20 Pages
English

Genome-wide expressions in autologous eutopic and ectopic endometrium of fertile women with endometriosis

-

Gain access to the library to view online
Learn more

Description

In order to obtain a lead of the pathophysiology of endometriosis, genome-wide expressional analyses of eutopic and ectopic endometrium have earlier been reported, however, the effects of stages of severity and phases of menstrual cycle on expressional profiles have not been examined. The effect of genetic heterogeneity and fertility history on transcriptional activity was also not considered. In the present study, a genome-wide expression analysis of autologous, paired eutopic and ectopic endometrial samples obtained from fertile women (n = 18) suffering from moderate (stage 3; n = 8) or severe (stage 4; n = 10) ovarian endometriosis during proliferative (n = 13) and secretory (n = 5) phases of menstrual cycle was performed. Methods Individual pure RNA samples were subjected to Agilent’s Whole Human Genome 44K microarray experiments. Microarray data were validated (P < 0.01) by estimating transcript copy numbers by performing real time RT-PCR of seven (7) arbitrarily selected genes in all samples. The data obtained were subjected to differential expression (DE) and differential co-expression (DC) analyses followed by networks and enrichment analysis, and gene set enrichment analysis (GSEA). The reproducibility of prediction based on GSEA implementation of DC results was assessed by examining the relative expressions of twenty eight (28) selected genes in RNA samples obtained from fresh pool of eutopic and ectopic samples from confirmed ovarian endometriosis patients with stages 3 and 4 (n = 4/each) during proliferative and secretory (n = 4/each) phases. Results Higher clustering effect of pairing (cluster distance, cd = 0.1) in samples from same individuals on expressional arrays among eutopic and ectopic samples was observed as compared to that of clinical stages of severity (cd = 0.5) and phases of menstrual cycle (cd = 0.6). Post hoc analysis revealed anomaly in the expressional profiles of several genes associated with immunological, neuracrine and endocrine functions and gynecological cancers however with no overt oncogenic potential in endometriotic tissue. Dys-regulation of three (CLOCK, ESR1, and MYC) major transcription factors appeared to be significant causative factors in the pathogenesis of ovarian endometriosis. A novel cohort of twenty-eight (28) genes representing potential marker for ovarian endometriosis in fertile women was discovered. Conclusions Dysfunctional expression of immuno-neuro-endocrine behaviour in endometrium appeared critical to endometriosis. Although no overt oncogenic potential was evident, several genes associated with gynecological cancers were observed to be high in the .

Subjects

Informations

Published by
Published 01 January 2012
Reads 10
Language English
Document size 1 MB

Khan et al. Reproductive Biology and Endocrinology 2012, 10:84
http://www.rbej.com/content/10/1/84
RESEARCH Open Access
Genome-wide expressions in autologous eutopic
and ectopic endometrium of fertile women with
endometriosis
1 1 2 1*Meraj A Khan , Jayasree Sengupta , Suneeta Mittal and Debabrata Ghosh
Abstract
Background: In order to obtain a lead of the pathophysiology of endometriosis, genome-wide expressional
analyses of eutopic and ectopic endometrium have earlier been reported, however, the effects of stages of severity
and phases of menstrual cycle on expressional profiles have not been examined. The effect of genetic
heterogeneity and fertility history on transcriptional activity was also not considered. In the present study, a
genome-wide expression analysis of autologous, paired eutopic and ectopic endometrial samples obtained from
fertile women (n=18) suffering from moderate (stage 3; n=8) or severe (stage 4; n=10) ovarian endometriosis
during proliferative (n=13) and secretory (n=5) phases of menstrual cycle was performed.
Methods: Individual pure RNA samples were subjected to Agilent’s Whole Human Genome 44K microarray
experiments. Microarray data were validated (P<0.01) by estimating transcript copy numbers by performing real
time RT-PCR of seven (7) arbitrarily selected genes in all samples. The data obtained were subjected to differential
expression (DE) and differential co-expression (DC) analyses followed by networks and enrichment analysis, and
gene set enrichment analysis (GSEA). The reproducibility of prediction based on GSEA implementation of DC results
was assessed by examining the relative expressions of twenty eight (28) selected genes in RNA samples obtained
from fresh pool of eutopic and ectopic samples from confirmed ovarian endometriosis patients with stages 3 and 4
(n=4/each) during proliferative and secretory (n=4/each) phases.
