The transcriptome of cervical ripening in human pregnancy before the onset of labor at term: identification of novel molecular functions involved in this process

Abstract

Objective:

The aim of this study was to identify changes in the cervical transcriptome in the human uterine cervix as a function of ripening before the onset of labor.

Study Design:

Human cervical tissue was obtained from women at term not in labor with ripe(n=11) and unripe(n=11) cervices and profiled using Affymetrix HGU133PLUS2.0 arrays. Gene expression was analyzed using a moderated t-test (False Discovery Rate 5%). Gene ontology and pathway analysis were performed. qRT-PCR was used for confirmation of selected differentially expressed (DE) genes.

Results:

1)91 genes were DE between ripe and unripe groups; 2)Cervical ripening was associated with enrichment of specific biological processes (e.g.cell adhesion, regulation of anatomical structure), pathways, and 13 molecular functions (e.g.extracelluar matrix (ECM)-structural constituent, protein binding, glycosaminoglycan binding); 3)qRT-PCR confirmed that 9 of 11 tested DE genes (determined by microarray) were up-regulated in a ripe cervix (e.g.MYOCD,VCAN,THBS1,COL5A1); 4)23 additional genes related to ECM metabolism and adhesion molecules were differentially-regulated (by qRT-PCR) in ripe cervices.

Conclusion:

1)This is the first description of the changes in the human cervical transcriptome with ripening before the onset of labor; 2)Biological processes, pathways and molecular functions were identified with the use of this unbiased approach; 3)In contrast to cervical dilation after term labor, inflammation-related genes did not emerge as differentially regulated with cervical ripening; 4)Myocardin was identified as a novel gene upregulated in human cervical ripening.

Background and Objective:

Cervical ripening is a critical component of the common terminal pathway of parturition, which also includes increased myometrial contractility and membrane/decidual activation.¹ The mechanisms of cervical change in pregnancy have been investigated in animals as well as in humans.^2–25 Critical hypothesis-driven studies of cervical biology have resulted in an improved understanding of the changes that occur in the cervical extracellular matrix during pregnancy and labor and delivery^{2, 3, 5, 6, 9–11, 18–20, 23, 26–36}. However, the current knowledge of the mechanisms involved in cervical change during pregnancy does not provide a complete understanding of the processes involved in human cervical ripening.

The use of high-dimensional biology techniques allow for the examination of the genome, transcriptome, proteome and metabolome¹⁵. The study of the transcriptome (changes in gene transcription) in a particular tissue provides a comprehensive, systematic, and unbiased description of genes differentially expressed in a specific condition or point in time¹⁵. The aim of this study was to characterize the cervical transcriptome in patients with a ripe cervix at term before the onset of labor compared to those with an unripe cervix.

Materials and Methods

Study Design

A cross-sectional study was performed in patients undergoing elective cesarean section at term with an unripe (n=11) and ripe cervix (n=11). As used in previous studies of cervical biology in pregnancy, a cervix with a Bishop score of ≥5 was defined as ripe¹¹. Patient inclusion criteria were: 1) term gestation (≥37 weeks), 2) no prostaglandin or oxytocin administration, 3) absence of histologic chorioamnionitis, 4) negative Neisseria gonorrhoeae and Chlamydia trachomatis determined by examination of cervical secretions, and 5) a normal Pap smear. Patients were invited to participate in a study which was approved by the Institutional Review Board, and provided written, informed consent. Patients underwent cervical biopsy following elective cesarean section without signs of labor. This procedure has been used extensively by investigators in the United States, Europe and other continents^{2, 3, 6, 7, 9–11, 13, 15, 16, 18, 23, 25–28, 31, 35, 37–40}. Half centimeter biopsies were obtained transvaginally from the anterior lip of the cervix at the 12 o’clock position and immediately snap frozen in liquid nitrogen or placed in RNAlater® (Ambion Inc., Austin, Texas) and stored at −70° C. No patients experienced complications from the cervical biopsy. The clinical and demographic data, obstetric and gynecological history, as well as pregnancy outcome were extracted from medical records.

