Hum. Reprod. Advance Access published February 7, 2014 Human Reproduction, Vol.0, No.0 pp. 1 –14, 2014 doi:10.1093/humrep/deu008
ORIGINAL ARTICLE Reproductive biology
Identification of new ovulation-related genes in humans by comparing the transcriptome of granulosa cells before and after ovulation triggering in the same controlled ovarian stimulation cycle M.L. Wissing 1,*, S.G. Kristensen2, C.Y. Andersen 2, A.L. Mikkelsen 1, T. Høst 1, R. Borup 3, and M.L. Grøndahl 2,4 1
The Fertility Clinic, Holbæk Sygehus, Holbæk, Denmark 2The Laboratory of Reproductive Biology, Rigshospitalet, Copenhagen, Denmark 3The Microarray Centre, Department of Genomic Medicine, Rigshospitalet, Copenhagen, Denmark 4The Fertility Clinic, Herlev Hospital, Copenhagen University Hospital, Copenhagen, Denmark *Correspondence address. Marie Louise Wissing. Holbæk Fertility Clinic, Smedelundsgade 60, 4300 Holbæk, Denmark. E-mail: mlwi@ regionsjaelland.dk,
[email protected]
Submitted on October 3, 2013; resubmitted on December 26, 2013; accepted on January 9, 2014
study question: Which genes and molecular mechanisms are involved in the human ovulatory cascade and final oocyte maturation? summary answer: Up-regulated genes in granulosa cells (GC) represented inflammation, angiogenesis, extracellular matrix, growth factors and genes previously associated with ovarian cancer, while down-regulated genes mainly represented cell cycle and proliferation.
what is known already: Radical changes occur in the follicle during final follicle maturation after the ovulatory trigger: these range from ensuring an optimal milieu for the oocyte in meiotic arrest to the release of a mature oocyte and remodeling into a corpus luteum. A wide range of mediators of final follicle maturation has been identified in rodents, non-human primates and cows. study design, size, duration: Prospective cohort study including 24 women undergoing ovarian stimulation with the long gonadotrophin-releasing hormone agonist protocol during 2010– 2012 at Holbæk Fertility Clinic. Nine paired samples of GC and 24 paired samples of follicular fluid (FF) were obtained before and after recombinant human chorionic gonadotrophin (rhCG) administration. participants/materials, setting, methods: Nine paired (nine arrays before rhCG and nine arrays after rhCG) samples of GC mRNA were amplified and hybridized to Affymetrix Human Gene 1.0 ST GeneChip arrays, compared and bioinformatically analyzed. Eleven selected genes were validated by quantitative reverse transcriptase PCR. FF hormones were analyzed by enzyme-linked immunosorbent assay. main results and the role of chance: Eleven hundred and eighty-six genes were differentially expressed (.2-fold, P,0.0001, false discovery rate ,0.0012) when comparing GC isolated before and 36 h after hCG, among those were genes known to be expressed at ovulation, i.e. ADAMTS1 and HAS2. Many new ovulation-related genes were revealed, such as CD24, ANKRD22, CLDN11 and FBXO32. FF estrogen, androstenedione and anti-Mu¨llerian hormone decreased significantly while progesterone increased, accompanied by radical changes in the expression of steroidogenic genes (CYP17A, CYP19A, HSD11B1 and HSD11B2, StAR). Genes related to inflammation, angiogenesis, extracellular matrix formation, growth factors and cancer were up-regulated while cell cycle genes were massively down-regulated. Seventy-two genes previously described in connection with ovarian cancer were among the highly regulated genes. In silico analysis for top upstream regulators of the ovulatory trigger suggested—besides LH—TNF, IGF1, PGR, AR, EGR1 (early growth response 1), ERK1/2 (extracellular signal regulated kinase 1/2) and CDKN1A (cyclin-dependent kinase inhibitor 1A) as potential mediators of the LH/hCG response.
limitations, reasons for caution: The present dataset was generated from women under hormonal stimulation. However, comparison with a macaque natural cycle whole follicle ovulation dataset revealed major overlap, supporting the idea that the ovulationrelated genes found in this study are relevant in the human natural cycle. & The Author 2014. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved. For Permissions, please email:
[email protected]
2
Wissing et al.
wider implications of the findings: These data will serve as a research resource for genes involved in human ovulation and final oocyte maturation. Ovulation-related genes might be good candidate biomarkers of follicle and oocyte health. Further, some of the ovulationrelated genes may serve as future ovarian cancer biomarkers. study funding/competing interest(s): Grants from the Research Fund of Region Sjælland are gratefully acknowledged. None of the authors declared any conflict of interest.
trial registration number: Not applicable. Key words: granulosa cell transcriptome / ovulation / hCG trigger / follicular fluid / human
Introduction The mid-cycle surge of luteinizing hormone (LH) is the physiological trigger of ovulation, while a bolus hCG is the golden standard used in connection with assisted reproduction. Ovulation essentially induces three distinct and highly complex processes comprising oocyte maturation, ovulation with release of an oocyte and transformation of the avascular granulosa cells (GC) into the highly vascularized granulosa-lutein cells. Rupture of the follicles has been characterized as a local inflammatory process during which the follicle releases the enclosed mature oocyte and subsequently undergoes profound remodeling leading to the formation of a corpus luteum (CL) to support possible implantation. The molecular processes behind the ovulatory cascade have only been studied in humans to a limited extent, because of limited access to follicles immediately before triggering of final oocyte maturation. Although some data are available from animal studies (Park et al., 2004; Espey, 2006; Stocco et al., 2007; Xu et al., 2011; Christenson et al., 2013), results may not readily extrapolate to humans. In this study, we describe for the first time a paired analysis of transcriptome changes in GC and hormonal changes in follicular fluid (FF) of the preovulatory follicles before and 36 h after recombinant human chorionic gonadotrophin (rhCG) triggering of final oocyte maturation from the same women in the same cycle following controlled ovarian stimulation (COS).
Materials and Methods This study was approved by The Danish Scientific Ethical Committee (SJ-156) and conducted in accordance with the Helsinki Declaration II. All participants gave informed consent before their inclusion in this study.
Participants A paired sample of FF before and after rhCG administration was obtained from each of 24 women aged (mean + SD) 27.9 + 3.4 years. Only those women who had developed more than eight follicles of at least 14 mm in diameter at the last control visit before rhCG triggering were enrolled. From 9 of the 24 women GC yielding good quality RNA were obtained prior to and after rhCG triggering (patient demographics and baseline characteristic are shown in Supplementary data, Table SI). The demographics and baseline characteristics in the nine women who contributed with paired samples of GC did not differ from the 15 women who only contributed paired samples of FF (data not shown).
Controlled ovarian stimulation with exogenous gonadotrophins The women were treated according to the standard long gonadotrophinreleasing hormone agonist protocol: pituitary desensitization with buserelin 0.5 mg (Suprefactw, Sanofi-Aventis, France) was started on Cycle day 21. COS with recombinant follicle-stimulating hormone (rFSH) (Puregonw, MSD, Denmark) was started after at least 14 days of desensitization. Six thousand five hundred international units rhCG (Ovitrellew, Merck Serono, Denmark) was administered 36 h before oocyte pick-up (OPU) when at least three follicles had reached the size of 17 mm.
Follicle puncture and isolation of GC FF and GC were sampled by transvaginal ultrasound guided puncture of a follicle 14– 17 mm in diameter at the last control visit before administration of rhCG. Follicle puncture was performed 13 h prior to rhCG administration for 10 women, 37 h for 13 women and 61 h for one woman. For the nine paired samples used for microarray analysis, four follicle punctures were performed at 13 h, four follicle punctures at 37 h and one follicle puncture at 61 h prior to hCG administration. The puncture was performed with a single lumen needle (Wallace Oocyte Recovery Systems, Smith Medical, Brisbane, Australia) attached to a syringe (10 ml BD DiscarditTM II, Becton Dickinson S.A. Fraga, Huesca, Spain) under transvaginal ultrasound guidance. After collection of FF, a new syringe containing flushing medium (Sydney IVF Follicle Flush buffer K-SIFB-100, Cook, Queensland, Australia) was used to flush the follicle if no or few GC aggregates were present in the FF. During OPU, FF from the first punctured follicle on each ovary was collected separately and used for collection of GC and FF. GC were isolated as described previously (Grøndahl et al., 2012).
FF analysis Immediately after GC collection, FF was isolated by centrifugation at 300 g for 10 min. The supernatant was transferred to 2 ml cryotubes and stored at 2808C until analysis. Estrogen, progesterone, androstenedione and testosterone were measured with enzyme-linked immunosorbent assays (NovaTec Immundiagnostica, Dietzenbach, Germany, DNOV 002, 003, 006 and 008, respectively) and performed according to manufacturer’s instructions. Anti-Mu¨llerian Hormone (AMH) was measured with AMH ELISA kit (AMH ELISA, AL-105-i Ansh Labs Houston, TX, USA) according to manufacturer’s instructions.
