Академический Документы
Профессиональный Документы
Культура Документы
Abstract
The molecular determinants of thyroid follicular nodules are incompletely understood and assessment of malignancy is
a diagnostic challenge. Since microRNA (miRNA) analyses could provide new leads to malignant progression,
we characterised the global miRNA expression in follicular adenoma (FA) and follicular carcinoma (FC). Comparison of
carcinoma and adenoma with normal thyroid revealed 150 and 107 differentially expressed miRNAs respectively. Most
miRNAs were down-regulated and especially miR-199b-5p and miR-144 which were essentially lost in the carcinomas.
Integration of the changed miRNAs with differentially expressed mRNAs demonstrated an enrichment of seed sites
among up-regulated transcripts encoding proteins implicated in thyroid tumourigenesis. This was substantiated by the
demonstration that pre-miR-199b reduced proliferation when added to cultured follicular thyroid carcinoma cells. The
down-regulated miRNAs in FC exhibited a substantial similarity with down-regulated miRNAs in anaplastic carcinoma
(AC) and by gene set enrichment analysis, we observed a significant identity between target mRNAs in FC and
transcripts up-regulated in AC. To examine the diagnostic potential of miRNA expression pattern in distinguishing
malignant from benign nodules we employed a supervised learning algorithm and leave-one-out-cross-validation. By this
procedure, FA and FC were identified with a negative predicted value of 83% (data generated by microarray platform) and
of 92% (data generated by qRT-PCR platform). We conclude that follicular neoplasia is associated with major changes in
miRNA expression that may promote malignant transformation by increasing the expression of transcripts encoding
tumourigenic factors. Moreover, miRNA profiling may facilitate the diagnosis of carcinoma vs adenoma.
Journal of Molecular Endocrinology (2012) 48, 1123
cells (Schmittgen 2008). Moreover, miRNAs are data indicated that miRNAs could promote malignant
aberrantly expressed or lost in a variety of cancers progression, we further employed classification algo-
(Rosenfeld et al. 2008). Many target mRNAs encode rithms to distinguish carcinoma from adenoma.
oncogenes and tumour suppressors and in this way
dysregulated miRNAs may play a causal role in
malignant progression. Not surprisingly, miRNAs are
therefore considered attractive candidates for classi- Materials and methods
fication of tumours. The role of miRNAs in thyroid
cancer is incompletely understood. A number of Thyroid tissue, FA and FC
miRNAs have previously been identified in various Thyroid samples originated from a consecutive series
thyroid tumours (He et al. 2005, Pallante et al. 2006, of patients with a solitary or prominent scintigraphically
Weber et al. 2006, Chen et al. 2008, Nikiforova et al. cold thyroid nodule operated on, and with a histological
2008). miR-197 and miR-346 are overexpressed in FC in diagnosis of an FA or FC. As a uniform histopathological
comparison to adenoma and in vitro studies suggested evaluation was essential, the diagnosis was made by a
that both miRNAs could have a significant impact on particular pathologist specialising in thyroid pathology.
tumour cell proliferation (Weber et al. 2006). All tumours were diagnosed and classified according to
In this study, we employed global miRNA analysis the WHO definition of histological criteria. Clinical data
to identify differentially expressed miRNAs in follicular are listed in Table 1. Surgically removed thyroid samples
thyroid carcinoma and adenoma. miRNAs were inte- were snap frozen at the Department of Pathology and
grated with differentially expressed mRNAs to identify stored at K120 8C over a 5-year period. The project was
and validate putative target mRNAs. The gene ontology approved by the ethics committee and informed consent
and gene set enrichment analysis (GSEA) of these was obtained from all patients. Twelve FA, twelve FC and
showed a significant enrichment of seed sites among ten normal thyroid (NT) specimens were included.
transcripts encoding proteins involved in thyroid NT specimens were obtained by a thyroid pathologist to
tumourigenesis and an overlap to transcripts up-regu- ensure that the tissue derived from macro- and micro-
lated in anaplastic carcinoma (AC) respectively. As the scopically normal tissue adjacent to the encapsulated
Table 1 Clinical data of patients with thyroid follicular neoplasia. Twelve patients with histopathologically verified follicular carcinomas (FC),
minimal and widely invasiveness, and twelve patients with follicular adenomas (FA). The table depicts diagnosis, age, sex, tumour size,
invasiveness of the examined tumours and status of known oncogenes. All tumours were negative for KRAS point mutation and examination of
BRAF showed only one positive carcinoma sample positive for BRAF point mutation; K601E (c.1801AOG p.Lys601Glu). Examination for
PAX8/PPARg translocation exhibited no positive samples
tumours. The number of tumours was balanced to for 841 human miRNAs, catalogued in the miRBase
provide optimal power estimates and a similar number Sequence Database (Release 11.0; http://microrna.
of samples in each diagnostic category. sanger.ac.uk/) and 428 proprietary human miRPlus
sequences not yet annotated in miRBase.
Furthermore, miRNA expression levels from 30
Total RNA and DNA isolation thyroid specimens (12 FC, 12 FA and 6 NT), were
Total RNA was isolated from frozen samples using generated using the miRCURY LNATM Universal RT
Trizol (Invitrogen) according to the manufacturers miRNA PCR panels I and II, V2 (Exiqon). Total RNA
protocol. RNA was precipitated using 100% iso- (40 ng) was reverse transcribed using the Universal
propanol. Purified RNA was subsequently quantified cDNA synthesis kit, mixed with SYBR Green master
on a NanoDrop ND-1000 Spectrophotometer (Nano- mix kit, and subsequently added to the pre-aliquoted
Drop Technologies, Wilmington, DE, USA) and miRNA PCR primer sets in two 384-well PCR plates
examined on a Bioanalyzer Nano RNA Chip (Agilent enabling profiling of 742 human miRNAs. All reagents
Technologies, Santa Clara, CA, USA) before labelling. were from Exiqon and their recommendations were
Each sample (1 mg) was pooled and used as common followed. Each plate contained an additional six primer
reference. DNA from tissues was isolated by lysing the sets for reference miRNAs and a set of negative
tissue in 20 ml proteinase K and 200 ml Tris/NaCl/ controls. The amplification curves were analysed using
EDTA/SDS, followed by overnight incubation at 55 8C. the Roche LC software, both for determination of
Finally, 5 M NaCl is added to the lysed tissues and DNA crossover point (Cp) and for melting curve analysis.
