microRNAs are biomarkers of oncogenic humanpapillomavirus infectionsXiaohong Wanga, Hsu-Kun Wangb, Yang Lia,c, Markus Hafnerd,e, Nilam Sanjib Banerjeeb, Shuang Tanga,1,Daniel Briskind,e, Craig Meyersf, Louise T. Chowb,2, Xing Xiec, Thomas Tuschld,e, and Zhi-Ming Zhenga,2

aTumor Virus RNA Biology Section, Gene Regulation and Chromosome Biology Laboratory, Center for Cancer Research, National Cancer Institute, NationalInstitutes of Health, Bethesda, MD 20892; bDepartment of Biochemistry and Molecular Genetics, University of Alabama at Birmingham, Birmingham, AL35294-0005; cDepartment of Gynecologic Oncology, Women’s Hospital, Zhejiang University School of Medicine, Hangzhou 310006, China; dHoward HughesMedical Institute and eLaboratory of RNA Molecular Biology, Rockefeller University, New York, NY 10065-6399; and fDepartment of Microbiology andImmunology, Penn State University College of Medicine, Hershey, PA 17033

Contributed by Louise T. Chow, January 29, 2014 (sent for review September 29, 2013)

Cellular and viral microRNAs (miRNAs) are the transcriptional prod-ucts of RNA polymerase II and are regulated by transcriptionalfactors for their differential expression. The altered expressionof miRNAs in many cancer types has been explored as a markerfor possible diagnosis and therapy. We report in this study thatoncogenic human papillomaviruses (HPVs) induce aberrant expres-sion of many cellular miRNAs and that HPV18 infection producesno detectable viral miRNA. Thirteen abundant host miRNAs werespecifically regulated by HPV16 and HPV18 in organotypic raftcultures of foreskin and vaginal keratinocytes as determined bymiRNA array in combination with small RNA sequencing. The in-crease of miR-16, miR-25, miR-92a, and miR-378 and the decreaseof miR-22, miR-27a, miR-29a, and miR-100 could be attributed toviral oncoprotein E6, E7, or both, all of which are known to targetmany cellular transcription factors. The examination of 158 cervicalspecimens, including 38 normal, 52 cervical intraepithelial neo-plasia (CIN), and 68 cervical cancer (CC) tissues, for the expressionof these eight miRNAs showed a remarkable increase of miR-25,miR-92a, and miR-378 with lesion progression but no obviouschange of miR-22, miR-29a, and miR-100 among the HPV-infectedtissues. Further analyses indicate that an expression ratio ≥1.5 ofmiR-25/92a group over miR-22/29a group could serve as a cutoffvalue to distinguish normal cervix from CIN and from CIN to CC.

oncogenes E6 and E7 | noncoding RNAs | regulatory RNAs |virol oncogenesis | DNA tumor viruses

Cervical cancer (CC) is the second most common canceramong women worldwide and is caused by persistent infec-

tion with oncogenic human papillomaviruses (HPVs). HPV in-fection has also been identified as a definitive human carcinogenfor the penis, vulva, vagina, anus, and oropharynx (including thebase of the tongue and tonsils) (1, 2). To date, 15 mucosal HPVtypes are deemed as oncogenic or high-risk (HR) HPVs, includingHPV16, HPV18, HPV31, HPV33, HPV35, HPV39, HPV45,HPV51, HPV52, HPV56, HPV58, HPV59, HPV68, HPV73, andHPV82 (3). Among the HR HPVs, HPV16 and HPV18 have acombined worldwide contribution to ∼70% of invasive CC (4).The productive HPV life cycle is tightly linked to squamous

cell differentiation, and organotypic raft cultures have been de-veloped to grow these viruses in vitro (5–7). HPV infection isinitiated when viral particles gain entry to undifferentiated basalepithelial cells through an abrasion or a wound, where viral earlygenes are up-regulated during wound healing. The virus-aidedexpansion of the infected cell population then establishes theinfection. The amplification of the extrachromosomal viral DNAand the expression of viral capsid proteins occur sequentially inthe middle and upper spinous and superficial cells (7). Persistent,active infection by HR HPVs can lead to cervical intraepithelialneoplasia (CIN) grade 1, 2, or 3 on the basis of increasing ab-normal depths of the proliferative cell compartment in the cer-vical epithelium (8). Histologically, the abnormal proliferativecells are restricted to the lower one-third of the epithelium in

CIN1. CIN2 and CIN3 are distinguished by the expansion of theneoplasia to the lower two-thirds (CIN2) or more (CIN3) of theepithelium. CIN3 may involve the full thickness of the epithe-lium and is sometimes referred to as cervical carcinoma in situ.CIN2 and CIN3 are developed in 10–20% of women with CIN1,and a fraction of CIN3 may progress to CC when the neoplasiainvades into the stroma underneath. Two viral oncoproteins, E6and E7, of the HR HPVs destabilize, respectively, two majorcellular tumor suppressor proteins, p53 and retinoblastomaprotein (pRB). Both viral proteins function to support the pro-ductive phase of the infection in the postmitotic differentiatedspinous cells. However, repeated elevated expression of HRHPV E6 and E7 in the undifferentiated basal epithelial cells orstem cells disrupts cell cycle regulation, inhibits cell differentia-tion, induces chromosome damage, and prevents apoptosis,resulting in cell immortalization and transformation, the basis ofHPV carcinogenesis (9, 10). Indeed, the expression of the HRHPV E6 and E7 is consistently elevated in CC.MicroRNAs (miRNAs), noncoding regulatory RNAs with a

length of ∼21 nt, are derived from RNA polymerase II tran-scripts of coding or noncoding genes, and their expression issubjected to transcriptional and posttranscriptional regulation.

