virus (CCGW), which is derived from CSGW butencodes a GFPreporter underthecontrolof a CMVpromoter.The virusinoculumwas pretreated with 20U ml21 RNase-free DNase I (Roche) for 1h at 37 uC in thepresence of 6mM MgCl2. In some experiments, cells were pretreated with IFN-afor24 h, or were treated with nevirapine startingat thetimeof infection. Cellswerecollectedat 2, 12 and24 h post-infection,washed with 13PBS, andtotalDNAwasextracted using the QIAamp DNA Blood Mini Kit (Qiagen). The resulting DNAsamples were used as template for quantitative PCR using FastStart UniversalSYBR Green Master Mix (Roche) and ABI 7500 Fast PCR system. The primerpairsused in thisstudy are as follows: forGFP (latereversetranscription products):

forward: 59-AAGTTCATCTGCACCACCGGCAA-3

9, reverse: 5

9-TGCACGCC

GTAGGTCAGG-39; for 2-LTR circles: forward: 59-GACTCTGGTAACTAGAGATCCCTC-39, reverse: 59-TTGGGAGTGAATTAGCCCTTCCA-3 9.

Western blotting. Cellsuspensions werenormalized for cell number, lysedin SDSsamplebuffer, separated by electrophoresison NuPage412% Bis-Tris gels(Novex)and blotted onto nitrocellulose membranes (GE Healthcare). Membranes wereincubated with rabbit anti-MX2 (Novus Biologicals) and mouse anti-tubulin(Sigma) antibodies, followed by incubation with goat anti-rabbit IRDye 800CWand goatanti-mouse IRDye 680RD, respectively (LI-COR Biosciences).A LI-COROdyssey scanner was used to detect and quantify fluorescent signals.

Microscopy.HOS cells expressing N-terminally HA-tagged MX2 and MX1 wereseeded onto 24-well gelatin-coated glass-bottomed dishes (MatTek) and immu-nostained using a combination of anti-HA (Covance) and anti-NUP98 (CellSignalingTechnology) antibodies followed by goat anti-mouse Alexa488 and goatanti-rabbit Alexa 594 secondary antibodies (Molecular Probes). Cells were visua-lized by deconvolution microscopy as described previously30. Image generation

and co-localizationanalysis and were completedwith the SoftWorx software suite

LETTER RESEARCH



6/14

(Applied Precision). Pearsonscoefficient values were derivedfor MX2 or MX1 (asa control) and NUP98 by analysis of optical sections coincident with the dorsalnuclear surface, for 610 individual cells.

24. Bainbridge, J. W. et al. In vivogene transfer to the mouse eye using anHIV-based lentiviral vector; efficient long-term transduction ofcorneal endothelium and retinal pigment epithelium. GeneTher. 8, 16651668(2001).

25. Hatziioannou,T., Cowan, S.,Von Schwedler,U. K.,Sundquist, W.I. & Bieniasz,P. D.Species-specific tropism determinants in the human immunodeficiency virustype1 capsid.J. Virol. 78,60056012 (2004).

26. Hatziioannou, T.,Cowan,S.,Goff,S. P., Bieniasz,P.D. & Towers,G. J.Restrictionofmultiple divergent retroviruses by Lv1 and Ref1.EMBO J.22, 385394 (2003).

27. Mitrophanous, K. et al.Stable gene transfer to the nervous system using anon-primate lentiviral vector. Gene Ther.6, 18081818 (1999).

28. Kemler,I., Barraza,R. & Poeschla, E. M. Mappingthe encapsidationdeterminantsof feline immunodeficiency virus.J. Virol. 76, 1188911903 (2002).

29. Ochsenbauer, C. et al. Generation of transmitted/founder HIV-1 infectiousmolecular clones and characterization of their replication capacity in CD4 Tlymphocytesand monocyte-derived macrophages.J. Virol.86,27152728 (2012).

30. Jouvenet, N. et al.Plasma membrane is the site of productive HIV-1 particleassembly. PLoS Biol.4, e435 (2006).

RESEARCH LETTER



7/14

Extended Data Figure 1| Candidate anti-HIV-1 genes from the microarrayanalysis. mRNA levels, determined using Illumina BeadChips and given inarbitrary units, for genes whose differential induction in undifferentiated and

PMA-treated THP-1, K562 and U937 cells correlated best with the anti-HIV-1effect of IFN-a.