One hundred and thirty-five miRNA assays were
is precipitated by adding 200 ml ice-cold 96% ethanol.
successfully assessed with sufficient signal (Cp !37, or
Total RNA for mRNA microarray was extracted from
5 Cp less than negative controls) in all samples.
pre-miRNA transfected cells using NucleoSpin RNA/
Analyses of global mRNA levels in RNA extracted
Protein from Macherey-Nagel according to their
from pre-miRNA transfected cells were performed
recommendations.
by the Affymetrix platform. Total RNA (250 ng) was
amplified and labelled using the Ambion WT
Detection of point mutations Expression Kit (Applied Biosystems) according to the
manufacturers instructions. The labelled samples were
Mutation analyses, including PCR-amplification and hybridised to the Human Gene 1.0 ST GeneChip array
sequencing, were performed on all 24 tumour samples (Affymetrix, Santa Clara, CA, USA). Arrays were washed
for KRAS and BRAF (V600E and K601E) and regular and stained with phycoerythrin-conjugated streptavidin
PCR-amplification, followed by gel-separation was used (SAPE) using the Affymetrix Fluidics Station 450,
to analyse the fusion point of PAX8/PPARg transloca- and subsequently scanned in the Affymetrix GeneArray
tion. For detection of possible amplicons, a positive 2500 scanner to generate fluorescent images, according
control was examined along with the 24 tumour to the Affymetrix GeneChip protocol. Cell intensity
samples. Primer sequences for KRAS, forward; files (CEL files) were generated in the GeneChip
5 0 -TGTAAAACGACGGCCAGTCGATACACGTCTGC- Command Console Software (AGCC; Affymetrix).
AGTCAA-3 0 and reverse; 5 0 -CAGGAAACAGCTATG- CEL files were modelled using the robust multichip
ACCCTCATGAAAATGGTCAGAGA-3 0 , BRAF, forward; average approach, followed by mean Probe Set
5 0 -TGCTTGCTCTGATAGGAAAATG-3 0 and reverse; Summarisation resulting in a single expression value
5 0 -GGATGGTAAGAATTGAGGCT-3 0 and PAX8/PPARg, for each gene. Modelled CEL files were generated with
forward; 5 0 -GCATTGACTCACAGAGCAGCA-3 0 and the software package Partek Genomics Suite 6.5.
reverse; 5 0 -AGACCACTCCCACTCCTTTG-3 0 (Kroll
et al. 2000).
Image analysis and normalisation of miRNA
expression
miRNA and mRNA expression analyses
Arrays were scanned in an Agilent DNA Microarray
miRNA expression levels were determined by micro- Scanner (Agilent Technologies) and resulting images
array analysis. Total RNA (1 mg) was labelled with were analysed with Genepix Pro 6.0 software (Molecular
fluorescent Hy3 (sample)/Hy5 (reference-sample) Devices). Background intensities were subtracted
dye from the miRCURY LNA microRNA Array Power from foreground intensities and within array LOESS
Labelling Kit (Exiqon, Vedbaek, Denmark) in accord- normalised, followed by Aquantile normalisation
ance with the manufacturers instructions. Using a between arrays as implemented in the Limma package
TECAN hybridisation station, labelled samples were in Bioconducter library (Gentleman et al. 2004) in
hybridised overnight to pre-printed miRCURY LNA R version 2.10 (R Development Core Team 2007). Mean
microRNA Array, v.11.0 (Exiqon), containing probes values of the quadruplicate probes for each of the
www.endocrinology-journals.org Journal of Molecular Endocrinology (2012) 48, 1123
14 M ROSSING, F N BENNEDBK, F C NIELSEN and others . miRNAs in follicular thyroid tumours
miRNAs were obtained and log2 ratios between sample Fishers exact test. GSEA of the target list was submitted to
and reference were used for further analyses. and performed using the open source software available
Normalisation of the results derived from the RT at http://www.broadinstitute.org/gsea/msigdb/index.
miRNA PCR panels was performed based on the average jsp. The gene list was loaded into the molecular signature
of the assays detected in all samples; Normalised database (MSigDB) and overlaps were detected in the
CpZaverage samples Cp (nZ135)Kassay Cp. The C2 currated gene analysis. To determine the association
normalised miRNA expression values were used for to the gene set, the hypergeometric distribution of
generating a diagnostic classifier between FC and FA overlapping genes over all genes in a gene set was
as described in the section Construction of classifier. calculated. For this analysis a cutoff was set at composite
score OC5 and only considering the differentially
expressed genes, a cutoff was set at P!0.01, qZ0.0036,
Class comparison analyses resulting in 287 probes. Heatmaps were generated as
Class comparison was performed by the Limma package supervised (two-way) cluster analysis.