Significance

Persistent infections with high-risk human papillomaviruses(HPVs) lead to development of cervical, penile, anal, and oro-pharyngeal cancers. The ability to diagnose HPV infectionshas been dependent on the detection of viral DNA, on virus-associated cytological and histological abnormalities, and ona few virus-induced host proteins. In this study, we identifieda subset of host microRNAs regulated specifically by HPV16 orHPV18 infection in in vitro model systems. The elevated ex-pression of miR-16, miR-25, miR-92a, and miR-378 and the de-creased expression of miR-22, miR-27a, miR-29a, and miR-100were attributed to viral oncoprotein E6 or E7. An expressionratio ≥1.5 of miR-25/92a group over miR-22/29a group wasfound to be informative in distinguishing normal cervix fromcervical intraepithelial neoplasia and cervical cancers.

Author contributions: X.W., C.M., L.T.C., X.X., T.T., and Z.-M.Z. designed research; X.W.,H.-K.W., Y.L., M.H., N.S.B., S.T., and D.B. performed research; H.-K.W., N.S.B., C.M., X.X.,and T.T. contributed new reagents/analytic tools; X.W., H.-K.W., Y.L., M.H., N.S.B., S.T.,D.B., C.M., L.T.C., X.X., T.T., and Z.-M.Z. analyzed data; and X.W., L.T.C., and Z.-M.Z. wrotethe paper.

Conflict of interest statement: T.T. is a cofounder of and scientific advisor to AlnylamPharmaceuticals and a scientific advisor to Regulus Therapeutics.1Present address: Division of Viral Products, Office of Vaccines Research and Review,Center for Biologics Evaluation and Research, US Food and Drug Administration,Bethesda, MD 20892.

2To whom correspondence may be addressed. E-mail: ltchow@uab.edu or zhengt@mail.nih.gov.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1401430111/-/DCSupplemental.

4262–4267 | PNAS | March 18, 2014 | vol. 111 | no. 11 www.pnas.org/cgi/doi/10.1073/pnas.1401430111

Dow

nloa

ded

by g

uest

on

Oct

ober

3, 2

020

Argonaute proteins, the core components of the RNA-inducedsilencing complex, bind the mature miRNAs and guide them toimperfect complementary sequences located primarily in the3′ UTR of target mRNAs, leading to translational repression ortarget degradation (11, 12). The human genome encodes ∼550miRNA genes to express about 1,000 miRNAs. miRNAs aredifferentially expressed in many human cell types (13) and targetabout 60% of genes (14). Aberrant expression of miRNAs hasbeen implicated in numerous disease states and has been exploredas a biomarker for possible diagnosis or prognosis of human dis-eases (15–17). Previously, our group and others found that HPVinfection regulates the expression of keratinocyte miRNAs (18–20) and that aberrant miRNA expression in CC and its derivedcell lines is important for cancer cell proliferation (21–24). How-ever, the different approaches used led to varied observations(25, 26). To date, there is no systematic study to identify hostmiRNAs as specific biomarkers for diagnosis of HPV infectionor progression. In this report, by using miRNA array analysis andsmall RNA sequencing (miRNA-Seq), we conducted a compre-hensive examination of miRNA profiles in human foreskin ker-atinocyte (HFK)- or human vaginal keratinocyte (HVK)-derivedraft cultures with or without productive HPV16 or HPV18 infec-tion. We identified a group of host miRNAs that are regulated byHPV18 E6 or E7. Altered expression of a subset of these miRNAswas also observed in HR HPV-infected CIN and CC tissues andmay serve as a biomarker for HPV infection or progression.

ResultsHPV16 and HPV18 Infection Regulates the Expression of a Subset ofHost miRNAs. To investigate HPV infection-induced genome-widechanges in miRNA expression, we first compared miRNA ex-pression profiles in HFK-derived raft cultures with or withoutHPV16 infection by miRNA array analysis (Fig. S1). As shown inFig. 1A (Left, panel 1), HPV16 infection led to decreased ex-pression of 20 host miRNAs (green), including miR-34a (18),and increased expression of 22 host miRNAs (red) (P < 0.05),including miR-16 (25). Northern blot analysis of miR-181a,miR-455-3p, miR-203, miR-375, and miR-16 confirmed the alteredexpression (Fig. 1B, Upper). To minimize potential experimentalvariances due to cell types, virus types, or detection methods, weexpanded the investigation to HPV18 infection in two types ofcells by using two analytical approaches. After initial verificationby Northern blotting of altered expression of miR-100 and miR-378 in HFK-derived raft cultures with HPV18 infection (Fig. 2B,Lower), we further compared miRNA expression profiles inHFK- and HVK-derived raft cultures with or without HPV18infection by using miRNA array analysis and miRNA-Seq (Fig.1A, Right, panels 1–3). Although each cell type with or withoutHPV infection displayed a slightly different expression profile ofhost miRNAs (Fig. 1A and Tables S1–S4), 13 abundant hostmiRNAs with altered expression were commonly detected byboth methods from raft cultures of HVKs or HFKs after HPV16or HPV18 infection (Fig. 1C). Eight miRNAs had increasedexpression, and five miRNAs had decreased expression (Table1). We confirmed by Northern blotting the altered expression ofmiR-16, miR-25, miR-22, and miR-29 in HVK-derived raft tis-sues infected with either HPV16 or HPV18 (Fig. S2). It is worthnoting that we did not find any HPV18-originated viral miRNAsin HVK- or HFK-derived raft cultures by miRNA-Seq.

