Objective
Distinct processes govern transition from quiescence to activation during term (TL) and preterm labor (PTL). We sought gene sets that are responsible for TL and PTL, along with the effector genes that are necessary for labor independent of gestation and underlying trigger.
Study Design
Expression was analyzed in term and preterm with or without labor (n = 6 subjects/group). Gene sets were generated with logic operations.
Results
Thirty-four genes were expressed similarly in PTL/TL but were absent from nonlabor samples (effector set); 49 genes were specific to PTL (preterm initiator set), and 174 genes were specific to TL (term initiator set). The gene ontogeny processes that comprise term initiator and effector sets were diverse, although inflammation was represented in 4 of the top 10; inflammation dominated the preterm initiator set.
Conclusion
TL and PTL differ dramatically in initiator profiles. Although inflammation is part of the term initiator and the effector sets, it is an overwhelming part of PTL that is associated with intraamniotic inflammation.
Successful parturition requires the synchronization of uterine events (myometrial activation, myometrial contraction, cervical ripening, cervical dilation, and rupture of the fetal membranes). The myometrium specifically must undergo a series of molecular and biochemical changes that transition it from the phase of quiescence, which is characterized by a loss of responsiveness to contractile agents, to the phase of activation with subsequent onset of labor. Many investigators have assumed the triggers of parturition are similar, regardless of the gestational age (GA), and have focused on the control of term labor as a surrogate for preterm labor. This approach has not met with great success.
Genomics helps expand and characterize the number of molecular pathways that potentially define an underlying condition. Since our initial report in 2000, oligonucleotide/complementary DNA microarrays have been used by many groups for the study of biologic processes that are involved in normal term labor. The results of such studies with the use of myometrium that is obtained from term laboring women suggest that contractile stimulators of labor include, but are not limited to, hormone receptors, cell adhesion molecules, interleukins (ILs), prostaglandins, and gap junctions. Other investigators have attempted to correlate human myometrial transcriptional levels at term with samples from idiopathic, spontaneous preterm labor and have concluded that the mechanisms underlying parturition at all stages of pregnancy are related. This is surprising because there are several well accepted causes of spontaneous preterm birth. Although molecules within these groups may well influence parturition during normal labor, there has been little study of myometrial gene expression in women with either preterm or dysfunctional term/preterm labor. Array type investigations have also been used to explore myometrial gene expression in the pregnant rodent. However, the resulting conclusions should be interpreted cautiously because the gestational and hormonal patterns of these species are not homologous to the human. Although these studies provide insight, they tend as a group to be biased by the limited number of genes that are included on the array and the number of genes that are selected for confirmatory study.
Rather than compare term labor to 1 of several models for preterm birth, we investigated human pregnancy and hypothesized that there should be a core set of genes the expression of which are necessary for the process of labor independent of GA and the underlying trigger for the labor. We further hypothesized this effector gene set should relate to myometrial contractility and the cellular activities necessary to sustain it. We also hypothesized there must be initiator gene sets that are responsible for the transition of the myometrium from quiescence to activation. But unlike the effector set that would be unaffected by the underlying labor stimulus, there should be separate initiator gene sets for term and preterm labor that reflect the underlying mechanism of labor. The potential result of identifying these gene sets could be the development of alternative treatment options that can be targeted at preterm and term or dysfunctional labor. The purpose of this investigation was to test our hypotheses in a series of women who were undergoing cesarean delivery at term or preterm in labor or absent labor.