(http://bioconductor.org/packages/release/bioc/
html/limma.html). For each miRNA standard errors
Construction of classifier
and fold changes were estimated by fitting a linear
model. Differentially expressed miRNAs were Normalised data from 24 samples consisting of the
determined by applying empirical Bayes moderated t- expression level of 545 different probes (377 annotated
statistics test (Wettenhall & Smyth 2004). miRNAs were miRNAs and 168 hsa miRPlus, all with an A value O7,
defined to be differentially expressed if they had a AZ12 (log2RedClog2Green)) were included in the
BenjaminiHochberg corrected P value below 0.05 and analysis. The classifier was constructed to classify FC
an absolute fold-change above 1.5. and FA. The validation of the classifiers performance
on the training set was done by leave-one-out-cross-
validation (LOOCV), that provides an unbiased
mRNA expression data and miRNA target predictions estimate of the prediction error by going over all
followed by pathway analysis training examples in turn using the complement
The global mRNA expression data originating from training set (of size 24K1Z23) to perform training
FA, FC and AC tissue samples was based on previous (including probe ranking, selection and model fitting)
work by our group (Borup et al. 2010). In addition, and using the last left out sample for prediction. In this
mRNA profiles of papillary and NT tissue samples were way, we obtained as many predictions as there were
down loaded from the Array Express, ID: E-GEOD-6004 samples in the training set. These predictions formed
and E-GEOD-7307 respectively. To predict mRNA unbiased estimates of the prediction error quantified in
targets, based on the observed miRNA profiles, we terms of the confusion matrix. The training of the
implemented the method developed by Gallagher et al. classifiers inside the LOO loop consisted of a univariate
(2010). Briefly, mRNA target predictions were based selection step followed by applying support vector
on the TargetScan miRNA target prediction database machine learning (SVM; Vapnik 1998). Probes were
in combination with the observed changes in miRNAs. ranked according to their differential expression by
Only miRNAs that exhibited an absolute change the Students t-test with a threshold of P!0.001
O1.5-fold and mRNA with an average expression (Dudoit et al. 2003). A permutation test was performed
intensity O40 in FCs were included in the analysis. to determine if the cross-validated misclassification
This method provides a weighted miRNA inhibitor rate is lower than expected by chance (Tusher et al.
score (sum of effects), predicting the transcripts, most 2001, Simon et al. 2007). In 1000 random permutations
likely to be regulated by miRNAs. As the majority of the of the class label, the entire cross-validation was
differentially expressed miRNAs were down-regulated, repeated for classifying the random classes of samples.
only mRNAs with a positive composite score were The proportion of the 1000 random permutations, that
considered in the analysis. A cutoff at C2 and a variance gives a smaller or similar cross-validation misclassifi-
filtering of 0.176 were used to remove unhybridised cation rate as obtained with the real data, determined
probes in our weighted inhibitor score list (composite the permutation P value. The statistical significance of
score). This resulted in the retrieval of 1381 probe sets, the error rate was determined for the SVM classifier.
representing 528 genes. To examine whether the
putative mRNA targets were associated with particular
Quantitative reverse transcription PCR of miRNA
biological functions or pathways, the predicted mRNAs
and mRNA
were examined in the Ingenuity Pathway Analysis soft-
ware package (Ingenuity Systems, www.ingeuity.com). cDNA was prepared from 25 ng total RNA from
The most significant pathways were identified by the 34 tumour samples using TagMan microRNA reverse
Journal of Molecular Endocrinology (2012) 48, 1123 www.endocrinology-journals.org
miRNAs in follicular thyroid tumours . M ROSSING, F N BENNEDBK, F C NIELSEN and others 15
transcription kit and TagMan microRNA assays to the E-96. Five hours post-transfection with pre-
containing predesigned primers for miR-221, miR- miRNAs, transfected cells were plated in quadruplicates
182, miR-96, miR-199a3p, miR-144*, miR-199b5p and and monitored for 5 days. The experiment was repeated
miR-1826 was added. MiR-191 was used for endogenous three times. For mRNA microarray analysis, total RNA
control. Quantitative reverse transcription PCR (qRT- was extracted 24 h after pre-miRNA transfections and
PCR) was performed by TagMan universal PCR master labelled as described in the section miRNA and mRNA
mix No AmpEras UNG, according to the manufac- expression analyses.
turers instructions, all from Applied Biosystems. Each
amplification reaction was performed in triplicate, and
median value of the three cycle threshold was used for
further analyses. For calculations of fold changes we Results
used the 2KDCt method (Schmittgen & Livak 2008). In
Tumour samples and oncogenic mutations
addition, we validated Exiqon miRPlus probe, miRPlus-
E-1078, used for classification of FC and FA. For this The thyroid samples originated from a consecutive
qRT-PCR analysis we employed the Exiqon microRNA series of patients and included tumours from 19 women
LNA PCR primer sets together with Universal cDNA and 5 men and all tumours were classified according
synthesis and SYBR green master mix, following the to the WHO definition of histological criteria. The
manufacturers instructions (Exiqon). median age was 44 years in carcinoma patients and 47
For validation of miRNA predicted transcripts 2 mg years in adenoma patients. The size of the tumours
total RNA from 24 tumour samples (12 FC and 10 NT) ranged from 1.5 to 10.5 cm and the median diameter
was reverse transcribed to cDNA using High-Capacity was 4 cm in the carcinoma patients and 3.75 cm in the
cDNA Reverse Transcription kit (Applied Biosystems). adenoma patients (Table 1). All tumours were
Synthesised cDNA (200 ng/ml) diluted 1/100-fold and examined for KRAS and BRAF mutations and this
2 ml aliquot used in 10 ml total PCR containing fast showed that only one carcinoma sample was positive for
SYBR green PCR Master Mix (Applied Biosystems) and BRAF and examination for PAX8/PPARg translocation
10 pmol/ml of each primer for target genes. Primer with PCR followed by gel-separation showed no positive
sequences for VPS26A, forward; 5 0 -GCTTGCCACC- amplicons (Table 1). Taken together, the thyroid
TATCCTGATG-3 0 and reverse; 5 0 -CTGGGTCCAATTC- specimens in the two diagnostic groups were com-
CTGTGAT-3 0 and ABCC1, forward; 5 0 -TAAAAGAGG- parable both with respect to the clinical features and
ATGCCCAGGTG-3 0 and reverse; 5 0 -ACCCTGTGATCC- the presence of oncogenic mutations.
