Retraction

MICROBIOLOGYRetraction for “Pan-viral-microRNA screening identifies inter-feron inhibition as a common function of diverse viruses,” byJennifer E. Cox, Lydia V. McClure, Andrei Goga, and ChristopherS. Sullivan, which appeared in issue 6, February 10, 2015, of ProcNatl Acad Sci USA (112:1856–1861; first published January 26,2015; 10.1073/pnas.1417891112).The authors wish to note the following: “After publication, we

were alerted to anomalous and duplicated immunoblot bands inFigs. 3D and 4C, for which we immediately contacted PNAS.Further investigation revealed additional band irregularities inboth the immunoblots and Northern blots presented, includingthose in Figs. 3E, 4D, 5 A, B, and D and Figs. S1 and S4B. Whenasked, the experimentalist could not provide key original non–computer-manipulated immunoblot and Northern blot data thatwere consistent with the experimental results being obtained asclaimed in the paper. We note that the three nonexperimentalistauthors were unaware that the data presented were not reflectiveof original Northern blot and immunoblot data. We continue toinvestigate how this occurred and whether any of the remainingdata are credible. For these reasons, all conclusions in the workmust be viewed with skepticism unless independently validated.Accordingly, we are retracting this paper. We deeply regretpublishing with these deficiencies and apologize to the readersfor not catching these problems sooner, and for all the wastedeffort this has caused.”

www.pnas.org/cgi/doi/10.1073/pnas.1504385112

E1692 | PNAS | March 31, 2015 | vol. 112 | no. 13 www.pnas.org

Dow

nloa

ded

by g

uest

on

Nov

embe

r 7,

202

0 D

ownl

oade

d by

gue

st o

n N

ovem

ber

7, 2

020

Dow

nloa

ded

by g

uest

on

Nov

embe

r 7,

202

0 D

ownl

oade

d by

gue

st o

n N

ovem

ber

7, 2

020

Dow

nloa

ded

by g

uest

on

Nov

embe

r 7,

202

0 D

ownl

oade

d by

gue

st o

n N

ovem

ber

7, 2

020

Dow

nloa

ded

by g

uest

on

Nov

embe

r 7,

202

0 D

ownl

oade

d by

gue

st o

n N

ovem

ber

7, 2

020

Pan-viral-microRNA screening identifies interferoninhibition as a common function of diverse virusesJennifer E. Coxa, Lydia V. McClurea, Andrei Gogab, and Christopher S. Sullivana,1

aDepartment of Molecular Biosciences, Center for Systems and Synthetic Biology, Center for Infectious Disease, and Institute for Cellular and MolecularBiology, The University of Texas at Austin, Austin, TX 78712-0162; and bDepartment of Cell & Tissue Biology, University of California, San Francisco,CA 94143-0512

Edited by Peter Palese, Icahn School of Medicine at Mount Sinai, New York, NY, and approved January 1, 2015 (received for review September 16, 2014)

Diverse viruses encode regulatory RNAs called microRNAs (miRNAs).Despite much progress, the functions of the majority of viral miRNAsremain unknown. Most previous studies have used biochemicalmethods to uncover targets of viral miRNAs, but it is unclear whatfraction of these targets is functionally important. Here, we applyan alternative strategy based on the premise that assorted viralmiRNAs will share functionality. Screening a library of >70 humanviral miRNAs showed that three unrelated miRNAs from distantlyrelated herpesviruses significantly inhibited IFN signaling. Strik-ingly, each of these miRNAs directly reduced expression of thecyclic AMP-responsive element-binding protein (CBP), which as partof the p300-CBP complex, mediates IFN signaling. We show thatboth 5′ and 3′ derivatives from Epstein–Barr virus (EBV) encodedmiR-BART-18 precursor miRNA (pre-miRNA) and the orthologouspre-miRNA from Rhesus lymphocryptovirus contribute to reduc-ing IFN signaling. Thus, through both convergent and divergentevolutionary mechanisms, varied herpesviral miRNAs share theability to decrease IFN signaling. Restoring miR-BART-18 to cellsinfected with an EBV miRNA mutant conveyed a cellular growthadvantage upon IFN treatment, and relevant miRNAs from otherherpesviruses were able to complement this activity. Blocking miR-BART-18 function in an EBV+ tumor cell line renders cells moresusceptible to IFN-mediated effects. These findings provide a mech-anism that can at least partially explain the resistance of some EBV-associated tumors to IFN therapy. Our work suggests that similarpan-viral-miRNA functional-based screening strategies are war-ranted for determining relevant activities of other viral miRNAs.

p300/CBP | latency | KSHV | HCMV | RNAi

MicroRNAs (miRNAs) are a class of small regulatory RNAsthat have attracted much attention as regulators of diverse

biological activities. Both host and viral miRNAs have beenimplicated in the control of processes relevant to virus infection,including the immune response, cell death, and regulation ofvirus gene expression (1–5). Viral miRNAs have generated ex-citement, in part, as a previously unrecognized class of geneproducts that could contribute to understanding the mechanismsof viral disease. Biochemical cross-linking studies have identifieddirect transcript targets of miRNAs and have provided muchinsight into viral miRNA biology (6–9). However, there remainmany unanswered important questions. For example, there islittle understanding of how viral miRNA genes evolve and thefunctions of the majority of miRNAs remain unknown. Like hostmiRNAs, individual viral miRNAs can bind to numerous targets,but is unclear what fraction of these targets are biologicallymeaningful (1, 10).The majority of viral miRNAs are found in viruses that take up

