muralidhar et al-2007-the journal of pathology

7/23/2019 Muralidhar Et Al-2007-The Journal of Pathology

1/10

Journal of Pathology

J Pathol2007;212: 368377

Published online30 April 2007 in Wiley InterScience

(www.interscience.wiley.com)DOI:10.1002/path.2179

Original Paper

Global microRNA profiles in cervical squamous cell

carcinoma depend on Drosha expression levelsB Muralidhar,1 LD Goldstein,2,3 G Ng,1 DM Winder,1 RD Palmer,1 EL Gooding,1 NL Barbosa-Morais,3

G Mukherjee,4 NP Thorne,2,3 I Roberts,1 MR Pett1 and N Coleman1,5*1Medical Research Council Cancer Cell Unit, Cambridge, CB2 0XZ, UK2Department of Applied Mathematics and Theoretical Physics, University of Cambridge, CB3 0WA, UK3Computational Biology Group, Department of Oncology, University of Cambridge, CB2 0XZ, UK4Department of Pathology, Kidwai Memorial Hospital, Bangalore, India5Department of Pathology, University of Cambridge, CB2 1QP, UK

*Correspondence to:Dr N Coleman, MedicalResearch Council Cancer Cell

Unit, Hills Road, Cambridge, CB22XZ, UK.E-mail: [email protected]

No conflicts of interest weredeclared.

Received: 29 December 2006

Revised: 27 February 2007

Accepted: 22 March 2007

Abstract

Gain of chromosome 5p is seen in over 50% of advanced cervical squamous cell carcinomas

(SCCs), although the genes responsible for the selective advantage provided by thisabnormality are poorly understood. In the W12 cervical carcinogenesis model, we observed

that 5p gain was rapidly selected over 15 population doublings and was associated with

the acquisition of a growth advantage and invasiveness. The most significantly upregulated

transcript following 5p gain was the microRNA (miRNA) processor Drosha. In clinically

progressed cervical SCC, Drosha copy-number gain was seen in 21/36 clinical samples and

8/10 cell lines and there was a significant association between Drosha transcript levels

and copy-number gain. Other genes in the miRNA processing pathway, DGCR8, XPO5

and Dicer, showed infrequent copy-number gain and over-expression. Drosha copy-number

and expression were not elevated in pre-malignant cervical squamous intraepithelial lesions.

Importantly, global miRNA profiling showed that Drosha over-expression in cervical SCC

appears to be of functional significance. Unsupervised principal component analysis of a

mixed panel of cervical SCC cell lines and clinical specimens showed clear separation

according to Drosha over-expression. miRNAs most significantly associated with Drosha

over-expression are implicated in carcinogenesis in other tissues, suggesting that theyregulate fundamental processes in neoplastic progression. Our evidence suggests that copy-

number driven over-expression of Drosha and consequent changes in miRNAs are likely

to be important contributors to the selective advantage provided by 5p gain in cervical

neoplastic progression.

Copyright 2007 Pathological Society of Great Britain and Ireland. Published by John

Wiley & Sons, Ltd.

Keywords: cervix; microRNA; Drosha; progression; SCC

Introduction

Cervical carcinoma remains the second most commoncause of cancer-related deaths in women world wide,with approximately 450 000 new cases each year.The majority are squamous cell carcinomas (SCCs),which arise through precursor squamous intraepitheliallesions (SILs). Infection by high-risk human papil-lomaviruses is a necessary but insufficient step forthe development of cervical SCC [1], with acquisi-tion of host genomic abnormalities also required formalignant progression. Copy-number gain and amplifi-cation of chromosome 5p occurs in 51% (mean value;

range: 3077%) of advanced stage cervical SCCs butnot in pre-malignancy [25], strongly suggesting animportant role in progression. Moreover, 5p gain is fre-quently seen in carcinomas at other anatomical sites,

including the head and neck [6], lung [7] and vulva

[8], implying that it is of broad relevance in oncoge-nesis.

