ZIKA VIRUS

Neurodevelopmental protein Musashi-1interacts with the Zika genome andpromotes viral replicationPavithra L. Chavali,1*† Lovorka Stojic,1* Luke W. Meredith,2 Nimesh Joseph,1

Michael S. Nahorski,3 Thomas J. Sanford,2 Trevor R. Sweeney,2 Ben A. Krishna,4

Myra Hosmillo,2 Andrew E. Firth,2 Richard Bayliss,5 Carlo L. Marcelis,6

Susan Lindsay,7 Ian Goodfellow,2 C. Geoffrey Woods,3 Fanni Gergely1‡

A recent outbreak of Zika virus in Brazil has led to a simultaneous increase in reports ofneonatal microcephaly. Zika targets cerebral neural precursors, a cell population essential forcortical development, but the cause of this neurotropism remains obscure. Here we reportthat the neural RNA-binding protein Musashi-1 (MSI1) interacts with the Zika genome andenables viral replication. Zika infection disrupts the binding of MSI1 to its endogenoustargets, thereby deregulating expression of factors implicated in neural stem cell function.Wefurther show that MSI1 is highly expressed in neural progenitors of the human embryonicbrain and is mutated in individuals with autosomal recessive primary microcephaly. SelectiveMSI1 expression in neural precursors could therefore explain the exceptional vulnerabilityof these cells to Zika infection.

Zika virus (ZIKV) recently emerged as a majorpublic health risk because of its devastatingeffect on fetal neurodevelopment (1–3). ZIKVwas first isolated in Uganda in 1947, andthe virus subsequently spread through Asia,

and from there to the Americas (4). A causal linkbetween ZIKV infection and congenital brainmalformations became apparent in 2016 after anoutbreak in Brazil (1). Brazilian ZIKV belongs tothe Asian lineage which affected New Caledoniaand French Polynesia, where cases of microceph-aly were reported retrospectively (5).Intrauterine infections can impair neurodevel-

opment (6), but ZIKV is highly neurotropic andinterferes specifically with fetal brain develop-ment, causing microcephaly, cortical malforma-tions, and intracranial calcifications (7–10). Wehypothesized that the single-stranded RNA fla-vivirus ZIKV may hijack RNA-binding factorspresent in the developing central nervous sys-tem (11). Host RNA-binding proteins are knownto interact with untranslated regions (UTRs) toregulate replication, translation, and stabilizationof viral genomes (11). In silico analysis of the

genomic RNA of the Brazilian ZIKV strain, PE243,revealed three consensus binding sites in the 3′UTR for the highly conserved Musashi familyof RNA-binding proteins, Musashi-1 (MSI1) andMusashi-2 (MSI2), both important translationalregulators in stem cells (12–15). Two sites were con-served between PE243 and the Ugandan MR766strains (sites 1 and 2), whereas the third (site 3)was found only in the Asian-lineage strains includ-ing PE243 (Fig. 1A and fig. S1, A and B). Bymappingthese sites onto a predicted secondary structureof ZIKV 3′UTR, we found all three sites to be pre-sent on stem-loop structures, which are consideredoptimal for MSI binding (16, 17). Moreover, a re-cent study revealed nucleotide substitutions prox-imal to sites 1 and 2 in the Asian-lineage strains,which could indicate positive selection for MSI1binding during ZIKV evolution (18).To address if the Musashi proteins interacted

with ZIKV, we first tested their binding to ZIKV3′UTR. RNA pull-down assays identified bindingof MSI1, but not MSI2, to the 3′UTR of PE243(Fig. 1B) (15). Mutating the three consensus MSI1sites in the 3′UTR of PE243 significantly weakenedthis interaction (Fig. 1, C and D, and fig. S1C). Wealso confirmed binding between MSI1 and the3′UTR of MR766 (Fig. 1C). To investigate whetherMSI1 also binds ZIKV 3′UTR in vivo, ultraviolet(UV) cross-linking immunoprecipitation (CLIP) ofRNA was performed from lysates of PE243-infectedU-251 glioblastoma cells, revealing a robust directinteraction between MSI1 and PE243 ZIKV RNA(Fig. 1E). Consistently, in ZIKV-infected cells, MSI1colocalized with double-stranded RNA (dsRNA), aviral replication intermediate, as visualized by con-focal and stimulated emission depletion (STED)super-resolution microscopy (Fig. 1, F and G).These data confirm an interaction between MSI1and ZIKV RNA, which is, at least in part, me-diated by the 3′UTR of the virus.

To investigate whether MSI1 had an effect onthe life cycle of ZIKV, we used RNA interferenceto deplete the protein in U-251 glioblastoma,SK-N-BE2c neuroblastoma, and H9-derived neu-ral stem cells (NSCs) and performed PE243 viralinfections. In all three cell types, MSI1 depletionled to a marked reduction in viral RNA levels(Fig. 2, A and B). We then generated MSI1knockouts (KOs) in U-251 cells by clustered regularlyinterspaced short palindromic repeats (CRISPR)–Cas9–mediated targeting of exons 8 or 6 of MSI1(KO1 and KO2, respectively; Fig. 2C and fig. S2).Control cells were obtained through clonal expan-sion of cells transfected with Cas9 alone. Bymeasuring viral RNA at different times afterPE243 infection, a marked reduction of viral loadwas seen in KO1 and KO2 cells at 24 and 48 hours(Fig. 2D). Whereas extensive cell death precludedRNA analysis in the controls at 72 hours, viral RNAwas comparable between the 48- and 72-hourtime points in the KO cells. Consistently, levelsof the viral dsRNA and flavivirus E protein, aswell as the infectious titer, were reduced in theKOs (Fig. 2E and fig. S3). Because MSI2 levelswere similar between control and KO cells(Fig. 2C), MSI1 and MSI2 are unlikely to havecomplete functional redundancy in ZIKV repli-cation. Replication of the MR766 strain was alsoimpaired in the KO cells (Fig. 2F). In summary,we identify MSI1 as an important factor for ZIKVreplication, both in primary and transformedneural cell lines.Because there was no discernible difference