Results: Higher clustering effect of pairing (cluster distance, cd=0.1) in samples from same individuals on
expressional arrays among eutopic and ectopic samples was observed as compared to that of clinical stages of
severity (cd=0.5) and phases of menstrual cycle (cd=0.6). Post hoc analysis revealed anomaly in the expressional
profiles of several genes associated with immunological, neuracrine and endocrine functions and gynecological
cancers however with no overt oncogenic potential in endometriotic tissue. Dys-regulation of three (CLOCK, ESR1,
and MYC) major transcription factors appeared to be significant causative factors in the pathogenesis of ovarian
endometriosis. A novel cohort of twenty-eight (28) genes representing potential marker for ovarian endometriosis
in fertile women was discovered.
Conclusions: Dysfunctional expression of immuno-neuro-endocrine behaviour in endometrium appeared critical to
endometriosis. Although no overt oncogenic potential was evident, several genes associated with gynecological
cancers were observed to be high in the expressional profiles in endometriotic tissue.
Keywords: Computational analysis, Endometriosis, Differential display, Gene expression, GSEA
* Correspondence: debabrata.ghosh1@gmail.com
1
Department of Physiology, All India Institute of Medical Sciences, New Delhi,
India
Full list of author information is available at the end of the article
© 2012 Khan et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.Khan et al. Reproductive Biology and Endocrinology 2012, 10:84 Page 2 of 20
http://www.rbej.com/content/10/1/84
Background Gynecology – AIIMS and showing evidence of endome-
Endometriosis is a complex disorder involving pathogen- triotic lesions, adhesions and endometriotic cyst were
esis and clinical presentation of ectopically implanted selected to participate in the present study. All the
endometrium [1]. It is generally assumed that elucidation patients were reportedly fertile and referred from the
of molecular expressional specificities of eutopic and ec- Pain Clinics, and had voluntarily agreed to donate their
topic endometrium may provide leads towards a better samples after understanding the purpose of the proposed
understanding of the pathophysiology of the disorder [2]. study. Signed informed consent was obtained from each
To this end, several studies exploring the differential ex- participant of this study. As shown in Figure 1, twenty-
pression of genes between autologous eutopic and ec- six (26) normally cycling and proven fertile women (age:
topic endometrium from patients with endometriosis 24–45 y) with history of pregnancy and with at least one
have been reported, however, with no specific compari- living biological offspring, and body mass indices within
2
son for stages of severity, fertility history and phases of normal ranges (20–22 k/m ) having ovarian endometri-
menstrual cycle [3-7], except a recent report [8]. More osis were selected for the present study. Confirmation of
over, it is notable that two types of endometriosis, namely ovarian endometriosis and exclusion of other types of
ovarian endometriosis and peritoneal endometriosis re- endometriosis was achieved from reports of pelvic im-
portedly show differential characteristics [4,9]. Further- aging based on ultrasound, MRI and/or diagnostic lapar-
more, there is evidence to support the idea that deep oscopy as described elsewhere [8]. Severity stages 3 and
infiltrating endometriosis also show differential patho- 4 of the disease condition were defined at the time of
physiology as compared to ovarian and peritoneal endo- surgical laparoscopy [8] according to rASRM protocol
metriosis [10,11]. In the present study, we examined a [12]. Selected subjects (n=18; shown as ‘E’ in Additional
genome-wide large-scale transcript survey of autologous, file 1: Table S1) contributed their eutopic (shown as ‘A’
paired eutopic and ectopic endometrial samples obtained in Figure 2) and ectopic (shown as ‘B’ in Figure 2) sam-
from fertile women suffering from moderate to severe ples during proliferative (days 9–14) phase (n=17) and
ovarian endometriosis, and excluded cases of peritoneal secretory (days 17–24) phase (n=8) of menstrual cycle
endometriosis and deep infiltrating endometriosis. We as described elsewhere [8]. Additional paired samples
assumed that the present model of subject selection collected from different group of subjects (n=8; shown
would reduce the impact of biological noise derived from as ‘Ep’ in Additional file 1: Table S1) with confirmed
genetic and pathogenetic heterogeneity and subfertility- ovarian endometriosis as described above and with clas-
associated variability on the transcriptional activity in the sified menstrual (proliferative: n=4; secretory: n=4)
target tissue. We report here for the first time that clus- phases and severity stages 3 (n=4) and 4 (n=4) were
tering effect of expressional arrays among eutopic and employed for validating the prediction as described
ectopic samples was higher for genetic homogeneity below. A small piece from each specimen was processed
(i.e. pairing of eutopic and ectopic samples from same for chemical fixation in neutral buffered formaldehyde
individuals) than that of clinical stages of severity and (4%, w/v) for subsequent confirmation of phase of cycle,
phases of menstrual cycle. Based on the present state of pathology and cell types from eutopic and ec-
transcriptomics data, we have also hypothesized that dys- topic samples, and the residual portions were trans-
functional immuno-neuro-endocrine behaviour in endo- ported on ice to the laboratory within 10 minutes of
metrium was associated with the pathogenesis of collection for further processing for RNA extraction.