We have taken an iterative approach to identifygenes on 5p that contribute to cervical carcinogene-sis. Initially, we exploited the W12 model, a cervi-cal keratinocyte cell line infected with HPV 16, theHPV type most commonly found in cervical SCCs[1]. In prolonged culture, W12 recapitulates cervicalneoplastic progression, both histologically and cytoge-netically [9]. Initial comparative genomic hybridiza-tion (CGH) investigations demonstrated that 5p gainwas present at passage-22 of W12 culture series-1

(W12ser-1p22) [10] but not at W12ser-1p19, indicat-ing rapid selection of this abnormality over only 15population doublings. Expression microarray analy-sis showed that the most significantly upregulated

Copyright 2007 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.www.pathsoc.org.uk


2/10

Global microRNA profiles in cervical squamous cell carcinoma 369

transcript on 5p in W12ser-1p22 vs W12ser-1p19 wasthe RNase III enzyme Drosha, a major microRNA(miRNA) processing gene. We subsequently used acombination of genomic technologies to show thatDrosha is frequently gained and over-expressed inadvanced cervical SCC, where it appears to be of func-tional significance by modifying miRNA expressionprofiles.

Materials and methods

Tissue samples

All tissue specimens were used with Local ResearchEthics Committee approval. Each sample was from adifferent patient. The specimens used were: (i) cervicalSCCs from the archives of the Kidwai MemorialInstitute of Oncology, Bangalore, India. Tumours

were staged according to the International Federationof Gynaecology and Obstetrics criteria for cervicalcarcinoma (http://www.figo.org/, accessed 30 March2007). We used frozen tissue samples from 36 pre-treatment SCCs, all of which were grade 3 andhigh stage (IIIA and above). (ii) Lesional epitheliummicrodissected from frozen sections of 25 cases ofhigh-grade SIL and 15 cases of low-grade SIL, aspreviously described [11]. In all cases at least 80% ofthe microdissected tissue was composed of abnormalepithelium. (iii) Samples of microdissected normalectocervical epithelium, obtained from hysterectomyspecimens for disease unrelated to the cervix [11].The epithelium was histologically normal and negativefor HPV DNA by nested PCR and reverse line blothybridization [11]. In total, seven samples were usedfor array analysis.

Cell culture

The cervical keratinocyte cultures used were as fol-lows: four different primary cultures of normal cer-vical keratinocytes (NCx1, NCx2, NCx5 and NCx11)generated from hysterectomy samples for disease unre-lated to the cervix; W12 series-1 pass-19 (W12ser-

1p19) and W12 series-1 pass-22 (W12ser-1p22) [10];the cervical SCC cell lines SiHa, CaSki, C33a, C4I,C4II, SW756, MS751, HT3, DOTC2 and ME180 (allfrom American Type Culture Collection). We used pre-viously published protocols for cell culture, metaphasepreparation, determination of colony forming effi-ciency and organotypic raft culture [9,10].

Chromosome-specific painting

Chromosome 5 specific paint was prepared by degen-erate oligonucleotide-primed PCR from flow sorted

chromosomes and labelled with digoxigenin-11-dUTP(Roche, Basel, Switzerland) [12]. After overnighthybridization, bound probe was detected using anti-digoxigenin FITC-conjugated Fab fragments (Roche),

counterstained with DAPI/antifade counterstain. Imageswere captured as described previously [12].

Array-CGH

Array platforms were constructed from 4134 opti-mized BAC clones covering the human genome.The mean gap size was 0.92 Mb [13]. GenomicDNA (gDNA) extraction, labelling and hybridizationwere performed as previously described [13]. For theSIL samples, gDNA was amplified by DOP-PCR asdescribed [11]. For the SCC samples the gDNA wasnot amplified. Test and reference gDNA (the latterfrom normal male peripheral blood lymphocytes), 1 gof each, were labelled with Cy3-dCTP or Cy5-dCTPusing random priming, with dye swapping (BioPrimePlus, Invitrogen).

For each spot and for each dye (channel), themedian of untransformed foreground pixel intensities

were background corrected using the median localbackground. The test : reference copy-number ratiosof corrected spot intensities were then log (base2) transformed and a global median normalizationapplied. Finally, the within-array normalized log ratioswere averaged over dye-swap paired experiments.

Thresholds were set at 1.2 for copy-number gainand 0.8 for copy-number loss, based on three stan-dard deviations of the mean of log ratios in fournormal : normal hybridizations (using gDNA from nor-mal ectocervical epithelium or normal female placentaversus gDNA from normal male peripheral blood lym-phocytes).