Isolation of mRNA The Absolutely RNA Nanoprep kit from Stratagene (Agilent Technologies, Waldbronn, Germany) was used to isolate total RNA from GC according to the manufacturer’’s instructions. The samples were analyzed for total RNA quality and level of degradation using an Agilent 2100 Bioanalyzer
3
Human granulosa cell transcriptome and ovulation
and RNA 6000 Pico LabChip according to the manufacturer’’s instructions (RNA 6000 Pico assay kit, Agilent Technologies, Waldbronn, Germany). All of the RNA samples showed two distinct peaks representing 18s and 28s rRNA which indicated good quality RNA.
Microarray analysis Total RNA (ranging from 1.5 –50 ng RNA, mean + SD 27 + 18.9 ng) from each sample was amplified using the Ovation Pico WTA v.2 RNA Amplification System from NuGENw Inc. (NuGENw, San Carlos, CA, USA) and biotin labeling was performed with the Encore Biotin Module (NuGENw). The labeled samples were hybridized to the Human Gene 1.0 ST GeneChip array (Affymetrix, Santa Clara, CA, USA). The arrays were washed and stained with phycoerythrin conjugated streptavidin (SAPE) using the Affymetrix Fluidics Stationw 450, and the arrays were scanned in the Affymetrix GeneArrayw 3000 7G scanner to generate fluorescent images, as described in the Affymetrix GeneChipw protocol. Cell intensity files (CEL files) were generated in the GeneChipw Command Consolew Software (AGCC) (Affymetrix).
Validation of differential expression by quantitative reverse transcriptase PCR (qRT-PCR) The following genes were tested by pre-designed TaqManw Gene Expression Assays (gene symbol/catalogue number): CD24 (Hs03044178_g1), AREG (Hs00950669_m1), HSD11B1 (Hs01547870_m1), HSD11B2 (Hs003886 69_m1), PTGS2 (Hs00153133_m1), COL6A3 (Hs00915125_m1), CYP19A1 (Hs00903413_m1), IGF2 (Hs04188276_m1), PTTG1 (Hs00851754_u1), CCNA2 (Hs00996788_m1) and INHBA (Hs01081598_m1) as well as the Endogenous Control Assays for human glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (Applied Biosystems/Life Technologies Europe, Nærum, Denmark). The data were normalized to GAPDH, and relative quantification according to the comparative cycle threshold (CT) Method (LightCycler480 Software, Roche, Denmark) was used to quantify gene expression. GAPDH was chosen as reference gene based on stable expression in GC over various follicle classes (Kristensen et al., 2013). Average cycle threshold (Ct) values for GAPDH (mean + SD) were 21 + 1.2 and 22 + 1.4 before and after rhCG, respectively.
Statistical analysis Differences in hormone concentration before and after rhCG were tested by Wilcoxon matched-pairs signed rank test. Differences in qRT-PCR gene expression between paired samples of GC were calculated by ratio paired t-test (GraphPad Prism versus 6, GraphPad Software, San Diego, CA, USA).
Differential gene expression analysis For statistical analysis the CEL files (nine representing GC before rhCG and nine representing GC 36 h after rhCG) were imported into the R (www. r-project.org/ (27 January 2014, date last accessed).) where they were robust multichip averaged (RMA) using quantile normalization and Median Polish summarization (Bolstad et al., 2003). The expression profiles in the two conditions were compared and genes were defined as differentially expressed when they were selected in the paired t-test, having P-values below 0.0001 and fold-change .2. All samples are minimum information about a microarray experiment (MIAME) compliant and were handled according to standard operation procedures in the Microarray Center. The 18 arrays were submitted to ArrayExpress at EMBL using MAGE-TAB submission. The accession number is E-MTAB-2203.
Gene function and pathway enrichment analysis The list of differentially expressed genes were subjected to gene function enrichment and pathway analysis in the Ingenuityw software (www. ingenuity.com (27 January 2014, date last accessed)) as described in Grøndahl et al. (2010, 2013). Furthermore, an upstream regulator analysis (IngenuitywSystems) was performed to reveal potential regulators of the observed effect of the rhCG triggering.
Validation of differential expression by external macaque ovulation data set The dataset GSE22776 was downloaded from http://www.ncbi.nlm.nih.gov/ geo/ (27 January 2014, date last accessed). The data set contains 22 samples from Macaque periovulatory follicles of which 10 samples were used in this study. These files are: GSM563219.CELGSM563220.CEL-GSM563221. CEL-GSM563222.CEL-GSM563223.CEL-GSM563224.CEL-GSM563233.CEL-GSM563234.CEL-GSM563235.CEL-GSM563236.CEL, representing follicles before and 36 h after exposure to an ovulatory rhCG dose. The CEL files were imported into R and RMA normalized. The expression matrix was exported to dChip (www.dchip.org (27 January 2014, date last accessed)) and used as basis of cluster visualization of genes, which were found to be differentially expressed in our data. Two thousand and seventy-nine redundant Macaque probe sets (a particular gene was represented more than once in the gene list) were overlapping between the Macaque Gene arrays and the present Human Gene arrays, based on the gene symbol of the 1186 unique genes that were defined as differentially expressed. This gene list was filtered to exclude genes (probe sets) in the primate data that showed only a small relative variation across samples (0.05 ,SD/mean ,1000.00). The 786 overlapping probe sets (513 unique genes) that were left after the filtering were used in a two-way hierarchical cluster visualization using the Qlucore Omics Explorer version 2.3 software (www.glucore.com) with average linkage (Supplementary data, Table SII).
Results Differentially expressed genes Unsupervised principal component analysis (PCA) revealed that the samples clustered within each of the two time frames and demonstrated general difference in the transcriptome in the GC before and after rhCG triggering (Fig. 1A). Setting the criteria for being differentially expressed in the paired t-test to P , 0.0001 (false discovery rate (FDR): 0.0012) and a fold change exceeding 2, a total of 1224 probe sets (1186 unique genes) were found to be differentially expressed with 572 genes being up-regulated and 614 genes down-regulated in the two types of GC (cluster diagram in Fig. 1B). Table I presents a selected list of differentially expressed genes including known reproduction genes, genes representing enriched functions and genes selected based on high fold change. The full list of differentially expressed genes (fold change .2, P , 0.0001) is shown in Supplementary data, Table SIII.
Confirmation of array data by qRT-PCR analysis of selected genes qRT-PCR of 11 selected genes in 7 paired samples of GC before/after rhCG demonstrated agreement with the microarray data (Fig. 2).
4
Wissing et al.
Figure 1 (A) Unsupervised PCA of human GC before and after rhCG triggering demonstrated general difference in the transcriptome before and after rhCG triggering. Blue labels represent GC samples before rhCG and red labels represent samples after rhCG. The single outlier among the ‘before rhCG’ sample did not differ according to baseline descriptive parameters in any way and the follicle puncture was performed 13 h prior to rhCG, but the amplified amount of RNA was only 1.5 ng, which was low compared with the other samples. This might explain why this sample behaved as an outlier (see Materials and methods for details). Horizontal axis corresponds to principal component 1 (PC1), vertical axis corresponds to PC2 and depth axis corresponds to PC3. The percent of variation explained by a particular PC is indicated in the axis label. (B) Hierarchical cluster diagram showing the 1186 genes selected as being differentially expressed between human GC before and after rhCG triggering (P , 0.0001 and fold change .2). The expression level for each gene is standardized to have a mean value of zero and a SD of 1. Black color represents the mean value zero, red color represents the gene expression level above the mean and green color represents expression below the mean. The intensity of the pseudo-color reflects the number of SDs from the mean.
Comparison with macaque whole follicle database before and 36 h after rhCG To examine the relevance of the present data to events in a natural cycle, the 1186 significantly differentially expressed genes reported in this paper were compared with the gene set generated by Xu et al. (2011) from macaques in their natural cycle before and 36 h after hCG triggering. This revealed that the macaque data included 786 overlapping probes (513 unique genes) with a differential expression pattern before and after rhCG, which resembled the expression pattern detected in our human data (Fig. 3 and Supplementary data, Table SII).
New ovulation-related genes in humans Many of the highly regulated genes were previously unrelated to ovulation, such as, for example, the highly up-regulated CD24 (CD24 protein), ANKDR22 (ankyrin repeat domain 22), TECRL (trans-2,3-enoyl-CoA reductase-like), FBXO32 (F-box protein 32), FLRT2 (fibronectin leucine rich transmembrane protein 2), CLDN11 (claudin 11) and the highly downregulated CEP55 (Centrosomal protein 55 kDa) and H19/mir675 (microRNA 675) (Table I).