Microarray Analysis

Microarray analysis was performed using the HGU133 PLUS 2.0 Affymetrix® arrays. Microarray statistical analysis included: 1) data preprocessing using the RMA algorithm⁴¹; 2) Calculation of nominal p-values (by combining the nominal p-values of all probesets of the same gene); 3) Genes with the false discovery rate (FDR) < 0.05 were considered statistically significant. Pathway analysis was performed on the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database using both an enrichment analysis and the Signaling Pathway Impact (SPIA) analysis^{42, 43}. Gene ontology analysis was performed using the GOstats package of Bioconductor⁴⁴.

Quantitative reverse transcription-polymerase chain reaction (qRT-PCR)

Real-time PCR-based human extracellular matrix and adhesion molecules RT2Profiler PCR Array (SA Bioscience Corporation, Frederick, MD) was used to screen the expression of 84 key related genes according to the manufacturer’s instructions. The validation of the results of the microarray for the myocardin gene was done individually by qRT-PCR analysis.

Results

Demographic and clinical characteristics of the study population are depicted in Table I.

Table I.

Study subjects. Patient demographics and clinical characteristics

	Term No Labor Unripe Cervix (n=11)	Term No Labor Ripe Cervix (n=11)	p
Age	28 (21–38)	29 (22–37)	NS
Parity	2 (0–4)	2 (0–5)	NS
Number of prior vaginal deliveries	0 (0–0)	0 (0–4)	NS
Gestational age at delivery (weeks)	39 (38–39)	39 (37–40)	NS
Bishop Score	2 (0–3)	7 (5–9)	p<0.0001
Cervical Dilation (cm)	0 (0–0.5)	2 (1–3.5)	p<0.0001

Results expressed as median (range)

NS, not significant.

Microarray analysis

Microarray analysis revealed that 91 genes were differentially expressed in the cervical tissue of patients not in labor with a ripe cervix when compared to those not in labor with an unripe cervix. Interestingly, 83 of 91 genes were up-regulated. The list of differentially expressed genes is presented in Table II, which describes the fold change and false discovery rate (FDR).

Table II.

Results of Microarray Analysis. Genes showing differential expression between Ripe versus Unripe Cervix in patients at term not in labor