HPV18 Regulates Host miRNA Expression over the Course of theProductive Infection. To validate the observation that the alteredexpression of host miRNAs is specifically attributable to HPVinfection, HFK-derived raft cultures with HPV18 infection onday 8, day 10, day 12, and day 16 were analyzed by miRNA-Seq(Fig. 2A). Viral DNA amplification peaked on days 12–14,whereas viral oncoprotein expression waned (7). Again, we didnot detect any HPV18-originated viral miRNAs from the raft

tissues at any of the four time points. However, relative to unin-fected HFK raft cultures, the altered expression of host miRNAswas obvious and could be classified into four groups (Table S5),with additional miRNAs showing fluctuations in expression overthe time course (Dataset S1). The miRNAs in group 1 and group2 exhibited increased and decreased expression over each of thetime points. The miRNAs in group 3 displayed elevated ex-pression in early HPV18 infection but reduced expression at thelater time points, and the reverse was true for the miRNAs ingroup 4, which had decreased expression at the earlier timepoints but increased expression at the later time of HPV18 in-fection. Nine of the 13 miRNAs with altered expression in Table 1were in group 1 and group 2 (Fig. 2A and Table S5) and areselectively graphed in Fig. 2B to show their persistent increase ordecrease over the course of HPV18 infection relative to theuninfected raft cultures. We also observed an initial increase, but

Fig. 1. HPV16 and HPV18 infections regulate the expression of a subset ofcellular miRNAs. (A, Left) Panels 1 and 2 illustrate increased (red) or de-creased (green) expression of individual miRNAs in HFK or HVK raft cultureswithout (HFK or HVK) or with HPV16 (HFK16) or HPV18 (HVK18) infection bymiRNA array analyses. (A, Right) Panels 3 and 4 illustrate increased (yellow)or decreased (blue) expression of individual miRNAs in HVK or HFK raftcultures without or with HPV18 (HVK18 or HFK18) infection by miRNA-Seqanalyses, with the heat map drawn from the top 95% of expressed miRNAsin each sample. (B) Verification of the altered expression of miRNAs inHPV16- and HPV18-infected HFK raft cultures by Northern blot analysis. (C)Venn diagram summarizes a subset of host miRNAs with altered expressionin the raft cultures among all four platforms (red arrows in A).

Wang et al. PNAS | March 18, 2014 | vol. 111 | no. 11 | 4263

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Oct

ober

3, 2

020

later decrease, of miR-143, a p53-sensitive target (27, 28) andRas responder (29). In contrast, miR-203, a skin differentiationmarker (30), displayed an initial decrease but a stepwise increasein HFK raft tissues over the time course of HPV18 infection(Fig. 2B), in agreement with the squamous differentiation ach-ieved in these raft cultures (Fig. S3A). Northern blotting con-firmed the increased expression of miR-25 and miR-92a and thedecreased expression of miR-27a and miR-29a in HFK-derivedraft tissues on day 10 after HPV16 or HPV18 infection (Fig. 2C).Together, these data suggest that the miRNAs in the first twogroups respond consistently to HPV18 infection; however, theresponses of the host miRNAs in group 3 and group 4 toHPV18 infection appear to be regulated by HPV activities (7)

and keratinocyte differentiation, as reported (31, 32). Basedon these results and on the data shown in Fig. 1C and Table 1, wesubsequently chose eight host miRNAs (miR-16, miR-22, miR-25,miR-27a, miR-29a, miR-92a, miR-100, and miR-378) for furtherinvestigation. These miRNAs have high signal intensity or reads inall testing platforms (Fig. S4) and are transcribed alone or in alarger cistron together with other miRNAs (cluster).

Host miRNAs and Oncogenic HPV E6 and E7. To investigate whetherthe increase or decrease of the eight miRNAs was regulated byviral oncoprotein E6 or E7, raft cultures derived from HFKstransduced with a retrovirus expressing HPV18 E6, HPV18 E7,HPV18 E6E7, or empty control retrovirus were examined by re-verse transcription real-time quantitative PCR (RT-qPCR). U6RNA was used as an internal loading control. miR-34a was usedas a positive control for the E6 activity, which reduces miR-34aexpression because this miRNA is regulated partially by p53 (18).As expected, E6 decreased the expression of miR-34a, whereasE7 did not (18). In contrast, HPV18 E7 significantly increased theexpression of miR-25, whereas HPV18 E6 had no effect (Fig. 3).These observations validated the specificity of E6 and E7 activities.In addition, HPV18 E7 had a stronger positive effect than HPV18E6 on the expression of miR-16 and miR-378, whereas both viraloncoproteins functioned similarly in inducing a moderate increaseof miR-92a expression (Fig. 3). Interestingly, miR-22, miR-27a,miR-29a, and miR-100 were all moderately down-regulated by ei-ther E6 or E7, with miR-22 and miR-27a being slightly more sus-ceptible to E7 than to E6 (Fig. 3). These results demonstrate thatthis set of host miRNAs respond toHPV infection and their alteredexpression could be attributed to viral oncoprotein E6 or E7.