ACCAGAAG-3 0 . b-Actin was measured for endogenous
control. The PCR comprised 40 cycles of a four-step
programme: step 1, 50 8C for 2 min, step 2, 95 8C for miRNA expression in FC and FA
2 min, step 3 (40 cycles), 95 8C for 10 s and 60 8C for The following class comparison analysis and miRNA
45 s and step 4, 95 8C for 15 s, 60 8C for 15 s and 95 8C target analysis are based on the derived microarray
for 15 s. qPCRs were performed in triplicate. expression data since this platform counts the largest
number of miRNAs. Class comparison analysis
miRNA transfections and proliferation experiments revealed 150 annotated and differentially expressed
human miRNAs 37 up-regulated and 113 down-
Human thyroid follicular R082 W-1 cancer cells were regulated miRNAs in FCs compared with NT.
grown at 37 8C, 5% CO2 in RPMI 1640 with GlutaMax The fold change ranged from 3.1- to K39-fold.
(Invitrogen) supplemented with 10% FBS (Invitrogen), Owing to the substantial 39-fold down-regulation of
1% penicillin and streptomycin. Cells were trans- miR-199b-5p (also named miR-199b); this miRNA
fected with 50 nM pre-miR-199b and pre-miR-144, is essentially lost in FC. MiR-144*, miR-199b-3p, miR-
both from Applied Biosystems and for control 199a-5p and miR-144 were also strongly reduced
miRNA, we used miRIDIAN microRNA mimic negative almost to the background. Of the most up-regulated
control#2 (Dharmacon/Thermo Scientific, hafayette, miRNAs, miR-221, miR-96 and miR-182 exhibited
CO, USA). Cells were transfected with Turbofect fold changes of 3.1, 2.9 and 2.6 respectively. The
transfection reagent (Dharmacon/Thermo Scientific) comparison of FA to NT tissue revealed 107
according to the manufacturers instructions. Cell differentially expressed miRNAs. Forty-two were
proliferation was studied real-time using the XCELLi- up-regulated and 65 were down-regulated. Finally,
gence system as described in the manufacturers the comparison of carcinoma to adenoma showed
instructions (Roche Applied Science and ACEA that 56 miRNAs were differentially expressed between
Bioscience). This allowed continuous measurements these groups. Twelve miRNAs were up-regulated
of the electronic impedance by detecting the physio- and 44 were down-regulated in the carcinoma. The
logical changes of the cells on the electrodes attached five most up- and down-regulated miRNAs in each
www.endocrinology-journals.org Journal of Molecular Endocrinology (2012) 48, 1123
16 M ROSSING, F N BENNEDBK, F C NIELSEN and others . miRNAs in follicular thyroid tumours
Table 2 Dysregulated miRNAs. The five most up- and down- Computational identification of putative target
regulated miRNAs in the three comparisons, follicular carcinomas mRNAs and pathway analysis
(FC) vs normal thyroid (NT), follicular adenomas (FA) vs NT
and FC vs FA are listed. miRNAs were defined to be differentially Results from the global expression profiling of NT and
expressed if they had a BenjaminiHochberg corrected P value FA and FC samples were used for integration of mRNA
below 0.05 and an absolute fold-change above 1.5
and miRNA array data. In a pilot analysis, we counted
Fold Adjusted predicted seed sites corresponding to the changed
miRNA change P value P value miRNAs from the differentially expressed transcripts
using the Targetscan miRNAmRNA integration soft-
FC vs NT ware in Partek Genomic Suite (http://www.partek.
miR-199b-5p K39 3.15!10K31 5.98!10K28 com/partekgs). Based on the fact that the vast majority
miR-144* K15.1 7.10!10K9 2.50!10K7 of mRNAs were down-regulated by associated miRNAs
miR-199b-3p K12.1 1.33!10K28 1.26!10K25
(Guo et al. 2010), only mRNAs and miRNAs that
miR-199a-5p K9 1.66!10K14 1.97!10K12
miR-144 K6.9 3.34!10K21 1.06!10K18 exhibited an inverse expression pattern were
miR-221 3.1 1.12!10K5 0.0001 considered. We detected a significant enrichment
miR-96 2.9 5.92!10K6 8.72!10K5 (Fishers exact test P!0.01, assuming that the total
miR-182 2.6 2.11!10K6 3.46!10K5 number of 904 miRNA could target 22.000 transcripts
miR-597 2.4 0.002 0.0114
miR-222 2.3 0.0007 0.0049 at a global level) of seed sites corresponding to down-
FA vs NT regulated miRNAs. The down-regulated miRNAs in
miR-199b-5p K26 4.69!10K27 8.92!10K24 the FC group exhibited putative seed sites in almost
miR-144* K8.1 6.76!10K6 0.0001 85% of the up-regulated transcripts, which distin-
miR-663 K6.2 5.48!10K12 4.51!10K10 guished carcinoma from NT and adenoma respectively.
miR-199b-3p K6.1 1.26!10K18 4.80!10K16
miR-142-3p K4.6 2.34!10K15 4.46!10K13 All limitations of the computational approach taken
ebv-miR-BART8 5.3 0.0083 0.0489 into consideration, the results suggest that the changed
miR-517a 4.5 0.0043 0.0285 miRNAs could have an impact on the observed changes
miR-512-3p 3.6 1.70!10K3 1.37!10K2 in the transcriptome during progression to carcinoma.
miR-301b 3.2 5.08!10K6 8.55!10K5
miR-518a-3p 2.60 6.36!10K3 3.94!10K2
To identify the biological pathways and further
substantiate the possibility that miRNA changes had an
FC vs FA
miR-512-3p K3.2 0.0032 0.0395 impact on the gene expression pattern in the carcinoma,
miR-886-5p K3 2.58!10K6 0.0003 we employed the previously reported ranking system
miR-450a K3 6.92!10K7 0.0001 for mRNA target identification (Gallagher et al. 2010).
miR-301b K2.6 5.78!10K5 0.0032 Following miRNA target identification, 528 of the
miR-429 K2.4 0.0001 0.0046
miR-637 2.1 7.09!10K5 0.0036 highest ranked transcripts with composite score OC2,
miR-631 2 0.0028 0.0364 was loaded into the geneontology and pathway analysis
miR-219-2-3p 1.8 0.0041 0.0467 feature of the Ingenuity Software. Almost 200 of the
miR-662 1.8 0.0017 0.0272 putative targets mRNAs encoded factors categorised in
miR-744 1.8 2.21!10K7 7.00!10K5
neoplasia, cancer and tumourigenesis (P values of
1.95!10K5, 2.16!10K5 and 2.25!10K5 respectively).