between ZIKV binding and entry into control andKO cells (Fig. 2, G and H), we asked if MSI1 couldregulate translation through ZIKV UTRs. To thisend, luciferase RNA flanked by the 5′ and 3′UTRsof PE243, was transfected with increasing amountsof MSI1 into human embryonic kidney (HEK)293T cells, which do not normally express MSI1(fig. S4). We observed a modest MSI1-driven in-crease in luciferase expression. The ability of MSI1to promote ZIKV UTR–driven translation in vitroraises the possibility that it performs a similarfunction in vivo. Alternatively, MSI1 might stabi-lize the viral RNA genome and/or regulate itscyclization or synthesis. In addition, given thepleiotropic roles of MSI1 in cellular pathways, itis plausible that MSI1-dependent regulation ofgene expression contributes to the ZIKV life cycle(19). However, MSI1 is unlikely to act throughgeneral pro- or antiviral pathways, as infectionwith H1N1 influenza virus was unaffected byMSI1 expression levels (fig. S5). In line with pub-lished work, we also find that MSI1 KO cells ex-hibited defective migration, increased doublingtime, and cell-cycle delay (fig. S6) (17, 20, 21). Be-cause ZIKV replication requires cyclin-dependentkinase activity, such pro-proliferative effects ex-erted by MSI1 might contribute to virus produc-tion (22). Nevertheless, the direct interactionbetween MSI1 and the ZIKV genome is consistentwith the hypothesis that the protein promotessome aspect of the viral life cycle.ZIKV predominantly infects neural progen-

itors in human fetal brain. We find MSI1 to beabundant in neural precursors of the ventricular

RESEARCH

Chavali et al., Science 357, 83–88 (2017) 7 July 2017 1 of 6

1Cancer Research UK Cambridge Institute, Li Ka ShingCentre, University of Cambridge, Robinson Way, CambridgeCB2 0RE, UK. 2Division of Virology, Department of Pathology,University of Cambridge, Addenbrooke’s Hospital, CambridgeCB2 2QQ, UK. 3Department of Medical Genetics, CambridgeInstitute for Medical Research, University of Cambridge, HillsRoad, Cambridge CB2 0XY, UK. 4Department of Medicine,University of Cambridge, Hills Road, Cambridge CB2 2QQ, UK.5Faculty of Biological Sciences, Astbury Centre for StructuralMolecular Biology, University of Leeds, Leeds LS2 9JT, UK.6Department of Human Genetics, Radboud University MedicalCentre, Nijmegen, Netherlands. 7Institute of Genetic Medicine,Newcastle University, International Centre for Life, CentralParkway, Newcastle upon Tyne NE1 3BZ, UK.*These authors contributed equally to this work. †Present address:Medical Research Council (MRC) Laboratory of Molecular Biology,Francis Crick Avenue, Cambridge CB2 0QH, UK. ‡Correspondingauthor. Email: fanni.gergely@cruk.cam.ac.uk

on February 14, 2020

http://science.sciencem

ag.org/D

ownloaded from

and subventricular zones of the human embryonicbrain, but absent from mature neurons (Fig. 3Aand fig. S7). Owing to its high levels in neural pro-genitors, and its ability to stimulate ZIKV replica-tion, MSI1 could be instrumental to ZIKV-inducedcytopathicity in the fetal brain. In addition, MSI1is required for neurodevelopment in both inver-tebrates and vertebrates, with MSI1-depletedzebrafish displaying microcephaly and mutantmice exhibiting a thin cerebral cortex and reducednumber of mature neural cell types, among othermorphological brain abnormalities (14, 19, 23–25).We have identified a consanguineous Turkish

family in which two siblings displayed clinicalfeatures suggestive of autosomal primary micro-cephaly (MCPH), a condition associated with aconsiderable reduction in cerebral cortex size,but a structurally normal brain (fig. S8, A and B)(26, 27). Exome sequencing uncovered potential-ly deleterious homozygous mutations in MSI1,ACACB, DKK4, and DTX3L (fig. S8, C to F, andtables S1 and S2). Of these, only MSI1 is knownto have neural functions, but because mutationswere present in four genes, the point mutationcausing the p.Ala184Val (A184V) substitution inMSI1 may not be the sole cause of MCPH in these

individuals (referred to as MSI1A184V). Neverthe-less, three lines of evidence show that A184V mu-tant MSI1 is functionally impaired. First, MSI1A184V

patient cells exhibit premature chromosome con-densation (PCC), the same phenotype as MSI1-deficient glioblastoma cells. Second, we showthat the A184V mutation impedes RNA bind-ing of MSI1, leading to deregulated expressionof its endogenous targets. Third, we find that theA184V mutant MSI1 is unable to support ZIKVreplication.The PCC phenotype seen in MSI1A184V-patient