Quantitative real-time PCR (qRT-PCR)

Transcript levels were measured in monolayer cellsand clinical samples using QuantiTect in one-stepSYBR Green RT-PCR reactions (Qiagen, Crawley,UK). Primer efficiencies were determined using seven-point serial dilutions of Universal Human ReferenceRNA (Stratagene, La Jolla, CA, USA). All reactions(including non-RT and no template controls) were runin triplicate using an Opticon-2 cycler (MJ Research,Waltham, MA, USA). Fluorescence was measured

at the last step of each cycle, and melting curvesconfirming single PCR products were obtained aftereach run.

Expression ratios were calculated using the compar-ative Ct method described by Pfaffl [14]. Values werenormalized using four housekeeping genes: beta-actin,glyceraldehyde-3-phosphate dehydrogenase, hydrox-ymethylbilane synthase and TATA box binding protein.Pooled RNA from four primary cultures of nor-mal cervical keratinocytes was used as the refer-ence sample. Based on the Pfaffl ratios observed (seeResults), Drosha over-expression was defined as an

expression fold-change greater than 2.0. Correlationbetween Drosha transcript over-expression and copy-number gain was examined using a one-sided Fisherstest.

J Pathol2007;212: 368 377 DOI: 10.1002/pathCopyright 2007 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.


3/10

370 B Muralidhar et al

miRNA microarray preparation

Test or reference total RNA (the latter pooled FirstChoice Human RNA Survey Panel, Ambion),2.53.0 g, was ligated to an RNA-linker p-rUrUr-UdA-Cy-dye (Dharmacon, Chicago, IL, USA), labelledat the 3-end with Cy3 (test RNA) or Cy5 (reference

RNA), using T4 RNA ligase overnight at 37

C, asdescribed elsewhere [15]. Unbound nucleotides/RNA-linkers were removed by ethanol precipitation. LabelledRNA was hybridized at 55 C to miRCURY LNAArray platforms (Exiqon, Vedbaek, Denmark) in anautomated hybridization station (Lucidea Slide Pro-cessor; GE Healthcare, Chalfont St. Giles, UK),using microarray hybridization solution (version 2; GEHealthcare) containing 10% formamide. Slides werescanned using a Genepix-4000B laser scanner (AxonInstruments Sunnyvale, CA, USA).

miRNA microarray data analysis

These data analyses were performed using the statisti-cal programming language R and functions availablefromBioconductor. Mean Cy3 and Cy5 signal intensi-ties were read into R. Poor spots, as reported in the rawdata file, and probes not annotated as miRNAs weregiven spot quality weight zero. The non-backgroundcorrected signal intensities were normalized using vsn[16]. Further analysis was based on the generalizedlog (base 2) ratios (M-values). Differential expres-sion was assessed using a moderated t-statistic, tak-ing into account spot quality weights and correlationbetween within-array replicate spots [17]. p-Values

were adjusted for multiple testing using Benjamini andHochbergs method [18].

For principal component analysis (PCA) and hier-archical clustering, the two data sets (cell linesand clinical samples) were pre-processed individu-ally. Non-zero weight replicate spots were averagedand probes were mean centred. Probes with all repli-cate spots of zero weight for a given array wereremoved. No further pre-processing was undertakenbefore combining the data sets. Unsupervised PCAwas performed using all 298 probes (282 miRNAs) inthe combined data set. Hierarchical clustering was per-

formed with Pearson correlation and average linkage,based on miRNAs selected for differential expressionbetween any of the three groups of interest in cell lines(adjustedp


4/10


Figure 1.(a) Array-CGH view of chromosome 5 for (i) W12ser-1p19 and (ii) W12ser-1p22, showing 5p gain in the latter. (b) Arepresentative metaphase spread of W12ser-1p22 hybridized with whole chromosome 5 paint (red), showing three copies ofnormal chromosome 5 (green arrowheads) and two copies of isochromosome 5p (yellow arrowheads). (c) and (d) Organotypic

raft culture showing representative images of (c) (i) and (ii) W12ser1-p19 and (d) (i) and (ii) W12ser1-p22. Both form dysplasticepithelia, with W12ser-1p22 reproducibly showing an ability to invade collagen (arrows)

progressed cervical SCC. Drosha copy number wasascertained using array-CGH analysis data from 46

cervical SCC samples (Ng et al, submitted). Droshagain was seen in 21/36 (58%) clinical specimensof pre-treatment SCC (all International Federation of

Gynaecology and Obstetrics stage III/IV) and 8/10SCC cell lines (Figure 2a). There were no cases of