Materials and Methods
Study design
Myometrium was obtained from the upper pole of the transverse lower uterine segment incision of 4 groups of women (n = 6 per group) at the time their primary cesarean section at Yale University: (1) preterm not in labor and no inflammation (PTNL; mean GA, 28.8 weeks; range, 25.4–32.5 weeks), (2) preterm in labor with inflammation (PTL; mean GA, 29.7 weeks; range, 25.1–32.6 weeks), (3) term not in labor (TNL; mean GA, 39.3 weeks; range, 38.4–41.0 weeks), and (4) term labor (TL; mean GA, 40.2 weeks; range, 39.0–41.2 weeks). The Yale University Institutional Review Board approved the protocol for sample collection, and all women provided informed written consent. Labor was defined by the presence of regular uterine contractions accompanied by progressive cervical dilation. The diagnosis of intraamniotic inflammation was based on an amniotic fluid mass restricted score of 3 or 4 plus >100 white blood cells/μL 3 in the context of a positive amniotic fluid culture in a sample that was obtained by transabdominal amniocentesis. These tests provided the most accurate tools currently available to maximize the likelihood of sample homogeneity. The mass restricted score provides qualitative information regarding the presence or absence of intraamniotic inflammation. Briefly, the score ranges from 0-4, depending on the presence (assigned a value of 1) or absence (assigned a value of 0) of each of 4 protein biomarkers. A score of 3-4 indicates inflammation, whereas a score of 0-2 excludes it. This biomarker pattern is predictive of preterm birth, histologic chorioamnionitis, and adverse neonatal outcome. A detailed description of the mass restricted method has been published previously. Indications for cesarean delivery in the PTNL group were all related to preeclampsia and in the TNL were all related to breech presentation. The indication for cesarean delivery in the TL group was an arrest of cervical dilation at ≥6 cm. No normal laboring patient at term underwent cesarean section delivery in this sample. Clinical data were retrieved from the medical records, and statistical analysis of patient demographics was performed using 1-way analysis of variance, followed by Newman Keuls Post Hoc test. All laboratory studies were performed at either the University of Kansas School of Medicine or the University of Maryland Baltimore School of Pharmacy.
Isolation of RNA
Total RNA was isolated using TRIzol Reagent (Invitrogen, Carlsbad, CA), according to the manufacturer’s protocol. The purity and integrity of each RNA sample was assessed by spectroscopy and formaldehyde-agarose gel electrophoresis.
Microarray preparation
Myometrial gene profiling was performed on each individual RNA sample with the Affymetrix GeneChip Human Genome U133 Plus 2.0 microarray (Affymetrix, Santa Clara, CA) that contained 38,500 human genes. RNA quality was reassessed before spotting with the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). Once the quality was confirmed, biotin-labeled complementary RNA was generated, and 20 μg of the sample was hybridized to the microarrays, using the manufacturer’s standard conditions. Image processing used an Affymetrix GeneArray 3000 scanner.
Microarray data processing and statistical analysis
Oligonucleotide microarrays were analyzed with the Affymetrix Expression Console. Gene expression levels were normalized by the R Statistical package and software that is available through the Bioconductor Project ( www.bioconductor.org ). The process normalizes gene expression using a background adjustment procedure and a sequence-specific expression method as described by Wu et al. Normalized microarray data were then subject to further discriminative analysis. A detection probability value was used to make a reliable call of gene expression (present, marginal, or absent). Genes that were present in <4 of the 6 patient samples were classified as absent overall and not considered further.
To identify the effector and initiator gene sets, we conducted a series of logic operations based on gene presence (detection probability value) and illustrated by the Venn diagram in Figure 1 . We first generated 4 groups of genes: group A = (PTL – PTNL); group C = (TL – TNL); group B = (group A – TNL); group D = (group C – PTNL).
The effector gene set was defined as those genes common to groups B and D. The effector set was further filtered to exclude genes whose expression during PTL changed ≥2-fold in either direction compared with TL, reasoning there was something unique to the disease state that influenced gene expression. The preterm initiator gene set was defined as those genes in group A exclusive to PTL (minus group C). Last, the term initiator gene set was defined as those genes in group C exclusive of TL (minus group A).