Entrez Gene	Symbol	Gene Name	Fold Change	FDR
6925	TCF4	transcription factor 4	1.45	<0.00001
1462	VCAN	versican	2.82	<0.00001
23433	RHOQ	ras homolog gene family, member Q	1.88	<0.00001
4673	NAP1L1	nucleosome assembly protein 1-like 1	1.22	0.0001
7078	TIMP3	TIMP metallopeptidase inhibitor 3 (Sorsby fundus dystrophy, pseudoinflammatory)	2.01	0.0001
780	DDR1	discoidin domain receptor tyrosine kinase 1	1.22	0.0003
54796	BNC2	basonuclin 2	1.45	0.0003
7035	TFPI	tissue factor pathway inhibitor (lipoprotein-associated coagulation inhibitor)	1.67	0.0005
23213	SULF1	sulfatase 1	1.81	0.0007
800	CALD1	caldesmon 1	1.44	0.0007
641700	ECSM2	endothelial cell-specific molecule 2	1.85	0.0007
1289	COL5A1	collagen, type V, alpha 1	1.69	0.0008
91624	NEXN	nexilin (F actin binding protein)	2.34	0.0009
5069	PAPPA	pregnancy-associated plasma protein A, pappalysin 1	1.60	0.0013
5350	PLN	phospholamban	1.95	0.0013
9509	ADAMTS2	ADAM metallopeptidase with thrombospondin type 1 motif, 2	1.53	0.0013
389136	VGLL3	vestigial like 3 (Drosophila)	2.32	0.0013
253827	MSRB3	methionine sulfoxide reductase B3	1.45	0.0018
5136	PDE1A	phosphodiesterase 1A, calmodulin-dependent	1.31	0.0019
6876	TAGLN	transgelin	2.00	0.0022
171024	SYNPO2	synaptopodin 2	1.51	0.0026
152137	CCDC50	coiled-coil domain containing 50	1.44	0.0029
7168	TPM1	tropomyosin 1 (alpha)	1.66	0.0029
2316	FLNA	filamin A, alpha (actin binding protein 280)	1.52	0.0029
1634	DCN	decorin	1.39	0.0039
7070	THY1	Thy-1 cell surface antigen	2.19	0.0039
274	BIN1	bridging integrator 1	1.40	0.0044
56999	ADAMTS9	ADAM metallopeptidase with thrombospondin type 1 motif, 9	1.72	0.0044
3107	HLA-C	major histocompatibility complex, class I, C	1.78	0.0044
2335	FN1	fibronectin 1	1.52	0.0044
4162	MCAM	melanoma cell adhesion molecule	1.68	0.0044
10082	GPC6	glypican 6	1.50	0.0049
2099	ESR1	estrogen receptor 1	1.18	0.0050
1284	COL4A2	collagen, type IV, alpha 2	1.99	0.0050
2308	FOXO1	forkhead box O1	1.78	0.0050
633	BGN	biglycan	1.45	0.0050
11010	GLIPR1	GLI pathogenesis-related 1 (glioma)	1.42	0.0051
348	APOE	apolipoprotein E	1.48	0.0051
57381	RHOJ	ras homolog gene family, member J	1.54	0.0060
3535	IGL@	immunoglobulin lambda locus	1.63	0.0061
87	ACTN1	actinin, alpha 1	1.50	0.0066
3398	ID2	inhibitor of DNA binding 2, dominant negative helix-loop-helix protein	1.36	0.0081
23208	SYT11	synaptotagmin XI	1.68	0.0081
90853	SPOCD1	SPOC domain containing 1	1.92	0.0088
221981	THSD7A	Thrombospondin, type I, domain containing 7A	1.61	0.0136
30008	EFEMP2	EGF-containing fibulin-like extracellular matrix protein 2	1.68	0.0145
3992	FADS1	fatty acid desaturase 1	1.58	0.0147
1292	COL6A2	collagen, type VI, alpha 2	2.04	0.0147
728215	LOC728215	similar to transmembrane protein 28	1.95	0.0202
3384	ICAM2	intercellular adhesion molecule 2	1.84	0.0204
7837	PXDN	peroxidasin homolog (Drosophila)	2.18	0.0205
10186	LHFP	lipoma HMGIC fusion partner	1.45	0.0235
10979	FERMT2	fermitin family homolog 2 (Drosophila)	1.86	0.0248
5125	PCSK5	proprotein convertase subtilisin/kexin type 5	1.49	0.0255
3915	LAMC1	laminin, gamma 1 (formerly LAMB2)	1.57	0.0255
1290	COL5A2	collagen, type V, alpha 2	2.21	0.0257
93649	MYOCD	Myocardin	3.13	0.0266
55752	SEPT11	septin 11	1.53	0.0266
1809	DPYSL3	dihydropyrimidinase-like 3	1.78	0.0266
4499	MT1M	metallothionein 1M	1.57	0.0266
55193	PBRM1	polybromo 1	1.13	0.0266
7057	THBS1	Thrombospondin 1	1.59	0.0270
79006	METRN	meteorin, glial cell differentiation regulator	1.53	0.0282
5175	PECAM1	platelet/endothelial cell adhesion molecule (CD31 antigen)	1.80	0.0297
116159	CYYR1	cysteine/tyrosine-rich 1	1.51	0.0302
2534	FYN	FYN oncogene related to SRC, FGR, YES	1.47	0.0313
84937	ZNRF1	zinc and ring finger 1	1.25	0.0317
10579	TACC2	transforming, acidic coiled-coil containing protein 2	1.17	0.0320
1281	COL3A1	collagen, type III, alpha 1 (Ehlers-Danlos syndrome type IV, autosomal dominant)	1.85	0.0322
202052	DNAJC18	DnaJ (Hsp40) homolog, subfamily C, member 18	1.25	0.0322
6614	SIGLEC1	sialic acid binding Ig-like lectin 1, sialoadhesin	1.41	0.0322
56648	EIF5A2	eukaryotic translation initiation factor 5A2	1.38	0.0325
23452	ANGPTL2	angiopoietin-like 2	1.49	0.0325
3908	LAMA2	laminin, alpha 2 (merosin, congenital muscular dystrophy)	1.73	0.0325
256435	ST6GALNAC3	ST6 (alpha-N-acetyl-neuraminyl-2,3-beta-galactosyl-1,3)-N-acetylgalactosaminide alpha-2,6-sialyltransferase 3	1.93	0.0328
83416	FCRL5	Fc receptor-like 5	2.28	0.0336
7026	NR2F2	nuclear receptor subfamily 2, group F, member 2	1.49	0.0336
4208	MEF2C	myocyte enhancer factor 2C	1.32	0.0336
55194	C1orf78	chromosome 1 open reading frame 78	1.39	0.0336
8406	SRPX	sushi-repeat-containing protein, X-linked	2.81	0.0336
649	BMP1	bone morphogenetic protein 1	1.19	0.0343
10150	MBNL2	muscleblind-like 2 (Drosophila)	1.29	0.0343
10205	MPZL2	myelin protein zero-like 2	1.60	0.0349
4256	MGP	Matrix Gla protein	1.96	0.0360
109	ADCY3	adenylate cyclase 3	1.50	0.0363
3487	IGFBP4	insulin-like growth factor binding protein 4	1.84	0.0371
26020	LRP10	low density lipoprotein receptor-related protein 10	1.23	0.0411
10439	OLFM1	olfactomedin 1	1.59	0.0445
5793	PTPRG	protein tyrosine phosphatase, receptor type, G	1.46	0.0453
2901	GRIK5	glutamate receptor, ionotropic, kainate 5	1.18	0.0453
23129	PLXND1	plexin D1	1.49	0.0454