Drosha copy-number loss. In contrast, there was infre-quent copy-number gain of other key genes in the

miRNA processing pathway, DGCR8 (8/46 17%),XPO5 (14/4630%) and Dicer (8/4617%)(Figure 2a). Drosha copy-number gain was not seen

by array-CGH analysis in any of 25 high-grade SILs

and 15 low-grade SILs, indicating that Drosha gain isa late step in cervical carcinogenesis.

qRT-PCR for gene expression was performed in 16cell samples and 10 clinical specimens where mRNAwas available (Figure 2b). The samples fell into two

distinct groups according to Pfaffl ratios for DroshamRNA levels, with ratios ranging from 2.86 to 7.67

in one group and from 0.70 to 1.29 in the othergroup. We therefore applied a Pfaffl ratio cut-off value

of 2.0 to define Drosha over-expressing and non-over-expressing samples. Increased Drosha transcriptlevels were associated with genomic copy-number

gain (p =0.0001; one-sided Fishers test; n = 26).



5/10


Figure 2.Array-CGH and quantitative real-time PCR expression data for Drosha, DGCR8, XPO5 and Dicer. (a) Genomic copynumber (as determined by array-CGH) in ten cervical SCC cell lines and 36 cervical SCC clinical samples. Thresholds were set at1.2 for gain (solid line) and 0.8 for loss (dashed line), based on normal : normal hybridizations. (b) Comparison of expression levelswith gene copy number in (from left to right on xaxis): 10 cervical SCC tumour samples; 10 cervical SCC cell lines; four normal

cervical keratinocyte primary cultures; and W12ser-1p19 and W12ser-1p22. Array-CGH ratios are in blue, with threshold forcopy-number gain (1.2) shown as a solid line. Expression Pfaffl ratios are in red, with threshold for over-expression (2.0) shownas a dashed line



6/10


Over-expression of DGCR8, XPO5 and Dicer occurredinfrequently and there was no correlation with copy-number (Figure 2b).

miRNA profiles in cervical SCC cluster accordingto Drosha expression status

Given that Drosha has a global processing role inmiRNA production [19], we next investigated thesignificance of Drosha over-expression in modify-ing miRNA expression profiles in cervical SCCs.

Combined miRNA expression data for ten SCC celllines, two primary cultures of normal ectocervical ker-atinocytes, two passages of W12 and eight clinical

specimens were subjected to unsupervised PCA. Prin-cipal components 2 and 3, accounting for 24% of the

variation in the combined data, showed clear separa-tion between samples with and without Drosha over-expression (Figure 3(i)).

We next sought to identify the miRNAs thatwere significantly differentially expressed accordingto Drosha status in cell lines. Sixteen miRNAs were

Figure 3. miRNA profiles in cervical SCC depend on Drosha expression status. Note that the key is applicable to all threepanels. (i) Principal components (PC) 2 and 3 for 282 miRNAs in cervical cell lines and clinical samples. (ii) Dendrogram showinghierarchical clustering of cervical cell lines and clinical samples based on the profiles of the 14 selected miRNAs. In the heatmap, red and blue indicate over- and under-expression compared with mean miRNA expression levels. (iii) Generalized log ratios(M-values) for four miRNAs found to be significantly up- or down-regulated in both SCC cell lines and SCC clinical specimens withDrosha over-expression (two-sided moderated t-test, adjusted p < 0.05). Circles represent M-values for four replicate probesper array



7/10


Table 1. miRNAs associated with Drosha over-expression in both SCC cell lines and SCC clinical samples. The estimatedfold-changes are between samples over-expressing Drosha and samples not over-expressing Drosha

Cervical SCC cell

lines (n = 3, 7)

Cervical SCC

clinical samples (n = 4, 3)

miRNA

Chromosome

position

Estimated

fold-change p-ValueEstimated

fold-change p-ValueDeregulation in

other cancers

miR-31 9: 21,502,114 21,502,184 8.75 1.6e-12 5.71 4.2e-06 Colorectal cancer [26]

miR-21 17: 55,273,409 55,273,480 0.38 6.6e-06 0.77 3.3e-02 Glioblastoma [27], breast cancer [27]

miR-193b 16: 14,305,325 14,305,407 0.31 3.4e-04 0.20 2.1e-10

miR29a 7: 130,212,046 130,212,109 0.33 3.6e-04 0.59 4.2e-06 Leukaemia cell line [28]