cells has been previously described in cells deficient

Chavali et al., Science 357, 83–88 (2017) 7 July 2017 2 of 6

Fig. 1. MSI1 interacts directly with the ZIKVRNA genome. (A) Schematic diagram of PE243ZIKV containing three putative MSI1 binding sitesin its 3′UTR. Sites shared with MR766 are red;the site specific to PE243 is blue. The polyproteincomprises the capsid (C), precursor membrane(prM), and envelope (E) proteins, in addition to thenonstructural (NS1-5) proteins. (B) RNA pull-downassays performed with the 3′UTR of PE243. Increasingconcentrations of in vitro transcribed biotinylatedPE243 RNA were incubated with U-251 cell extracts,and RNA-protein complexes were captured on strep-tavidin beads. Representative Western blots wereprobed with antibodies against MSI1 and the RNA-binding proteins MSI2 and hnRNP Q/R.Corresponding protein and RNA inputs are shown onthe right. bp, base pair. (C) RNA pull-down assaysperformed with the WTor triple mutant (D123) 3′UTRof PE243. Note that PE243-3′UTR_D123 lacks allthree MSI1 binding sites depicted in Fig. 1A (see fig.S1C for further details). Increasing concentrationsof in vitro transcribed biotinylated RNA wereincubated with U-251 cell extracts, and RNA-proteincomplexes were captured on streptavidin beads.Representative Western blots probed with antibodyagainst MSI1 are shown together with correspondingprotein and RNA inputs. (D) Densitometric analysisof MSI1 levels from Western blots of RNA pull-downassays, an example of which is shown in Fig. 1C. Theamount of MSI1 precipitated by PE243-3′UTR_D123is expressed as a percentage of MSI1 precipitatedby the same concentration of PE243-3′UTR_WT.n = 3 biological replicates. Bar charts depict mean± SEM. (E) CLIP analysis from mock- or ZIKVPE243-infected U-251 cells. Western blot showsimmunoprecipitations (IPs) by rabbit immuno-globulinG (IgG)andMSI1 antibodies frommock- andPE243-infected U-251 cells after UVcross-linking.Input (5%) represents whole-cell extract.Western blotwas probed with antibodies against MSI1. Graphbelow shows quantitative polymerase chain reaction(qPCR) performed on bound RNA from IP. CLIPvalues are presented as a percentage of input aftersubtraction of the IgG background. Glyceraldehyde-phosphate dehydrogenase (GAPDH) serves asnegative control. n = 3 biological replicates. A primerpair against 9519 to 9681 bp of ZIKVgenome wasused in these qPCRs (table S4). (F) Immuno-fluorescence of mock- or PE243-infected U-251 cells.MSI1 (green) and dsRNA (red) signals are detectedbyconfocalmicroscopy.DNA is detected byHoechst stain (blue). Framedarea is shownat highermagnificationbelow. (G) Immunofluorescenceof aPE243-infectedU-251cell.MSI1 (green)anddsRNA(red)signalsaredetectedbySTEDsuper-resolutionmicroscopy.Outlinesof thecell andnucleusare indicated inwhite and blue, respectively.Framed area is shown at higher magnification to the right. P values were obtained from Student’s t test, unpaired, two-tailed: *P < 0.05; ***P < 0.0005.

MSI1 dsRNA (ZIKV)

4µm

MSI1

dsRNA (ZIKV)

2µm

dsRNA (ZIKV) MSI1

dsRNA (ZIKV)

MSI1

MSI1 dsRNA (ZIKV)

DNA

Mock PE243

dsRNA (ZIKV)MSI1 MSI1 dsRNA

5µm

2µm

1 5 100

20

40

60

80

PE243-3 UTR_Δ123

% M

SI1

bou

nd r

elat

ive

to P

E24

3-3’

UT

R_W

T

***

p=0.09

*

RNA (pmol)

1 5 10

WT Δ123

1 5 10

MR766-3 UTR

50

37

37

RNA

MSI1

MSI1

PE243-3 UTR

RNA (pmol)

500bp

RNA500bp

37kDa

75

37

hnRNP Q/R

MSI2

MSI1

RNA (pmol)

PE243-3 UTR

0lysate 1 5 10 20 1 5 10 20

RNA pulldown

Input

10668 (2)10478 (1)

10773 (3)

PE243 genome

EprM NS1-5C

ZIKV polyprotein

MSI1 consensus : G/A(U)1-3AG

RNA pulldown

Input

0

5

10

15

CLI

P (

% in

put)

MockPE243

GAPDH ZIKV

PE243Mock

Inpu

tIg

G MSI1

Inpu

tIg

GM

SI1

37 MSI1

kDa

kDa

**

RESEARCH | REPORTon F

ebruary 14, 2020

http://science.sciencemag.org/

Dow

nloaded from

of the MCPH-associated protein microcephalin(MCPH1) (fig. S8G) (28, 29). Because theMCPH1locus is unaffected in MSI1A184V patients, wespeculated that MSI1 could control chromosomecondensation by regulating MCPH1 expression.To determine if MCPH1 was a MSI1 target, weperformed RNA immunoprecipitation (RIP) undernative conditions. In addition to its known targetsNUMB, p21WAF1, and the MCPH-associated geneCDK6/MCPH12, MSI1 co-precipitated with twoisoforms of MCPH1 (MCPH1_S and MCPH_L),

despite their divergent 3′UTRs, but did notinteract with other MCPH genes tested (fig. S9,A and B) (12, 30, 31). CLIP and RNA electropho-retic mobility shift assays suggest a direct inter-action between MSI1 and MCPH1_L (Fig. 3, Band C). MSI1 can act as a translational suppressor(e.g., for NUMB and p21WAF1) or activator (e.g.,for CDK6) (12, 21, 30). Consistent with a role forMSI1 in translational activation of MCPH1, weobserved low MCPH1 protein levels in MSI1A184V-patient and MSI1-deficient U-251 cells and a reduc-

tion in polysome-associated MCPH1 transcriptsin KO cells (Fig. 3, D and E, and figs. S9C and S10).Given that MSI1 interacts with ZIKV RNA, andthat viral RNA is abundant in the infected cell, theviral genome could compete with endogenous tar-gets for bindingMSI1. Indeed, upon ZIKV infectionof U-251 cells, we observed a marked reduction inthe interaction between MSI1 and its target RNAs,including MCPH1 and NUMB, accompanied bychanges in their protein levels that mirrored thoseof MSI1 KO cells (Fig. 3, F and G).