Genes are ranked in order of false discovery rate (FDR)

Direction of fold change denotes change in ripe cervix

Gene Ontology enrichment analysis⁴⁵ was used to gain insight into the biology defined by differential gene expression. This analysis revealed that 11 biological processes and 13 molecular functions were significantly enriched (see Table III). The focal adhesion and extracellular matrix-receptor interaction pathways were found to be significant by both SPIA and enrichment analysis (FDR <0.05).

Table III.

Gene Ontology Analysis

Biological Process Category	Genes in Significant List (91 genes)	Genes on Array	p value*
Cell adhesion	12	396	<0.0001
Regulation of anatomical structure	6	71	<0.0001
Regulation of locomotion	5	67	0.005
Multicellular organismal development	25	1891	0.005
Cell motility	6	136	0.008
Phosphate transport	5	82	0.009
Localization	24	1777	0.01
Regulation of cellular component organization and biogenesis	7	207	0.01
Regulation of body fluid levels	5	110	0.02
Skin development	2	7	0.03
Blood coagulation	4	69	0.03
Molecular Function Category	Genes in Significant List(91 genes)	Genes on Array	p value*
extracellular matrix structural constituent	10	89	2.11×10−8
protein binding	57	6680	0.0057
structural molecule activity	13	648	0.0063
integrin binding	4	43	0.0063
glycosaminoglycan binding	5	100	0.0127
polysaccharide binding	5	103	0.0127
pattern binding	5	114	0.0173
heparin binding	4	78	0.0304
actin filament binding	3	37	0.0304
actin binding	7	284	0.0304
cytoskeletal protein binding	8	403	0.0495

^*p values were derived using a hypergeometric distribution and were subsequently FDR corrected

qRT-PCR

Of the 85 genes tested by qRT-PCR (84 in the RT2Profiler PCR Array and myocardin), thirty-two genes showed differential mRNA expression in patients with a ripe cervix when compared to those with an unripe cervix (See Table IV). We were able to confirm the findings of microarray for 9 of 11 genes whose expression was also up-regulated by microarray analysis in patients with a ripe cervix. Myocardin (MYOCD), versican (VCAN), thrombospondin 1 (THBS1), and collagen, type V, alpha1 (COL5A1) were among the genes upregulated in patients with a ripe cervix (e.g. p<0.05) in both the microarray and qRT-PCR assays (Figures 1 and 2). Cadherin 1, type 1, E-cadherin (epithelial), (CDH1), was significantly downregulated in the microarray analysis; this result was confirmed by qRT-PCR. GAPDH was significantly downregulated in women with a ripe cervix based upon qRT-PCR analysis. Furthermore, 21 additional genes related to extracellular matrix metabolism and adhesion molecules were up-regulated in ripe cervices as demonstrated by the multiplex PCR array. (e.g. ADAM metallopeptidase with thrombospondin type I motif, 8 (ADAMTS8), vascular adhesion molecule 1 (VCAM1), collagen types IV, alpha 2 (COL4A2), and thrombospondin 1, TIMP 1) (See Table IV). The results of the qRT-PCR for all 32 genes matched the direction of change or demonstrated significance as suggested by the microarray data.

Table IV.