Steroidogenesis and gonadotrophin receptors Gene function enrichment analysis The top ‘Molecular and Cellular Function’ and the top ‘Disease and Disorder’ enriched in the differentially expressed genes were Cell Cycle (P ¼ 1.7E225; 132 genes) and Cancer (P ¼ 1.5E225; 435 genes), respectively. Seventy-two genes previously described in connection with ovarian cancer were among the highly regulated genes according to the Ingenuityw enrichment analysis (Supplementary data, Table SIV). Expression of cell cycle-related genes was highly down-regulated 36 h after the bolus of rhCG while inflammation and coagulation related genes were highly up-regulated after rhCG. The ‘Top Canonical Pathway analysis’ listed in Table II likewise underlined these central mechanisms represented in the up- and down-regulated genes. In the following sections, the results are highlighted with special emphasis on ovarian physiology and new genes in reproduction. For a brief overview, please refer to Fig. 4.
The FF hormone concentrations before and after rhCG triggering are listed in Table III. The FF estrogen concentration decreased significantly after rhCG with a concomitant decrease in CYP19A1 (cytochrome P450, family 19, subfamily A, polypeptide 1) expression. The FF androstenedione concentration decreased significantly after rhCG, accompanied by a decrease in the expression of the theca cell-specific CYP17A1 (cytochrome P450, family 17, subfamily A, polypeptide 1). The FF progesterone concentration and StAR expression (P ¼ 0.00031, fold change 2.7) increased after rhCG as expected, while at time points before and 36 h after rhCG, gene expression of genes promoting progesterone synthesis HSD3B2 (hydroxy-delta-5-steroid dehydrogenase, 3 beta- and steroid delta-isomerase 2), LDLR (low density lipoprotein receptor) and CYP11A1 (cytochrome P450, family 17, subfamily A, polypeptide 1) were unaltered. Testosterone concentrations before and after rhCG were unaltered as was HSD17B (17b-hydroxysteroid dehydrogenase) expression.
5
Human granulosa cell transcriptome and ovulation
Table I Selected list of differentially expressed genes in GC 36 h after rhCG triggering as compared with GC before rhCG administration. Gene symbol
Gene title
P-value
Mean expression before hCG
Mean expression after hCG
Fold change
Function
............................................................................................................................................................................................. CD24
CD24 molecule
0.00001
19
1540
80.1
TECRL
trans-2,3-enoyl-CoA reductase-like
0.00000
14
520
36.7
Ovarian cancer Lipid metabolic process
ANKRD22
Ankyrin repeat domain 22
0.00000
26
724
28.0
Unknown
SLCO2A1
Solute carrier organic anion transporter family, member 2A1
0.00001
71
1924
27.2
Prostaglandin/thromboxane transporter activity
FBXO32
F-box protein 32
0.00000
26
635
24.0
Protein ubiquitinylation
FLRT2
Fibronectin leucine rich transmembrane protein 2
0.00000
41
897
21.8
Cell adhesion/extracellular matrix
AREG
Amphiregulin
0.00001
50
1074
21.6
Growth factor
ADAMTS9
ADAM metallopeptidase with thrombospondin type 1 motif,9
0.00001
42
876
20.8
Metallopeptidase activity
ADAMTS1
ADAM metallopeptidase with thrombospondin type 1 motif,9
0.00000
40
818
20.6
Ovulation
CLDN11
Claudin-11
0.00000
61
1134
18.6
Integral membrane protein/ tight junction strands
PTX3
Pentraxin 3
0.00001
39
617
15.9
Extracellular matrix
NTS
Neurotensin
0.00007
61
951
15.7
Extracellular matrix
EFNB2
Ephrin-B2
0.00000
34
538
15.7
Angiogenesis
HSD11B1
Hydroxysteroid (11-b) dehydrogenase 1
0.00002
193
2789
14.5
Glucocorticoid biosynthetic process
SEMA3A
Semaphorin-3A
0.00001
84
1142
13.6
Neovascularization
SERPINA3
Serpin peptidase inhibitor, clade A (alpha-1 antiproteinase, antitrypsin), member
0.00002
40
546
13.5
Plasma protease inhibitor/ inflammatory response
PTGS2
Prostaglandin-endoperoxide synthase 2
0.00001
116
1559
13.4
Prostaglandin biosynthesis
HAS2
Hyaluronan synthase 2
0.00000
24
258
10.8
Extracellular matrix
COL6A3
Collagen, type VI, alpha 3
0.00002
66
671
10.1
Extracellular matrix
CD68
CD68 molecule
0.00001
81
811
10.1
Scavenger receptor family/ recruitment of macrophages
ITGA5
Integrin, alpha 5
0.00000
56
553
9.9
Angiogenesis/leukocyte migration
c-FOS
FBJ murine osteosarcoma viral oncogene homolog
0.00001
157
1419
9.0
Transcription factor
RGS2
Regulator of G-protein signaling 2
0.00008
189
1473
7.8
GTPase activating proteins
SEMA6A
Semaphorin 6A
0.00001
54
413
7.6
Neovascularization
SEMA3C
Semaphorin 3C
0.00002
44
329
7.5
Neovascularization
FN1
Fibronectin 1
0.00004
350
2543
7.3
Extracellular matrix
F2R
Coagulation factor II (thrombin) receptor
0.00001
44
305
6.9
Blood coagulation/ inflammatory response
TFPI
Tissue factor pathway inhibitor
0.00010
151
1045
6.9
Blood coagulation, extrinsic pathway
GSTM3
Glutathione S-transferase mu 3
0.00000
160
1096
6.9
Response to estrogen signaling
PHLDA1
Pleckstrin homology-like domain, family A, member 1
0.00003
91
629
6.9
(Anti)-apoptosis
EREG
Epidermal growth factor
0.00053a
47
319
6.8
Growth factor, ovulation
COL5A2
Collagen, type V, alpha 2
0.00000
26
171
6.7
Extracellular matrix
FGF12
Fibroblast growth factor 12
0.00000
50
324
6.4
Cell growth, tissue repair and invasion
Continued
6
Wissing et al.
Table I Continued Gene symbol
Gene title
P-value
Mean expression before hCG
Mean expression after hCG
Fold change
Function
............................................................................................................................................................................................. TGFB1
Transforming growth factor b1
0.00001
30
190
6.3
Epidermal growth factor receptor signaling pathway
PLA2G4A
Phospholipase A2, group IVA
0.00000
49
297
6.1
Arachidonic acid production
MMP19
Matrix metallopeptidase 19
0.00000
30
175
5.8
Extracellular matrix
NCAM1
Neural cell adhesion molecule 1
0.00002
22
126
5.7
Cell movement
EGR1
Early growth response 1
0.00007
279
1530
5.5
Transcriptional regulator
OXTR
Oxytocin receptor
0.00002
64
318
5.0
Menstrual cycle, pregnancy and lactation
TIMP1
TIMP metallopeptidase inhibitor 1
0.00003
1293
5949
4.6
Extracellular matrix
F3
Coagulation factor III (thromboplastin, tissue factor)
0.00007
63
281
4.5
Coagulation
EGR2
Early growth response 1
0.00005
50
229
4.5
Transcriptional regulator
PTGES
Prostaglandin E synthase
0.00003
498
1805
3.6
Prostaglandin synthesis
ALOX5AP
Arachidonate 5-lipoxygenase-activating protein
0.00002
24
86
3.6
Leukotriene synthesis
ALOX4
Arachidonate 5-lipoxygenase
0.0004a
23
74
3.3
Leukotriene synthesis
IL6ST
Interleukin 6 signal transducer
0.00001
817
2696
3.3
Cytokine receptor complex
RUNX1
Runt-related transcription factor 1
0.00002
93
300
3.2
Transcription factor
PLXNB2
Plexin B2
0.00002
101
296
2.9
Neuron development
ANGPT1
Angiopoietin 1
0.00008
11
31
2.8
Vascular development and angiogenesis
JUN
jun proto-oncogene
0.00003
105
229
2.2
Cell– cell contact
SLC24A6
Solute carrier family 24 (sodium/potassium/ calcium exchanger), member 6
0.00001
166
78
22.1
Sodium/calcium exchanger
ESR2
Estrogen receptor 1
0.00000
95
44
22.2
Steroid binding
ESR1
Estrogen receptor 2
0.00002
102
40
22.5
Steroid binding
CYP19A1
Cytochrome P450, family 19, subfamily A, polypeptide 1
0.00004
4237
1591
22.7
Steroidogenesis
FOXL2
Forkhead box L2
0.00000
380
130
22.9
Transcription factor/ovarian follicle development
PCNA
Proliferating cell nuclear antigen
0.00000
478
153
23.1
Cell cycle
PGRMC1
Progesterone receptor membrane component 1
0.00000
1714
519
23.3
Steroid binding
NR5A1 b
Nuclear receptor subfamily 5, group A, member 1
0.00000
270
79
23.4
Steroidogenesis
FAM84A
Family with sequence similarity 84, member A
0.00002
509
256
23.5
Cancer
AMHR2
Anti-Mullerian hormone receptor, type II
0.00000
247
66
23.7
Folliculogenesis
GJC1
Gap junction protein, gamma 1
0.00003
328
85
23.9
Cell– cell contact
LHCGR
Luteinizing hormone choriogonadotrophin receptor
0.00000
723
147
24.9
Gonadotrophin binding
TIMP3
TIMP metallopeptidase inhibitor 3
0.00000
501
93
25.4
Extracellular matrix
MCM10
Minichromosome maintenance complex component
0.00000
113
20
25.6
LPL
Lipoprotein lipase
0.00001
3052
531
25.7
Lipid uptake
GSTA1
Glutathione S-transferase a 1
0.00007
4970
929
25.3
Glutathione metabolic process
IGF2
Insulin-like growth factor 2
0.00003
4219
656
26.4
Growth factor, mitosis
TNNI3
Troponin I type 3
0.00001
2048
289
27.1
Continued
7
Human granulosa cell transcriptome and ovulation
Table I Continued Gene symbol
Gene title
P-value
Mean expression before hCG
Mean expression after hCG