comparison are listed in Table 2 and the complete When we explored the biological functions, a group
list of miRNAs that changed more than twofold is of 17 transcripts (APC, ATM, CCDC6, CCND1, COPS2,
listed in Supplementary Table 1 (see section on ETS1, FN1, GABBR2, LIFR, NRAS, PPARGC1A, PTPN4,
supplementary data given at the end of this article). RAP2A, S1PR1, SMAD2, TGFBR1 and VEGFA; PZ4.42
Fifty-eight miRNAs were differentially expressed in !10K5) encoded factors previously described in thyroid
both the carcinoma and adenoma compared with cancer. Corresponding to the down-regulation of miR-
NT (Fig. 1A). The overlap was marked among the NAs, 9 out of 11 target transcripts within the thyroid
down-regulated miRNAs, where 49 of the 65 down- cancer specific mRNAs were increased (82%; P!0.05).
regulated miRNAs in the adenoma were identical, To further examine the putative biological signi-
compared with 9 out of 42 for the up-regulated ficance of miRNA changes, we performed a GSEA
miRNAs. With the exception of miR-199b-5p, that (http://www.broadinstitute.org/gsea/msigdb/index.
were progressively down-regulated in carcinoma, jsp). The target list was filtered by a composite score
the shared miRNAs exhibited basically the same OC5 and only differentially expressed genes were
levels in adenoma and carcinoma compared with NT included (P!0.01, qZ0.0036) in the analysis, as shown
(Fig. 1B). Taken together, the results imply that in Fig. 2B. This resulted in 287 probe sets of which 260
changes in miRNA expression predominantly occur probe sets were up-regulated. The 260 probes corre-
during transition from NT epithelial cell to adenoma sponded to 135 different genes, which were uploaded
and not during malignant transformation. into GSEA and underwent a C2 currated genes analysis.
Journal of Molecular Endocrinology (2012) 48, 1123 www.endocrinology-journals.org
miRNAs in follicular thyroid tumours . M ROSSING, F N BENNEDBK, F C NIELSEN and others 17
algorithm SVM and LOOCV to generate a classifier. miRNAs, a PCA plot was generated (Fig. 6B). Both the
The optimal signature for classification of FC and FA negative predictive value (NPV) and the positive
consist of two miRNAs, miR-1826 and miR-Eplus-1078, predictive value (PPV) for carcinoma is 83%. The
and based on expression values of the two classification LOOCV was used to compute misclassification rate.
NT FC PC AC
A B
2 (11%)
1 (23%)
3 (9%)
C D
E
2
Figure 2 Heatmaps of predicted miRNA target transcripts in thyroid tumours. (A) depicts
an unfiltered principal component analysis (PCA) of the gene expression in follicular
carcinoma (FC) and normal thyroid tissue (NT). The heatmap in (B) shows the relative
expression of the 260 up-regulated probe sets, representing 135 gene symbols, derived
from the miRNA predicted targets with a composite score OC5 and filtered, so only
differentially expressed genes were included (P!0.01, qZ0.0036) in the analysis. The
heatmap of (C) illustrates the relative expression level of the overlapping 17 transcripts
(GSK3B, FXR1, WAPAL, ITCH, NUPL1, SLC7A11, DDX3X, KPNA4, STK4, MAP3K7IP3,
SSX2IP, TBL1XR1, SIX4, EIF5B, PRPF40A, NARG1 and USP31) represented by 82
probes from the Rodrigues_thyroid_carcinoma_anaplastic_up-regulated_cluster based
on the gene set enrichment analysis. The relative expression of the entire Rodrigues cluster
in FC and NT is depicted in the heatmap of (D). The expression pattern of the 135 miRNA
target transcripts in FC, papillary (PC) and anaplastic carcinoma (AC) is shown in (E).
130
Doubling time
for classification of FC and FA consists of 14 miRNAs,
110 Control miRNA 13 h miR-19a, -501-3p, -17, -335, -106b, -15a, -16, -374a, -542-5p,
miR-199b-5p 16 h (P=0004)
90 miR-144 14 h (P=0056)
-503, -320a, -326, -330-5p and let-7i. A PCA plot based
Cell index
Table 3 miR-199b-5p target transcripts in follicular cancer specimens. A complete list of 39 target transcripts of miR-199b-5p, matching
the unique identifier probes as illustrated in the two-way cluster, Fig. 4. The listed targets are a subset of the initially 190 computational
predicted targets of miR-199b-5p that were correlated to the down-regulated transcripts in follicular thyroid cancer cells upon transfecting
with pre-miR-199b. This comparison resulted in 56 identical transcripts, i.e. 29.5% of the computational predicted targets were actual
targets. The expression pattern of the 56 transcripts was examined in the data from the histopathological follicular thyroid cancer
specimens and upon filtering the data (P!0.05) the final list resulted in 39 transcripts. Nearly all of the 39 transcripts (87%) were
significantly up-regulated as expected in the follicular thyroid cancer samples
AP1G1, ARHGAP12, ARHGAP21, CCDC43, CELSR1, C1GALT1, C9orf5, ECE1, ERLIN1, ETS1, EXTL3, FZD6, GIT1, GPD2, IPO8,
LARP4, LIN7C, MAP3K11, MGAT4B, MPP5, NLK, NPAS2, PARP12, PCYOX1, PLXND1, POMGNT1, PPARGC1A, PPFIBP1, PPP1R2,
RBBP4, R3HDM2, SLC24A3, SOS2, STK4, TAF9B, TSPAN6, VPS26A, WDTC1, ZNF468
neoplasia are incompletely understood. Consequently, up-regulated in AC. The resemblance to AC may not be
efforts have been devoted towards defining biomarkers surprising, because there is a substantial overlap
such as miRNAs that would allow the clinicians to between the miRNAs that change in FC and those
distinguish carcinoma from adenoma as well as other altered in AC (Braun et al. 2010). Moreover, in a PCA of
thyroid cancers and obtain information of the molecu- all genes, FCs tend to distribute as a continuum ranging
lar pathways that drive thyroid tumourigenesis. from relatively well-differentiated FCs to FCs that are
Compared with NT, FA and FC exhibited widespread located in close proximity to ACs in contrast to papillary