long-term persistent infections (11). In particular, most reportedviral miRNAs are encoded by herpesviruses (1, 4, 11–13). In-fection with herpesviruses is associated with serious disease, in-cluding several different cancers. Epstein–Barr virus (EBV), themost common herpesviral human tumor virus, is associated withlymphocyte and solid tissue cancers (14). Like most herpesvirusesstudied, EBV encodes numerous—at least 25—pre-miRNAs thatgive rise to even more bioactive derivatives (15–18). EBV tumors

mostly undergo latent infection, a state of limited viral gene ex-pression without virus production. In some latent states, noncodingRNAs are the only abundant viral gene products expressed (19–22).Independent studies implicate oncogenic activity of viral miRNAsin at least some EBV-associated tumors (13, 23–26). Developingmore effective therapies against EBV-associated tumors could savethousands of lives each year (27). Recently, adaptive immune re-sponse-based therapies have shown much promise (28–30). Incontrast, innate immune therapies based on treatment with type IIFNs have shown limited success (31–33). Positive responses haveonly been observed in a subset of patients and tumor types (30, 33).Deciphering the functions of EBV miRNAs has utility not only inunderstanding EBV biology and disease but potentially also tumorresponse to treatment. In this study, we use a functional-basedscreen of all known human herpesviral miRNAs, and additionalhuman viral miRNAs, to uncover an EBVmiRNA that inhibits IFNsignaling and associated growth arrest. This activity is shared withmiRNAs from closely related and distantly related herpesviruses,and our work indicates that this shared activity occurs at least in partthrough a common host target. These findings expand the un-derstanding of viral miRNA function and herpesvirus latency, andmay help explain the response of tumors to IFN therapy.

ResultsFunctional Screen Identifies Inhibition of IFN Activity by HerpesviralmiRNAs.miRNA target identification strategies have led to muchinsight into viral miRNA function, but it remains unclear howcomprehensive these approaches are, and what fractions of theidentified targets are virologically relevant. To complement theseapproaches, we undertook a “functions-first” study based on thepremise that similar to regulatory viral proteins, miRNAs from

Significance

As part of the host–pathogen evolutionary arms race, virusesmust develop mechanisms to evade host defenses. Viral miRNAs,small regulatory RNAs, are likely key weapons in this struggle.Most previous studies attempted to understand viral miRNAfunctionality starting with target identification. Here, we usean alternative “functions-first” approach to test the hypothesisthat miRNAs from distantly related viruses share commonfunctions. Our work identifies evasion of the antiviral cytokineinterferon as a common activity of divergent herpesviral miRNAs.This approach expands our understanding of latent infection andmay at least partly explain the resistance of some herpesviral-associated tumors to interferon-mediated therapy. This workdemonstrates the utility of functional-based screening for uncov-ering relevant viral miRNA activities.

Author contributions: J.E.C., A.G., and C.S.S. designed research; J.E.C. performed research;L.V.M. contributed new reagents/analytic tools; J.E.C. and C.S.S. analyzed data; and J.E.C.and C.S.S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: Chris_sullivan@austin.utexas.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1417891112/-/DCSupplemental.

1856–1861 | PNAS | February 10, 2015 | vol. 112 | no. 6 www.pnas.org/cgi/doi/10.1073/pnas.1417891112

See Retraction Published , 201 March 23 5

different viruses will possess shared functionality. We constructeda miRNA expression library expressing 70 high-confidence hu-man miRNAs from the polyomavirus, anellovirus, and herpesvi-rus families, plus an additional 27 constructs for less-confidenthuman viral miRNAs and viruses that only rarely infect humans(Table S1). We tested a subset of 30 of these constructs andconfirmed via Northern blot analysis that transfection of ourvectors gave rise to physiological miRNA levels (Fig. S1). Wechose to screen for miRNA-mediated inhibition of IFN as a po-tential widespread activity of viruses for two reasons: (i) proteinsthat hinder the IFN response are nearly universal among mam-malian viruses, and (ii) we have already demonstrated that theunrelated viruses, Torque teno virus tth8 (TTV-tth8) and SimianFoamy virus of African green monkey (SFVagm) encode miRNAsthat inhibit type I IFN signaling (34, 35). Thus, we conducted ascreen to identify those human viral miRNAs that block IFN-mediated signaling.HEK293T cells were cotransfected with an IFN-responsive

luciferase reporter and individual members of our viral miRNAlibrary, or a positive control vector expressing miR-132. Forty-eight hours posttransfection, IFN was added for 14 h and thena luciferase assay was conducted (Fig. 1A). As expected, miR-132, which has previously been shown to inhibit IFN signaling(35, 36) scored in our assay by reducing the IFN-induced tran-scription of the firefly-Luc reporter by 35%. Additionally, threeherpesviral miRNAs consistently scored as having a ≥25% re-duction (Fig. 1B). These miRNAs were EBV miR-BART-18,Kaposi’s sarcoma-associated herpesvirus (KSHV) miR-K12, andhuman cytomegalovirus (HCMV) miR-US25-2. Subsequentanalysis centered on these three herpesviral miRNAs.