Fold change
Function
............................................................................................................................................................................................. 0.00000
296
40
27.4
Glucocorticoid biosynthetic process
Histone cluster 1, H2bh
0.00000
1166
156
27.5
Cell cycle related
Cell division cycle 20
0.00000
278
37
27.6
Cell cycle
FSHR
Follicle-stimulating hormone receptor
0.00005
719
86
28.3
Gonadotrophin binding
CCNB1
Cyclin B1
0.00000
805
87
29.3
Cell cycle
CCNB2
Cyclin B2
0.00000
328
34
29.8
Cell cycle
LRP8
Low density lipoprotein receptor-related protein 8
0.00000
2085
211
29.9
Lipid uptake
H19/MIR675
MicroRNA 675
0.00001
3997
401
210.0
Post-transcriptional regulation of gene expression
PTTG1
Pituitary tumor-transforming 1
0.00000
1889
182
210.4
Cell cycle
HSD11B2
Hydroxysteroid (11-beta) dehydrogenase 2
HIST1H2BH CDC20
b
CYP17A1
CCNA2
a
Cytochrome P450, family 17, subfamily A, polypeptide 1
0.00047
403
38
210.6
Steroidogenesis
Cyclin A2
0.00000
537
41
213.2
Cell cycle
TOP2A
Topoisomerase (DNA) II alpha 170 kDa
0.00001
1113
81
213.7
Cell cycle
INHBA
Inhibin b A
0.00000
1606
110
214.6
Regulator of FSH secretion, growth factor
CEP55
Centrosomal protein 55 kDa
0.00000
830
47
217.5
Cell cycle
PBK
PDZ binding kinase
0.00000
278
14
220.1
Mitosis
Genes were selected based on fold change and/or function. GC, granulosa cells. a Marginally significant. b CYP17A1 and NR5A1 are expressed in theca cells exclusively, indicating minor theca cell contamination of the GC preparations.
HSD11B1 (hydroxysteroid 11b dehydrogenase 1) was up-regulated while HSD11B2 (Hydroxysteroid 11b dehydrogenase 2) was downregulated. The expression of FSHR (follicle-stimulating hormone receptor) and LHCGR (luteinizing hormone receptor) decreased after rhCG, as did INHBA (inhibin bA) (Table I).
Exit from cell cycle Cyclins were down-regulated concomitantly with an increase in cell cycle inhibitors (CDKN1A, alias p21cip, cyclin-dependent kinase inhibitor 1A). Other genes marking actively dividing cells, such as TOP2A (topoisomerase 2 alpha), CDC20 (cell division cycle 20) and the MCM’s (minichromosome maintenance complex components) were dramatically down-regulated (Table I and Supplementary data, Table SIII).
Inflammation, angiogenesis and coagulation Many inflammation-related genes were highly up-regulated after rhCG, such as genes involved in prostaglandin synthesis: PTSG2 (prostaglandin endoperoxide synthase 2), PTGES (prostaglandin E synthase), PLA2G4A (phospholipase A2, group 4A) and the leukotriene-producing enzymes from the lipoxygenase pathway ALOX5 (arachidonate 5 lipoxygenase) and ALOX5AP (arachidonate 5-lipoxygenase-activating protein) (Table I and Supplementary data, Table SIII) Genes involved in interleukin signaling and acute phase response were also highly represented among the up-regulated genes (Table II).
Genes promoting angiogenesis and invasiveness, such as CD24 and EFNB3 (ephrin B3), were among the highly up-regulated genes. ANGPT1 (angiopoietin 1) was also up-regulated while VEGFA (vascular endothelial growth factor A) was down-regulated (Table I). Further, genes involved with coagulation, such as F3 (coagulation factor III) and F2R (coagulation factor II receptor), were up-regulated (Table I).
Extracellular matrix remodeling and cell-to-cell contact Extracellular matrix protein genes such as HAS2 (hyaluron synthase 2), PTX3 (pentraxin 3), FN1 (fibronectin 1), ADAMTS1 and ADAMTS9 (ADAM metallopeptidase with thrombospondin type 1 motif, 1 and 9) were highly up-regulated (Table I). MMP19 (matrix metallopeptidase) and TIMP1 (TIMP metallopeptidase inhibitor 1) were up-regulated while TIMP3 were highly down-regulated. COL6A3 (collagen 6A3), FLRT2 (fibronectin leucine rich transmembrane protein 2) and CLDN11 (claudin 11), previously unrelated to ovulation, were highly increased.
Growth factors The genes encoding the EGF-like ligands AREG (amphiregulin) and EREG (epiregulin) were highly up-regulated after rhCG as was TGF- b1 (transforming growth factor b1). FF AMH concentration decreased significantly (Table III) while AMH gene expression remained unchanged and AMHR2 (anti-Mu¨llerian hormone receptor II) expression decreased 3.7-fold (Table I). IGF2 (insulin-like growth factor 2) and its genomic
8
Wissing et al.
Figure 2 Validation of microarray results by qRT-PCR in 11 selected genes in 7 paired samples of GC prior to and 36 h after rhCG triggering. All validated genes were significantly differentially expressed before compared with after rhCG in the qRT-PCR experiment (P-values, ratio paired t-test): CD24: P ¼ 0.0069, AREG: P ¼ 0.0015, PTGS2: P ¼ 0.0004, COL6A3: P ¼ 0.0007, HSD11B1: P ¼ 0.0001, HSD11B2: P ¼ 0.0132, INHBA: P ¼ 0.0012, CCNA2: P ¼ 0.0037, PTTG1: P ¼ 0.0049, IGF2: P ¼ 0.0064 and CYP19A1: P ¼ 0.0010. GC, granulosa cells.
neighbor H19/mir675 (microRNA 675) were significantly downregulated after rhCG (Table I).
Transcription factors and upstream regulators EGR1 and EGR2 (early growth response) were highly induced by rhCG as was JUN (jun proto-oncogene) and c-FOS (FBJ murine osteosarcoma viral oncogene homolog). FOXL2 (forkhead box L2) was significantly downregulated (Table I). The in silico upstream regulator analysis suggested FSH, LH as well as PGR (progesterone receptor), ESR1 (estrogen receptor 1) and AR (androgen receptor) as top upstream regulators of the rhCG trigger induced changes. Furthermore, various pro-inflammatory cytokines (i.e. TNF (tumor necrosis factor), IL6 (interleukin-6)), growth factors (i.e. IGF1, IGF2 (insulin-like growth factors 1 and 2), EGF (epidermal growth factor), PGDF b (platelet-derived growth factor b) as well as cell cycle regulators (i.e. CCND1 (cyclin D1), CDKN1A (cyclindependent kinase inhibitor 1A) and transcription factors (i.e. TP53, TP63 (tumor protein 53/63) and FOXO3 (Forkhead box O3) were suggested as top upstream regulators (Table IV and for the full list Supplementary data, Table SV).
Discussion This human transcriptome dataset will serve as an easily accessible source for the study of ovulation-related genes in humans. A unique and highly significant difference was demonstrated in human GC expressed genes obtained from the same women in the same cycle in two hallmarks of folliculogenesis: before and after the ovulatory surge of rhCG for final oocyte maturation. A total of 1186 genes were highly significantly differentially expressed, among those were many genes
Figure 3 Comparison between the 1186 genes differentially expressed in human GC before and 36 h after rhCG and genes differentially expressed in the natural cycle in macaque whole follicles at time 0 and 36 h after rhCG triggering (Xu et al., 2011). An overlap of 786 genes was revealed, shown here as a hierarchical cluster diagram of the overlapping differentially expressed genes. Purple corresponded to a macaque follicle sample before hCG and orange corresponded to an unruptured macaque follicle sample 36 h after hCG. Black color represents the mean value zero, red color represents the gene expression level above the mean and green color represents expression below the mean. The intensity of the pseudo-color reflects the number of SDs from the mean. GC, granulosa cells.
previously not related to ovulation and interestingly, several of those are reported as related to cancer. The criteria for differential expression were strict (FC . 2, P , 0.0001, FDR , 0.012); hence, some important ovulation-related genes might have been overlooked. However, our approach in this study was to describe the most significantly differentially expressed ovulation-related genes and pathways. In addition to qRT-PCR confirmation, our data and model were further validated by a high overlap to differently expressed genes observed in whole follicles from macaque monkeys collected at the same time points as in our study (Xu et al., 2011). The monkeys were in their natural menstrual cycle and the high overlap may indicate that the data obtained in the present study also represent changes occurring in natural ovulation in humans. This overlap is an important external validation of the data set since the aspiration of a follicle prior to hCG injection hypothetically could induce the formation of an early CL with local and systemic effects potentially influencing the GC transcriptome at 36 h after hCG. In the following sections, the data will be discussed in the context of known ovarian physiology as well as extended in light of the many new ovulation-related genes found in this study.