Expression of miR-16, miR-22, miR-25, miR-27a, miR-29a, miR-92a,miR-100, and miR-378 in HR HPV-Infected Cervical Lesions. To eval-uate whether the altered expression of miRNAs in oncogenicHPV-infected raft tissues could be observed in CINs and CCinfected with HR HPVs, we examined by RT-qPCR the ex-pression of miR-16, miR-25, miR-92a, miR-378, miR-22, miR-27a, miR-29a, and miR-100 in 38 normal cervical tissues withoutHPV infection and in 13 CIN1+2, 39 CIN3, and 68 CC tissueswith HR HPV infection. As shown in Fig. 4A, statistically significantelevation of miR-25, miR-92a, and miR-378 was observed amongHPV-infected tissue groups, consistent with the results obtainedfromHPV-infected raft tissues. Expression of miR-16 also displayeda trend of increase in CIN3 and CC tissues, but this increase mighthave been distorted by the extremely high levels of miR-16 in onesample in the CIN3 group and three samples in the CC group (Fig.4A). An increased level of miR-378 was observed in CIN3 (P =0.021) and CC (P = 0.035) over the normal cervical tissues, but notin CC over CIN3 (P = 0.184). In contrast, we found no obviouschange of miR-22, miR-27a, miR-29a, and miR-100 in CIN1+2 andCIN3 or of miR-22, miR-29a, and miR-100 in CC over the normalcervical tissues. However, a higher level of miR-27a was observed inCC compared with CIN3, CIN1+2, or normal cervical tissues. Inthis and previous studies (22), miR-29a expression was decreasedfrom normal tissue to CIN and CC, and this trend was verified byNorthern blotting using RNAs extracted from a small group ofrandomly selected samples (Fig. S5). However, this decrease wasnot statistically significant (P = 0.416) (Fig. 4A).Based on the mean values and statistical analysis, miR-25 and

miR-92a were subsequently combined as a group; miR-22 andmiR-29a were combined as another group; and the individual ex-pression levels were averaged in each group for normal, CIN1+2,CIN3, and CC tissues. Comparison between two groups of miRNAlevels showed a stepwise increase of the miR-25/92a from nor-mal (1.6) to CIN1+2 (2.2), CIN3 (3.9), and CC (8.2) (Fig. 4B),whereas the small fluctuation of miR-22/29a did not exhibit atrend. Given that individual miRNA levels varied from one sampleto another, we further converted the miRNA expression level

Fig. 2. HPV18 regulates the expression of host miRNAs over the course ofinfection. (A) Increased (yellow) or decreased (blue) expression of individualmiRNAs in day 8 (D8), D10, D12, and D16 HFK raft cultures without (HFK) orwith HPV18 (HFK18) infection and examined by miRNA-Seq. The heat map wasdrawn from the top 90% of expressed miRNAs in each sample, with red arrowsfor the miRNAs with altered expression in Table 1 over the course of infection.(B) Expression of selectedmiRNAs (Table S5) in HFK raft cultures over the courseof HPV18 infection. (C) Verification of increasedmiR-25 andmiR-92a expressionand decreased miR-27a and miR-29a expression in the 10-d raft tissues derivedfrom HFKs with or without HPV16 or HPV18 infection by Northern blot analysis.Three separate tissue samples labeled as 1, 2, and 3 in each raft tissue groupwere used for the assay. U6 RNA was used as a loading control. Bar graphsbelow the blot show the relative level (mean ± SD) of each miRNA, after beingnormalized to U6 RNA, from three separate raft tissues in each raft group.

4264 | www.pnas.org/cgi/doi/10.1073/pnas.1401430111 Wang et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

3, 2

020

from individual tissues to the expression ratio between miR-25/92a and miR-22/29a groups for a better view of the dataset. Athreshold ratio was set at 1.5 to evaluate the possibility of usingthese four miRNAs for diagnosis of CINs and CC. As summa-rized in Table 2, only 5 of 38 (13.2%) normal cervical tissues hada ratio equal to or higher than 1.5, but 5 of 13 CIN1+2 (38.5%),19 of 39 CIN3 (48.7%), and 60 of 68 (88.2%) CC tissues hada ratio equal to or above 1.5 (P < 0.0001). This increase wassignificant even for CIN1+2 tissue samples compared with nor-mal cervical samples (P = 0.047). Collectively, our data stronglysuggest that the expression ratio between these two miRNAgroups could be useful for the diagnosis of cervical lesions andprogression to CC due to oncogenic HPV infections.

DiscussionIn this report, host miRNAs specifically regulated by oncogenicHPV16 and HPV18 infection in HFK- and HVK-derived raftcultures were comprehensively investigated by both miRNA ar-ray analysis and miRNA-Seq. We identified 13 host miRNAs

responsive to oncogenic HPV regulation, including nine miRNAsfrom six miRNA clusters (miR-15/16/195/457, miR-106b/93/25,miR-17-5p/18a/19a/20a/19b/92a, miR-224/452, miR-23a/24/27a, andmiR-100/let-7a) and four noncluster miRNAs (miR-22, miR-29a,miR-210, and miR-378). The previously reported miRNAs (miR-21, miR-34a, miR-146a, miR-143/145, miR-203, and miR-218)that are altered in some CCs (21, 22, 33) or in HPV infection (18,20, 34) are not among the 13 miRNAs identified in this study,nor are miRNAs altered in HPV31-infected HFK rafts (35),because none of them were consistently altered in all four of ourassay platforms. We note that HPV18 infection did modulate othercellular miRNAs in HFK rafts (Dataset S1), as previously found inHPV31 infection (35). From the panel of 13 miRNAs, eight wereselected to represent each cluster and noncluster miRNA for fur-ther investigation. The expression of miR-224 displayed fluctuationover the time course of HPV18 infection; thus, miR-224 was ex-cluded from further study. miR-210 was also excluded because itsexpression is induced by hypoxia (36) and both HR and low-riskHPV E7 proteins enhance HIF-1α stability (37, 38).We note that miR-25, miR-27a, miR-92a, and miR-378 are on-