Herpesvirus miRNAs Inhibit IFN-Induced Growth Arrest. We havepreviously shown that a lymphoblastoid cell line (LCL), GM19240,immortalized with EBV strain B95.8, undergoes growth inhibitionupon treatment with type I IFN at levels comparable to reportedtherapeutic dosages (32–35). As B95.8 is a deletion mutant viruslacking 20 of the 25 pre-miRNAs (including miR-BART-18) (Fig.2A) (37), we asked if restoring miR-BART-18 expression toGM19240 cells increased cell growth in the presence of IFN.Similar to the positive control miR-132, miR-BART-18 partiallyhindered the growth inhibitory effects of IFN treatment (Fig. 2B).Therefore, miR-BART-18 can reduce IFN signaling activity in thecontext of EBV latent infection. Next we asked whether miRs K12and US25-2 could functionally complement miR-BART-18 inpromoting cellular growth in the presence of IFN. Indeed, both ofthese miRNAs were sufficient to enhance growth of GM19240cells upon IFN treatment (Fig. 2B), albeit to a lesser degree thanmiR-BART-18. To further elucidate the mechanism by whichmiR-BART-18 provides a growth advantage, LCLs grown in thepresence of IFN were assayed for cell viability and alterations incell cycle rate. miR-BART-18 expression provided no measurableprotection from IFN-mediated cell death compared with irrele-vant miRNA or control empty vector (Fig. S2A). However, cellproliferation analysis demonstrated that upon IFN treatment,miR-BART-18–expressing LCLs divide at a faster rate than cellsexpressing either an irrelevant miRNA or a control vector (Fig. 2Cand Fig. S2B). Taken together, these results demonstrate thatdiverse human herpesviruses encode miRNAs that inhibit IFN-mediated signaling, and this has the functional consequence ofconditionally promoting cell growth in latently infected cells.

CBP Is a Direct Target of Herpesviral miRNAs.To identify targets thatcould account for the IFN inhibitory effects of miR-BART-18,we integrated previously published immunoprecipitation (7, 38)and mRNA steady-state level microarray studies (39, 40), andasked which putative targets overlapped between the studies.Then, using DAVID analysis (41), we determined which of theseoverlapping candidate targets were functionally implicated in theantiviral response (Fig. 3A). This analysis identified five putativemiR-BART-18 target transcripts (Fig. S3A). To determine whichof these putative targets are directly regulated by miR-BART-18,

we generated and screened 3′UTR luciferase reporters for reg-ulation by miR-BART-18. Of the five putative targets, only thecyclic AMP-responsive element binding protein (CBP) 3′UTRreporter was significantly down-regulated by cotransfection ofplasmid-expressing miR-BART-18 (Fig. S3B). To determine ifthis regulation was direct, we generated a reporter with twonucleotides mutated in the putative miR-BART-18 5p dockingsite. These mutations abolished regulation, thereby confirmingthat miR-BART-18 directly regulates CBP transcripts (Fig. 3C).Next, we examined the effect of miR-BART-18 expression onendogenous CBP protein levels. HEK293T cells were transfectedwith a vector expressing miR-BART-18 and total protein washarvested. Immunoblot analysis showed that expression of miR-BART-18 decreased CBP protein levels relative to actin (Fig.3D). Unlike typical pre-miRNAs, pre-miR-BART-18 has beenreported to give rise to both abundant 5′ (5p) and 3′ derivate (3p)miRNAs, depending on cellular context (39, 42, 43). We con-firmed that expression of pre-miR-BART-18 gives rise to ap-proximately equally abundant 5p and 3p derivatives in HEK293Tcells (Fig. S4B). To determine if both the 5p and 3p miRNAs canregulate CBP transcripts, or alternatively, whether miR-BART-185p exclusively regulates CBP levels, we used specific molecularsponges against either the 5p or 3p miR-BART-18 miRNAs.Each sponge was confirmed to specifically inhibit 5p or 3pmiRNA activity via luciferase assay performed on 5p- or 3p-specific luciferase reporters (Fig. S5A). We further demonstrated

Fig. 1. Functional screen identifies three unrelated herpesviral miRNAsthat inhibit IFN signaling. (A) Schematic of screening procedure. Pre-miRNAexpression vectors (n = 97) transfected into HEK293T cells give rise to maturemiRNA derivatives. After type I IFN treatment, IFN signaling is measured viafirefly-luciferase reporter activity. (B) IFN signaling activity in the presence ofeach miRNA. Luciferase levels were normalized to cells transfected withcontrol empty vector. Error bars represent SEM of three independentexperiments each performed in triplicate. The positive control, miR-132, andthe viral miRNAs, EBV miR-BART-18, KSHV miR-K12, and HCMV miR-US25-2,are indicated in the inset. **P ≤ 0.01.

Cox et al. PNAS | February 10, 2015 | vol. 112 | no. 6 | 1857

MICRO

BIOLO

GY

See Retraction Published , 201 March 23 5

that the sponges were effective, because the mean fluorescent in-tensity of eGFP, whose expression derives from the coding portionof the sponge transcripts (44, 45), is lower only in cells that expressmiR-BART-18 (Fig. S5B). Cotransfection of the individual spongevectors with a miR-BART-18 expression vector indicated that CBPprotein levels are partially restored only in the presence of the 5psponge (Fig. 3E). Combined, these results demonstrate that miR-BART-18 5p directly inhibits CBP expression.Because miRs BART-18, K12, and US25-2 scored to a similar

degree in both the IFN signaling and growth promotion assays,we interrogated whether there are any putative miRNA bindingsites in the CBP transcript for miRs K12 and US25-2. In fact,bioinformatic analysis identified nonoverlapping complementaryregions in the 3′UTR of CBP to the seed regions of both miRsK12 and US25-2 (Fig. 3B). As seed regions are established asplaying a particularly important role in mediating direct in-teraction between the miRNA and target, we generated luciferasereporters with either the WT 3′UTR of CBP or identical versionsexcept with the candidate miRNA seed docking sites mutated.Cotransfection of vectors expressing either miR-K12 or miR-US25-2 with WT CBP 3′UTR reporters resulted in down-regu-lation of luciferase levels, and this was ablated when the re-spective putative miRNA docking sites were mutated (Fig. 3C).These data suggest that miRs K12 and US25-2 directly regulateCBP transcripts. To determine if these miRNAs could alter en-dogenous CBP protein levels, we transfected cells with vectorsexpressing either miR-K12 or miR-US25-2 and then immuno-blotted for CBP levels. In contrast to the empty vector-transfectedcontrol cells, both miRs K12 and US25-2 reduced endogenous CBPprotein levels (Fig. 3D). These data demonstrate that like miR-BART-18, miRs K12 and US25-2 directly regulate CBP levels.