9
Human granulosa cell transcriptome and ovulation
Table II Ingenuityw canonical pathways enriched (P < 0.001) in the list of genes down-regulated and up-regulated in GC 36 h after comparing with before rhCG triggering. Ingenuityw canonical pathways
P-value
Ratioa
Genes
............................................................................................................................................................................................. Pathways enriched in the genes down-regulated in GC after hCG triggering Cell cycle control of chromosomal replication
6.6E210
10/30
MCM5, MCM6, CDC45, MCM2, CDC6, CDK6, ORC6, MCM4, CDK2, MCM7, ORC1
Role of BRCA1 in DNA damage response
1.4 E29
15/63
FANCG, BARD1, RBBP8, PLK1, CHEK1, RAD51, FANCB, FANCD2, MSH2, RFC4, BRCA2, BRIP1, BRCA1, FANCA, E2F2
Mitotic roles of polo-like kinase
2.3 E29
15/64
KIF23, CDC25C, ESPL1, CDC20, WEE1, PRC1, CCNB2, PLK1, CDK1, CCNB1, PLK4, FBXO5, PKMYT1, KIF11, CDC25A
Estrogen-mediated S-phase entry
3.1 E29
10/27
CCNA2, CCNE2, TFDP1, ESR2, ESR1, CDK1, E2F2, CDK2, SKP2, CDC25A
G2/M DNA damage checkpoint regulation
1.6 E28
12/48
CDC25C, CKS2, WEE1, CKS1B, CCNB2, PKMYT1, PLK1, BRCA1, CDK1, SKP2, CHEK1, CCNB1
ATM signaling
9.1 E28
13/61
CDC25C, CCNB2, CBX5, CDK1, CHEK1, CCNB1, RAD51, SMC2, FANCD2, H2AFX, BRCA1, CDK2, CDC25A
Cyclins and cell cycle regulation
3.7 E27
14/87
CCNE2, TFDP1, WEE1, CDK6, CCNB2, CDKN2C, CDK1, SKP2, CCNB1, CCNA2, CCND3, CDK2, E2F2, CDC25A
GADD45 signaling
2.1 E26
7/21
PCNA, CCNE2, CCND3, BRCA1, CDK1, CDK2, CCNB1
Role of CHK proteins in cell cycle checkpoint control
2.5 E26
11/57
PCNA, CDC25C, RFC4, CLSPN, PLK1, BRCA1, CDK1, E2F2, CDK2, CDC25A, CHEK1
Pathways enriched in the genes upregulated in GC after hCG triggering IL-6 signaling
5.4 E28
12/122
HSPB3, SOCS3, PIK3C2B, IL18, JUN, MAPK14, RRAS, TNFRSF1A, MAP4K4, STAT3, IL1R1, MCL1
Role of tissue factor in cancer
1.0 E23
11/109
PIK3C2B, MAPK14, PTK2B, RRAS, ITGAV, RPS6KA3, PLAUR, PLCB1, CYR61, F3, PRKCA
Coagulation system
1.2 E23
6/35
F2R, PROS1, F5, PLAUR, TFPI, F3
Platelet-derived growth-factor signaling
1.1 E23
9/79
SYNJ2, PIK3C2B, JAK1, JUN, RRAS, CAV1, STAT3, OCRL, PRKCA
Role of JAK family kinases in IL-6-type cytokine signaling
1.3 E23
5/25
SOCS3, JAK1, MAPK14, OSMR, STAT3
Acute phase response signaling
1.6 E23
14/172
SOCS3, SERPING1, TNFRSF1A, RRAS, C1S, SERPINA3, STAT3, IL1R1, C1R, IL18, SOD2, JUN, MAPK14, OSMR
Extrinsic prothrombin activation pathway
1.7 E23
4/16
PROS1, F5, TFPI, F3
IL-10 signaling
1.9 E23
8/71
SOCS3, IL18, JAK1, JUN, MAPK14, MAP4K4, STAT3, IL1R1
Thrombopoietin signaling
2.5 E23
7/59
PIK3C2B, JUN, RRAS, IRS2, PRKCH, STAT3, PRKCA
a
Number of genes differentially expressed/number of genes in the pathway. GC, granulosa cells.
Steroidogenesis and gonadotrophin receptors To the best of our knowledge this study is the first to show concentrations of steroids in follicles from the same women collected just prior to and 36 h after ovulation induction. This provided detailed information on the profound changes that occurred in steroidogenesis in connection with ovulation and the hormonal profile of the FFs supported the gene expression data. The expression of CYP17A1 and NR5A1 (nuclear receptor subfamily 5, group A, member 1) indicated minor theca cell contamination, which, however, did not seem to affect the overall results of this study confirming recent bovine data on minor theca cell contamination of GC preparations from follicle aspirates (Christenson et al., 2013). As in humans, macaques ovulate around 36 h after hCG administration. In macaque whole follicles, an increase in StAR and HSD3B2 was found 12 h after hCG induced ovulation, but decreased to pre-hCG levels 36 h after hCG in non-ruptured follicles and increased again in ruptured follicles, revealing a biphasic pattern of the gene expression of the progestogenic pathway (Xu et al., 2011). As we found unaltered levels of HSD3B2, LDLR (low density lipoprotein receptor), CYP11A1 36 h after ovulation but increased levels of progesterone in FF, we speculate that
in humans there could be a similar biphasic expression pattern of the progestogenic pathway in the periovulatory interval. The early increase of the progestogenic pathway could be important for locally induced actions of progesterone in the follicle whereas the later increase after follicle rupture facilitates CL formation and progesterone production for systemic as well as local purposes (Xu et al., 2011). Additional radical changes in steroidogenesis occur during the periovulatory period. The decrease of CYP171 (catalyzing androgen production) and CYP19A (converting androstenedione to estrogen) accompanied FF hormone changes and reflected the shift from estrogenic to progestogenic follicles after rhCG. The opposite expression patterns of HSD11B1 (up-regulation; promoting cortisol production from cortisone) and HSD11B2 (downregulation, promoting cortisone production from cortisol) confirmed previous findings in humans (Yding Andersen et al., 1999) and in macaque whole follicles (Fru et al., 2006). The very high concentration of biologically active cortisol in FF may serve to attenuate the local inflammatory reaction in the pre-ovulatory follicle preparing for ovulation (Andersen, 2002).
10
Wissing et al.
Figure 4 The human ovulatory cascade. Schematic overview of the molecular processes elicited by rhCG triggering for final oocyte maturation in humans, based on transcriptomic profiling of human GC and hormonal analyses of FF obtained before and 36 h after rhCG triggering. The ovary cartoon is a modification of a previously published cartoon (Grøndahl et al., 2013).
Table III Hormone concentration in FF before and 36 h after rhCG triggering. FF hormones
Before rhCG
36 h Post-rhCG
P-value
........................................................................................ Estrogen (nmol/l)
4651 + 2433
1302 + 1105
,0.0001
Progesterone (nmol/l)
2604 + 2211
18661 + 15688
,0.0001
132 + 225
42 + 23
,0.0001
Testosterone (nmol/l)
32 + 23
25 + 5
NS
AMH (pmol/l)
85 + 60
33 + 21
,0.0001
Androstendione (nmol/l)
Data are shown as mean + SD. P-values , 0.05 are considered significant (Wilcoxon matched-pairs signed rank test).
The decreased expression of FSHR and LHCGR was expected (Xu et al., 2011; Jeppesen et al., 2012) and probably reflected receptor downregulation (Hirsh et al., 2005).
Exit from cell cycle The LH/hCG surge induces down-regulation of cell cycle genes, in order to decrease GC proliferation and promote final GC differentiation (Stocco et al., 2007). Down-regulation of cell cycle genes and proliferation marker genes as CDC20, TOP2A and the MCM’s were in line with a previous report showing a decline of macaque GC entering S-phase 24 h after rhCG (Fru et al., 2007) and inhibition of GC mitosis in
macaque ovaries by a bolus of rhCG (Chaffin et al., 2001). Additionally, expression of cell cycle inhibitor genes as CDKN1A was increased after rhCG as reported in rodents (Stocco et al., 2007) and macaques (Chaffin et al., 2001). In macaques, CDKN1A increased early (12 h) after the ovulatory trigger whereas CDKN1B expression increased later (at 36 h) (Chaffin et al., 2001). CDKN1B expression was also dominant in rodent corpora lutea (Stocco et al., 2007). This suggested that CDKN1A is important for the initial decline in GC mitosis while CDKN1B might maintain cell cycle arrest in the luteinized cells of the CL. This could explain why we only observed an up-regulation of CDKN1A.