cogenic miRNAs and miR-16, miR-22, miR-29, and miR-100 aretumor-suppressive miRNAs (39–42) and that they are modulatedby p53, E2F, and c-Myc (25, 43–45). Although miR-16, miR-25,and miR-92a from three different miRNA clusters and nonclustermiR-378 all had increased expression, the expression of the miR-106b/93/25 cluster produced fromMCM7 transcript (46) appears tobe the most sensitive to oncogenic HPV infection, with increasedexpression in each of the three miRNAs from this cluster in allassays (Table 1). In contrast, miR-92a and miR-16 were the majormiRNAs with significantly increased expression within the respec-tive clusters of miR-17-5p/18a/19a/20a/19b/92a and miR-15/16.We found that both viral E6 and E7 could enhance miR-92a ex-

pression but that only viral E7 was responsible for the up-regulationof miR-25 (Fig. 3). These observations are consistent with ourunderstanding that the expression of both the miR-106b/93/25and miR-17-5p/18a/19a/20a/19b/92a clusters is suppressed byp53 through an indirect mechanism via inhibition of E2F1 ex-pression (44, 47, 48). Their up-regulation can then be attributedto the abilities of the HR HPV E6 to destabilize p53 and of theviral E7 to target the pRB/p130 protein family, resulting in therelease of E2F from suppressive transcriptional complexes.Also, HR HPV E6 and E7 increase MCM7 expression throughE2F-dependent and E2F-independent pathways (49), and c-Myctransactivated by E2F1 promotes transcription and expression ofthe miR-17-92a cluster (47), as well as miR-378 (50). Free E2F1also transactivates the expression of the miR-15/16 cluster (51).

Table 1. Thirteen host miRNAs specifically regulated by HPV infections

Name Expression miR-array (HFK16/HFK) miR-Seq (HFK18/HFK) miR-array (HVK18/HVK) miR-Seq (HVK18/HVK)

miR-16 Up +3.1 +3.2 +1.8 +1.8miR-25 Up +4.1 +1.6 +2.7 +2.0miR-92a Up +1.3 +1.9 +2.1 +1.6miR-93 Up +3.0 +2.3 +3.6 +2.4miR-106b Up +2.0 +1.9 +1.6 +2.4miR-210 Up +2.1 +1.4 +2.3 +2.5miR-224 Up +2.7 +4.1 +2.4 +2.4miR-378 Up +4.0 +1.9 +4.2 +3.7miR-22 Down −10.3 −2.1 −3.8 −2.2miR-24 Down −2.0 −1.5 −1.9 −1.4miR-27a Down −3.0 −2.0 −3.7 −2.6miR-29a Down −8.3 −2.5 −3.6 −1.6miR-100 Down −1.9 −2.8 −2.7 −1.3

Total RNAs were isolated from day 10 raft cultures of HFKs or HVKs. Thirteen miRNAs were identified by four test platforms (Fig. 1C)that are altered in four HPV-infected raft cultures relative to control raft cultures. The numbers indicate the fold increase (+) ordecrease (−) of the signal intensity [miRNA (miR)-array assay] or relative miRNA reads (miR-Seq) of the raft tissue with HPV infectiondivided by that of the raft tissue without HPV infection. Details are provided in Tables S1 and S2.

Fig. 3. HPV18 E6 and E7 oncoproteins regulate the expression of hostmiRNAs. Expressions of eight host miRNAs selected from Table 1 were ex-amined by RT-qPCR in 11-d raft cultures of HFKs transduced with HPV18 E6,E7, or E6E7 retrovirus or an empty control retrovirus vector. miR-34a wasused as a positive control for HPV18 E6 activity (18), whereas U6 RNA wasused as an internal loading control. The bar graph shows the relative foldincrease (above 0) or decrease (below 0) (mean ± SD) of each miRNA in theraft tissues expressing E6, E7, or E6E7 over the empty vector after beingnormalized to U6 RNA from two independent experiments.

Wang et al. PNAS | March 18, 2014 | vol. 111 | no. 11 | 4265

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Oct

ober

3, 2

020

Both HR HPV E6 and E7 interact with c-Myc and augmentc-Myc transactivation activities (52, 53). However, c-Myc sup-presses the expression of many miRNAs (44) by binding tomiRNA promoter, including those of miR-15/16, miR-23a/24/27a, miR-22, miR-29a, and miR-34a, for example (54–57). More-over, c-Myc–induced miR-17-5p and miR-20a from the miR-17-92a cluster can inhibit E2F1 translation (43) and affect the ex-pression of E2F1-dependent miRNAs (45, 58). Our data indicatethat the expression of oncogenic miRNAs is superimposed overthat of the tumor-suppressive miRNAs as a net gain by HR HPVinfection. It is possible that the posttranscriptional regulationcould also play a role in this net gain, because p53 interactswith Drosha/p68 complex to facilitate Drosha-mediated primary-miRNA processing of certain miRNAs (27) and some miRNAsderived from intron regions are coupled to RNA splicing (59, 60).