Estimation was based on generalized log (base 2) intensity ratios. Of the SCC cell lines, three samples were without Drosha over-expression

and seven were with Drosha over-expression. Of the clinical specimens, four samples were without Drosha over-expression and three were with

Drosha over-expression. p-Values are based on a two-sided moderated t-test, adjusted for multiple testing using Benjamini and Hochbergs method [18].

differentially expressed in SCC cell lines with Droshaover-expression (n = 7) compared with normal ker-atinocytes (n = 2) or SCC cell lines without Drosha

over-expression (n =3) (see Supplementary Table 1,available at http://www.interscience.wiley.com/jpages/0022-3417/suppmat/path.2179.html). Two furthermiRNAs (let-7f and miR-141) were differentiallyexpressed between normal keratinocytes and the SCCcell lines without Drosha over-expression.

Four of the 16 miRNAs were also significantlydifferentially expressed according to Drosha over-expression in the SCC clinical specimens (four notover-expressing; three over-expressing; Table 1 andFigure 3(iii)).There was no association between themiRNA transcript changes and copy number at the

genomic loci of the precursor molecules (data notshown), indicating that the differential expression wasnot due to genomic imbalance.

Of the 16 miRNAs associated with Drosha expres-sion status and two miRNAs differentially expressedin normal keratinocytes, complete data were avail-able for 14 miRNAs (shown in the vertical axis ofFigure 3(ii) and Supplementary Table 1, available athttp://www.interscience.wiley.com/jpages/0022-3417/suppmat/path.2179.html). Hierarchical clusteringbased on the 14 miRNAs showed the three dis-tinct sample groups, representing: SCC cell lineswith Drosha over-expression; SCC cell lines with-

out Drosha over-expression; and normal ectocervicalkeratinocytes. Importantly, using the same set of miR-NAs, W12ser-1p22 clustered with the SCC cell linesover-expressing Drosha while W12ser-1p19 clusteredwith the normal ectocervical keratinocytes. Moreover,inclusion of miRNA data from the independent eightclinical specimens in which Drosha expression levelswere known (three over-expressing Drosha; five notover-expressing Drosha, including one normal cervix)resulted in the same cluster groups being retained(Figure 3(ii)).

Levels of three miRNAs (miR31, miR193b and

miR203) were validated in cell lines and clinicalsamples by qRT-PCR (Figures 4(i) and (ii)). The samethree miRNAs were also examined in four clinicalsamples where Drosha expression levels were known,

but where miRNA array hybridization had failed.Levels in these specimens were similar to other clinicalsamples with equivalent Drosha status (Figure 4(iii)).

Discussion

The RNase III endonuclease Drosha and its partnerDGCR8 form a microprocessor that is essential forthe initial stages of miRNA biogenesis [20]. Thiscomplex cleaves pri-miRNAs, consisting of a termi-nal loop, a hairpin stem and 5 and 3 single-strandedRNA extensions, to pre-miRNAs (60 70-nucleotidestem loop structures) in the nucleus. Pre-miRNAs arethen transferred by Exportin-5 (encoded by XPO5)

into the cytoplasm [21], where they are further cleavedby another RNase III endonuclease, Dicer, to gener-ate mature21-nucleotide miRNA duplexes [22]. Onestrand of the mature miRNA duplex is then loaded intoa multimeric RNA-induced silencing complex (RISC),which may elicit translational repression or degrada-tion of mRNA targets [23]. Our data show that Droshatranscript levels increase in advanced cervical SCCsthrough a copy-number driven mechanism and thathigh levels of Drosha transcription are associated witha change in global miRNA profile. DGCR8, XPO5and Dicer did not display similar genomic gain orover-expression, suggesting that other members of the

miRNA processing machinery are not responsible forthe changes in miRNA profiles seen. In a recent study,Drosha over-expression was shown to be associatedwith increased cell proliferation and poor prognosisin patients with oesophageal SCC [24]. This suggeststhat Drosha over-expression may also be involved inthe progression of SCCs at sites other than the cervix,although miRNA profiles in the oesophageal sampleswere not examined. Consistent with our findings, nosignificant alterations in DGCR8, XPO5 and Dicerexpression were found in the oesophageal SCC spec-imens.

miRNAs control critical functions across various bio-logical processes, one of which is carcinogenesis [25].miRNA fingerprinting profiles distinguish betweencancers from different lineages and within a single