Inflammation In line with ovulation being compared with an acute inflammatory reaction in the ovary (Espey, 2006) we demonstrated an up-regulation of genes involved in inflammation, such as immune cell trafficking, interleukin-signaling, TNFa activation and prostaglandin synthesis. In primates, prostaglandins are critical for ovulation, but not for luteinisation (Duffy and Stouffer, 2002). In the present data set, PTGS2 was found to be heavily up-regulated (.13-fold). The specific PTGS2 inhibitor celecoxib has been shown to increase the incidence of delayed rupture of follicles and unruptured follicles in humans, but not as effectively as in monkeys, suggesting that in humans, there are other pathways ensuring follicle rupture (Duffy and Stouffer, 2002). In humans, the lipoxygenase pathway could, as for other species (Downey et al., 1998; Mikuni et al., 1998) play a role in ovulation as well, as we found an up-regulation of
11
Human granulosa cell transcriptome and ovulation
Table IV In silico upstream regulator analysis. Upstream regulator gene symbol
Gene title
Fold changea
Molecule type
Predicted activation state
Activation z-scoreb
P-value of overlap
Number of target molecules in dataset
............................................................................................................................................................................................. TP53
Tumor protein p53
CCND1
Cyclin D1
TGFB1
Transforming growth factor, beta 1
FOXM1
Forkhead box M1
CDKN1A
Cyclin-dependent kinase inhibitor 1A (p21, Cip1)
S100A6 PGR
Transcription regulator
Activated
Other 6.278 210.083
Growth factor
Activated
Transcription regulator
Inhibited
Kinase
Activated
S100 calcium binding protein A6
Transporter
Inhibited
Progesterone receptor
Ligand-dependent nuclear receptor
Activated
FSH
Follicle-stimulating hormone
Complex
TNF
Tumor necrosis factor
Cytokine
3.290
Activated
6.354
1.09E231
106
21.709
6.39E226
50
3.202
2.70E225
70
24.347
1.89E221
24
3.047
8.16E216
23
22.840
1.22E211
15
2.060
8.34E211
28
20.343
1.65E210
47
3.586
8.82E210
60
TP63
Tumor protein p63
Transcription regulator
20.150
1.14E209
29
LH
Luteinizing hormone
Complex
20.146
3.51E209
38
IGF1
Insulin-like growth factor 1 (somatomedin C)
Growth factor
3.162
5.04E208
14
BRCA1
Breast cancer 1, early onset
Transcription regulator
0.456
4.15E207
13
AR
Androgen receptor
Ligand-dependent nuclear receptor
1.167
8.47E207
20
PDGF BB
Platelet-derived growth factor beta
Complex
Activated
4.242
1.07E206
18
Other
Activated
VTN
Vitronectin
ESR1
Estrogen receptor 1
CTNNB1
Catenin (cadherin-associated protein), beta 1
EGF
Epidermal growth factor
IGF2
Insulin-like growth factor 2 (somatomedin A)
IL6R
Interleukin 6 receptor
CD24 (human)
CD24 molecule
IL6
Interleukin 6 (interferon, beta 2)
FOXO3
Forkhead box O3
EGR1
Early growth response 1
24.419
4.009 22.519
2.214
5.34E206
6
Ligand-dependent nuclear receptor
20.785
1.81E205
27
Transcription regulator
21.387
5.47E205
21
1.85E204
10
Growth factor
2.76E204
3
Transmembrane receptor
5.35E204
5
21.508
6.35E204
17
0.768
1.06E203
16
2.616
1.25E203
9
0.428
0.00132
6
Growth factor 26.427
80.104
Activated
Activated
Other Cytokine Transcription regulator
5.5
Activated
Transcription regulator
2.292
Selected list of suggested upstream regulators of the rhCG signal in human granulosa cells undergoing final follicle maturation. a If the suggested upstream regulator was among the differentially expressed genes in the present study, fold change was listed. b A positive z-score indicated activation and a negative z-score indicated inhibition.
ALOX5 and ALOX5AP after rhCG, but this remains to be further investigated. The increased expression of coagulation factors (F2R, F3), promoting coagulation in the microvasculature, extended previous findings in mice of increased F3 expression shortly after hCG triggering (Carletti and Christenson, 2009). The coagulation factors might prevent microbleeding during ovulation.
Angiogenesis and cancer-related genes More than 400 of the differentially expressed genes were connected to cancer in the biofunction enrichment analysis suggesting that CL formation is an example of a normal physiological process with gene expression similarities to tumor formation without leading to cancer in the right environment. Vasculogenesis and endothelial cell migration functions were highly enriched in the genes up-regulated after rhCG suggesting a
12 shift towards an invasive phenotype. EFNB2 (up-regulated .15-fold in GC 36 h after rhCG) and other ephrins were previously reported to be expressed in human luteinized GC (Xu et al., 2006). Blocking EFNB2 with highly specific antibodies inhibited angiogenesis, lymphangiogenesis and tumor growth in a xenografted mice model (Abe´ngozar et al., 2012). Thus, ephrins may be important for the induction of angiogenesis in the developing CL in humans. The reduced VEGFA (vascular endothelial growth factor A) expression could resemble the pattern of VEGFA expression in the primate periovulatory follicle, with a temporary decrease in expression after hCG followed by a rise after follicle rupture (Xu et al., 2011). Many ovulation-related genes have previously been associated with cancer and ovarian cancer. This suggests that some forms of ovarian cancer could originate from ovulatory molecular processes that have gone astray, supported by the fact that ovarian epithelial tumors developed and maintained a milieu rich in pro-inflammatory chemokines and cytokines (reviewed in Maccio` and Madeddu, 2012) which resembled the local inflammatory reaction in the follicle during ovulation. CD24 expression in ovarian cancer was related to tumor aggressiveness, especially cell invasion and chemotactic migration (Kang et al., 2013). The massive up-regulation of CD24 expression following rhCG exposure was new and unexpected. It enforces an association between CD24, vascularization and angiogenesis characterizing CL formation. The present observation calls for further investigations to determine the precise role of CD24 in these processes. CLDN11, highly up-regulated by rhCG and not previously related to ovulation, was overexpressed in ovarian cancer, promoting cancer cell invasion (Aravindakshan et al., 2006 (mice); English and Santin, 2013). TGF-b1 has been proposed as a local regulator of microvascular angiogenesis (Maroni and Davis, 2011) and has also been proposed as a facilitator of CL formation in pigs (Sriperumbudur et al., 2010). Interestingly, ovarian epithelial cancer cell lines have shown to be refractory to TGF-b1 signaling through loss of expression of TGF-b1/SMAD4 targets FBXO32 and RUNX1 T (Chou et al., 2010). FBXO32 was among the highly up-regulated genes described here for the first time as being involved in human ovulation. Hence, TGF-b1 and its target genes could serve as a control mechanism of the invasive process of CL formation. An activating FOXL2 mutation has been described in adult ovarian epithelial tumors (Ko¨bel et al., 2009), and besides the target genes StAR and CYP19A1, FOXL2 has pleiotropic effects in TGF-b/SMAD signaling. The down-regulation of FOXL2 36 h after rhCG further underlined the tightly regulated ovulation process.
Extracellular matrix remodeling and cell to cell contact An abundance of proteases have been implicated in ovulation, such as the Matrix metalloproteinases (MMPs), Tissue inhibitor of metalloproteinase (TIMPs), A Disintegrin And Metalloproteinase with Trombospondin Motifs (ADAMTSs), serine proteases and cathepsin L. ADAMTS1 has been shown to play a central role in human ovulation (Yung et al., 2010). Here, we report for the first time that ADAMTS9 is also massively induced by rhCG in human GC, in line with findings in macaque (Peluffo et al., 2011). ADAMTS1, cathepsin L, MMP-1 and TIMP-1 are up-regulated by progesterone in the primate ovary (Chaffin, 1999). We found MMP5 and TIMP1 to be up-regulated while TIMP3 was down-regulated, underlining the complexity and species differences of the process.
Wissing et al.
Other extracellular matrix genes that were highly up-regulated after the ovulatory surge, such as COL6A3 have previously been reported in human theca cells (Iwahashi et al., 2000) and in murine cumulus cells and GC in response to rhCG (Adriaenssens et al., 2009). We have shown for the first time in human GC that COL6A3 expression increased massively after the ovulatory surge. COL6A3 could be an important factor in the ovulatory process and in CL formation. ANKRD22, coding for a membrane integrated protein ankyrin repeat domain 22 with no known biological function, was highly up-regulated in the post-rhCG GC. A recent comparison of the transcriptome of corresponding cumulus and GC on individual preovulatory follicle level found ANKRD22 expression to be low in cumulus cells, while up to 17-fold higher expressed in GC (Grøndahl et al., 2012). This indicates a specialized, yet unknown function in the GC-compartment.