Chavali et al., Science 357, 83–88 (2017) 7 July 2017 3 of 6

Moc

kP

E24

3

0

20

40

60

80

100

Control KO1

Control KO1n=75 n=40

dsRNA (ZIKV)DNA (Hoechst)/dsRNA (ZIKV)

Control KO1

dsRNA (ZIKV)DNA (Hoechst)/dsRNA (ZIKV)

n=137 n=3275µm

dsR

NA

sta

inin

g vo

lum

e(µ

m3 /

cell)

200

400

600

800

0

% c

ells

pos

itive

for

dsR

NA

sta

inin

g

****

***

10 4

10 5

10 3

U251SK-N-BE2c

siRNA

36h

PE243

36h

Vira

l RN

A c

opie

s pe

r ng

RN

A

siMSI1

siCon

siMSI1

siCon

102

103

104

105

106

107

0 8 24 48 72Time (h)

siCon

siCon

siMSI1

siMSI1

–

–135

42kDa

U251SK-N-BE2c

siCon

siMSI1

H9-NSC

MSI1

kDa–42

p150–135

27.85kb

98761 5 141110

MSI1

(KO2) (KO1)

guide RNA guide RNA

15

H9-NSC

2

4

6

0

8 *

siMSI1

siCon

Vira

l RN

A c

opie

s pe

r ng

RN

A (

X10

4 )

Vira

l RN

A c

opie

s pe

r ng

RN

A

MSI1 (CT)

MSI1 (NT)

p150

Contro

l

KO1KO2

42

42

135

25

25

kDaPar

enta

lU251

ParentalControl

KO1

KO2U251

PE243

72h

0 8 24 48 72

siRNA PE243

72h 48h

p150

MSI1

Paren

tal

Contro

lKO1

KO2

10 4

10 3

10 2

10 1

10 0

n.s

**

Paren

tal

Contro

lKO1

KO2

104

105

103

106 +MR766

0

0.1

0.2

0.3

1234

Luc

ifera

se a

ctiv

ity

(RLU

x106 )

Paren

tal

Contro

lKO1

KO2

ControlZIKV (MoMLV)ZIKV (NL43)VSVg

Vira

l RN

A c

opie

s pe

r ng

RN

A

Vira

l RN

A c

opie

s pe

r ng

RN

A

MSI2 42

Fig. 2. MSI1 is required for effective replication of ZIKV. (A) Graph onleft depicts viral RNA copies in control siRNA (siCon)– and MSI1 siRNA(siMSI1)–treated SK-N-BE2c neuroblastoma cells and U-251 cells afterinfection with PE243 [multiplicity of infection (MOI): 1 focus-forming units/cell(FFU/cell), 72 hours]. Graph on right shows viral RNA copies in MSI1 siRNA–treated H9-derived NSCs after infection with PE243 (MOI: 1 FFU/cell,48 hours). In all viral replication assays, ZIKV was quantified by TaqManassay as described in materials and methods. n = 3 biological replicates.(B) Representative Western blots of cell lines treated with control and MSI1siRNAs from Fig. 2A. Blots were probed with antibodies against MSI1 orp150 as a loading control. (C) Location of the guide RNAs used for CRISPR/Cas9-mediated editing of the MSI1 locus in U-251 cells. For further details,see text and fig. S2. Western blots of parental, control, KO1, and KO2 celllines probed with antibodies against N- and C-termini of MSI1 (NT or CT)and MSI2. p150 serves as a loading control. (D) Kinetics of PE243 viral RNAcopies after infection in U-251 cells of different genotypes at the indicatedtime points. Note that cell death precluded collection of RNA from parentaland control cells at 72 hours (MOI: 3 FFU/cell, 72 hours). (E) Confocalmicroscopy images of PE243-infected control and KO1 U-251 cells

immunostained with antibodies against dsRNA (green) and Hoechst DNAstain (red) after mock or PE243 infection (MOI: 3 FFU/cell, 48 hours). Graphon top right shows percentage of cells containing dsRNA signal, whereasbox plot on bottom right depicts total dsRNA-staining volume per cell. Onlycells with detectable dsRNA signal were included in the latter analysis.Boxes: 25th to 75th percentile; whiskers: 5 to 95% range; line: median.(F) Viral RNA copies in U-251 cells of different genotypes after infection withMR766 (MOI: 3 FFU/cell, 48 hours). n = 3 biological replicates. (G) Virus-binding assays performed under conditions that prevent internalization.PE243 infection was performed in U-251 cells of different genotypes(MOI: 3 FFU/cell, 1 hour). n = 3 biological replicates. Bar charts depictmean ± SEM. (H) Pseudotyped particle–infectivity assay in U-251 cells ofdifferent genotypes. HIV (pNL4-3.luc.R-E-) or Moloney murine leukemiavirus (MoMLV) pseudotyped virus expressing a luciferase reporter, witheither PE243 ZIKV, VSVG, or a negative control envelope used to determineviral entry events (VSVG, vesicular stomatitis virus glycoprotein). RLU,relative light units. n = 3 biological replicates. Bar charts depict mean ± SEM.P values were obtained from Student’s t test, unpaired, two-tailed: *P < 0.05;**P < 0.005; ****P < 0.0001; n.s., not significant.