Results of qRT-PCR: 32 genes with differential expression between Ripe versus Unripe Cervix in patients at term not in labor

Symbol	Gene Name	Fold Change	P value
MYOCD	myocardin	3.6	*0.002
VCAN	versican	3.4	*0.012
THBS1	thrombospondin 1	2.1	*0.006
FN1	fibronectin 1	1.9	*0.031
COL4A2	collagen, type IV, alpha 2	1.9	*0.013
COL6A2	collagen, type VI, alpha 2	1.7	*0.013
LAMC1	laminin, gamma 1	1.7	*0.018
LAMA2	laminin, alpha 2	1.7	*0.013
COL5A1	collagen, type V, alpha 1	1.6	*0.037
CTNND2	catenin, delta 2	5.5	0.008
MMP3	matrix metallopeptidase 3 (stromelysin 1, progelatinase)	5.3	0.042
SELE	selectin	3.1	0.036
ADAMTS8	ADAM metallopeptidase with thrombospondin type 1 motif, 8	2.9	0.010
CTGF	connective tissue growth factor	2.8	0.032
COL8A1	collagen, type VIII	2.6	0.020
VCAM1	vascular cell adhesion molecule 1	2.5	0.017
ITGB3	integrin, beta 3 (platelet glycoprotein IIIa, antigen CD61)	2.3	0.012
SELP	selectin P (granule membrane protein 140kDa, antigen CD62)	2.1	0.017
TIMP1	TIMP metallopeptidase inhibitor 1	2.0	0.017
VTN	vitronectin	2.0	0.039
ITGA5	integrin, alpha 5 (fibronectin receptor, alpha polypeptide)	2.0	0.004
COL15A1	collagen, type XV, alpha 1	2.0	0.015
SGCE	sarcoglycan, epsilon	1.9	0.025
COL6A1	collagen, type VI, alpha 1	1.7	0.043
KAL1	Kallmann syndrome 1 sequence	1.7	0.012
LAMB1	laminin, beta 1	1.6	0.012
SPARC	secreted protein, acidic, cysteine-rich (osteonectin)	1.6	0.008
ITGB1	integrin, beta 1 (fibronectin receptor, beta polypeptide)	1.6	0.033
ADAMTS1	ADAM metallopeptidase with thrombospondin type 1 motif, 10	1.6	0.042
CDH1	cadherin 1, type 1, E-cadherin (epithelial)	-1.5	0.002
TGFB1	transforming growth factor, beta 1	1.5	0.048
TIMP2	tissue inhibitor of metalloproteinase 2	1.5	0.025

All genes in this table, except for CDH1, show higher expression levels in Ripe cervix compared to Unripe cervix.

^*significant by microarray and PCR analysis; fold change ≥ 1.5 and p < 0.05

Direction of fold change denotes change in ripe cervix

Figure 1:

Results of qRT-PCR assay of versican in cervical tissue in patients at term without labor: Unripe Cervix versus Ripe Cervix

Figure 2:

Results of qRT-PCR assay of myocardin in cervical tissue in patients at term without labor Unripe Cervix versus Ripe Cervix

Comment

Principal findings of the study:

1) This genome wide study has demonstrated that ninety-one genes were differentially expressed in patients at term not in labor with a ripe cervix when compared to those with an unripe cervix (83 up-regulated and 8 down-regulated); 2) Gene Ontology analysis indicated that cervical ripening was associated with enrichment of specific biological processes (e.g. cell adhesion, regulation of anatomical structure, regulation of locomotion, and phosphate transport) and 13 molecular functions (e.g. extracellular matrix structural constituent, protein binding, glycosaminoglycan binding, heparin binding 3) Pathway analysis identified involvement of focal adhesion, extracellular matrix-receptor interaction, cell communication and cell adhesion molecule pathways in the transcriptome differences between ripe and unripe cervices; 4) Genes previously reported to be involved in cervical remodeling [e.g. versican, biglycan, decorin] were up-regulated in the cervical tissue of patients with a ripe cervix; 5) This study identifies a new set of genes involved in cervical ripening, such as myocardin, ADAMTS8 and catenin. Many other genes not previously known to be differentially regulated with cervical ripening were also identified; and 6) In contrast to cervical dilation after term labor,^{12, 15} inflammation-related genes did not emerge as differentially expressed with cervical ripening.