Growth factors Epidermal growth factor (EGF)-family members are induced by LH/hCG and propagate the LH/hCG stimulus in a autocrine and paracrine manner (Conti et al., 2006). Both AREG (amphiregulin) and EREG (epiregulin) expression was highly increased after rhCG, as expected (Zamah et al., 2010). INHBA (Inhibin bA) was significantly reduced, extending the findings of Hua et al., (2008) who found decreased INHBA in response to hCG in murine follicles cultured in vitro. The insulin-like growth factors (IGF1 and 2) are important intra-ovarian growth factors. IGF2 enhances FSH-induced GC proliferation and steroidogenesis (Mazerbourg et al. 2003). IGF2 and the genomic neighbor H19/mir675 were substantially reduced 36 h after rhCG administration in the present study, confirming previous findings in primate ovary of reduced Igf2 expression after hCG (Brogan et al. 2010). IGF2-H19 is a tandem gene, whose dual expression is regulated by imprinting: IGF2 is a growth-promoting peptide while H19/mir675 gives rise to several non-coding microRNAs that inhibit cell proliferation (Ratajczak, 2012). H19/mir675 was previously shown to reduce Igf1 receptor mRNA (Keniry et al. 2012), and interestingly, IGF1R was increased 1.4 times (P ¼ 0.0005) in the present study. The dramatic downregulation of H19/mir675 after rhCG suggests that H19/mir675 regulation plays an important role during ovulation that remains to be further elucidated. The role of AMH in the late phases of folliculogenesis remains unsettled (Andersen et al., 2010). The decrease in AMHR2 expression confirmed the effect of final maturation in human cumulus–oocyte complexes with lower AMHR2 expression in cumulus cells from expanded versus non-expanded cumulus cells (Grøndahl et al., 2011) and lower AMHR2 expression after hCG in rat cumulus–oocyte complexes (Agca et al., 2012). Higher FF AMH at oocyte retrieval has been linked to decreased pregnancy chance after in vitro fertilisation (Desforges-Bullet et al., 2010), suggesting that reduced AMH signaling is important for final follicular maturation. Our findings of decreased AMH in FF after rhCG support recent findings of decreased serum AMH levels through the periovulatory interval and the luteal phase (Hadlow et al., 2013).
Oocyte competence marker genes are found in the follicular to luteal transition expression profile Several of the genes differentially expressed, before and after rhCG, have been suggested as follicular health markers. Hamel et al. (2010)
13
Human granulosa cell transcriptome and ovulation
performed three consecutive gene expression studies using GC obtained in connection with OPU, comparing GC from follicles yielding oocytes that resulted in live birth versus GC from follicles resulting in no pregnancy. Interestingly, four of the nine proposed follicle health markers overlapped with our gene set of differentially expressed genes before and after rhCG (RGS2, CYP19A1, UGP2 (UDP-glucose pyrophosphorylase 2) and PHLDA1 (pleckstrin homology-like domain, family A, member 1)). Furthermore, we observed an overlap of our differentially expressed genes and reported cumulus cell marker genes: EFNB2 (Wathlet et al., 2012), PTGS2, PTX3, HAS2 (Gebhardt et al., 2011), RGS2, ANKRD22 and PLIN2 (perilipin 2) (Feuerstein et al., 2012). This overlap underlined the importance of the final maturational process induced by hCG/LH and downstream signals for the competence of the oocyte to undertake further development, and indicated that ovulation-related genes may serve as good markers of follicle and oocyte quality.
Conclusion One thousand one hundred and eighty-six genes were significantly differentially expressed when comparing GC isolated before and 36 h after rhCG administration and GC from these two hallmarks of folliculogenesis represent highly distinct gene expression profiles. Known ovulationrelated genes were confirmed and many new ovulation-related genes were discovered, such as CD24, FBXO32, ANKRD22, CLDN11 and H19/mir675. Up-regulated genes represented inflammation, angiogenesis, extracellular matrix and growth factors. Down-regulated genes represented cell cycle and proliferation. Seventy-two of the regulated genes had previously been related to ovarian cancer. FF hormone analyses confirmed the shift from estrogenic to progestogenic and further, a significant decrease in FF AMH after rhCG was documented. These data will serve as an easily accessible resource of GC genes involved in human ovulation and final oocyte maturation. The identification of enriched genes, pathways and intermediate regulators of the final follicle maturation is essential for understanding the complicated ovulatory process in more detail and the function of all the individual components involved.
Supplementary data Supplementary data are available at http://humrep.oxfordjournals.org/.
Acknowledgements The staff at Holbæk Fertility Clinic are gratefully acknowledged for help with collection of study material.
Authors’ roles M.L.W. planned the study, recruited patients, collected GC and FF, interpreted results and drafted the manuscript. S.G.K. extracted RNA and performed qRT-PCR. C.Y.A. conceived the idea and interpreted results. T.H.T. collected materials. A.L.M. recruited patients and collected materials. R.B. performed bio-informatical analyses of the microarray data. M.L.G. interpreted the results and drafted the manuscript. All authors have revised and approved the final version of the manuscript.
Funding Grant from Research Fund of Region Sjælland is gratefully acknowledged.
Conflict of interest None declared.
References Abe´ngozar MA, de Frutos S, Ferreiro S, Soriano J, Perez-Martinez M, Olmeda D, Marenchino M, Can˜amero M, Ortega S, Megias D et al. Blocking ephrinB2 with highly specific antibodies inhibits angiogenesis, lymphangiogenesis, and tumor growth. Blood 2012;119:4565 – 4576. Adriaenssens T, Mazoyer C, Segers I, Wathlet S, Smitz J. Differences in collagen expression in cumulus cells after exposure to highly purified menotropin or recombinant follicle-stimulating hormone in a mouse follicle culture model. Biology Reprod 2009;80:1015– 1025. Agca C, Yakan A, Agca Y. Estrus synchronization and ovarian hyperstimulation treatments have negligible effects on cumulus oocyte complex gene expression whereas induction of ovulation causes major expression changes. Mol Reprod Dev 2012;80:102 – 117. Andersen CY. Possible new mechanism of cortisol action in female reproductive organs: physiological implications of the free hormone hypothesis. J Endocrinol 2002;173:211 –217. Andersen CY, Schmidt KT, Kristensen SG, Rosendahl M, Byskov AG, Ernst E. Concentrations of AMH and inhibin-B in relation to follicular diameter in normal human small antral follicles. Hum Reprod 2010;25:1282 – 1287. Aravindakshan J, Chen X, Sairam MR. Differential expression of claudin family proteins in mouse ovarian serous papillary epithelial adenoma in aging FSH receptor-deficient mutants. Neoplasia 2006;8:984 – 994. Bolstad BM, Irizarry RA, Astrand M, Speed TP. A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 2003;19:185 – 193. Brogan RS, Mix S, Puttabyatappa M, VandeVoort CA, Chaffin CL. Expression of the insulin-like growth factor and insulin systems in the luteinizing macaque ovarian follicle. Fertil Steril 2010;93:1421 – 1429. Carletti MZ, Christenson LK. Rapid effects of LH on gene expression in the mural granulosa cells of mouse periovulatory follicles. Reproduction 2009; 137:843– 855. Chaffin CL. Expression of matrix metalloproteinases and their tissue inhibitor messenger ribonucleic acids in macaque periovulatory granulosa cells: time course and steroid regulation. Biol Reprod 1999;61:14 – 21. Chaffin CL, Schwinof KM, Stouffer RL. Gonadotropin and steroid control of granulosa cell proliferation during the periovulatory interval in rhesus monkeys. Biol Reprod 2001;65:755– 762. Chou J-L, Su H-Y, Chen L-Y, Liao Y-P, Hartman-Frey C, Lai Y-H, Yang H-W, Deatherage DE, Kuo C-T, Huang Y-W et al. Promoter hypermethylation of FBXO32, a novel TGF-beta/SMAD4 target gene and tumor suppressor, is associated with poor prognosis in human ovarian cancer. Lab Invest 2010; 90:414– 425. Christenson LK, Gunewardena S, Hong X, Spitschak M, Baufeld A, Vanselow J. Research resource: pre-ovulatory LH surge effects on follicular theca and granulosa transcriptomes. Mol Endocrinol 2013; 27:1153– 1171. Conti M, Hsieh M, Park J-Y, Su Y-Q. Role of the epidermal growth factor network in ovarian follicles. Mol Endocrinol 2006;20:715 – 723. Desforges-Bullet V, Gallo C, Lefebvre C, Pigny P, Dewailly D, Catteau-Jonard S. Increased anti-Mu¨llerian hormone and decreased FSH levels in follicular fluid obtained in women with polycystic ovaries at the time of follicle puncture for in vitro fertilization. Fertil Steril 2010;94:198–204.