Network analysis and biologic classification
Identification and visual analysis of gene networks were performed using MetaCore analytical suite (version 2.0; GeneGo, St. Joseph, MI). There were few direct interactions identified among the individual proteins of the effector, preterm initiator, and term initiator genes. Therefore, we analyzed the possible networks that could be built from such relatively large gene sets using an algorithm that generates subnetworks that are highly saturated with input genes from each dataset (ie, effector set, preterm initiator set, and term initiator set). In this case, the network algorithm takes a list of root nodes (ie, genes from effector, preterm initiator, and term initiator sets) and for each 1 creates shortest paths networks to other root nodes in the list and stops the network expansion at a size of 50 nodes (standard number predefined in MetaCore). The resulting networks are evaluated and ranked according to their statistical significance (probability values). This high trust probability value calculation was also used to evaluate the network’s relevance to gene ontology (GO) biologic processes classification. In general, GO is a method of standardizing the representation of gene and gene product attributes across species and databases. Three GO functional ontologies (processes, molecular functions, and localizations) are used in Metacore software for enrichment analysis. The advantage of using this network algorithm is that it may find a well-connected cluster of root nodes without any predefined restrictions and, as a result, offers more flexibility in the identification of possible connections. These interactions were then assigned to specific biologic processes, cellular components, and/or molecular functions to further characterize the underlying condition to yield insight into the underlying mechanisms. We selected some genes of interest for confirmation by quantitative real time–polymerase chain reaction (qRT-PCR) based on their participation in subnetworks with a low probability value. Other genes that were selected from the input list were selected for qRT-PCR analysis based on their statistically significant association with GO processes.
qRT-PCR
Total RNA from the original samples used for the microarrays was used to quantify each gene. Primer sequences for amplifications were based on previously published complementary DNA sequences with the Beacon Designer program (BioRad, Hercules, CA). SYBR green (BioRad) was used for amplicon detection. All primer sets were tested to ensure the efficiency of amplification over a wide range of template concentrations. PCR reactions were carried out in the iQ5 RT-PCR detection system (BioRad). A melting curve was performed after amplification to ensure all samples exhibited a single amplicon. Gene expression of each individual myometrial sample was first normalized to the expression of 18s ribosomal RNA (internal control) by the ΔCT method and then compared with the average of the corresponding control samples. Messenger RNA (mRNA) relative expression is equal to 2-ΔΔCT and determined by the following equations: (1) ΔΔCT = ΔCT1 (each individual sample) – ΔCT2 (control group mean); 920 ΔCT1 = gene CT value – 18S CT value of the same sample; and (3) ΔCT2 = average of (gene CT value of control group – 18S CT value of control group). The point at which the fluorescence crosses the threshold is called the CT value and is read by the RT-PCR instrument as a specific parameter.
Results
Clinical characteristics of myometrial samples
The clinical characteristics of the pregnancies included are listed in Table 1 . There were no statistically significant differences in GA or birthweight among women at term (no labor vs labor) or preterm (no labor vs labor). There were no statistically significant differences in maternal age as determined by 1-way analysis of variance. Thus, the observed differences in genes in the initiator groups are most likely the result of labor (term or preterm) rather than any variation in GA or maternal age.
| Group | Gestational age, wk a | Maternal age, y a | Birthweight, g a | Indication for cesarean delivery | Histologic chorioamnionitis |
|---|---|---|---|---|---|
| Term not in labor | 39 ± 1 | 30 ± 3 | 3490 ± 618 | Breech | NA |
| Term labor | 40 ± 1 | 28 ± 6 | 3490 ± 333 | Failure to progress | NA |
| Preterm not in labor | 29 ± 3 b , c | 26 ± 9 | 971 ± 380 b , c | Preeclampsia | Stage 0 |
| Preterm labor | 30 ± 3 b , c | 34 ± 6 | 1299 ± 344 b , c | Preterm premature rupture of membranes/spontaneous preterm labor | Stage III |
a Data are presented as mean ± SD;
b P < .05, based on Newman-Keuls test compared with term not in labor;
c P < .05, based on Newman-Keuls test compared with term labor.