Both miR-BART-18 5p and 3p Contribute to IFN Signaling Repression.Our observation that unrelated miRNAs down-regulate CBPand IFN-mediated signaling implies that different human her-pesviruses have converged on this strategy as a common way toevade a host defense. Given the presumed importance of suchregulation, one prediction from these findings is that closelyrelated viruses should also display similar regulatory activities.

To test this, we screened pre-miR-24 from Rhesus lymphoc-ryptovirus (rLCV), which is orthologous to the premiR-BART-18 locus (Fig. S4A) (16, 46). Indeed, rLCV-miR-24 inhibitedIFN-mediated signaling to a similar degree as miR-BART-18(Fig. 4B). Our above work (Fig. 3) established a role for miR-BART-18 5p in directly regulating CBP, but it does not rule outcontributions of the 3p derivative in evading IFN-mediated sig-naling via other targets. To test this possibility, we used oursponges that target either the miR-BART-18 5p or 3p miRNAs(Fig. S5A). As expected, cotransfection of the 5p sponge with themiR-BART-18 expression vector substantially restored luciferaselevels in the IFN-mediated signaling assays. The 3p sponge didnot restore reporter levels to the same degree as the 5p sponge,but significant activity was still observed (Fig. 4A). Combiningboth sponges resulted in restoration of almost 100% of the lu-ciferase activity, thereby suppressing nearly all IFN-inhibitoryactivity of the pre-miR-BART-18 (Fig. 4A). These results suggestthat although the 5p derivative accounts for the majority of IFNinhibitory activity of pre-miR-BART-18, there is also an additionalsmaller contribution from the 3p miRNA. Based on our abovestudies (Fig. 3E), the target of the 3p miRNA is unlikely to beCBP, thereby implicating additional target transcripts in the IFNsuppressive activity of miR-BART-18. To our knowledge, thismakes pre-miR-BART-18 one of the few pre-miRNAs to havedocumented identical bioactivity from both the 5p and 3p deriv-atives. We performed similar sponge studies on rLCV miR-24and observed the same trend, suggesting that this atypical featureof pre-miR-BART-18 is evolutionarily conserved (Fig. 4B). Thus,both closely related and more distantly related herpesviruses ex-press IFN-inhibitory miRNAs.Our above studies suggest that targeting CBP accounts for the

major IFN inhibitory effects of miR-BART-18, with smalleradditive contributions from additional targets. To better definethe potential magnitude of the contribution of targeting CBP,two different experiments were performed. First, we combinedknockdown of CBP with BART-18 expression. We treated cellswith an siRNA specific for CBP transcripts and this resulted inspecific knockdown of CBP proteins levels (Fig. 4C). When cellswith reduced CBP levels were treated with IFN, a significant

Fig. 2. Herpesviral miRNAs block IFN-inducedgrowth arrest. (A) B95.8 is an EBV deletion mutantthat does not encode 17 of the 22 BART pre-miRNAs(light gray arrows), including miR-BART-18 (blackarrow). (B) LCLs (GM19240) latently infected withEBV strain B95.8, exhibit growth inhibition upontype I IFN treatment. GM19240 cells stably expressingmiR-BART-18, miR-K12, or miR-US25-2 pre-miRNAswere treated with daily IFN. Growth curves of threeindependent experiments in the absence or presenceof IFN were calculated over 5 d. Error bars representSEM. (C) LCLs expressing miR-BART-18 or controlswere treated with CFSE and subsequently culturedwith or without IFN. The number of cellular divisionswere determined through FACS analysis. A plot ofmean division number versus time is shown for twoindependent experiments performed in triplicate.

1858 | www.pnas.org/cgi/doi/10.1073/pnas.1417891112 Cox et al.

See Retraction Published , 201 March 23 5

reduction in IFN signaling was observed (Fig. 4C). However,coexpression of miR-BART-18 and the CBP siRNA did notresult in an enhanced inhibition of IFN signaling. Second, we

coexpressed miR-BART-18 and a version of CBP that is resistantto miR-BART-18 5p regulation. Cotransfection of a CBP ex-pression construct (lacking its natural 3′UTR and associatedmiR-BART-18 docking site) with the miR-BART-18 expressionvector resulted in near full restoration of interferon-stimulatedresponse element (ISRE) activity (Fig. 4D). Combined, theseresults show that although other targets contribute, repressingCBP transcripts accounts for a large fraction of the IFN-inhibitoryactivity associated with miR-BART-18.