RESEARCH | REPORTon F

ebruary 14, 2020

http://science.sciencemag.org/

Dow

nloaded from

MSI1 interacts with target transcripts throughits two RNA-recognitionmotifs (RRMs) (19, 30, 32).Nuclear magnetic resonance studies show thatthe conserved Ala184 within the RRM2 is anRNA-binding residue (33). Modeling based onthe crystal structure of the RNA-binding proteinHRP1 with RNA indicates that the A184V mu-tation impairs the MSI1-RNA interaction (Fig. 4A).To evaluate the effect of A184V on MSI1 functionin cells, transgenes encoding wild-type (WT) andA184VMSI1 (MsiWT andMsiAV, respectively) were

randomly integrated into KO1 or KO2 U-251 cells,and single clones were isolated (KO1/2-MsiWTor KO1/2-MsiAV). MSI1 protein expression in theseclones was quantified with respect to endogenousprotein levels in parental cells (Fig. 4B). Whencompared to KO2-MsiWT cells, MSI1 RIP recov-ered two- to fourfold fewer target transcriptsfrom KO2-MsiAV, indicative of reduced RNAbinding by the A184V mutant (Fig. 4C). NUMBand MCPH1 protein levels changed accordingly(Fig. 4B). These results were recapitulated in

HEK293T cells expressing Msi1WT or Msi1AV(fig. S11). MSI1A184V, MSI1 KO, KO1/2-MsiAV,and MSI1-depleted cells all exhibited suboptimalMCPH1 protein levels and PCC, prompting us toinvestigate if a functional link existed betweenthese phenotypes (fig. S9, C to F). MCPH1 over-expression reduced PCC frequency in MSI1-depleted cells, thereby confirming a role for MSI1in chromosome condensation through translation-al control of MCPH1 (fig. S8, G and H). Therefore,the A184V mutation impairs binding of MSI1 to

Chavali et al., Science 357, 83–88 (2017) 7 July 2017 4 of 6

MSI1 β-III tubulin

10 pcw

12 pcw

MSI1 Nestin

MSI1 β-III tubulin DNA

SVZ

CP

VZ

MSI1 Nestin DNA

50µm

50µm

VZ

SVZ

MCPH1_L 3’UTR

free probe:

complex:MCPH1_L 3’UTR+GST-MSI1

GST-M

SI1

GST

GST

GST-MSI1

MSI1

Input IgG MSI1IP

42kDa

4

10

20

30

40

50

GAPDH

MCPH1_

L

CDK6p2

1

NUMB

0

8

p=0.05

CLI

P (

% in

put)

32

MCPH1110

4632

kDa

CDK6

MSI1 46

CDK4 32

p150150

Paren

t (A18

4V/+

)

Patien

t

(A18

4V/A

184V

) Mock

PE243

MR76625

42 –

–

kDa

MSI1

PE243 MR766Mock

Inpu

tIg

GM

SI1In

put

IgG

MSI1

Inpu

tIg

GM

SI1

ZIKV

RIP

(%

inpu

t)

ZIKV RNA

02

46

01

23

NUMBNUMB MCPH1_S

01

5.15.0

MCPH1_L

60

24

8

CDK6 GAPDH

01

5.15.0

01

5.15.0

0.2

Moc

k

36 CDK6

MCPH111072

PE243

NUMB70

p2125

α-tub50

MSI1

25

42

kDa

Paren

tal

Contro

l

KO1KO2

kDa

NUMB

MSI1

70

40

MCPH1_LMCPH1_S70

110

P C K K

p150135

25 p21

Fig. 3. MSI1 is enriched in neural progenitors and regulates micro-cephalin (MCPH1) expression. (A) Immunohistochemistry of humanembryonic brain at postconception week (pcw) 10 and 12. Tissue sectionsstained with antibodies against MSI1 (red) combined with neuron-specificb-III tubulin (green), or the apical neural progenitor marker Nestin (green).DNA is detected by 4′6-diamidino-2-phenylindole (DAPI) (blue). Note thatMSI1 is enriched in neural progenitors at the ventricular and subventricularzones (VZ and SVZ, respectively) but is absent from the cortical plate(CP). (B) MSI1 CLIP from U-251 cells with genotypes as indicated. CLIPwas performed with rabbit IgG or MSI1 antibodies. Graphs show qPCRs ofbound transcripts. n = 3 biological replicates. (C) RNA electrophoreticmobility shift assay (EMSA) analysis to detect binding between MCPH1_L3′UTR and purified GST-MSI1 recombinant protein (GST, glutathioneS-transferase). Coomassie staining of corresponding purified proteins isshown below. (D) Western blots of parental, control, KO1, and KO2 celllines. Blots were probed with antibodies as indicated. p150 serves as

loading control. (E) Western blots of whole-cell lysates from MSI1A184V

parent-of-patient– and patient-derived primary lymphocytes. Blots wereprobed with antibodies as indicated. MSI1 levels are unchanged. (F) MSI1RIP from mock- and ZIKV-infected U-251 cells (MOI: 1 FFU/cell, 72 hours).Rabbit IgG or MSI1 antibodies were used for IP. Input corresponds to10% of whole-cell extract. Western blot was probed with MSI1 antibody.Graphs below show amounts of bound RNAs, including ZIKV genome andendogenous target transcripts. RIP values are presented as a percentageof input after subtraction of the IgG background. MR766 and PE243were quantitated by TaqMan assay and endogenous transcripts with SYBRqPCR. Bar charts depict mean ± SEM. n = 3 biological replicates. (G)Western blots of PE243-infected U-251 cells (MOI: 1 FFU/cell, 72 hours).Blots were probed with antibodies as indicated. a-tubulin serves as loadingcontrol. MSI1 positively regulates MCPH1 and CDK6 and negatively regulatesNUMB and p21 protein levels. P values were obtained from Student’s t test,unpaired, two-tailed: *P < 0.05; **P < 0.005; n.s., not significant.