Meaning of the study:

Disorders of cervical ripening complicate term (e.g. arrest of dilatation or protracted dilatation) and preterm (e.g. premature cervical dilation in the midtrimester) pregnancies. Samples of human cervical biopsy specimens have been obtained from patients in preterm and term labor, preterm and term non-labor and non-pregnant patients after a hysterectomy to describe the cervical extracellular matrix and its relationship to abnormal labor.^{2–13, 15, 16, 18, 25, 28, 46–48} Animal studies have also provided insight into the changes in extracellular matrix that occur in the uterine cervix during pregnancy.^{17, 19–22, 28, 33, 34, 36, 49–54}. Studies of cervical biopsy tissue have been conducted in pregnant women as early as 1960.¹⁸ Yet, the precise mechanism of cervical ripening in human pregnancy has not been fully elucidated. The current study represents the first description of the changes in the human cervical transcriptome in unripe versus ripe cervices. Some of our results confirm differential expression of genes previously implicated in cervical ripening, such as those involved in extracellular matrix metabolism and cell adhesion molecules. Interestingly, the novel finding of increased expression of myocardin in patients with a ripe cervix when compared to those with an unripe cervix was demonstrated. In addition, inflammation-related genes were not differentially expressed in patients with cervical ripening. The results reported herein characterize the processes involved in cervical ripening in humans before the onset of labor. An unbiased microarray analysis of the cervical tissue was carried out, followed by confirmation of selected genes by the use of qRT-PCR. A separate study must be conducted in order to confirm our results with an independent set of samples. Such studies are not easy to conduct because of the difficulties in obtaining these samples.

The cervix is comprised of smooth muscle and extracellular matrix, which consists of collagen, elastin, proteoglycans, and glycoproteins such as fibronectins.^55,56 The proteoglycans found in the cervix include decorin, fibromodulin, biglycan, versican, aggrecan, and heparan sulfate proteoglycan.^{35, 38, 57, 58} Examination of cervical biopsies from non-pregnant women with a history of cervical insufficiency suggests that increased distensibility of the cervix during pregnancy can be a result of pre-pregnancy decreased collagen concentration and perhaps a pre-pregnancy increased smooth muscle content, i.e. a congenital abnormality of the cervix^{27, 59, 60}. Some investigators have suggested that the period of cervical ripening can be divided into a ‘slow’ and a ‘fast’ phase.³⁵ The current study suggests that the ‘slow’ ripening process is up regulated in women with a Bishop score above 5 before the start of labor, whereas there is no indication of activation of the ‘fast’ ripening process involving inflammatory mediators. Furthermore, the up regulation of myocardin suggests that high muscle content in the cervix should be considered as a possible etiology for a ripe cervix.

Known and novel processes involved in cervical ripening during pregnancy Collagen types IV, V, and VI

Our study demonstrated an increased mRNA expression (both in microarray and qRT-PCR) of collagen types IV (alpha 2), V (alpha1), and VI (alpha 2) in patients with a ripe cervix. Collagens type IV, V, VI have not previously been studied during cervical ripening. Collagen type IV, alpha 2 is the major structural component of basement membranes and interacts with laminin and proteoglycans^{32, 61–63}. Collagen type V has been implicated as a critical determinant of fibril structure and matrix organization⁶⁴, while collagen type VI is found in most tissues and interacts with type IV collagens and the basement membrane and the surrounding matrix⁶⁵. The major structural collagens, types I and III, that dominate the cervix quantitatively, were not differentially regulated in this study, despite the fact that there is a known decrease in their concentrations in the cervix during pregnancy at term³¹.

Proteoglycans

Versican a large extracellular matrix proteoglycan, was up-regulated 3-fold by both microarray and qRT-PCR analysis in patients with a ripe cervix. Biglycan (1.5 fold) and decorin (1.4 fold) were also upregulated in patients with a ripe cervix as demonstrated by microarray analysis. These proteoglycans have many functions within the extracellular matrix which include effects upon collagen disorganization, cell adhesion, migration, and proliferation⁶⁶.

Of interest, the molecular function term ‘heparin binding’ was significant after Gene Ontology analysis of the differentially regulated genes. Recently, the role of heparin in cervical remodeling has been examined. Ekman-Ordeberg and colleagues have demonstrated that low molecular weight heparin increased IL-8 secretion in cervical fibroblasts⁶⁷. This area of research has great potential for targeting the mechanisms involved in cervical change during pregnancy.