We observed increased expression of miR-16, miR-25, miR-92a, and miR-378 in CIN and CC tissues with HR HPV in-fection, consistent with what we observed in raft tissues withHPV16 or HPV18 infection. In particular, the striking increaseof miR-25 and miR-92a correlated with the progression of thecervical lesions, making them credible biomarkers of CINs andCC. Furthermore, we found that the expression ratio of miR-25/92a vs. miR-22/29a increased even more significantly in CIN1+2tissues over the normal cervical tissues as an early indicator ofHPV infection of the cervix.We note that the expression profiles of miR-22, miR-27a,

miR-29a, and miR-100 in clinical samples are slightly differentfrom those in HR HPV-infected and HPV18 oncogene-expressingraft cultures. Several reasons may account for the variation. First,the HPV-infected raft cultures have pathological changes similarto CIN1 only and maintain cell differentiation (Fig. S3A). Celldifferentiation could affect miRNA expression and, hence, theregulation of miRNAs in productive HPV infections (20, 25).Indeed, the expression of miR-21, and presumably miR-27a andmiR-205, in monolayer cell cultures was altered by cell differ-entiation in the presence of calcium (Fig. S3 B–D). Second, raftcultures comprise a pure population of keratinocytes, whereaspatient tissues contain many additional cell types, including fibro-blasts, fat cells, endothelial cells, and infiltrating immune cells.Third, tissue sampling is another contributing factor, particularlyfor cervical precancer lesions, which are surrounded by normaltissues. The biopsies might be a mixture of normal and diseasedtissues, thereby leading to a slightly different miRNA profilefrom sample to sample and from that of the raft cultures. Ananalysis of additional patient samples could help to clarify theconclusions. Also, because the sample sizes of CIN1+2 in ourstudy are relatively small, future studies of additional and serialpatient specimens over a period of time would be important todetermine if there is an miRNA ratio conversion in correlationwith CIN progression or regression.

Materials and MethodsPrimary keratinocyte cultures and their derived raft tissues with or withoutHPV16 or HPV18 infection were prepared according to our standard pro-tocols. Total RNA purified from raft tissues was used for miRNA microarrayand miRNA-Seq analyses. The resultant miRNA profiles from each tissue werevalidated by Northern blotting and TaqMan miRNA RT-qPCR assays and werefurther verified in the institutional review board-approved human cervicaltissues with or without HR HPV infections. A detailed discussion of thematerials and methods used is provided in SI Materials and Methods.

ACKNOWLEDGMENTS. We thank Jeffrey Strathern for his support andcritical reading of our manuscript. This study was supported by the Intra-mural Research Programs of the National Cancer Institute, Center for CancerResearch and National Institutes of Health Grants R01 CA83679 (to L.T.C.)and R01 AI57988 (to C.M.), Natural Science Foundation of China Grant NSFC81172475 (to X.X.), and Natural Science Foundation of Zhejiang Province ofChina Grant LQ13H160003 (to Y.L.).

Fig. 4. Expression of selected miRNAs in normal cervix and cervical lesions.(A) Expression of miR-16, miR-25, miR-92a, miR-378, miR-22, miR-27a, miR-29a, and miR-100 in 38 normal cervical tissues and in 13 CIN1+2, 39 CIN3, and68 CC tissues were examined by RT-qPCR. U6 RNA was used as an internalcontrol. Each dot in the dot plot represents the detection level of individualmiRNA in a sample from each group after being normalized to U6 RNA.Mean indicates an average level of the miRNA in all samples examined ineach group. (B) Mean expression levels of two miRNA groups (miR-25/92aand miR-22/29a) in normal cervical, CIN1+2, CIN3, and CC tissues. P < 0.0001for miR-25/92a progressive expression among the four tissue groups (F test).

Table 2. Expression ratio of two miRNA groups can be used fordiagnosis of CIN and CC

miR-25/92a vs. miR-22/29a

<1.5 ≥1.5

Pathology Total samples Samples % Samples %

Normal 38 33 86.8 5 13.2*CIN1+2 13 8 61.5 5 38.5CIN3 39 20 51.3 19 48.7CC 68 8 11.8 60 88.2

*P < 0.0001 (F test) among four groups with a ratio equal to or greaterthan 1.5.

4266 | www.pnas.org/cgi/doi/10.1073/pnas.1401430111 Wang et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

3, 2

020

1. Moscicki AB, et al. (2012) Updating the natural history of human papillomavirus andanogenital cancers. Vaccine 30(Suppl 5):F24–F33.

2. Chaturvedi AK, et al. (2011) Human papillomavirus and rising oropharyngeal cancerincidence in the United States. J Clin Oncol 29(32):4294–4301.

3. Muñoz N, et al.; International Agency for Research on Cancer Multicenter CervicalCancer Study Group (2003) Epidemiologic classification of human papillomavirustypes associated with cervical cancer. N Engl J Med 348(6):518–527.

4. de Sanjose S, et al.; Retrospective International Survey and HPV Time Trends StudyGroup (2010) Human papillomavirus genotype attribution in invasive cervical cancer:A retrospective cross-sectional worldwide study. Lancet Oncol 11(11):1048–1056.