8/10


lineage [25], although the mechanisms by which suchprofiles change in neoplastic tissue are poorly under-

stood. We propose that Drosha over-expression dis-turbs the process of miRNA biogenesis and representsone of the mechanisms by which miRNA profiles alterduring carcinogenesis. Unsupervised PCA revealed

that a large amount of variation in miRNA expressionprofiles in cervical SCCs can be explained by Drosha

over-expression. After multiple testing correction, 16miRNAs were identified as significantly differentiallyexpressed according to Drosha over-expression in theSCC cell lines, of which four were also significantly

differentially expressed in the clinical samples. Inter-estingly, three of these four miRNAs have previouslybeen implicated in carcinogenesis in other tissues

[2628]. miR31, which showed a mean fold-changeincrease in samples over-expressing Drosha (compared

with cases not over-expressing Drosha) of 5.71 inclinical specimens and 8.75 in cell lines, may be of

particular importance. Increased levels of miR31 havebeen seen in high-stage colorectal carcinoma [26], sug-

gesting that this miRNA may contribute to a moreaggressive phenotype in several malignancies.

Our study shows that miRNAs may either increaseor decrease when Drosha is over-expressed. Further-more miRNA levels do not appear to change in

only one direction during tumourigenesis at variousanatomical sites [25]. We have considered three non-exclusive reasons for this. Firstly, some pri-miRNAsmay be more sensitive to changes in Drosha lev-

els than others. During miRNA biogenesis, adeno-

sine deaminases that act on RNAs (ADARs) may editpri-miRNAs in a site-specific manner [29], resultingin their degradation by Tudor-SN, a component ofRISC [30], preventing Drosha/DGCR8 processing and

mature miRNA formation. ADARs convert adenosineto inosine in double-stranded RNA. It may be thatcertain pri-miRNAs are more susceptible to editing byADARs than others. If so, only those miRNAs whose

primary transcript is not edited would show increasedlevels upon Drosha over-expression. In addition, asyet undiscovered mechanisms of pri-miRNA modifi-cation may also contribute to selective Drosha/DGCR8

cleavage.Secondly, the exact mechanism by which Drosha/

DGCR8 recognizes pri-miRNAs to generate pre-miRNAs is not fully understood. Since there is nostrong sequence bias in pri-miRNAs, it seems rea-

sonable that structural components of the pri-miRNAare important in targeting Drosha/DGCR8 cleavage.Differences in the size of the terminal loop [31] andproperties of the single-stranded RNA segments at the

base of the stem of the pri-miRNA [31,32] have beenfound to alter the efficiency of Drosha/DGCR8 cleav-age. Certain pri-miRNAs may have terminal loops and

single-stranded stems that allow them to be cleavedmore efficiently by Drosha/DGCR8. If so, a greater

quantity of the respective mature miRNA productsmay be generated when Drosha is over-expressed.

Figure 4. TaqMan quantitative real-time PCR analysis ofselected miRNAs. (i) and (ii) Validation of microarray data(dashed bars) by real time PCR (solid bars) for miRNAs in:(i) SCC cell lines HT3 (no Drosha over-expression) and CaSki(Drosha over-expression) and (ii) clinical SCC samples T34 (noDrosha over-expression) and T55 (Drosha over-expression).(iii) miRNA expression levels in clinical samples of normal cervix

(NCxB) and cervical SCC (G30, G2 and G22) where microarrayanalysis had failed. Data for cases T34 and T55 are included forcomparison

Thirdly, miRNAs may decrease the transcriptionof, or degrade the mRNA of, transcription factorsinvolved in the production of other pri-miRNAs.Since the production of different pri-miRNAs maybe controlled by different transcription factors, theincrease of certain mature miRNAs (by mechanismssuch as those described above) may result in decreasedlevels of other miRNAs, by suppressing essential

transcription factors required for the production oftheir relevant pri-miRNAs.Alterations in the machinery involved in miRNA

production have been previously shown to occur



9/10


during carcinogenesis at various anatomical sites[24,33,34], though, to our knowledge, these alterationshave not previously been linked to changes in miRNAprofiles. We conclude from our study that specificchanges in miRNA profiles in cervical SCCs dependon Drosha over-expression and that this is likely to bean important mechanism underlying the significanceof 5p amplification in malignant progression in thecervix, and potentially at other sites. Consequently,Drosha over-expression may alter the levels of spe-cific miRNAs during cervical carcinogenesis, resultingin progressive disease.