RESEARCH | REPORTon F

ebruary 14, 2020

http://science.sciencemag.org/

Dow

nloaded from

Chavali et al., Science 357, 83–88 (2017) 7 July 2017 5 of 6

RNA

A184 V184

RNA

0 0.4 0.8 1.2 1.6

NUMB

MSI1

MCPH1_SMCPH1_L

Paren

tal

KO1KO2

KO1-M

si1W

T

KO2-M

si1W

T

KO1-M

si1AV

KO2-M

si1AV

Contro

l

kDa40 MSI1

α-tubulin50

NUMB70

MCPH1_LMCPH1_S70

110

KO2-Msi1WT

KO2-Msi1AVParental

Inpu

tIg

GM

SI1In

put

IgG

MSI1

Inpu

tIg

GM

SI1

MSI142

kDa

KO2-Msi1WT

Parental

KO2-Msi1 AV

***

** **

−30

−20

−10

0

5

−40−45

n.s

Paren

tal

Contro

lKO1

KO2

KO2-M

siWT

KO1-M

siWT

KO2-M

siAV

KO1-M

siAV

PE243 in U251 cells (MOI:3)

Cel

l via

bilit

y (%

cha

nge)

******

**n.s

104

105

106

103

Paren

tal

Contro

lKO1

KO2

KO2-M

siWT

KO1-M

siWT

KO2-M

siAV

KO1-M

siAV

PE243 in U251 cells (MOI:3)

Cel

l via

bilit

y (%

cha

nge)

PE243 in HEK293 cells (MOI:3)

-30

-20

-10

0

1015

n.s * ** n.s n.s

***

A184V

RRM2RRM1118620

362

MSI1 (wild-type)

MSI1 (A184V)

20

15

10

5

0

% Z

IKV

-infe

cfed

cel

ls

HEK-Con

trol

Paren

tal H

EK

HEK-MsiW

T1

HEK-MsiW

T2

HEK-MsiA

V1

HEK-MsiA

V2

PE243 in HEK293 cells (MOI:1)

HEK-Con

trol

Paren

tal H

EK

HEK-MsiW

T1

HEK-MsiW

T2

HEK-MsiA

V1

HEK-MsiA

V2

Vira

l RN

A c

opie

s pe

r ng

RN

A

RIP

(%

inpu

t)

CDK6

01

23

4

n.s

MCPH1_LMCPH1_L

00.

51

1.5 n.s

GAPDH

00.

51

1.5 n.s

NUMB

00.

51

1.5

2 n.s

MCPH1_S

00.

51

1.5 n.s

Fig. 4. The A184V MCPH mutation disrupts RNA binding by MSI1 andimpairs the ability of MSI1 to drive ZIKV replication. (A) Structuralmodel of MSI1 (blue) and HRP1-RNA complex (gray-orange) predicts that theA184V mutation impairs the interaction with RNA because of a steric clash.(B) Western blots of cell lines stably expressing WTor A184V MSI1transgenes. Cell lines were derived from either KO1 or KO2 cells as specified.Blots were probed with antibodies as indicated. p150 serves as loadingcontrol. Graph shows signal intensities of each protein normalized to parentalcells. (C) MSI1 RIP from U-251 cells with genotypes as indicated. Rabbit IgGor MSI1 antibodies were used for IP. Input corresponds to 10% of whole-cellextract. Western blot was probed with MSI1 antibody. Graphs below showqPCRs of bound transcripts. n = 3 biological replicates. (D) Quantification of

viral RNA copies in U-251 cells with the indicated genotypes after infection withPE243 (MOI: 3 FFU/cell, 48 hours). n = 3 biological replicates. (E) Changes insurvival of U-251 cells with different genotypes after infection with PE243(MOI: 3 FFU/cell, 48 hours). n = 3 biological replicates. (F) ZIKV infection ofHEK293T cells as measured by fluorescence-activated cell sorting (FACS)analysis of flavivirus protein E. FACS was performed on control, MsiWT, orMsiAV transgene-expressing cells after infection with PE243 (MOI: 1 FFU/cell,48 hours). n = 2 biological replicates. Bar charts depict mean ± SEM.(G) Changes in survival of control, MsiWT, or MsiAV transgene-expressingHEK293Tcells after infection with PE243 (MOI: 3 FFU/cell, 48 hours). n = 3biological replicates. P values were obtained from Student’s t test, unpaired,two-tailed: *P < 0.05; **P < 0.005; n.s., not significant.