Metalloproteinases

Matrix metalloproteinases (MMPs) are major regulators of the extracellular matrix⁶⁸ and have been implicated as possible mediators in the cervical remodeling process by cleaving one or more constituents of the extracellular matrix^{50, 69}. In ripe cervices, matrix metalloproteinase-3 (MMP-3, stromelysin) was upregulated 5-fold by qRT-PCR but not by microarray analysis. MMP-3 degrades fibrillin, a glycoprotein that is critical for the formation of elastic fibers in connective tissue⁷⁰. In addition, administration of antiprogesterone in a rabbit model results in augmentation of MMP-3 in the uterine cervix⁷¹. The finding reported here, of a 5-fold increase in MMP-3 mRNA expression in the cervical tissue of patients with a ripe cervix when compared to patients with an unripe cervix, suggests a role for MMP-3 in human cervical ripening.

Furthermore, MMP3 has been shown to cleave fibrinogen, cross-linked fibrin, the cell adhesion molecule E-cadherin, and exhibits proteolytic activity on laminin, alpha-2-macroglobulin, fibronectin, casein, and alpha-1-antitrypsin ^72–74. In the current study, E-cadherin (CDH1), a calcium dependent cell-cell adhesion molecule was significantly downregulated (by qRT-PCR) while laminin and fibronectin (by both the microarray and qRT-PCR analysis) were upregulated in patients with a ripe cervix when compared to those with an unripe cervix in.

In addition, the molecular function categories of actin filament binding, actin binding, and cytoskeletal binding were among those that were significant in the current study. The interaction of E-cadherin with the actin cytoskeleton has been shown to be directly regulated by the epidermal growth factor receptor in a breast cancer cell line ⁷⁵. Further study is required to elucidate the mechanism of action of E-cadherin as it relates to MMP-3, fibronectin, laminin and actin in human cervical ripening.

Delta-2-catenin (CTNND2) mRNA was increased by 5-fold in patients with a ripe cervix when compared to those with an unripe cervix. Delta-2-catenin is involved in cell adhesion and movement⁷⁶ and has not previously been described as playing a key role in cervical ripening or remodeling in pregnancy. In addition, delta-catenin interacts with E-cadherin and beta-catenin and has been implicated in the organization of cell-cell junctions.⁷⁷The precise role of delta-2-catenin in cervical ripening is unknown, and future research in this area is warranted.

We found that ADAMTS8 and ADAMTS1 mRNA expression were increased in the cervical tissue of patients with a ripe cervix when compared to those with an uripe cervix.. ADAMTS (a disintegrin and metalloproteinase with thrombospondin motifs) is thought to participate in the degradation of extracellular matrix.^78,79. Members of the ADAMTS family are secreted enzymes, and several of these bind to the extracellular matrix. The proteases ADAMTS-4, −5, and −8 degrade aggrecans and hence have been designated as “aggrecanases”⁷⁹. ADAMTS8 cleaves aggrecan at the aggrecanase-susceptible Glu373-Ala374 peptide bond⁸⁰. In addition, ADAMTS-1 and −4 cleave versican^81,82. Ruscheinsky demonstrated an up-regulation of ADAMTS1 in the cervix before birth in a mouse model⁸³. Thus, the cleavage of aggrecan and versican by ADAMTS-1 and −8 might contribute to the changes in proteoglycans observed during the process of cervical ripening before the onset of labor in humans.

Cervical ripening and lack of differential expression of inflammation-related genes

Liggins was the first to propose that cervical ripening can be likened to an inflammatory process.³ This concept was based largely upon histological observations of the uterine cervix after cervical ripening. However, recently, the role of inflammation in cervical ripening has been debated.^{14, 84,7, 10, 13} Sakamoto et al reported that there was no correlation between the degree of clinical cervical ripening and IL-8 concentrations in cervical tissue. In contrast, IL-8 concentrations in the cervical tissue increased after labor and delivery.¹⁰ In a mouse model with a transgene insertion on chromosome 6, Word et al¹⁴ reported that parturition did not occur despite uterine contractions because of a rigid non-elastic cervix at term. Unexpectedly, cervical ripening was not observed following the infiltration with neutrophils and macrophages in the cervical tissue.¹⁴ Similarly, Timmons and Mahendroo have challenged the importance of the influx of inflammatory cells as a major regulatory event of cervical ripening using steroid 5 alpha-reductase type 1 null mice (Srd5a1^−/−)⁸⁴. The investigators concluded that cervical ripening does not require activation of a typical inflammatory response, because macrophages, eosinophils, and myloperoxidase activity did not increase during cervical ripening. Moreover, depletion of neutrophil numbers (after injection of a rat anti-mouse monoclonal antibody directed against Ly6G (GR1), an antigen on the surface of mature murine neutrophils) before birth has no effect on the timing or success of parturition.⁸⁴ Of importance, experiments conducted in tissues collected before the onset of labor and after vaginal delivery demonstrated overexpression of genes involved in neutrophil chemotaxis, apoptosis and extracellular matrix regulation.^{12, 15} However, these studies should not be interpreted to represent the biology of cervical ripening because the observed alterations in gene expression may be due to the process of parturition (dilatation, remodeling, etc.) rather than the events that prepare the cervix for the onset of labor.