changes in their miRNA expression. We confirmed cancers that form their own homogenous group
previously reported up-regulations of miR-197, -346, (Borup et al. 2010). The common changes in miRNAs
-187, -221, -222, -224 and -155 in carcinoma and may also be interpreted as an indication that FC and
up-regulation of miR-339, -210, -328 and -342 in ACs share a common tumourigenic pathway that differs
adenoma (Weber et al. 2006, Nikiforova et al. 2008) from the papillary cancers. Moreover, loss of miRNAs is
and identified a number of previously undescribed likely to represent an early event in follicular neoplasia
thyroid miRNAs including miR-199b-5p, miR-144*, since the majority of the miRNAs are also down-
miR-199b-3p, miR-199a-5p, miR-144, miR-96, miR-182 regulated in adenoma. Taken together, the results
and miR-597, to mention the most striking. Of imply that the observed changes in miRNA signatures
particular significance, among the novel thyroid do have functional consequences in the tumours and
miRNAs, miR-199b-5p was found to be lost in the may participate in the tumourigenesis. This was further
carcinoma. Loss of miR-199b-5p (also known as miR- reinforced by the demonstration that miR-199b-5p
199b) has previously been described in chorioncarci-
noma (Chao et al. 2009) and in meduloblastoma, where
the loss increases metastasis (Garzia et al. 2009). MiR-96 4
NT
was among the few up-regulated miRNAs in the
35 FC
carcinoma and this is in line with urothelial carcinomas,
where miR-96 is a promising tumour marker in urine 3
(Yamada et al. 2010). MiR-182 was also noteworthy in
that increased expression of miR-182 has also been 25
observed in malignant melanomas and gliomas (Segura
2
et al. 2009, Jiang et al. 2010), where it is associated with
metastasis and poor prognosis. 15
However, in order to substantiate the biological
significance of the changed miRNAs, we performed a 1
weighted target identification (Gallagher et al. 2010),
05
followed by a molecular pathway analysis and an analysis
of the expression of the target mRNAs, since w80% of 0
miRNA regulations are reflected by changes in mRNA ABCC1 VSP26A
expression levels (Guo et al. 2010). Moreover, the Figure 5 Relative expression level of two predicted miR-199b-5p
function of miR-199b-5p and miR-144 was examined by targets by QRT-PCR. Relative expression level of two of the
introducing the two miRNAs in thyroid carcinoma cells. computational predicted targets of miR-199b-5p, VPS26A and
The pathway analyses depicted an enrichment ABCC1 examined in the tumour set (12 follicular carcinoma (FC)
of transcripts encoding proteins directly involved in and 10 normal thyroid (NT) samples) also used for miRNA
analysis. VPS26A and ABCC1 showed a relative increase of
thyroid carcinogenesis and tumourigenesis, and, in the 3- (PZ0.007) and 3.5 (PZ0.0008)-fold in FC vs NT respectively.
GSEA, we discovered a significant overlap with genes Blue colour bars represent NT and green ones represent FC.
A B
2 (11%)
3 (75%)
1 (98%)
1 (36%)
3 (0%)
2 (2%)
C 2 (9%) D 2 (13%)
1 (33%)
1 (10%)
3 (7%) 3 (12%)
N FA FC
Figure 6 Principal component analysis. (A) shows the projection of follicular carcinomas
(FC) and follicular adenomas (FA) and normal thyroid (NT) employing all miRNAs derived
from the microarray analysis. The projection of FC and FA employing the expression
values of only miR-1826 and miR-Eplus-1078 is seen in (B). (C) shows the projection of
FC and FA and NT employing all miRNAs derived from the qRT-PCR panels. In (D) the
projection of FC and FA employing the expression values of the 14 miRNAs that was
found to be the optimal signature for classification of FC is seen.
reduced cell-growth and that almost 30% of the platform respectively. The qRT-PCR platform provided
computational predicted targets were in fact down- a better separation of FA and FC than the microarray
regulated by pre-miR199b in the cultured thyroid platform, which is reflected by higher accuracy. From a
cells and correspondingly up-regulated in the thyroid clinical point of view, the predicted probabilities
carcinoma lacking the miRNA. MiR-144 had no
significant effect on proliferation and we cannot Table 4 Predictive probabilities of miRNA classifier. Predictive
probabilities for follicular adenoma (FA) and follicular carcinoma
exclude that loss of this miRNA reflects alterations (FC) samples using SVM classifier, Cost Z10, gamma Z0.01.
in the vascularisation of the carcinoma (Rasmussen
et al. 2010). Predictive Predictive
The tumours collected from consecutively referred Sample probability Sample probability
patients whose sex and age were in accordance with that
of larger epidemiological studies. Furthermore, there FC_01 0.04 FA_01 0.73
was no preponderance of oncogenic mutations in the FC_02 0.2 FA_02 0.66
FC_03 0.07 FA_03 0.84
tumour sets. To exploit if the miRNAs could be useful FC_04 0.41 FA_04* 0.45
to depict FC from adenoma, we generated a diagnostic FC_05 0.18 FA_05 0.9
signature from two different technical platforms, the FC_06* 0.67 FA_06 0.94
qRT-PCR assays was included as an independent FC_07 0.13 FA_07 0.93
FC_08* 0.62 FA_08 0.87
validation (Git et al. 2010). Although the results FC_09 0.3 FA_09 0.96
need to be confirmed by independent studies, the FC_10 0.46 FA_10 0.8
performance of either platform was acceptable, as the FC_11 0.22 FA_11 0.5
classifiers exhibited an NPV of 83 and 92% for FC_12 0.08 FA_12* 0.1
malignancies with the microarray and qRT-PCR-based *Misclassified samples (FA: nZ2, FC: nZ2)
derived from each individual sample are essential since Braun J, Hoang-Vu C, Dralle H & Huttelmaier S 2010 Downregulation
they provide a quantitative value (on the reliability) to of microRNAs directs the EMT and invasive potential of anaplastic
thyroid carcinomas. Oncogene 29 42374244. (doi:10.1038/onc.