endometriosis. Additionally, we did not observe an overt
oncogenic potential in the expressional profiles in endo- Experimental procedure
metriotic tissue, however, several genes associated with The methodological details of RNA extraction followed
gynecological cancers were highly expressed in the euto- by the estimation of its yield and purity using standard
pic and ectopic endometrium. Finally, a novel cohort of electrophoretic and spectrophometric protocols and its
28 genes was identified, the expression of which carry po- RIN score using the Agilent 2100 Bioanalyzer, RNA
tential marker value for endometriosis in fertile women. 6000 Nano LabChip kit and Agilent 2100 Expert Soft-
A flow diagram of the experimental design is shown in ware (Agilent Technologies, Santa Clara, CA, USA) have
Figure 1. been given elsewhere [8,13]. Individual RNA samples
from eutopic and ectopic tissue samples (n=18) from
Methods confirmed stages 3 (n=8) and 4 (n=10) collected during
Subjects and tissue samples proliferative (n=13) and secretory (n=5) phases and
The present study was approved by the Ethics having RIN scores >8.0 were subjected to whole tran-
Committee on the Use of Human Subjects, All India scriptome array experiment using the Agilent Whole
Institute of Medical Sciences (AIIMS), New Delhi. The Human Genome 60-mer 4X44K microarray according to
patients enrolled in the Department of Obstetrics and the manufacturer’s recommendations. Thus, seven (7)Khan et al. Reproductive Biology and Endocrinology 2012, 10:84 Page 3 of 20
http://www.rbej.com/content/10/1/84
AimIdentification of expressional pathways and specific cohort of genes
associated with the pathogenesis of endometriosis
Out Patient Departments
Subject
Pain clinics followed by
Obstetrics and GynecologyObstetrics and Gynecology
Subject
recruitment
Pelvis Imaging: Ultrasound, MRI, Laparoscopy
Diagnosis
Subject selection
based on clinical Selection
history and data
Tissue Surgical laparoscopy: Staging (rASRM, 1997);
collection & Paired eutopic(A) and ectopic(B) tissue collection
stagingstaging
Sample classification
based on clinical data:
Fertile with known
Classificationseverity stages and
cycle phases
Sample (A and B) processing Methodology
Tissue fixation, paraffin Total RNA extraction: Quality
blocking, and histology check and quantification
RT-PCR
Sample selection [28 genes]
(n=26, paired)(n=8, paired)(n=8, paired)
Whole genome microarrayhybridization
RT-PCR Analysis and image analysis (n=18, paired) [7 genes]
Data
Array validationanalysis
Cluster analysis to identify co- Differential expression (DE) analysis:
expressed genes and differential Effects of tissue location, severity
stage and cycle phaseanalysis of co-expressed (DC) genes
Post-hoc functional analysis using gene ontology (GO) pathways & networks Functional
interpretation
and gene set enrichment analysis (GSEA)
Expressional cohort of marker genes
Figure 1 Flow diagram of the experimental design showing overall aim and work plan of the present study.
samples could not be used either for insufficient RNA Clara, CA, USA) for further analysis. Pearson’s correl-
yield or RIN scores (see Additional file 1: Table S1 for ation coefficients done to assess the reliability of data
the subject details of the selected samples). Hybridized obtained from two separate hybridization runs for same
arrays were scanned with Agilent’s G2505B microarray RNA preparation for four (4) eutopic and ectopic sam-
scanner system and the raw data were imported into ples confirmed the reproducibility assurance (P<0.01)
GeneSpring 11.5.0 software (Agilent Technologies, Santa among hybridizations. Analysis of the data retrieved
Prediction analysisKhan et al. Reproductive Biology and Endocrinology 2012, 10:84 Page 4 of 20
http://www.rbej.com/content/10/1/84
(A)
14000
11897
(10116)12000
919110000
(5361)
8189
(7274)
8000 71486969 (6128)(3680) 6156
(4927) 5687
(3345)6000 4985
(4140)
4000 3186
(1456) 2670
(2375) 2246
(2032)
2000
960838 (848)462 380(589) 42 1360 7 02 (418) (359) 56 0 1
(2) (37) (7) (122) (0) (51) (1)(0) (0)
0
2.0 1.5 1.0 0.5 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Expression (log )2
(B)
Figure 2 General descriptive characteristics of expressions in eutopic and ectopic endometrium. (A) Histogram of frequency distribution
of probes (genes) for different groups of expression levels (in log ) in eighteen (18) autologous, paired eutopic (blank bar) and ectopic2
(hatched bar) endometrial samples. (B) Two-way representation of unsupervised hierarchical cluster analysis (HCA) of the expression levels (in
logarithmic scale) of all the target probes/genes (Y-axis) in each sample (each column), eutopic labelled as A and ectopic labelled as B from
all subjects (n=18) and their clustering based on expressional distance (Pearson correlation coefficient) between samples in dendrogram
formation (X-axis). Each horizontal line represents a single probe, and each column represents a single sample. Relative expression of each
probe is colour-coded: high (red) and low (blue), as indicated in the colour legend. Categorical annotations of each sample are shown in the
X-axis. The samples cluster by cycle phase and severity stages, as shown by the bar at the bottom of the heat map: proliferative phase (black
bar), secretory phase (crossed white bar), stage 3 (red crossed bar) and stage 4 (blank red bar). A, eutopic; B, ectopic; P, proliferative phase; S,
secretory phase; 3, stage 3; 4, stage 4.