In the present study, increased mRNA expression (based upon microarray and qRT-PCR analysis) of fibronectin 1 (FN1), laminin, gamma 1, laminin alpha 2, collagen type IV alpha 2, collagen type V alpha 1, and collagen type VI alpha 2 was demonstrated in patients with a ripe cervix when compared with those with an unripe cervix. These genes are known to be involved in the focal adhesion, extracellular matrix interaction and cell communication pathways. In addition, the novel genes encoding for Delta-2-catenin and myocardin were upregulated in patients with a ripe cervix. In contrast, genes involved in the inflammatory pathway were not differentially regulated based upon microarray analysis. It is possible that the changes represented in the current study are those of early cervical ripening.

A novel gene involved in cervical ripening - myocardin

Myocardin mRNA expression was up-regulated in both microarray and qRT-PCR analysis (3.6 fold) in the cervical tissue of patients with a ripe cervix. This is a new factor possibly involved in cervical ripening that has never been described before as playing a role in this process. Myocardin, expressed in smooth and cardiac muscle lineages, has been named as a serum response factor transcriptional coactivator.^85–87 Myocardin activates smooth muscle differentiation, can carry out this function in non-muscle cells, and has been described as a ‘master regulator’ of smooth muscle gene expression⁸⁸. Although the uterine cervix only contains 10–15% smooth muscle,²⁶ cervical ripening may not only include changes in the extracellular matrix, but also alterations in the smooth muscle component of the cervix. These findings require further investigation into the mechanism, localization, and significance of myocardin in cervical change in the pregnant uterine cervix.

Human cervical ripening

The traditional view is that cervical ripening occurs during the last few weeks of pregnancy prior to the onset of labor. Indeed, the Bishop score, which is widely used to assess the state of cervical ripening, was first introduced as a method to predict the likelihood that a patient would go into spontaneous labor based upon digital examination of the cervix (effacement, dilatation, consistency and position). Although attempts have been made to generate an objective definition of cervical ripening, clinical examination remains the standard (Bishop score or modification of this system). The clinical diagnosis of cervical ripening in animals presents challenges. The conduction of this study in pregnant women in which the Bishop score has been determined allows examination of the relationship between cervical ripening in the human and the transcriptome.

More importantly, the mechanism of action of specific treatments (e.g. prostaglandins or mechanical devices) is not known. Cervical ripening is likely the result of several processes which may involve more than changes in the extracellular matrix. The understanding of normal cervical ripening is a first step to understanding premature or protracted cervical ripening.

Strengths, Limitations and Future investigations

The current study represents the first description of the changes in the human cervical transcriptome in unripe versus ripe cervices. Contrary to former hypothesis driven studies on cervical ripening, this method is unbiased thus allowing for discovery of new pathways. Some of our results confirm differential expression of genes previously implicated in cervical ripening before start of labor. Furthermore novel genes, including myocardin were suggested.

Our results provide several hypotheses for future exploration. The validation of the reported results by the use of a second set of samples will be required. In addition, the analysis of a larger sample size would be optimal. The role of myocardin in cervical ripening must be validated and further characterized by the performance of studies of immunohistochemistry and protein expression. In addition, analysis of the cervical tissue will allow for examination of smooth muscle staining. Furthermore, histological examination of the tissue of patients with a ripe cervix should be analyzed for the presence of infiltration of neutrophils and macrophages in order to further elucidate the role of inflammation in cervical ripening. Of importance, follow-up studies must include a demonstration of changes in protein expression for those genes that have been reported to be significantly altered in patients with a ripe cervix.