14 Downey BR, Mootoo JE, Doyle SE. A role for lipoxygenase metabolites of arachidonic acid in porcine ovulation. Anim Reprod Sci 1998;49:269– 279. Duffy DM, Stouffer RL. Follicular administration of a cyclooxygenase inhibitor can prevent oocyte release without alteration of normal luteal function in rhesus monkeys. Hum Reprod 2002;17:2825 – 2831. English DP, Santin AD. Claudins overexpression in ovarian cancer: potential targets for clostridium perfringens enterotoxin (CPE) based diagnosis and therapy. Int J Mol Sci 2013;14:10412 – 10437. Espey LL. Comprehensive analysis of ovarian gene expression during ovulation using differential display. Methods Mol Biol 2006;317:219 – 241. Feuerstein P, Puard V, Chevalier C, Teusan R, Cadoret V, Guerif F, Houlgatte R, Royere D. Genomic assessment of human cumulus cell marker genes as predictors of oocyte developmental competence: impact of various experimental factors. PLoS One 2012;7:e40449. Fru KN, VandeVoort CA, Chaffin CL. Mineralocorticoid synthesis during the periovulatory interval in macaques. Biol Reprod 2006;75:568– 574. Fru KN, Cherian-Shaw M, Puttabyatappa M, VandeVoort CA, Chaffin CL. Regulation of granulosa cell proliferation and EGF-like ligands during the periovulatory interval in monkeys. Hum Reprod 2007;22:1247 – 1252. Gebhardt KM, Feil DK, Dunning KR, Lane M, Russell DL. Human cumulus cell gene expression as a biomarker of pregnancy outcome after single embryo transfer. Fertil Steril 2011;96:47 – 52.e2. Grøndahl ML, Yding Andersen C, Bogstad J, Nielsen FC, Meinertz H, Borup R. Gene expression profiles of single human mature oocytes in relation to age. Hum Reprod 2010;25:957 – 968. Grøndahl ML, Nielsen ME, Dal Canto MB, Fadini R, Rasmussen IA, Westergaard LG, Kristensen SG, Yding Andersen C. Anti-Mu¨llerian hormone remains highly expressed in human cumulus cells during the final stages of folliculogenesis. Reprod Biomed Online 2011;22:389 – 398. Grøndahl ML, Andersen CY, Bogstad J, Borgbo T, Boujida VH, Borup R. Specific genes are selectively expressed between cumulus and granulosa cells from individual human pre-ovulatory follicles. Mol Hum Reprod 2012;18:572 – 584. Grøndahl ML, Borup R, Vikesa˚ J, Ernst E, Andersen CY, Lykke-Hartmann K. The dormant and the fully competent oocyte: comparing the transcriptome of human oocytes from primordial follicles and in metaphase II. Mol Hum Reprod 2013;19:600– 617. Hadlow N, Longhurst K, McClements A, Natalwala J, Brown SJ, Matson PL. Variation in antiMu¨llerian hormone concentration during the menstrual cycle may change the clinical classification of the ovarian response. Fertil Steril 2013;99:1791 – 1797. Hamel M, Dufort I, Robert C, Le´veille´ M-C, Leader A, Sirard M-A. Identification of follicular marker genes as pregnancy predictors for human IVF: new evidence for the involvement of luteinization process. Mol Hum Reprod 2010;16:548– 556. Hirsh L, Ben-Ami I, Freimann S, Dantes A, Tajima K, Kotsuji F, Amsterdam A. Desensitization to gonadotropic hormones: a model system for the regulation of a G-protein-coupled receptor with 7-transmembrane spanning regions. Biochem Biophys Res Commun 2005;326:1–6. Hua VK, Fleming SD, Illingworth P. Effects of protein kinase A and C inhibitors on follicular inhibin and activin during ovulation. Reprod Biomed Online 2008;17:642 – 651. Iwahashi M, Muragaki Y, Ooshima A, Nakano R. Type VI collagen expression during growth of human ovarian follicles. Fertil Steril 2000;74:343 – 347. Jeppesen JV, Kristensen SG, Nielsen ME, Humaidan P, Dal Canto M, Fadini R, Schmidt KT, Ernst E, Yding Andersen C. LH-receptor gene expression in human granulosa and cumulus cells from antral and pre-ovulatory follicles. J Clin Endocrinol Metab 2012;97:E1524– E1531. Kang KS, Choi YP, Gao M-Q, Kang S, Kim BG, Lee JH, Kwon MJ, Shin YK, Cho NH. CD24(+) ovary cancer cells exhibit an invasive mesenchymal phenotype. Biochem Biophys Res Commun 2013;432:333 – 338.
Wissing et al.
Keniry A, Oxley D, Monnier P, Kyba M, Dandolo L, Smits G, Reik W. The H19 lincRNA is a developmental reservoir of mir-675 that supresses growth and Igf1r. Nat Cell Biol 2012;14:659 –665. Kristensen SG, Andersen K, Clement CA, Franks S, Hardy K, Andersen CY. Expression of TGF-beta superfamily growth factors, their receptors, the associated SMADs and antagonists in five isolated size-matched populations of preantral follicles from normal human ovaries. Mol Hum Reprod doi:10.1093/molehr/gat089. Ko¨bel M, Gilks CB, Huntsman DG. Adult-type granulosa cell tumors and FOXL2 mutation. Cancer Res 2009;69:9160– 9162. Maccio` A, Madeddu C. Inflammation and ovarian cancer. Cytokine 2012; 58:133 – 147. Maroni D, Davis JS. TGFB1 disrupts the angiogenic potential of microvascular endothelial cells of the corpus luteum. J Cell Sci 2011; 124:2501 – 2510. Mazerbourg S, Bondy CA, Zhou J, Monget P. The insulin-like growth factor system: a key determinant role in the growth and selection of ovarian follicles? a comparative species study. Reprod Domest Anim 2003; 38:247 – 258. Mikuni M, Yoshida M, Hellberg P, Peterson CA, Edwin SS, Bra¨nnstro¨m M, Peterson CM. The lipoxygenase inhibitor, nordihydroguaiaretic acid, inhibits ovulation and reduces leukotriene and prostaglandin levels in the rat ovary. Biol Reprod 1998;58:1211 –1216. Park J-Y, Su Y-Q, Ariga M, Law E, Jin S-LC, Conti M. EGF-like growth factors as mediators of LH action in the ovulatory follicle. Science 2004; 303:682– 684. Peluffo MC, Murphy MJ, Baughman ST, Stouffer RL, Hennebold JD. Systematic analysis of protease gene expression in the rhesus macaque ovulatory follicle: metalloproteinase involvement in follicle rupture. Endocrinology 2011;152:3963 – 3974. Ratajczak MZ. Igf2-H19, an imprinted Tandem gene, is an important regulator of embryonic development, a guardian of proliferation of adult pluripotent stem cells, a regulator of longevity and a “passkey” to cancerogenesis. Folia Histochem Cytobiol 2012;50:171 – 179. Sriperumbudur R, Zorrilla L, Gadsby JE. Transforming growth factor-beta (TGFbeta) and its signaling components in peri-ovulatory pig follicles. Anim Reprod Sci 2010;120:84 –94. Stocco C, Telleria C, Gibori G. The molecular control of corpus luteum formation, function, and regression. Endocr Rev 2007;28:117 – 149. Wathlet S, Adriaenssens T, Segers I, Verheyen G, Janssens R, Coucke W, Devroey P, Smitz J. New candidate genes to predict pregnancy outcome in single embryo transfer cycles when using cumulus cell gene expression. Fertil Steril 2012;98:432 – 439. Xu Y, Zagoura D, Keck C, Pietrowski D. Expression of Eph receptor tyrosine kinases and their ligands in human Granulosa lutein cells and human umbilical vein endothelial cells. Exp Clin Endocrinol Diabetes 2006; 114:590– 595. Xu F, Stouffer RL, Mu¨ller J, Hennebold JD, Wright JW, Bahar A, Leder G, Peters M, Thorne M, Sims M et al. Dynamics of the transcriptome in the primate ovulatory follicle. Mol Hum Reprod 2011;17:152 –165. Yding Andersen C, Morineau G, Fukuda M, Westergaard LG, Ingerslev HJ, Fiet J, Byskov AG. Assessment of the follicular cortisol:cortisone ratio. Hum Reprod 1999;14:1562 – 1568. Yung Y, Maman E, Konopnicki S, Cohen B, Brengauz M, Lojkin I, Dal Canto M, Fadini R, Dor J, Hourvitz A. ADAMTS-1: a new human ovulatory gene and a cumulus marker for fertilization capacity. Mol Cell Endocrinol 2010; 328:104– 108. Zamah AM, Hsieh M, Chen J, Vigne JL, Rosen MP, Cedars MI, Conti M. Human oocyte maturation is dependent on LH-stimulated accumulation of the epidermal growth factor-like growth factor, amphiregulin. Hum Reprod 2010;25:2569 – 2578.