5. Meyers C, Frattini MG, Hudson JB, Laimins LA (1992) Biosynthesis of human papillo-mavirus from a continuous cell line upon epithelial differentiation. Science 257(5072):971–973.

6. Dollard SC, et al. (1992) Production of human papillomavirus and modulation of theinfectious program in epithelial raft cultures. OFF. Genes Dev 6(7):1131–1142.

7. Wang HK, Duffy AA, Broker TR, Chow LT (2009) Robust production and passaging ofinfectious HPV in squamous epithelium of primary human keratinocytes. Genes Dev23(2):181–194.

8. Massad LS, et al.; 2012 ASCCP Consensus Guidelines Conference (2013) 2012 updatedconsensus guidelines for the management of abnormal cervical cancer screening testsand cancer precursors. Obstet Gynecol 121(4):829–846.

9. Scheffner M, Werness BA, Huibregtse JM, Levine AJ, Howley PM (1990) The E6 on-coprotein encoded by human papillomavirus types 16 and 18 promotes the degra-dation of p53. Cell 63(6):1129–1136.

10. Roman A, Munger K (2013) The papillomavirus E7 proteins. Virology 445(1-2):138–168.

11. Bartel DP (2009) MicroRNAs: Target recognition and regulatory functions. Cell 136(2):215–233.

12. Helwak A, Kudla G, Dudnakova T, Tollervey D (2013) Mapping the human miRNAinteractome by CLASH reveals frequent noncanonical binding. Cell 153(3):654–665.

13. Landgraf P, et al. (2007) A mammalian microRNA expression atlas based on small RNAlibrary sequencing. Cell 129(7):1401–1414.

14. Friedman RC, Farh KK, Burge CB, Bartel DP (2009) Most mammalian mRNAs areconserved targets of microRNAs. Genome Res 19(1):92–105.

15. Lu J, et al. (2005) MicroRNA expression profiles classify human cancers. Nature435(7043):834–838.

16. Mitchell PS, et al. (2008) Circulating microRNAs as stable blood-based markers forcancer detection. Proc Natl Acad Sci USA 105(30):10513–10518.

17. Calin GA, et al. (2005) A MicroRNA signature associated with prognosis and pro-gression in chronic lymphocytic leukemia. N Engl J Med 353(17):1793–1801.

18. Wang X, et al. (2009) Oncogenic HPV infection interrupts the expression of tumor-suppressive miR-34a through viral oncoprotein E6. RNA 15(4):637–647.

19. Wang X, Meyers C, Guo M, Zheng ZM (2011) Upregulation of p18Ink4c expression byoncogenic HPV E6 via p53-miR-34a pathway. Int J Cancer 129(6):1362–1372.

20. Melar-New M, Laimins LA (2010) Human papillomaviruses modulate expression ofmicroRNA 203 upon epithelial differentiation to control levels of p63 proteins. J Virol84(10):5212–5221.

21. Wang X, et al. (2008) Aberrant expression of oncogenic and tumor-suppressive mi-croRNAs in cervical cancer is required for cancer cell growth. PLoS ONE 3(7):e2557.

22. Li Y, et al. (2011) Progressive miRNA expression profiles in cervical carcinogenesis andidentification of HPV-related target genes for miR-29. J Pathol 224(4):484–495.

23. Wang F, et al. (2011) miR-375 is down-regulated in squamous cervical cancer andinhibits cell migration and invasion via targeting transcription factor SP1. Am J Pathol179(5):2580–2588.

24. Xu J, et al. (2013) Suppressed miR-424 expression via upregulation of target geneChk1 contributes to the progression of cervical cancer. Oncogene 32(8):976–987.

25. Zheng ZM, Wang X (2011) Regulation of cellular miRNA expression by human pap-illomaviruses. Biochim Biophys Acta 1809(11-12):668–677.

26. Kaczkowski B, Morevati M, Rossing M, Cilius F, Norrild B (2012) A Decade of GlobalmRNA and miRNA Profiling of HPV-Positive Cell Lines and Clinical Specimens. OpenVirol J 6:216–231.

27. Suzuki HI, et al. (2009) Modulation of microRNA processing by p53. Nature 460(7254):529–533.

28. Shi M, et al. (2012) Glucocorticoid regulation of a novel HPV-E6-p53-miR-145 pathwaymodulates invasion and therapy resistance of cervical cancer cells. J Pathol 228(2):148–157.

29. Kent OA, et al. (2010) Repression of the miR-143/145 cluster by oncogenic Ras initiatesa tumor-promoting feed-forward pathway. Genes Dev 24(24):2754–2759.

30. Jackson SJ, et al. (2013) Rapid and widespread suppression of self-renewal by mi-croRNA-203 during epidermal differentiation. Development 140(9):1882–1891.

31. Nissan X, et al. (2011) miR-203 modulates epithelial differentiation of human em-bryonic stem cells towards epidermal stratification. Dev Biol 356(2):506–515.

32. Lal A, et al. (2009) miR-24 Inhibits cell proliferation by targeting E2F2, MYC, and othercell-cycle genes via binding to “seedless” 3′UTR microRNA recognition elements. MolCell 35(5):610–625.

33. Martinez I, et al. (2008) Human papillomavirus type 16 reduces the expression ofmicroRNA-218 in cervical carcinoma cells. Oncogene 27(18):2575–2582.

34. McKenna DJ, McDade SS, Patel D, McCance DJ (2010) MicroRNA 203 expression inkeratinocytes is dependent on regulation of p53 levels by E6. J Virol 84(20):10644–10652.