BART-18 Reduces IFN-Mediated Effects in an EBV+ Tumor Cell Line.Although we confirmed physiologically relevant levels of miR-BART-18 in the above studies, we wished to determine whethermiR-BART-18 expressed from the EBV genome can inhibit IFNactivity. Akata A.15 cells are a Burkitt’s lymphoma cell line latentlyinfected with EBV (47, 48). Consistent with previous studies (40,49), we were able to confirm that Akata A.15 cells express miR-BART-18 5p (Fig. 5A). We transduced Akata A.15 cells witha vector containing both sponges against miR-BART-18 5p and 3p,and selected for cells expressing the construct. The sponges wereactive in these cells because immunoblot analysis showed increasedCBP levels compared with cells transduced with control spongeconstructs (Fig. 5B). Next, we added IFN daily and measured thegrowth rate of cells over the course of 7 d. This analysis showed asignificant reduction in growth of the BART-18 sponge cells com-pared with cells expressing a control sponge (Fig. 5C). As a controlfor possible off-target effects of the sponge, LCLs deficient in miR-BART-18 expression were transduced with miR-BART-18 spongesand these showed no difference in cell growth (Fig. S5C). We nextassayed for differences in IFN-signaling. The Akata cells expressingthe control sponge showed an induction of the IFN-simulated gene(ISG-15) as assayed by Northern blot (Fig. 5D). However, theAkata cells expressing the BART-18 sponge showed significantlymore induction, consistent with miR-BART-18 inhibiting IFN sig-naling. Thus, an endogenously expressed viral miRNA contributesto inhibition of IFN activity in an EBV tumor cell line.

DiscussionThere has been recent progress in understanding viral miRNAfunctions, but many important questions remain. For example,most viral miRNAs lack a detailed understanding of their rele-vant targets and functions, and little is known about how viralmiRNAs evolve. With notable exceptions (e.g., ref. 50), most

Fig. 3. Herpesviral miRNAs down-regulate CBP. (A) Meta-analysis of RNA:target cross-linking studies and mRNA expression studies identified potentialtargets of miR-BART-18. Targets that overlapped in at least three of the fourstudies were then filtered for those associated with the antiviral response.(B) The 3′UTR of CBP contains predicted docking sites for miRs BART-18, K12,and US25-2. (C) Reporter constructs containing WT CBP 3′UTR, or the CBP3′UTR with mutations in the predicted miRNA binding sites, were indi-vidually cotransfected into HEK293T cells with relevant miRNA expressionvector or empty vector control. The luciferase assay was normalized relativeto empty vector. The SEM of three independent experiments performed intriplicate is shown. (D) Immunoblot analysis for CBP protein levels was per-formed on HEK293T transfected with miRNA expression vectors or emptyvector. (E) Immunoblot analysis of HEK293T cells that were cotransfectedwith miR-BART-18 expression vector and a 5p or 3p sponge. β-Actin served asa load control. *P ≤ 0.05; **P ≤ 0.01.

Fig. 4. Both 5p and 3p pre-miR-BART-18 derivativescontribute to IFN signaling repression. (A) Molecularsponges directed against either miR-BART-18 5p or3p were cotransfected with miR-BART-18 expressionvector and the IFN-stimulated signaling assays wereperformed. Sponges with mutations in the miRNA-docking site served as a control for direct regula-tion. The SEM based on three independent experi-ments, each from the average of three technicalreplicates, is shown. (B) Molecular sponges againsteither rLCV-miR-24 5p or 3p were cotransfectedwith miR-rLCV-24 expression vector then IFN-stim-ulated assays were performed. Sponges withmutations in the miRNA-docking site served ascontrols. (C) HEK293T cells were cotransfected witha miR-BART-18 expression vector and either a con-trol siRNA or an siRNA against CBP, and then IFNsignaling assays were performed using luciferase-based ISRE reporters. Luciferase data are normal-ized to cells cotransfected with empty vector andcontrol siRNA. SEM of three independent experi-ments performed in triplicate are shown. *P ≤ 0.05;**P ≤ 0.01. NS indicates that there is no significantdifference. (Lower) siRNA activity was confirmed via immunoblot. (D) HEK293T cells were cotransfected with a CBP expression vector or an empty vector andmiR-BART-18 expression vector or a control miRNA and ISRE reporter assays performed; “776” refers to a control vectors expressing an irrelevant SV40 strain776 microRNA. Luciferase data are normalized to cells transfected with empty vector alone. SEM of three independent experiments performed in triplicateare shown. **P ≤ 0.01. (Lower) Expression of exogenous CBP was confirmed through immunoblot analysis.

Cox et al. PNAS | February 10, 2015 | vol. 112 | no. 6 | 1859

MICRO

BIOLO

GY

See Retraction Published , 201 March 23 5

previous studies relied on miRNA target identification as thestarting point for uncovering viral miRNA functions. In contrast,we used a “functions-first” approach and uncover that distantlyrelated human herpesviruses encode miRNAs that suppress theIFN response. We integrated these findings with previouslyreported candidate viral miRNA target datasets and show thatCBP is a common target of herpesviral miRNAs, and that tar-geting CBP allows for suppression of IFN-mediated signaling.The fact that distantly related viruses encode miRNAs of

unrelated sequence that score for the same activity implies thatthese viruses have converged on the same strategy to combata host defense. The fact that EBV and the closely related rLCVboth encode sequence-related pre-miRNAs that give rise to twodifferent miRNAs with the same functionality, implies that thesepre-miRNAs diverged from a common ancestor that also sharedthese atypical features. Thus, through both convergent and di-vergent evolutionary mechanisms, herpesviruses regulate CBPand the IFN response.Although reducing CBP levels clearly can reduce IFN activity,