Identification of effector set and the term initiator and preterm initiator sets
Sixty-seven genes were present in both PTL and TL but were absent from all nonlabor samples (n = 6; groups B and D). This group was termed the effector set because the genes comprising it represent a distinct group that might act to sustain the labor itself rather than contributing to its onset. This effector set was then further filtered to exclude genes that showed at least a 2-fold difference between PTL and TL based on the assumption that there was something about the disease process that led to differential regulation. This left 34 genes in the effector set ( Table 2 ).
| Gene symbol | Gene name | Entrez identification no. |
|---|---|---|
| ABRA | Actin-binding rho activating protein | 137735 |
| ALDH16A1 | Aldehyde dehydrogenase 16 family, member a1 | 126133 |
| ANKS3 | Ankyrin repeat and sterile alpha motif domain containing 3 | 124401 |
| ATRIP/TREX1 | ATR-interacting protein/Three prime repair exonuclease 1 | 84126 |
| BAIAP2L1 | Bai1-associated protein 2-like 1 | 55971 |
| CADM2 | Cell adhesion molecule 2 precursor | 253559 |
| CCDC36 | Coiled-coil domain containing 36 | 339834 |
| CFB | Complement factor b | 629 |
| CORO2A | Coronin, actin binding protein, 2a | 7464 |
| D4S234E | DNA segment on chromosome 4 (unique) 234 expressed sequence | 27065 |
| EGFL7 | Egf-like-domain, multiple 7 | 51162 |
| FPGS | Folylpolyglutamate synthase | 2356 |
| GPATCH3 | G patch domain containing 3 | 63906 |
| H2AFY2 | H2a histone family, member y2 | 55506 |
| HNT | Neurotrimin | 50863 |
| HTRA4 | Htra serine peptidase 4 | 203100 |
| IL8RB | Interleukin 8 receptor, beta | 3579 |
| KCNAB2 | Potassium voltage-gated channel, shaker-related subfamily, beta member 2 | 8514 |
| KRTAP8-1 | Keratin associated protein 8-1 | 337879 |
| LY6G5C | Lymphocyte antigen 6 complex, locus g5c | 80741 |
| MPP3 | Membrane protein, palmitoylated 3 (maguk p55 subfamily member 3) | 4356 |
| P2RY8 | Purinergic receptor p2y, g-protein coupled, 8 | 286530 |
| PLAC8L1 | Plac8-like 1 | 153770 |
| PLCXD1 | Phosphatidylinositol-specific phospholipase c, x domain containing 1 | 55344 |
| POLR3D | Polymerase (RNA) iii (DNA directed) polypeptide d, 44 kd | 661 |
| RRAD | Ras-related associated with diabetes mellitus | 6236 |
| SIRPB2 | Signal-regulatory protein beta 2 | 284759 |
| SKIV2L | Superkiller viralicidic activity 2-like (s. Cerevisiae) | 6499 |
| SLPI | Secretory leukocyte peptidase inhibitor | 6590 |
| ST3GAL3 | St3 beta-galactoside alpha-2,3-sialyltransferase 3 | 6487 |
| ZNF552 | Zinc finger protein 552 | 79818 |
| ZNF718 | Zinc finger protein 718 | 255403 |
| ZNF74 | Zinc finger protein 74 (cos52) | 7625 |
| ZNF788 | Zinc finger protein 788 | 388507 |
Although the effector gene set was by definition independent of GA, we suspected there must exist gene sets that enabled the initiation of the labor either term or preterm. We detected 49 genes that were specific to preterm labor alone (preterm initiator set; Table 3 ) and 174 genes that were specific to TL alone (term initiator set; Table 4 ). In light of the GAs of the preterm subjects, it is highly likely that the selected gene sets represent, or at least include, molecular groups that are responsible for initiating the transition from myometrial quiescence to activation in their respective categories.