from separate chips with the same RNA samples yielded analysis of expression arrays were performed with the
QC statistics highly concordant with that of the manu- help of GeneSpring software 11.5.0. Analysis of variance
facturer, and it revealed more than 95% confidence level. followed by pair-wise differential (>3-fold at P<0.01) ex-
pression (DE) for specific genes between eutopic and ec-
Data analysis topic samples, as well as, between proliferative and
Unsupervised and supervised hierarchical clustering ana- secretory phases, and between clinical stage 3 and stage
lysis (HCA), and non-hierarchical K-mean cluster 4 of severity for eutopic endometrium, and for ectopic
Number of probes (genes)Khan et al. Reproductive Biology and Endocrinology 2012, 10:84 Page 5 of 20
http://www.rbej.com/content/10/1/84
endometrium, respectively were done using multiple individual RNA samples obtained from eight (8) add-
comparison tests as described elsewhere [14]. itional subjects giving paired eutopic and ectopic samples
with confirmed endometriosis stages 3 (n=4) and 4
Post-hoc analysis (n=4) during proliferative (n=4) and secretory (n=4)
Networks and enrichment analysis were done using gene were examined using real time RT-PCR technology. The
lists obtained from the above analyses and based on a subject details are shown in Additional file 1: Table S1.
priori setting of a cut-off threshold (pFDR(p)=0.05) with The genes selected based on GSEA implementation of
the help of the GeneSpring11.5.0 software and Metacore DC results were employed to test the predictability func-
platform (GeneGo, St. Joseph, MI, USA). The K-mean tion of the expression of those genes. The details of RNA
clusters were further used for differential co-expression methodologies are given above. GAPDH was selected as
(DC) analyses and analyzed in terms of Gene Ontology an endogenous control based on its observed expres-
(GO) enriched categories using GeneSpring11.5.0 soft- sional consistency in arrays on data analysis. All primers
ware. Gene Set Enrichment Analysis (GSEA) version 3.7 were designed on the Beacon Designer software7.0 (Lab-
was applied to each of the K-mean clusters independently ware Scientific Inc., Milipitas, CA, USA) based on SYBR
to examine at FDR≤0.25 for not less than 10 genes for a green chemistry and obtained from Qiagen (Cologne,
set with a maximum of 1000 permutation whether pre- Germany). QuantiTect Reverse Transcription kit for
annotated BROAD gene sets [15]: C1 (cytogenetic sets), cDNA synthesis and QuantiFast SYBR green PCR kit for
C2 (functional sets), C3 (regulatory sets), C4 (cancer PCR amplification from Qiagen (Cologne, Germany)
neighborhood sets), and C5 (gene ontology sets) could were used according to the protocol given by the manu-
identifyanyinterestinginformationintheDCsets[16]. facturer. The estimates of relative expression ratios be-
tween groups and copy numbers for target transcripts in
Quantification of candidate gene expression by real time complex RNA samples were obtained as described above.
RT-PCR
In order to validate the microarray data, relative expres- Results
sion of arbitrarily chosen seven (7) selected genes (ATX, The data sets are available at NCBI-GEO website [19].
DDHD1, DYNLT1, FTH1, LAMR1, MIER2, and WDR87) A distribution histogram of the number of probes and
in eutopic and ectopic samples collected from all genes for different ranges of expression in autologous,
patients were performed using Taqman multiplexing paired eutopicand ectopicsamplesobtainedfrom eighteen
technology on iCycler iQTm real time RT-PCR detection (18) fertile women with confirmed ovarian endometriosis
system (BioRad, Hercules, CA, USA). GAPDH was is shown in Figure 2A. Total numbers and per cent esti-
selected as an endogenous control based on its observed mates of probes/genes expressed in eutopic and ectopic
expressional consistency in arrays on data analysis. Pri- samples in optimized scale are shown in Table 1. On aver-
mers and probes were designed on Beacon Designer age, ~75% and ~50% of expressed genes showed marked
software7 (Labware Scientific Inc., Milipitas, CA, USA) signal in eutopic and ectopic samples, respectively.