35. Gunasekharan V, Laimins LA (2013) Human papillomaviruses modulate microRNA 145expression to directly control genome amplification. J Virol 87(10):6037–6043.

36. Biswas S, et al. (2010) Hypoxia inducible microRNA 210 attenuates keratinocyteproliferation and impairs closure in a murine model of ischemic wounds. Proc NatlAcad Sci USA 107(15):6976–6981.

37. Nakamura M, et al. (2009) Hypoxia-specific stabilization of HIF-1alpha by humanpapillomaviruses. Virology 387(2):442–448.

38. Bodily JM, Mehta KP, Laimins LA (2011) Human papillomavirus E7 enhances hypoxia-inducible factor 1-mediated transcription by inhibiting binding of histone deacety-lases. Cancer Res 71(3):1187–1195.

39. Xiong J, Du Q, Liang Z (2010) Tumor-suppressive microRNA-22 inhibits the tran-scription of E-box-containing c-Myc target genes by silencing c-Myc binding protein.Oncogene 29(35):4980–4988.

40. Zheng YS, et al. (2012) MiR-100 regulates cell differentiation and survival by targetingRBSP3, a phosphatase-like tumor suppressor in acute myeloid leukemia. Oncogene31(1):80–92.

41. Lee DY, Deng Z, Wang CH, Yang BB (2007) MicroRNA-378 promotes cell survival,tumor growth, and angiogenesis by targeting SuFu and Fus-1 expression. Proc NatlAcad Sci USA 104(51):20350–20355.

42. Farazi TA, Spitzer JI, Morozov P, Tuschl T (2011) miRNAs in human cancer. J Pathol223(2):102–115.

43. O’Donnell KA, Wentzel EA, Zeller KI, Dang CV, Mendell JT (2005) c-Myc-regulatedmicroRNAs modulate E2F1 expression. Nature 435(7043):839–843.

44. Hua Y, et al. (2013) miRConnect 2.0: Identification of oncogenic, antagonistic miRNAfamilies in three human cancers. BMC Genomics 14:179.

45. Brosh R, et al. (2008) p53-Repressed miRNAs are involved with E2F in a feed-forwardloop promoting proliferation. Mol Syst Biol 4:229.

46. Poliseno L, et al. (2010) Identification of the miR-106b∼25 microRNA cluster asa proto-oncogenic PTEN-targeting intron that cooperates with its host gene MCM7 intransformation. Sci Signal 3(117):ra29.

47. Woods K, Thomson JM, Hammond SM (2007) Direct regulation of an oncogenic mi-cro-RNA cluster by E2F transcription factors. J Biol Chem 282(4):2130–2134.

48. Sylvestre Y, et al. (2007) An E2F/miR-20a autoregulatory feedback loop. J Biol Chem282(4):2135–2143.

49. Shai A, Brake T, Somoza C, Lambert PF (2007) The human papillomavirus E6 oncogenedysregulates the cell cycle and contributes to cervical carcinogenesis through twoindependent activities. Cancer Res 67(4):1626–1635.

50. Feng M, et al. (2011) Myc/miR-378/TOB2/cyclin D1 functional module regulates on-cogenic transformation. Oncogene 30(19):2242–2251.

51. Ofir M, Hacohen D, Ginsberg D (2011) MiR-15 and miR-16 are direct transcriptionaltargets of E2F1 that limit E2F-induced proliferation by targeting cyclin E. Mol CancerRes 9(4):440–447.

52. Veldman T, Liu X, Yuan H, Schlegel R (2003) Human papillomavirus E6 and Mycproteins associate in vivo and bind to and cooperatively activate the telomerase re-verse transcriptase promoter. Proc Natl Acad Sci USA 100(14):8211–8216.

53. Wang YW, Chang HS, Lin CH, Yu WC (2007) HPV-18 E7 conjugates to c-Myc andmediates its transcriptional activity. Int J Biochem Cell Biol 39(2):402–412.

54. Chang TC, et al. (2008) Widespread microRNA repression by Myc contributes to tu-morigenesis. Nat Genet 40(1):43–50.

55. Li X, et al. (2013) c-MYC-regulated miR-23a/24-2/27a cluster promotes mammarycarcinoma cell invasion and hepatic metastasis by targeting Sprouty2. J Biol Chem288(25):18121–18133.

56. Zhang X, et al. (2012) Coordinated silencing of MYC-mediated miR-29 by HDAC3 andEZH2 as a therapeutic target of histone modification in aggressive B-Cell lymphomas.Cancer Cell 22(4):506–523.

57. Zhang X, et al. (2012) Myc represses miR-15a/miR-16-1 expression through re-cruitment of HDAC3 in mantle cell and other non-Hodgkin B-cell lymphomas. On-cogene 31(24):3002–3008.

58. Bueno MJ, et al. (2010) Multiple E2F-induced microRNAs prevent replicative stress inresponse to mitogenic signaling. Mol Cell Biol 30(12):2983–2995.

59. Westholm JO, Lai EC (2011) Mirtrons: MicroRNA biogenesis via splicing. Biochimie93(11):1897–1904.

60. Melamed Z, et al. (2013) Alternative splicing regulates biogenesis of miRNAs locatedacross exon-intron junctions. Mol Cell 50(6):869–881.

Wang et al. PNAS | March 18, 2014 | vol. 111 | no. 11 | 4267

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Oct

ober

3, 2

020