Acknowledgements

This work was funded by the UK Medical Research Council

and Cancer Research UK. NPT and NLB-M are supported by

CRUK grants with Professor Simon Tavare. LDG is EPSRC

funded. We thank Professor Barbara Weber for providing the

BAC arrays used in this study and Professor Margaret Stanley

for access to the W12 cervical carcinogenesis model. NPT

and LDG thank John Marioni for helpful discussions on data

analysis.

Supplementary material

Supplementary material may be found at the webaddress http://www.interscience.wiley.com/jpages/0022-3417/suppmat/path.2179.html

References

1. Walboomers JM, Jacobs MV, Manos MM, Bosch FX, Kum-

mer JA, Shah KV, et al. Human papillomavirus is a neces-

sary cause of invasive cervical cancer worldwide. J Pathol

1999;189:1219.

2. Atkin NB. Cytogenetics of carcinoma of the cervix uteri: a review.

Cancer Genet Cytogenet1997;95:3339.

3. Atkin NB, Baker MC, Fox MF. Chromosome changes in 43

carcinomas of the cervix uteri. Cancer Genet Cytogenet

1990;44:229241.

4. Heselmeyer K, Macville M, Schrock E, Blegen H, Hellstrom AC,

Shah K, et al. Advanced-stage cervical carcinomas are defined

by a recurrent pattern of chromosomal aberrations revealing high

genetic instability and a consistent gain of chromosome arm 3q.

Genes Chromosomes Cancer1997;19:233240.5. Kirchhoff M, Rose H, Petersen BL, Maahr J, Gerdes T, Lund-

steen C, et al. Comparative genomic hybridization reveals a

recurrent pattern of chromosomal aberrations in severe dyspla-

sia/carcinoma in situ of the cervix and in advanced-stage cervical

carcinoma.Genes Chromosomes Cancer1999;24:144150.

6. Jin C, Jin Y, Wennerberg J, Annertz K, Enoksson J, Mertens F.

Cytogenetic abnormalities in 106 oral squamous cell carcinomas.

Cancer Genet Cytogenet2006;164:4453.

7. Garnis C, Lockwood WW, Vucic E, Ge Y, Girard L, Minna JD,

et al. High resolution analysis of non-small cell lung cancer cell

lines by whole genome tiling path array CGH. Int J Cancer

2006;118:15561564.

8. Huang FY, Kwok YK, Lau ET, Tang MH, Ng TY, Ngan HY.

Genetic abnormalities and HPV status in cervical and

vulvar squamous cell carcinomas. Cancer Genet Cytogenet

2005;157:4248.

9. Pett MR, Alazawi WO, Roberts I, Dowen S, Smith DI, Stan-

ley MA, et al. Acquisition of high-level chromosomal instability

is associated with integration of human papillomavirus type 16 in

cervical keratinocytes. Cancer Res 2004;64:13591368.

10. Pett MR, Herdman MT, Palmer RD, Yeo GS, Shivji MK, Stan-

ley MA, et al. Selection of cervical keratinocytes containing inte-

grated HPV16 associates with episome loss and an endogenous

antiviral response. Proc Natl Acad Sci USA 2006;103:38223827.

11. Alazawi W, Pett M, Strauss S, Moseley R, Gray J, Stanley M,

et al. Genomic imbalances in 70 snap-frozen cervical squamous

intraepithelial lesions: associations with lesion grade, state ofthe HPV16 E2 gene and clinical outcome. Br J Cancer

2004;91:20632070.

12. Roberts I, Wienberg J, Nacheva E, Grace C, Griffin D, Cole-

man N. Novel method for the production of multiple colour chro-

mosome paints for use in karyotyping by fluorescence in situ

hybridisation. Genes Chromosomes Cancer1999;25:241250.

13. Ng G, Huang J, Roberts I, Coleman N. Defining ploidy-specific

thresholds in array comparative genomic hybridization to improve

the sensitivity of detection of single copy alterations in cell lines.

J Mol Diagn 2006;8:449458.