LETTER RESEARCH



8/14

Extended Data Figure 2| Additional candidate anti-HIV-1 genes from themicroarrayanalysis. mRNA levels, determined usingIllumina BeadChipsandgiven in arbitrary units, for genes whose differential induction in

undifferentiated and MA-treated THP-1, K562 and U937 cells correlated tosome degree with the anti-HIV-1 effect of IFN-a.

RESEARCH LETTER



9/14

Extended Data Figure 3| Induction of MX2 by IFN-ain primary CD41

T cells and macrophages. Western blot analysis of MX2 and tubulinexpression in purified CD41 T cells, activated with PHA or anti-CD3/CD28,and macrophages treated for 24h with the indicated doses of IFN-a. Numbersbelow each lane indicate fluorescence intensity associated with the MX2 band.The second more rapidly migrating MX2 species that was detectedinconsistently is of unknown provenance, and may represent a proteolyticbreakdown product, or may arise through the use of an alternative start codonat amino acid 25, generating an MX2 protein that lacks the NLS.

LETTER RESEARCH



10/14

Extended Data Figure 4| MX2 is required for the full antiviral activityof IFN-ain HOS cells. a, Western blot analysis of MX2 expression HOS cells

transduced with vectors expressing control orMX2-targeted shRNAs,and treated with IFN-a. Numbers below each lane indicate fluorescenceintensity associated with the MX2 band. b, Infectious titre of an HIV-1GFPreporter virus determined using the shRNA-containing HOS cells from a,with or without IFN-atreatment. Titres are mean1 s.d.,n5 3 technicalreplicates,Pvalues calculated using unpairedt-test, representative ofthree experiments.

RESEARCH LETTER



11/14

Extended Data Figure 5| MX2 activity reduces levels of nuclear HIV-1 DNA in K562 cells. Quantitative PCR analysis of reverse transcript (left) and 2-LTRcircle (right) abundance in empty-vector untreated (none) nevirapine-treated or MX2-expressing K562 cells.

LETTER RESEARCH



12/14

Extended Data Figure 6| Localization of MX2 at nuclear pores.a, Deconvolution microscopic images (single optical sections) ofimmunofluorescently stained NUP98(red), haemagglutinin(HA)-tagged MX2(green, expressed using CSIB vectors) and DAPI (49,6-diamidino-2-phenylindole)-stained DNA (blue) in HOS cells. The top set of panels is anoptical section approximately through the centre of the vertical dimension ofthe nucleus, whereas the middle and bottom panels are an optical section

approximately coincident with the dorsal surface of the nucleus. The bottompanelsare an expanded view of a portion of thecentre panels. Scale bars, 10mm(top), 5mm (middle) and 1mm (bottom).b, Pearsons coefficient forcolocalization of MX1 or MX2 and NUP98. Each data point represents anindividual cell and the horizontal bar is the mean (n56 for MX1,n5 10 forMX2).

RESEARCH LETTER



13/14

Extended Data Figure 7| The N57S capsid mutation reduces HIV-1sensitivity to IFN-ain HOS cells. Infectivity of wild-type and N57SCA-mutant HIV-1GFP reporter viruses in untreated and IFN-a-treated HOScells. Titres are mean1 s.d,n5 3 technical replicates, representative ofthree experiments. Fold inhibition is the ratio of the titres on untreated andIFN-a-treated cells.

LETTER RESEARCH



14/14

Extended Data Figure 8| Effect of MX2 on HIV-1 and murine leukaemiavirus infection in dividing and non-dividing cells. a, MLVGFP reportervirus infection of dividing and non-dividing (aphidicolin-treated) vectoror MX2-expressing HOS cell clones.b, HIV-1GFP reporter virus infection ofdividing and non-dividing (aphidicolin-treated) vector or MX2-expressingK562 cell clones.c, MLVGFP reporter virus infection of dividing andnon-dividing (aphidicolin-treated) vector or MX2-expressing K562 cell clones.

RESEARCH LETTER

Date post:	04-Jun-2018
Category:	Documents
Upload:	sofia-busquet
View:	215 times
Download:	0 times

Kane M, Mx2 HIV Nature 2013

Documents