and obtained from Qiagen (Cologne, Germany) (see
Additional file 2: Table S2 for the details). The ratio of Table 1 Descriptive analysis of array data
estimated efficiency of the primers for the selected genes Parameter Estimate
and GAPDH was ~1.0. An optimized kit (QuantiTect Per chip Per cent
multiplex PCR kit, Qiagen, Cologne, Germany) was used
Number of probes 41000
to synthesize cDNA from respective RNA (5 μg) sam-
(genes) (29421)
ples. Relative expression ratios between groups were cal-
a-ΔΔCt Number of hybridized probes (genes)culated by using 2 method [17]. Quantification of
Eutopic 35646 87copy numbers for target transcripts in complex RNA
samples was obtained as described elsewhere [18]. Com- (25987) (88)
parison between fold change data obtained from real Ectopic 35587 87
time RT-PCR and microarray image analysis for selected
(26222) (89)
seven (7) genes revealed a high degree of concordance b
Number of high expressed probes (genes)
and pattern similarity in expression profile. Concordance
Eutopic 23267 65correlation test between real time RT-PCR based quanti-
(19168) (74)tative data and microarray data for the seven (7) genes
showed a high degree of correlation (P<0.01) [8]. Ectopic 15912 45
In order to test the reproducibility of prediction (13681) (52)
derived from analysis of microarray results, the relative aHybridization signal more than mean optimized background signal±2SD.
bexpressions of twenty eight (28) selected genes in >0 in normalized log scale.2Khan et al. Reproductive Biology and Endocrinology 2012, 10:84 Page 6 of 20
http://www.rbej.com/content/10/1/84
Unsupervised HCA yielded marked segregation of samples was only moderate in samples which were classified based
into two major clustering branches with clustering cohe- on either severity stages (cd: 0.5) or phases of menstrual
sionbeinghighest(clusterdistance,cd:0.1)betweenpaired cycle (cd: 0.6) (Fig. 2B). Supervised HCA revealed that the
samples from same subjects. However, clustering cohesion ectopic location of tissue had a higher clustering effect (cd:
(A)
Eutopic-to-ectopic 7
34Gene Fold Change 48
3Symbol Stage 3 Stage 4
LAMC2 4.3 3.5 3 3452
Stage 3RASEF 3.7 6.9 Stage 4
TACSTD2 3.1 3.1
20 113
Gene Fold Change 12320
Proliferative SecretorySymbol
EGR3 3.7 -3.4 4 139129
ERBB3 3.2 3.7 Proliferative Secretory
LAMC2 3.4 5.7
MATN4 3.1 5.1
230 714
(B) 50 19
Eutopic 230 50 683 Ectopic
Clinical stages 3-to-4
Fold Change Fold Change Fold Change
Gene Gene Gene
Symbol Eutopic Ectopic Eutopic Ectopic Symbol Eutopic EctopicSymbol
ALAS1 -5.9 13.7 PPFIA1 4.9 6.4FAM93B 17.9 10.8
ALDH1L2 14.2 14.8 PRR5 -5.6 -7.3FAM154B 6.0 13.6
ALMS1P 7.6 7.9 FBXO9 6.0 10.2 RBMX 6.4 6.3
FLOT1 9.2 9.3AP1B1 12.9 6.8 RFX5 7.7 7.5
GJC3 9.1 19.9 C1QTNF3 7.0 7.1 S1PR5 7.3 15.5
GK5 5.2 5.3
C11orf64 17.4 20.9 SOCS5 9.5 9.9
GOT2 7.1 6.1
C6orf 10.9 33.2 STAT2 6.4 5.6
HDHD1A 6.1 3.7
C20orf12 4.7 3.8 TBCEL 5.9 6.3
KCNK12 15.5 9.3
CALCOCO2 4.5 6.1 TP53INP2 3.9 8.2
KRT222 6.0 10.3
CCNF 9.6 12.8 WBSCR17 5.4 3.8
MAGEA10 9.5 13.1CCNT2 4.0 5.6 ZNF135 9.0 4.4
NFIB 6.9 8.3CSRNP1 5.5 14.4 ZNF257 7.8 17.7
NOB1 5.0 6.0DAGLB 3.1 6.5 ZNF274 6.3 15.2
NOS1AP 12.9 13.8DNAH7 7.5 14.2 ZNF343 4.7 9.0
NRM 6.7 4.0EAF1 6.9 12.8 ZNF551 3.9 4.4
EFCAB6 7.9 16.8 NT5C1B 16.5 34.0 ZRANB2 4.1 6.0
EVC2 7.2 7.8 PKNOX2 5.2 8.1
(C)
Proliferative-to-secretory phases
21 Gene Fold Change
9558Eutopic EctopicSymbol
4ADAM8 -3.6 -3.1 75 91
DCAF4 3.5 -3.3 Eutopic Ectopic
PDLIM5 -3.7 -3.9
-3.9UGDH 4.9
Figure 3 Venn analysis of distribution of differentially expressed (DE) genes in eutopic-to-ectopic analysis. Distribution of DE genes in
(A) eutopic and ectopic samples of stage 3 and stage 4, and proliferative and secretory phases, (B) stage 3-to-stage 4 for eutopic and ectopic
samples, and (C) proliferative-to-secretory phases for eutopic and ectopic samples. Common genes among comparative groups are detailed in