RESEARCH | REPORTon F

ebruary 14, 2020

http://science.sciencemag.org/

Dow

nloaded from

RNA, which leads to reduced MCPH1 expressionand a concomitant increase in PCC frequency.Notably, defective chromosome condensation hasbeen recently found to cause MCPH (34).Given that the A184Vmutation impedes binding

of MSI1 to RNA, we next probed the effect ofA184V mutant MSI1 on ZIKV replication. Tothis end, viral RNA levels and cell viability wereassayed in KO1/2, KO1/2-MsiWT, and KO1/2-MsiAVU-251 cells infected with PE243. Complementationof KO1/2 cells with MsiWT increased both ZIKVRNA levels and cell death (Fig. 4, D and E). Bycontrast, MsiAV was unable to support ZIKVreplication and showed minimal impact on cellviability. Additionally, in HEK293T cells, expres-sion of MsiWT, but not MsiAV, increased viralRNA and cell death upon infection (Fig. 4, F andG, and fig. S12). These findings also imply that MSI1expression increases susceptibility of HEK293Tcells to ZIKV infection (35). Furthermore, we notedan apparent dose-dependent effect of MSI1 onviral replication; those U-251 and HEK293T clonesthat express higher levels of MSI1 displayed greaterviral RNA levels and increased cell death (Fig. 4, Band D to G, and figs. S11 and S12).Our study raises the question of whether MSI1

could have functions in other flaviviruses. Wehave surveyed putative MSI1 binding sites in anumber of flaviviruses by mapping the consen-sus site [A/GU(1 to 3)AG] onto predicted second-ary structures of flaviviridae 3′UTRs obtainedfrom a recent publication (table S3) (16). Althoughseveral flaviviruses harbor consensus MSI1 sitesin appropriate structural landscapes, whetherMSI1 is relevant to the biology of these virusesremains to be established. Furthermore, where-as our data are consistent with a role for MSI1in ZIKV neurotropism and pathology, multiplefactors must collude in ZIKV infection of the fetalbrain, not least of which are viral entry receptorsthat allow the virus to cross the placental barrier(36). Viruses such as human cytomegalovirus andRubella can also access the developing fetal brain,but whether MSI1 contributes to their replication

or pathogenesis is unknown and would requirefurther study.This work suggests that high MSI1 expression

levels in neural precursors could be a key con-tributor to the fetal neurotropism exhibited byZIKV (2, 10, 37) (fig. S13). Intriguingly, MSI1 is alsohighly expressed in the retina and testis, other tis-sues deemed vulnerable to ZIKV infection (38–41).Although our study provides new insight into thepotential pathogenic mechanisms of ZIKV, furtherwork will be required to determine if the mod-ification or interference of the MSI1-ZIKV inter-action results in neuronal attenuation of ZIKV.

REFERENCES AND NOTES

1. J. Mlakar et al., N. Engl. J. Med. 374, 951–958 (2016).2. A. S. Oliveira Melo et al., Ultrasound Obstet. Gynecol. 47, 6–7 (2016).3. F. R. Cugola et al., Nature 534, 267–271 (2016).4. S. N. Slavov, K. K. Otaguiri, S. Kashima, D. T. Covas, Braz.

J. Med. Biol. Res. 49, e5420 (2016).5. S. Cauchemez et al., Lancet 387, 2125–2132 (2016).6. K. M. Adams Waldorf, R. M. McAdams, Reproduction 146,

R151–R162 (2013).7. C. Li et al., Cell Stem Cell 19, 120–126 (2016).8. H. Li et al., Cell Stem Cell 19, 593–598 (2016).9. J. J. Miner, M. S. Diamond, Cell Stem Cell 18, 559–560 (2016).10. J. Dang et al., Cell Stem Cell 19, 258–265 (2016).11. Z. Li, P. D. Nagy, RNA Biol. 8, 305–315 (2011).12. T. Imai et al., Mol. Cell. Biol. 21, 3888–3900 (2001).13. C. L. Donald et al., PLOS Negl. Trop. Dis. 10, e0005048 (2016).14. S. Sakakibara et al., Dev. Biol. 176, 230–242 (1996).15. S. Sakakibara, Y. Nakamura, H. Satoh, H. Okano, J. Neurosci.

21, 8091–8107 (2001).16. S. M. Villordo, C. V. Filomatori, I. Sánchez-Vargas, C. D. Blair,

A. V. Gamarnik, PLOS Pathog. 11, e1004604 (2015).17. P. J. Uren et al., Mol. Cell. Biol. 35, 2965–2978 (2015).18. Z. A. Klase et al., PLOS Negl. Trop. Dis. 10, e0004877 (2016).19. R. G. Fox, F. D. Park, C. S. Koechlein, M. Kritzik, T. Reya,

Annu. Rev. Cell Dev. Biol. 31, 249–267 (2015).20. P. R. de Araujo et al., Am. J. Pathol. 186, 2271–2278 (2016).21. C. Battelli, G. N. Nikopoulos, J. G. Mitchell, J. M. Verdi,

Mol. Cell. Neurosci. 31, 85–96 (2006).22. M. Xu et al., Nat. Med. 22, 1101–1107 (2016).23. Y. Kaneko et al., Dev. Neurosci. 22, 139–153 (2000).24. S. Sakakibara et al., Proc. Natl. Acad. Sci. U.S.A.99, 15194–15199 (2002).25. J. T. Eppig, J. A. Blake, C. J. Bult, J. A. Kadin, J. E. Richardson,

Nucleic Acids Res. 43, D726–D736 (2015).26. E. C. Gilmore, C. A. Walsh, Wiley Interdiscip. Rev. Dev. Biol. 2,

461–478 (2013).27. P. L. Chavali, M. Pütz, F. Gergely, Philos. Trans. R. Soc. London

B Biol. Sci. 369, 20130468 (2014).28. M. Trimborn et al., Am. J. Hum. Genet. 75, 261–266 (2004).