it remains likely that additional miRNA targets contribute. Inparticular, the fact that both the 5p and 3p derivatives of miR-BART-18 and rLCV-miR-24 contribute to IFN inhibition impliesadditional host targets contribute to suppression of IFN activity.Nevertheless, coexpression of anti-CBP siRNAs and miR-BART-18did not result in additional suppression of IFN activity and restoringCBP expression in miR-BART-18–expressing cells recovered mostof the IFN-responsive activity in these cells (Fig. 4C). These findingssuggest that targeting CBP accounts for the majority of IFN sup-pressive activity of miR-BART-18, or that CBP activity is down-stream of any additional relevant targets. In addition to promotingIFN-mediated signaling, it is well established that CBP also par-ticipates in transactivation of type I IFN. Thus, targeting CBP mayhelp herpesviruses evade both the induction and effector phases ofthe IFN response. Conversely, herpesviral miRNA-targeting of CBPcould have benefits to these viruses in addition to evading the an-tiviral IFN response. For example, recruitment of active CBP/p300promotes lytic activation of herpes simplex virus-1 (51). Thus,miRNA-mediated suppression of CBP may promote latency byrestricting lytic activation. In support of this notion, it is interestingthat both miRs BART-18 and US25-2 have been implicated inpromoting latent infection (40, 52, 53). Thus, viral miRNA regu-lation of CBP could represent a multipronged approach to en-hancing viral fitness.Combined with previous publications showing that a retroviral

and an annelloviral miRNA block IFN activity (34, 35), ourcurrent work demonstrates that miRNA-mediated IFN inhibitionis a common activity that is shared among related and disparateviruses. Because inhibition of the IFN response has been linkedwith cancer (54), our findings suggest that these viral miRNAscould contribute to maintenance of tumors associated with EBVand KSHV. In addition to any direct role in cancer, our obser-vation that miR-BART-18 inhibits IFN signaling, and the asso-ciated cell growth arrest, may partly account for the resistance ofsome EBV-associated tumors to IFN treatment in the clinic. Fi-nally, our work further emphasizes the utility of functional-basedassays for uncovering relevant viral miRNA activities and targets.

Materials and MethodsCell Lines. HEK293T cells were maintained in DMEM supplemented with 10%FBS and pen-strep. Akata BL lines A.2 (EBV−) and A.15 (EBV+) were main-tained in RPMI 1640 supplemented with 10% FBS and 10% (vol/vol) pen-strep. Human LCLs, GM19240 cells, were maintained in RPMI 1640 supple-mented with 15% FBS and 10% (vol/vol) pen-strep.

IFN Signaling Assay. Viral miRNA expression vectors were cotransfected withan IFN response reporter into HEK293T cells. Cells were treated with 10 μuniversal IFN and after 14 h a dual-luciferase assay was performed. Inductionof the IFN response is shown relative to an empty-vector control. miR-BART-18 or control sponges were cotransfected in indicated experiments. ffLucvalues were normalized to rLuc to control for transfection efficiencies. In-duction of IFN signaling is shown relative to empty vector. Further detailshave been previously described (34, 35).

Growth Assay. LCLs were transduced with lentiviral vectors and selected withpuromycin to express herpesviral miRNAs, control miRNAs or vector control.Akata A.15 cells were transduced and selected with puromycin to expressmiR-BART-18 sponge or control sponge. LCL and Akata A.15 cell lines weretreated with 750 μ of daily IFN. A growth curve was calculated over a 5- or7-d period in the absence or presence of IFN.

Fig. 5. BART-18 reduces IFN mediated effects in an EBV+ tumor cell line. (A)Northern blot analysis for miR-BART-18 5p of total RNA harvested from EBV+

Akata A.15 cells and EBV− Akata A.2 cells. (B) Immunoblot analysis of CBPprotein levels from cell Akata A.15 cells expressing a miR-BART-18–specific orcontrol sponge. (C) Akata A.15 cells expressing a miR-BART-18–specific orcontrol sponge were treated daily with IFN treatment for 7 d. Growth curvesof each cell line in the presence or absence of IFN was determined. Error barsrepresent SEM of three independent experiments performed in triplicate.*P ≤ 0.05. (D) Northern blot analysis of ISG-15 induction of IFN-treated AkataA.15 cells expressing either a miR-BART-18–specific or control sponge. ISG-15levels were quantified and normalized to actin as a load control. SEM wascalculated from three independent experiments.

1860 | www.pnas.org/cgi/doi/10.1073/pnas.1417891112 Cox et al.

See Retraction Published , 201 March 23 5

Cell Division Assay. LCLs expressing miR-BART-18 or controls were treated withcarboxyfluorescein diacetate succinimidyl ester (CFSE), a dye that decreaseswith each cellular division. FACS analysis was performed to determine thenumber of divisions in cells treated with 750 μ of daily IFN or a BSA control.

MicroRNA Target Reporter Assay. Luciferase-based reporter constructs werecotransfected with miRNA expression vectors or empty-vector controls intoHEK293T cells. After 24 h, a dual-luciferase assay was performed. Luciferasevalues are normalized relative to empty-vector control.

ISG Induction. Akata A.15 cells expressing miR-BART-18 or control spongeswere treated with 100 μ universal IFN. RNA was harvested at several timepoints posttreatment and resolved using a formaldehyde gel. ISG-15 and actinprobes were radioactively labeled. All primers and oligos are listed in Table S2.

Western Blot Analysis. Western blot analysis was performed as previouslydescribed (34).

Small RNA Northern Blot Analysis. Small RNA Northern blot analysis wasperformed as previously described (18).