respective tables along with the vector of regulation and fold changes. The number of genes with relative up-regulation and down-regulation
are shown by respective arrows. For details of DE genes, see Additional file 3: Table S3. Note that the areas in the Venn distribution analysis are
not drawn to scale.Khan et al. Reproductive Biology and Endocrinology 2012, 10:84 Page 7 of 20
http://www.rbej.com/content/10/1/84
0.2) than that of phases of cycle (cd: 0.3), but not than that clusters showed overt patterns for menstrual cycle
oftheclinicalstagesofseverity(cd:0.1). phases and severity stages. A large number of genes
belonging to cluster 2 (K2) showed over-expression in
Differential expression (DE) severity stage 4 secretory phase endometrium (Fig. 4B).
Additional file 3: Table S3 gives the list of the genes The co-expressed genes in cluster 3 (K3) and cluster 4
along with their differential expression (DE) patterns (K4) showed very similar patterns with an overall higher
under different categories based on expressional arrays expression in stage 3 as compared to stage 4 endomet-
in autologous, paired eutopic and ectopic samples rium samples irrespective of cycle phases.
obtained from 18 fertile women with ovarian endometri- Table 4 shows the pathways-based enrichment analysis
osis. Figure 3 shows the number of genes with DE in dif- of groups of genes in four (4) K-mean clusters revealing
ferent categories of comparison and the lists of common differential co-expression (DC) profiles between paired
genes in it. Table 2 highlights the enriched categories of eutopic and ectopic endometrial tissues. It essentially
pathways for the common genes from above-mentioned substantiated the observation obtained from DE analysis
DE analysis between eutopic and ectopic endometrium. that transcriptomic signals related to cell cycle, signal
It appeared that different signaling pathways associated transduction, cytoskeleton remodeling, apoptosis and
with immune response, several neuronal processes, and survival, chemotaxis, cell adhesion, and immune re-
ERBB family signaling pathways were commonly sponse were affected in the pathogenesis process of
selected. A summary of DE analysis of the non-common endometriosis.
genes showing differential display under different cat-
egories and their enrichment analysis are shown in Gene-set enrichment analysis (GSEA)
Table 3. Collectively, it appeared that informational flow Table 5 provides a summary of the results of GSEA im-
for a wide array of pathways involving cellular signaling, plementation on co-expressed genes with differential
apoptosis and survival, cytoskeleton remodeling, chemo- display (DC) in the four K-mean clusters. In K4, one (1)
taxis, cell adhesion, immune response and several neuro- cytoband i.e. C1 set and two (2) gene ontology i.e. C5
physiological processes were affected. sets were selected. More over, three (3) DC gene sets –
one each in K1, K2 and K4, respectively – were selected
K-mean clusters and differential co-expression (DC) under BROAD regulatory gene motif sets, C3. It is not-
As shown in Figure 4, K-mean cluster analysis identified able that two (2) selected regulatory motif sets belonging
four clusters of expression patterns and profiles based to K1 and K2 were significantly (p<0.0001) associated
on normalized hybridization signals for all expressed with ectopic sample as evident from their negative nor-
genes in all samples. The genes in cluster 1 (K1) did not malized enrichment scores (NES). Further, four (4) DC
show any specific expression pattern, while other three gene sets – two (2) each in K1 and K4, respectively –
aTable 2 Enriched common genes showing differential changes under different categories of comparisons
Description of comparison Gene in enriched Enriched pathways (p-value)