14. Herdman MT, Pett MR, Roberts I, Alazawi WO, Teschen-

dorff AE, Zhang XY, et al. Interferon-{beta} treatment of cervical

keratinocytes naturally infected with human papillomavirus 16

episomes promotes rapid reduction in episome numbers and emer-

gence of latent integrants. Carcinogenesis 2006;27:234153.15. Miska EA, Alvarez-Saavedra E, Townsend M, Yoshii A, Ses-

tan N, Rakic P,et al. Microarray analysis of microRNA expression

in the developing mammalian brain. Genome Biol 2004;5:R68.

16. Huber W, von Heydebreck A, Sultmann H, Poustka A, Vingron M.

Variance stabilization applied to microarray data calibration and

to the quantification of differential expression. Bioinformatics

2002;18(Suppl 1):S96S104.

17. Smyth GK. Linear models and empirical Bayes methods for

assessing differential expression in microarray experiments. Stat

Appl Genet Mol Biol 2004;3:Article3.

18. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a

practical and powerful approach to multiple testing. J R Statist Soc

B 1995;289300.

19. Lee Y, Ahn C, Han J, Choi H, Kim J, Yim J, et al. The

nuclear RNase III Drosha initiates microRNA processing. Nature2003;425:415419.

20. Gregory RI, Yan KP, Amuthan G, Chendrimada T, Doratotaj B,

Cooch N,et al. The microprocessor complex mediates the genesis

of microRNAs. Nature 2004;432:235240.

21. Yi R, Qin Y, Macara IG, Cullen BR. Exportin-5 mediates the

nuclear export of pre-microRNAs and short hairpin RNAs. Genes

Dev 2003;17:30113016.

22. Lee YS, Nakahara K, Pham JW, Kim K, He Z, Sontheimer EJ,

et al. Distinct roles for Drosophila Dicer-1 and Dicer-2 in the

siRNA/miRNA silencing pathways. Cell 2004;117:6981.

23. Meister G, Tuschl T. Mechanisms of gene silencing by double-

stranded RNA. Nature 2004;431:343349.

24. Sugito N, Ishiguro H, Kuwabara Y, Kimura M, Mitsui A, Kure-

hara H, et al. RNASEN regulates cell proliferation and affects

survival in esophageal cancer patients. Clin Cancer Res2006;12:73228.

25. Lu J, Getz G, Miska EA, Alvarez-Saavedra E, Lamb J, Peck D,

et al. MicroRNA expression profiles classify human cancers.

Nature2005;435:834838.

26. Bandres E, Cubedo E, Agirre X, Malumbres R, Zarate R,

Ramirez N, et al. Identification by real-time PCR of 13 mature

microRNAs differentially expressed in colorectal cancer and non-

tumoral tissues. Mol Cancer2006;5:29.

27. Esquela-Kerscher A, Slack FJ. Oncomirs microRNAs with a

role in cancer. Nat Rev Cancer2006;6:259269.

28. Yu J, Wang F, Yang GH, Wang FL, Ma YN, Du ZW, et al.

Human microRNA clusters: genomic organization and expression

profile in leukemia cell lines. Biochem Biophys Res Commun

2006;349:5968.

29. Yang W, Chendrimada TP, Wang Q, Higuchi M, Seeburg PH,

Shiekhattar R, et al. Modulation of microRNA processing and

expression through RNA editing by ADAR deaminases. Nat Struct

Mol Biol 2006;13:1321.



10/10


30. Scadden AD. The RISC subunit Tudor-SN binds to hyper-edited

double-stranded RNA and promotes its cleavage. Nat Struct Mol

Biol 2005;12:489496.

31. Zeng Y, Yi R, Cullen BR. Recognition and cleavage of primary

microRNA precursors by the nuclear processing enzyme Drosha.

EMBO J2005;24:138148.

32. Han J, Lee Y, Yeom KH, Nam JW, Heo I, Rhee JK, et al.

Molecular basis for the recognition of primary microRNAs by the

Drosha-DGCR8 complex. Cell 2006;125:887901.

33. Karube Y, Tanaka H, Osada H, Tomida S, Tatematsu Y, Yanagi-

sawa K, et al. Reduced expression of Dicer associated with poor

prognosis in lung cancer patients. Cancer Sci 2005;96:111115.

34. Chiosea S, Jelezcova E, Chandran U, Acquafondata M, McHale T,

Sobol RW, et al. Up-regulation of dicer, a component of the

microRNA machinery, in prostate adenocarcinoma. Am J Pathol

2006;169:18121820.


muralidhar et al-2007-the journal of pathology

Documents