ACKNOWLEDGMENTS. We thank Dr. Jeffery Sample (The Pennsylvania StateMedical School) for providing cell reagents. This work was supported byGrant R01AI077746 from the National Institutes of Health (to C.S.S.); aBurroughs Wellcome Investigators in Pathogenesis Award (to C.S.S.);Grant RP110098 from the Cancer Prevention and Research Institute ofTexas (to C.S.S.); a University of Texas at Austin Institute for Cellular andMolecular Biology Fellowship (to C.S.S.); and Congressionally DirectedMedical Research Programs Era of Hope Scholar W81XWH-12-1-0272 andNIH U19 CA179512 (to A.G.).

1. Kincaid RP, Sullivan CS (2012) Virus-encoded microRNAs: An overview and a look tothe future. PLoS Pathog 8(12):e1003018.

2. Cullen BR (2013) MicroRNAs as mediators of viral evasion of the immune system. NatImmunol 14(3):205–210.

3. Boss IW, Renne R (2010) Viral miRNAs: Tools for immune evasion. Curr Opin Microbiol13(4):540–545.

4. Pfeffer S, Voinnet O (2006) Viruses, microRNAs and cancer. Oncogene 25(46):6211–6219.5. Skalsky RL, Cullen BR (2010) Viruses, microRNAs, and host interactions. Annu Rev

Microbiol 64:123–141.6. Haecker I, et al. (2012) Ago HITS-CLIP expands understanding of Kaposi’s sarcoma-

associated herpesvirus miRNA function in primary effusion lymphomas. PLoS Pathog8(8):e1002884.

7. Gottwein E, et al. (2011) Viral microRNA targetome of KSHV-infected primary effu-sion lymphoma cell lines. Cell Host Microbe 10(5):515–526.

8. Skalsky RL, Kang D, Linnstaedt SD, Cullen BR (2014) Evolutionary conservation ofprimate lymphocryptovirus microRNA targets. J Virol 88(3):1617–1635.

9. Riley KJ, et al. (2012) EBV and human microRNAs co-target oncogenic and apoptoticviral and human genes during latency. EMBO J 31(9):2207–2221.

10. Grundhoff A, Sullivan CS (2011) Virus-encoded microRNAs. Virology 411(2):325–343.11. Cox JE, Sullivan CS (2014) Balance and stealth: The role of noncoding RNAs in the

regulation of virus gene expression. Annu Rev Virol 1:89–109.12. Cullen BR (2011) Herpesvirus microRNAs: Phenotypes and functions. Curr Opin Virol

1(3):211–215.13. Zhu Y, Haecker I, Yang Y, Gao S-J, Renne R (2013) γ-Herpesvirus–encoded miRNAs and

their roles in viral biology and pathogenesis. Curr Opin Virol 3(3):266–275.14. Kutok JL, Wang F (2006) Spectrum of Epstein-Barr virus-associated diseases. Annu Rev

Pathol 1:375–404.15. Pfeffer S, et al. (2005) Identification of microRNAs of the herpesvirus family. Nat

Methods 2(4):269–276.16. Cai X, et al. (2006) Epstein-Barr virus microRNAs are evolutionarily conserved and

differentially expressed. PLoS Pathog 2(3):e23.17. Zhu JY, et al. (2009) Identification of novel Epstein-Barr virus microRNA genes from

nasopharyngeal carcinomas. J Virol 83(7):3333–3341.18. Grundhoff A, Sullivan CS, Ganem D (2006) A combined computational and micro-

array-based approach identifies novel microRNAs encoded by human gamma-her-pesviruses. RNA 12(5):733–750.

19. Brooks LA, Lear AL, Young LS, Rickinson AB (1993) Transcripts from the Epstein-Barrvirus BamHI A fragment are detectable in all three forms of virus latency. J Virol 67(6):3182–3190.

20. Edwards RH, Marquitz AR, Raab-Traub N (2008) Epstein-Barr virus BART microRNAsare produced from a large intron prior to splicing. J Virol 82(18):9094–9106.

21. Lerner MR, Andrews NC, Miller G, Steitz JA (1981) Two small RNAs encoded byEpstein-Barr virus and complexed with protein are precipitated by antibodies frompatients with systemic lupus erythematosus. Proc Natl Acad Sci USA 78(2):805–809.

22. Qiu J, et al. (2011) A novel persistence associated EBV miRNA expression profile isdisrupted in neoplasia. PLoS Pathog 7(8):e1002193.

23. Seto E, et al. (2010) Micro RNAs of Epstein-Barr virus promote cell cycle progressionand prevent apoptosis of primary human B cells. PLoS Pathog 6(8):e1001063.

24. Vereide DT, et al. (2014) Epstein-Barr virus maintains lymphomas via its miRNAs.Oncogene 33(10):1258–1264.

25. Lopes LF, et al. (2013) Epstein-Barr virus (EBV) microRNAs: Involvement in cancerpathogenesis and immunopathology. Int Rev Immunol 32(3):271–281.

26. Feederle R, et al. (2011) A viral microRNA cluster strongly potentiates the trans-forming properties of a human herpesvirus. PLoS Pathog 7(2):e1001294.

27. Thompson MP, Kurzrock R (2004) Epstein-Barr virus and cancer. Clin Cancer Res 10(3):803–821.

28. Rooney CM, Leen AM, Vera JF, Heslop HE (2014) T lymphocytes targeting native re-ceptors. Immunol Rev 257(1):39–55.

29. Rooney C, Leen A (2012) Moving successful virus-specific T-cell therapy for hematopoieticstem cell recipients to late phase clinical trials. Mol Ther — Nucleic Acids 1(11):49–52.