29. A. P. Jackson et al., Am. J. Hum. Genet. 71, 136–142 (2002).30. F. M. Cambuli et al., Stem Cells 33, 3621–3634 (2015).31. M. S. Hussain et al., Hum. Mol. Genet. 22, 5199–5214 (2013).32. H. Kawahara et al., J. Cell Biol. 181, 639–653 (2008).33. T. Nagata et al., J. Mol. Biol. 287, 315–330 (1999).34. C. A. Martin et al., Genes Dev. 30, 2158–2172 (2016).35. R. Hamel et al., J. Virol. 89, 8880–8896 (2015).36. A. S. Richard et al., Proc. Natl. Acad. Sci. U.S.A. 114,

2024–2029 (2017).37. J. B. Brault et al., EBioMedicine 10, 71–76 (2016).38. H. M. Lazear et al., Cell Host Microbe 19, 720–730 (2016).39. S. Salinas et al., J. Virol. 91, e02144-16 (2017).40. K. Susaki et al., Exp. Eye Res. 88, 347–355 (2009).41. W. Ma et al., Cell 167, 1511–1524.e10 (2016).

ACKNOWLEDGMENTS

The authors are indebted to A. Kohl (Centre for Virus Research,University of Glasgow) and L. J. Pena and R. Oliveira de Freitas França,Fiocruz Recife, Pernambuco, Brazil, for the provision of PE243 ZIKVRNA used to generate the virus stock. We thank and acknowledgeS. Lisgo for the provision of human embryonic histology sectionsthrough the Human Developmental Biology Resource (HDBR) at theUniversity of Newcastle funded by a joint UK Medical Research Council(MRC)/Wellcome Trust grant (099175/Z/12/Z). We thank L. Smithfor her assistance with homology modeling, G. van Zande for hishelp, and the patients’ families for their participation. The NationalResearch Ethics Service Committee, East of England–CambridgeCentral, UK (C.G.W., REC 05/Q0108/402) approved the informedconsent to enter the study. We further thank T. Hussain for help withpolysome fractionations and J. Sinclair and A. Git for technical advice.We thank K. J. Patel and members of the Gergely lab for usefuldiscussions and comments. We are grateful for help from the CancerResearch UK Cambridge Institute (CRUK CI) Core Facilities. I.G. andA.E.F. are Wellcome Trust Senior Fellows. I.G. was supported byresearch grants 097997/Z/11/A and 097997/Z/11/Z, and A.E.F. wassupported by grant 106207. M.S.N. was funded by the WellcomeTrust (200183/Z/15/Z), and T.R.S. is a Wellcome Trust Henry DaleFellow (202471/Z/16/Z). This workwasmade possible by funding fromCRUK C14303/A17197 to F.G. and C24461/A12772 to R.B. F.G. andC.G.W. acknowledge support from the National Institute for HealthResearch (NIHR) Cambridge Biomedical Research Centre, theUniversity of Cambridge, and Hutchison Whampoa Ltd. All data tounderstand and assess the conclusions of this research are availablein the main paper and supplementary materials.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/357/6346/83/suppl/DC1Materials and MethodsFigs. S1 to S13Tables S1 to S4References (42–56)

8 February 2017; accepted 16 May 2017Published online 1 June 201710.1126/science.aam9243

Chavali et al., Science 357, 83–88 (2017) 7 July 2017 6 of 6

RESEARCH | REPORTon F

ebruary 14, 2020

http://science.sciencemag.org/

Dow

nloaded from

replicationNeurodevelopmental protein Musashi-1 interacts with the Zika genome and promotes viral

C. Geoffrey Woods and Fanni GergelySweeney, Ben A. Krishna, Myra Hosmillo, Andrew E. Firth, Richard Bayliss, Carlo L. Marcelis, Susan Lindsay, Ian Goodfellow, Pavithra L. Chavali, Lovorka Stojic, Luke W. Meredith, Nimesh Joseph, Michael S. Nahorski, Thomas J. Sanford, Trevor R.

originally published online June 1, 2017DOI: 10.1126/science.aam9243 (6346), 83-88.357Science 

, this issue p. 83; see also p. 33Scienceexpressed at high levels in the mouse testis, which may explain the sexual transmission of this virus.cells. This interaction could put a pregnant woman at risk of giving birth to a microcephalic child. Furthermore, MSI1 is abnormal brain development (see the Perspective by Griffin). MSI1 also binds ZIKV RNA to amplify viral replication inmutations in a neural precursor protein, Musashi-1 (MSI1), impede RNA binding to neural stem cell targets, resulting in

found that in congenital microcephaly,et al.virus damages brain development in the fetus is enigmatic. Chavali Microcephaly has been the terrifying hallmark of the recent outbreak of Zika virus (ZIKV) in the Americas. How the

Inherited microcephaly exposes Zika culprit

ARTICLE TOOLS http://science.sciencemag.org/content/357/6346/83

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2017/05/31/science.aam9243.DC1

CONTENTRELATED

http://stm.sciencemag.org/content/scitransmed/9/388/eaag0538.fullhttp://stm.sciencemag.org/content/scitransmed/8/330/330ed2.fullhttp://stm.sciencemag.org/content/scitransmed/8/369/369ra179.full

REFERENCES

http://science.sciencemag.org/content/357/6346/83#BIBLThis article cites 56 articles, 14 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Science. No claim to original U.S. Government WorksCopyright © 2017 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of

on February 14, 2020

http://science.sciencem

ag.org/D

ownloaded from