(Number of genes) category (Gene symbol)
Eutopic-to-ectopic ERBB3 Activation of astroglia proliferation (0)
bStage 3 & Stage 4 (3) CDK5 mediated cell death and survival (0)
bProliferative & Secretory (4) ERBB family signaling (0)
Membrane bound ESR1 interaction with (<0.01)
growth factor signaling
Ligand-independent activation of ESR1 and ESR2 (<0.01)
ERBB3, LAMC2 Alpha6/beta-4 integrins in carcinoma progression (<0.01)
Stages 3-to-4 STAT2 Immune response involving IL-15 and IFN signaling (<0.02)
cEutopic & Ectopic (50)
Proliferative-to-Secretory Angiotensin signaling via STATs (<0.03)
dEutopic & Ectopic (4) NOS1AP nNOS signaling in neuronal process (<0.03)
AP1B1 Immune response involving regulation of (<0.04)
T cell function by CTLA-4
SOCS5 Immune response involving IL-4 signaling (<0.04)
GOT2 GABA biosynthesis and metabolism (<0.05)
a b c d
see Figure 2, A, B and C.Khan et al. Reproductive Biology and Endocrinology 2012, 10:84 Page 8 of 20
http://www.rbej.com/content/10/1/84
aTable 3 Estimates and enriched categories of differentially regulated non-common genes
Specific analysis Nature of differential Top enriched pathways [Gene symbol(s) of (p-value)
change [Number major candidate(s)]
of genes]
Pooled Up-regulated [50] WNT signaling [NRCAM, WNT16] (0)
DNA damage-induced responses [CHEK1] (<0.01)
and apoptosis
Role of 14-3-3 proteins in cell [CHEK1] (<0.02)
cycle regulation
Cadherins mediated cell adhesion [CHP] (<0.03)
Endothelial cell contacts by [CHP] (<0.03)
non-junctional mechanisms
Role of SCF complex in cell cycle [CHEK1] (<0.03)
regulation
ATM/ATR regulation of cell cycle [CHEK1] (<0.04)
nNOS signaling in neuronal synapses [RASD1] (<0.03)
Activation of astroglial cell [ERBB3] (<0.04)
proliferation by ACM3
G-protein signaling in RhoA [ARHGAP26] (<0.04)
regulation pathway
CDK5 in apoptosis and survival [ERBB3] (<0.04)
ERBB-family signaling [ERBB3] (<0.05)
Regulation of ElF2 activity [CSNK1G1] (<0.05)
associated with translation
Ligand-independent activation of [ERBB3] (<0.05)
ESR1 and ESR2
Non-genomic action of [WNT16] (<0.05)
androgen receptor
Down-regulated [41] Regulation of glucose and [APOE] (0)
lipid metabolism
GDNF signaling [ITGB1] (<0.04)
Immune response involving antigen [HLA-C] (0.05)
presentation by MHC class I
Chemotaxis involving CCR4-induced [ITGB1] (<0.05)
leukocyte adhesion
Stage 3 Up-regulated [4] No specific enriched category identified
Down-regulated [48] Cytoskeleton remodeling involving [RALGDS] (<0.01)
RalB and RalA regulation pathway
Clathrin coated vesicle formation [MYO1D] (<0.02)
Transcriptional silencing involving [PFDN5] (<0.02)
HP1 family
G-protein signaling involving interaction [RALGDS] (<0.03)
among Ras-family GTPases and
K-RAS/N-RAS/H_RAS regulation pathway
Stage 4 Up-regulated [31] Cell contraction involving relaxin [ADCY6, EDNRA, RXFP1] (0)
and GPCRs
Development involving [ADCY6, EDNRA] (0)
endothelin-1/EDNRA signaling
DNA damage induced apoptosis [NBN] (<0.01)
and DNA repair
Beta-2 adrenergic dependent [ADCY6] (<0.01)
CFTR expression
Regulation of lipid metabolism [PPARA] (<0.02)
Alpha-1 adrenergic receptor signaling [ADCY6] (<0.02)Khan et al. Reproductive Biology and Endocrinology 2012, 10:84 Page 9 of 20
http://www.rbej.com/content/10/1/84
aTable 3 Estimates and enriched categories of differentially regulated non-common genes (Continued)
Mu- and kappa-type opioid receptor [ADCY6] (<0.03)
mediated physiological process
Mucin expression via IL-6, IL-17 [TRAF3IP2] (<0.04)
signaling pathways
G-protein signaling [ADCY6] (<0.04)
Down-regulated [3] Transport from Golgi and ER to the [PPIA] (0)
apical membrane
Intracellular cholesterol and [PPIA] (<0.01)
sphingolipids transport
Proliferative phase Up-regulated [109] RAS regulation pathway [BCR, RASGRF1] (0)
TC21 pathway [BCR,] (0)