the extent in the accuracy of the diagnosis based on the 2010.169)
classification. According to this, we found that most Chao A, Tsai CL, Wei PC, Hsueh S, Chao AS, Wang CJ, Tsai CN, Lee YS,
samples are classified with high accuracy. Wang TH & Lai CH 2009 Decreased expression of microRNA-199b
Since we showed that miRNA-based classification increases protein levels of SET (protein phosphatase 2A inhibitor)
of histopathological follicular thyroid specimens is in human choriocarcinoma. Cancer Letters 291 99107. (doi:10.
1016/j.canlet.2009.10.005)
possible, the next obvious step is to examine whether Chen YT, Kitabayashi N, Zhou XK, Fahey TJ III & Scognamiglio T 2008
it is feasible to implement miRNA-based classification as MicroRNA analysis as a potential diagnostic tool for papillary
an additional preoperative diagnostic tool. Taking the thyroid carcinoma. Modern Pathology 21 11391146. (doi:10.1038/
limited sensitivity and reproducibility of the histopatho- modpathol.2008.105)
Curado MP & Edwards B 2007 Cancer Incidence in Five Continents Vol. IX.
logical diagnosis into account, the consistency between
MP Curada, B Edwards, HR Shin, H Storm, J Ferlay, M Heanue &
miRNA-based classification and the pathological diag- P Boyle (eds). IARC Scientific Publication. No. 160, Lyon: IARC.
nosis is surprisingly high. This could reflect the fact Dudoit S, Shaffer JP & Boldrick JC 2003 Multiple hypothesis testing in
that all samples were examined by the same dedicated microarray experiments. Statistical Science 18 71103. (doi:10.1214/
endocrine pathologist. Studies of inter-observer ss/1056397487)
Faquin WC 2008 The thyroid gland: recurring problems in histologic
variations amongst pathologists in assessment of
and cytologic evaluation. Archives of Pathology and Laboratory Medicine
follicular lesions have demonstrated an observer 132 622632.
variation for FC of 27%, where the carcinomas tended Franc B, de la Salmoniere P, Lange F, Hoang C, Louvel A, de
to be misdiagnosed as adenomas (Hirokawa et al. Roquancourt A, Vilde F, Hejblum G, Chevret S & Chastang C 2003
2002, Kakudo et al. 2002). In a similar study an overall Interobserver and intraobserver reproducibility in the histopathol-
ogy of follicular thyroid carcinoma. Human Pathology 34 10921100.
agreement between American and Japanese path-
(doi:10.1016/S0046-8177(03)00403-9)
ologists of 33 and 52% respectively, was found Friedman RC, Farh KK, Burge CB & Bartel DP 2009 Most mammalian
(Hirokawa et al. 2002). mRNAs are conserved targets of microRNAs. Genome Research 19
All results considered, we conclude that thyroid 92105. (doi:10.1101/gr.082701.108)
follicular neoplasia is accompanied by major changes Gallagher IJ, Scheele C, Keller P, Nielsen AR, Remenyi J, Fischer CP,
Roder K, Babraj J, Wahlestedt C, Hutvagner G et al. 2010 Integration
in miRNA expression that may be directly implicated in of microRNA changes in vivo identifies novel molecular features of
malignant transformation and may facilitate diagnosis muscle insulin resistance in type 2 diabetes. Genome Medicine 2 9.
of follicular thyroid cancer. (doi:10.1186/gm130)
Garzia L, Andolfo I, Cusanelli E, Marino N, Petrosino G, De Martino D,
Esposito V, Galeone A, Navas L, Esposito S et al. 2009 MicroRNA-
Supplementary data 199b-5p impairs cancer stem cells through negative regulation of
HES1 in medulloblastoma. PLoS ONE 4 e4998. (doi:10.1371/
This is linked to the online version of the paper at http://dx.doi.org/ journal.pone.0004998)
10.1530/JME-11-0039. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S,
Ellis B, Gautier L, Ge Y, Gentry J et al. 2004 Bioconductor: open
software development for computational biology and bioinfor-
matics. Genome Biology 5 R80. (doi:10.1186/gb-2004-5-10-r80)
Declaration of interest Git A, Dvinge H, Salmon-Divon M, Osborne M, Kutter C, Hadfield J,
Bertone P & Caldas C 2010 Systematic comparison of microarray
The authors declare that there is no conflict of interest that could be profiling, real-time PCR, and next-generation sequencing tech-
perceived as prejudicing the impartiality of the research reported. nologies for measuring differential microRNA expression. RNA 16
9911006. (doi:10.1261/rna.1947110)
Guo H, Ingolia NT, Weissman JS & Bartel DP 2010 Mammalian
Funding microRNAs predominantly act to decrease target mRNA levels.
Nature 466 835840. (doi:10.1038/nature09267)
This study was supported by the Agnes and Knut Mrk Foundation, He H, Jazdzewski K, Li W, Liyanarachchi S, Nagy R, Volinia S, Calin GA,
the Research Foundation of Herlev University Hospital, the Toyota Liu CG, Franssila K, Suster S et al. 2005 The role of microRNA genes
Foundation and the Novo Nordisk Foundation. in papillary thyroid carcinoma. PNAS 102 1907519080. (doi:10.
1073/pnas.0509603102)
Hegedus L, Bonnema SJ & Bennedbaek FN 2003 Management of
simple nodular goiter: current status and future perspectives.