30. Straathof KCM, Bollard CM, Rooney CM, Heslop HE (2003) Immunotherapy forEpstein-Barr virus-associated cancers in children. Oncologist 8(1):83–98.

31. Gutterman JU (1994) Cytokine therapeutics: Lessons from interferon alpha. Proc NatlAcad Sci USA 91(4):1198–1205.

32. Davis CL, et al. (1998) Interferon-alpha treatment of posttransplant lymphoproliferativedisorder in recipients of solid organ transplants. Transplantation 66(12):1770–1779.

33. Kiladjian J-J, Mesa RA, Hoffman R (2011) The renaissance of interferon therapy forthe treatment of myeloid malignancies. Blood 117(18):4706–4715.

34. Kincaid RP, Burke JM, Cox JC, de Villiers E-M, Sullivan CS (2013) A human torque teno virusencodes a microRNA that inhibits interferon signaling. PLoS Pathog 9(12):e1003818.

35. Kincaid RP, Chen Y, Cox JE, Rethwilm A, Sullivan CS (2014) Noncanonical microRNA(miRNA) biogenesis gives rise to retroviral mimics of lymphoproliferative and im-munosuppressive host miRNAs. MBio 5(2):e00074.

36. Lagos D, et al. (2010) miR-132 regulates antiviral innate immunity through suppres-sion of the p300 transcriptional co-activator. Nat Cell Biol 12(5):513–519.

37. Lin JC, Raab-Traub N (1987) Two strains of Epstein-Barr virus (B95-8 and a P3HR-1subclone) that lack defective genomes induce early antigen and cause abortive in-fection of Raji cells. J Virol 61(6):1985–1991.

38. Dölken L, et al. (2010) Systematic analysis of viral and cellular microRNA targets incells latently infected with human gamma-herpesviruses by RISC immunoprecipita-tion assay. Cell Host Microbe 7(4):324–334.

39. Marquitz AR, Mathur A, Chugh PE, Dittmer DP, Raab-Traub N (2014) Expressionprofile of microRNAs in Epstein-Barr virus-infected AGS gastric carcinoma cells. J Virol88(2):1389–1393.

40. Qiu J, Thorley-Lawson DA (2014) EBV microRNA BART 18-5p targets MAP3K2 to fa-cilitate persistence in vivo by inhibiting viral replication in B cells. Proc Natl Acad SciUSA 111(30):11157–11162.

41. Huang W, Sherman BT, Lempicki RA (2009) Systematic and integrative analysis oflarge gene lists using DAVID bioinformatics resources. Nat Protoc 4(1):44–57.

42. Cosmopoulos K, et al. (2009) Comprehensive profiling of Epstein-Barr virus microRNAsin nasopharyngeal carcinoma. J Virol 83(5):2357–2367.

43. Yang H-J, et al. (2013) Comprehensive profiling of Epstein-Barr virus-encoded miRNAspecies associated with specific latency types in tumor cells. Virol J 10:314–339.

44. Stern P, et al. (2008) A system for Cre-regulated RNA interference in vivo. Proc NatlAcad Sci USA 105(37):13895–13900.

45. Ebert MS, Neilson JR, Sharp PA (2007) MicroRNA sponges: Competitive inhibitors ofsmall RNAs in mammalian cells. Nat Methods 4(9):721–726.

46. Umbach JL, Strelow LI, Wong SW, Cullen BR (2010) Analysis of Rhesus rhadinovirusmicroRNAs expressed in virus-induced tumors from infected Rhesus macaques. Vi-rology 405(2):592–599.

47. Takada K, et al. (1991) An Epstein-Barr virus-producer line Akata: Establishment ofthe cell line and analysis of viral DNA. Virus Genes 5(2):147–156.

48. Ruf IK, Rhyne PW, Yang C, Cleveland JL, Sample JT (2000) Epstein-Barr virus smallRNAs potentiate tumorigenicity of Burkitt lymphoma cells independently of an effecton apoptosis. J Virol 74(21):10223–10228.

49. Amoroso R, et al. (2011) Quantitative studies of Epstein-Barr virus-encoded micro-RNAs provide novel insights into their regulation. J Virol 85(2):996–1010.

50. Nachmani D, Stern-Ginossar N, Sarid R, Mandelboim O (2009) Diverse herpesvirusmicroRNAs target the stress-induced immune ligand MICB to escape recognition bynatural killer cells. Cell Host Microbe 5(4):376–385.

51. Du T, Zhou G, Roizman B (2013) Modulation of reactivation of latent herpes simplexvirus 1 in ganglionic organ cultures by p300/CBP and STAT3. Proc Natl Acad Sci USA110(28):E2621–E2628.

52. Stern-Ginossar N, et al. (2009) Analysis of human cytomegalovirus-encoded microRNAactivity during infection. J Virol 83(20):10684–10693.

53. Qi M, et al. (2013) Over-expression of human cytomegalovirus miR-US25-2-3p down-regulates eIF4A1 and inhibits HCMV replication. FEBS Lett 587(14):2266–2271.

54. Hertzog PJ, Hwang SY, Kola I (1994) Role of interferons in the regulation of cellproliferation, differentiation, and development. Mol Reprod Dev 39(2):226–232.

Cox et al. PNAS | February 10, 2015 | vol. 112 | no. 6 | 1861

MICRO

BIOLO

GY

See Retraction Published , 201 March 23 5