Regulation of CDC42 activity [BCR, FGFR1] (<0.01)
Sin3 and NuRD mediated [CHD3, SIN3A] (<0.01)
transcription regulation
GDNF family signaling [GFRA2, NRTN] (<0.01)
Phospholipid metabolism [GPD2, NRTN] (<0.02)
Immune response involving [IRF1, TRAF3IP2] (<0.02)
CD40 signaling
Down-regulated [20] Cytoskeleton remodeling involving α-1A
adrenergic receptor
Dependent inhibition of PI3K and [LAMB1, MYL12B] (<0.01)
regulation of actin by Rho GTPases
Cell contraction involving δ-type [MYL12B] (<0.01)
opioid receptor, S1P2 receptor, ACM
Development associated MAG dependent [MYL12B] (<0.01)
inhibition of neurite outgrowth
Development associated with TGF-beta [TPM1] (<0.01)
dependent induction of EMT via
RhoA, PI3K and ILK
Cell adhesion involving histamine [MYL12B] (<0.01)
H1 receptor
Cell adhesion and chemotaxis [LAMB1, MYL12B] (<0.01)
involving integrin
Chemotaxis involving inhibitory [MYL12B] (<0.01)
action of lipoxins on
IL-8 and leukotriene B4-induced
neutrophil migration
GPCRs in platelet aggregation [MYL12B] (<0.02)
Immune response involving CCR3 [MYL12B] (<0.02)
signaling in eosinophils
Oxidative phosphorylation [UQCR11] (<0.03)
Secretory phase Up-regulated [17] Transport involving RAN [TNPO1] (<0.01)
regulation pathway
Immune response involving [PGR] (<0.01)
MIF-JAB1 signaling
nNOS signaling in neuronal synapses [RASD1] (<0.02)
and circadian rhythm
Cell cycle associated spindle assembly [TNPO1] (<0.02)
and chromosome separation
Regulation of lipid metabolism [TNPO1] (<0.02) of glycogen metabolism [AGL] (<0.02)
Progesterone mediated maturation [PGR] (<0.02)Khan et al. Reproductive Biology and Endocrinology 2012, 10:84 Page 10 of 20
http://www.rbej.com/content/10/1/84
aTable 3 Estimates and enriched categories of differentially regulated non-common genes (Continued)
Cell adhesion associated [MME] (<0.03)
ECM remodeling
TGF-beta receptor signaling [TNPO1] (<0.03)
in development
Down-regulated [122] Cell contraction involving δ-type [MYL9] (0)
opioid receptor
Development associated [DPYSL2, ROBO3] (<0)
Slit-Robo signaling
Insulin mediated regulation [ElF4EBP1, PPP1CC] (<0.01)
of translation
Leukotriene 4 biosynthesis [GGT5, LTA4H] (<0.01)
and metabolism
Chemotaxis involving inhibitory [MYL9, RAC2) (<0.01)
action of lipoxins on IL-8 and
leukotriene B4-induced
neutrophil migration
Endoplasmic reticulum stress [ATF4, PPP1CC] (<0.01)
response pathway
Immune response involving CCR3 [MYL9, RAC2] (<0.03)
signaling in eosinophils
GTP-XTP metabolism [GUK1, NME3, POLR3H] (<0.03)
Cytoskeleton remodeling via RalB [RALGDS] (<0.04)
regulation pathway
Eutopic Up-regulated [182] Cytoskeleton remodeling involving [CHRM4, GNAQ] (0)
ACM3 and ACM4
G-protein signaling involving regulation [CHRM4, GNAQ] (0)
of cAMP levels by ACM
Transcription involving Tubby signaling [GNAQ] (0)
and HP1 family
Regulation of lipid metabolism involving [GNAQ, PTGS2] (0)
G-alpha(q) regulation
Cell contacts by non-junctional mechanisms [ITGA5, MAG1, PECAM1] (<0.01)
NMDA –dependent neurophysiological [GNAQ, GRIN2A] (<0.01)
process
Cell cycle at metaphase check point [CBX3, INCENP] (<0.01)
G-protein signaling involving Rap1A [MAGI1, RAPGEF1] (<0.02)
regulation pathways
Regulation of translation through [EIF4A2] (<0.03)
EIF4F activity
Regulation of translation by alpha-1 [EIF4A2, GNAQ] (<0.03)
adrenergic receptors
Development involving [GNAQ, NPPB] (<0.03)
endothelin-1/EDNRA signaling
Cytoskeleton remodeling via [GNAQ, RAPGEF1] (<0.04)
FAK signalin
Immune response involving PGE2 [GNAQ, PTGS2] (<0.03)
common pathways
Immune response involving IL-17 [CXCL3, PTGS2] (<0.04)
signaling pathways
Cell contraction via oxytocin signaling [GNAQ, PTGS2] (<0.04)
Transcription via PPAR pathway [MED1, PTGS2] (<0.04)
Regulation of lipid metabolism through [GNAQ, PTGS2] (<0.05)
alpha-1 adrenergic receptors signaling
via arachidonic acid