| Genes symbol | Gene name | Entrez identification no. |
|---|---|---|
| ABCA3 | ATP-binding cassette, sub-family a (abc1), member 3 | 21 |
| ABP1 | Amiloride binding protein 1 (amine oxidase [copper-containing]) | 26 |
| ARSA | Arylsulfatase a | 410 |
| CD3E | Cd3e antigen, epsilon polypeptide (tit3 complex) | 916 |
| DGKQ | Diacylglycerol kinase, theta 110 kd | 1609 |
| GJA3 | Gap junction protein, alpha 3, 46 kd (connexin 46) | 2700 |
| GSTT2 | Glutathione s-transferase theta 2 | 2953 |
| HLA-DOA | Major histocompatibility complex, class ii, do alpha | 3111 |
| HYAL1 | Hyaluronoglucosaminidase 1 | 3373 |
| IFNG | Interferon, gamma | 3458 |
| IRF5 | Interferon regulatory factor 5 | 3663 |
| LLGL1 | Lethal giant larvae homolog 1 (drosophila) | 3996 |
| LYL1 | Lymphoblastic leukemia derived sequence 1 | 4066 |
| MYCL1 | V-myc myelocytomatosis viral oncogene homolog 1, lung carcinoma derived (avian) | 4610 |
| PDE1C | Phosphodiesterase 1c, calmodulin-dependent 70 kd | 5137 |
| MAPK11 | Mitogen-activated protein kinase 11 | 5600 |
| PTPRZ1 | Protein tyrosine phosphatase, receptor-type, z polypeptide 1 | 5803 |
| ZNF208 | Zinc finger protein 208 | 7757 |
| ABCA7 | ATP-binding cassette, sub-family a (abc1), member 7 | 10347 |
| CXCL13 | Chemokine (c-x-c motif) ligand 13 (b-cell chemoattractant) | 10563 |
| OGFR | Opioid growth factor receptor | 11054 |
| ELOVL2 | Elongation of very long chain fatty acids (fen1/elo2, sur4/elo3, yeast)-like 2 | 54898 |
| CPXM | Carboxypeptidase x (m14 family) | 56265 |
| AARSL | Alanyl-tRNA synthetase like | 57505 |
| TINAGL1 | Tubulointerstitial nephritis antigen-like 1 | 64129 |
| DIO3OS | Deiodinase, iodothyronine, type iii opposite strand | 64150 |
| NDRG4 | Ndrg family member 4 | 65009 |
| VPS33A | Vacuolar protein sorting 33a (yeast) | 65082 |
| LRFN4 | Leucine rich repeat and fibronectin type iii domain containing 4 | 78999 |
| TLE6 | Transducin-like enhancer of split 6 (e[sp1]) homolog, drosophila) | 79816 |
| TIGD6 | Tigger transposable element derived 6 | 81789 |
| KRTAP4-8 | Keratin associated protein 4-8 | 83898 |
| CHCHD6 | Coiled-coil-helix-coiled-coil-helix domain containing 6 | 84303 |
| ZNF341 | Zinc finger protein 341 | 84905 |
| CCDC97 | Coiled-coil domain-containing protein 97 | 90324 |
| STK11IP | Serine/threonine kinase 11 interacting protein | 114790 |
| SORCS1 | Sortilin-related vps10 domain containing receptor 1 | 114815 |
| CATSPER2 | Cation channel, sperm associated 2 | 117155 |
| LRGUK | Leucine-rich repeats and guanylate kinase domain containing | 136332 |
| FAM69B | Family with sequence similarity 69, member b | 138311 |
| BEAN | Brain expressed, associated with nedd4 | 146227 |
| GRASP | Grp1 (general receptor for phosphoinositides 1)-associated scaffold protein | 160622 |
| ZNF579 | Zinc finger protein 579 | 163033 |
| DENND1B | Denn/madd domain containing 1b | 163486 |
| SAMD14 | Sterile alpha motif domain containing 14 | 201191 |
| ALS2CL | Als2 c-terminal like | 259173 |
| STK32C | Serine/threonine kinase 32c | 282974 |
| CATSPER2P1 | Cation channel, sperm associated 2 pseudogene | 440278 |
| GSTT2B | Similar to glutathione s-transferase theta 2 (gst class-theta 2) | 653689 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