References Endocrine Reviews 24 102132. (doi:10.1210/er.2002-0016)
Hirokawa M, Carney JA, Goellner JR, DeLellis RA, Heffess CS, Katoh R,
Bartel DP 2004 MicroRNAs: genomics, biogenesis, mechanism, and Tsujimoto M & Kakudo K 2002 Observer variation of encapsulated
function. Cell 116 281297. (doi:10.1016/S0092-8674(04)00045-5) follicular lesions of the thyroid gland. American Journal of Surgical
Bartel DP 2009 MicroRNAs: target recognition and regulatory Pathology 26 15081514. (doi:10.1097/00000478-200211000-
functions. Cell 136 215233. (doi:10.1016/j.cell.2009.01.002) 00014)
Borup R, Rossing M, Henao R, Yamamoto Y, Krogdahl A, Godballe C, Jiang L, Mao P, Song L, Wu J, Huang J, Lin C, Yuan J, Qu L, Cheng SY &
Winther O, Kiss K, Christensen L, Hogdall E et al. 2010 Molecular Li J 2010 miR-182 as a prognostic marker for glioma progression
signatures of thyroid follicular neoplasia. Endocrine-Related Cancer and patient survival. American Journal of Pathology 177 2938.
17 691708. (doi:10.1677/ERC-09-0288) (doi:10.2353/ajpath.2010.090812)
Kakudo K, Katoh R, Sakamoto A, Asa S, DeLellis RA, Carney JA, Schmittgen TD & Livak KJ 2008 Analyzing real-time PCR data by
Naganuma H, Kameyama K & Takami H 2002 Thyroid gland: the comparative C(T) method. Nature Protocols 3 11011108.
international case conference. Endocrine Pathology 13 131134. (doi:10.1038/nprot.2008.73)
(doi:10.1385/EP:13:2:131) Segura MF, Hanniford D, Menendez S, Reavie L, Zou X, Alvarez-Diaz S,
Kroll TG, Sarraf P, Pecciarini L, Chen CJ, Mueller E, Spiegelman BM & Zakrzewski J, Blochin E, Rose A, Bogunovic D et al. 2009 Aberrant
Fletcher JA 2000 PAX8PPARgamma1 fusion oncogene in human miR-182 expression promotes melanoma metastasis by repressing
thyroid carcinoma [corrected]. Science 289 13571360. (doi:10. FOXO3 and microphthalmia-associated transcription factor.
1126/science.289.5483.1357) PNAS 106 18141819. (doi:10.1073/pnas.0808263106)
Nikiforova MN, Tseng GC, Steward D, Diorio D & Nikiforov YE 2008 Simon R, Lam A, Li MC, Ngan M, Menenzes S & Zhao Y 2007 Analysis
MicroRNA expression profiling of thyroid tumours: biological of gene expression data using BRB-ArrayTools. Cancer Informatics 3
significance and diagnostic utility. Journal of Clinical Endocrinology 1117.
and Metabolism 93 16001608. (doi:10.1210/jc.2007-2696) Tusher VG, Tibshirani R & Chu G 2001 Significance analysis of
Pallante P, Visone R, Ferracin M, Ferraro A, Berlingieri MT, Troncone microarrays applied to the ionizing radiation response. PNAS 98
G, Chiappetta G, Liu CG, Santoro M, Negrini M et al. 2006 51165121. (doi:10.1073/pnas.091062498)
MicroRNA deregulation in human thyroid papillary carcinomas. Vapnik V 1998 Statistical Learning Theory. New York, NY, USA: Wiley-
Endocrine-Related Cancer 13 497508. (doi:10.1677/erc.1.01209) Interscience.
Platt JC 1999 Probabilities for SV machines. In Advances in Large Weber F, Teresi RE, Broelsch CE, Frilling A & Eng C 2006 A limited set
Classifiers, pp 7285. Eds PJ Bartlett, B Scholkopf, D Schuurmans & of human MicroRNA is deregulated in follicular thyroid carcinoma.
AJ Simola. Cambridge, MA: MIT Press. Journal of Clinical Endocrinology and Metabolism 91 35843591.
Rasmussen KD, Simmini S, Abreu-Goodger C, Bartonicek N, Di GM,
(doi:10.1210/jc.2006-0693)
Bilbao-Cortes D, Horos R, Von LM, Enright AJ & OCarroll D 2010 The
Wettenhall JM & Smyth GK 2004 limmaGUI: a graphical user interface
miR-144/451 locus is required for erythroid homeostasis. Journal of
for linear modeling of microarray data. Bioinformatics 20 37053706.
Experimental Medicine 207 13511358. (doi:10.1084/jem.20100458)
(doi:10.1093/bioinformatics/bth449)
R Development Core Team. R: A Language and Environment for
Yamada Y, Enokida H, Kojima S, Kawakami K, Chiyomaru T, Tatarano S,
Statistical Computing, R Foundation for Statistical Computing,
Yoshino H, Kawahara K, Nishiyama K, Seki N et al. 2010 MiR-96
Vienna, Austria, ISBN 3-900051-07-0. URL. Downloads. 2007.
and miR-183 detection in urine serve as potential tumour markers
Rosenfeld N, Aharonov R, Meiri E, Rosenwald S, Spector Y, Zepeniuk
M, Benjamin H, Shabes N, Tabak S, Levy A et al. 2008 MicroRNAs of urothelial carcinoma: correlation with stage and grade, and
accurately identify cancer tissue origin. Nature Biotechnology 26 comparison with urinary cytology. Cancer Science 102 522529.
462469. (doi:10.1038/nbt1392) (doi:10.1111/j.1349-7006.2010.01816.x)
Schmid KW & Farid NR 2006 How to define follicular thyroid carcinoma?
Virchows Archiv 448 385393. (doi:10.1007/s00428-006-0162-0)
Schmittgen TD 2008 Regulation of microRNA processing in develop- Received in final form 23 September 2011
ment, differentiation and cancer. Journal of Cellular and Molecular Accepted 1 November 2011
Medicine 12 18111819. (doi:10.1111/j.1582-4934.2008.00483.x) Made available online as an Accepted Preprint 2 November 2011