help button home button Endocrine Society Endocrinology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Endocrinology, doi:10.1210/en.2006-0683
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental Data
Right arrow All Versions of this Article:
148/3/1059    most recent
Author Manuscript (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Copyright Permission
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Winn, V. D.
Right arrow Articles by Fisher, S. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Winn, V. D.
Right arrow Articles by Fisher, S. J.
Endocrinology Vol. 148, No. 3 1059-1079
Copyright © 2007 by The Endocrine Society

Gene Expression Profiling of the Human Maternal-Fetal Interface Reveals Dramatic Changes between Midgestation and Term

Virginia D. Winn, Ronit Haimov-Kochman, Agnes C. Paquet, Y. Jean Yang, M. S. Madhusudhan, Matthew Gormley, Kui-Tzu V. Feng, David A. Bernlohr, Susan McDonagh, Lenore Pereira, Andrej Sali and Susan J. Fisher

Departments of Obstetrics, Gynecology, and Reproductive Sciences (V.D.W., R.H.-K., M.G.), Cell and Tissue Biology (R.H.-K., K.-T.V.F., S.M., L.P., S.J.F.), and Medicine (A.C.P., Y.J.Y.), Lung Biology Center, Departments of Biopharmaceutical Sciences and Pharmaceutical Chemistry (M.S.M., A.S.) and Anatomy and Pharmaceutical Chemistry (S.J.F.), California Institute for Quantitative Biomedical Research (M.S.M., A.S.), University of California, San Francisco, San Francisco, California 94143; and Department of Biochemistry, Molecular Biology, and Biophysics (D.A.B.), University of Minnesota, Minneapolis, Minnesota 55455

Address all correspondence and requests for reprints to: Virginia D. Winn, M.D., Ph.D., University of Colorado Health Sciences Center, Reproductive Science, Mail Stop 8309, 12800 East 19th Avenue, P.O. Box 6511, Aurora, Colorado 80045. E-mail: virginia.winn{at}uchsc.edu.


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
Human placentation entails the remarkable integration of fetal and maternal cells into a single functional unit. In the basal plate region (the maternal-fetal interface) of the placenta, fetal cytotrophoblasts from the placenta invade the uterus and remodel the resident vasculature and avoid maternal immune rejection. Knowing the molecular bases for these unique cell-cell interactions is important for understanding how this specialized region functions during normal pregnancy with implications for tumor biology and transplantation immunology. Therefore, we undertook a global analysis of the gene expression profiles at the maternal-fetal interface. Basal plate biopsy specimens were obtained from 36 placentas (14–40 wk) at the conclusion of normal pregnancies. RNA was isolated, processed, and hybridized to HG-U133A&B Affymetrix GeneChips. Surprisingly, there was little change in gene expression during the 14- to 24-wk interval. In contrast, 418 genes were differentially expressed at term (37–40 wk) as compared with midgestation (14–24 wk). Subsequent analyses using quantitative PCR and immunolocalization approaches validated a portion of these results. Many of the differentially expressed genes are known in other contexts to be involved in differentiation, motility, transcription, immunity, angiogenesis, extracellular matrix dissolution, or lipid metabolism. One sixth were nonannotated or encoded hypothetical proteins. Modeling based on structural homology revealed potential functions for 31 of these proteins. These data provide a reference set for understanding the molecular components of the dialogue taking place between maternal and fetal cells in the basal plate as well as for future comparisons of alterations in this region that occur in obstetric complications.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
SURVIVAL AND GROWTH of the fetus require normal development of the placenta, which in humans involves the formation of a transient organ with both maternal and fetal contributions. Specifically, invasive cytotrophoblasts (CTBs), components of anchoring chorionic villi, attach to and invade the maternal decidua. A subset of these cells remodel the uterine vasculature, which they also occupy (Fig. 1Go). This process primarily occurs during the second trimester of pregnancy. The region in which maternal and fetal cells coexist is termed the basal plate or maternal-fetal interface, and its proper formation and function are required for normal pregnancy outcome. At a cellular level, many unusual processes occur in this area. For example, invasive CTBs execute a novel epithelial-to-mesenchymal transition that enables vascular mimicry (1, 2). Perhaps most remarkably, the maternal immune system tolerates the invasion of the hemiallogeneic fetal cells for the duration of pregnancy.


Figure 1
View larger version (43K):
[in this window]
[in a new window]

  		FIG. 1. Diagram of the human maternal-fetal interface. A, Representation of the human placenta after delivery. The placental surface that was adjacent to the uterine wall is termed the basal plate. The boxed area denotes the region biopsied for these studies. B, View of the basal plate at the cellular level. This chimeric region of the placenta is composed of both maternal and fetal components: extravillous (invasive) cytotrophoblasts (dark gray), decidual cells (light gray), remodeled vasculature (both invasive CTBs and maternal endothelium) and maternal immune cells (white).

 
Over the past several decades, a great deal of information has been gained about placental development by taking a candidate molecule approach (3). By analogy with cells and tissues that perform similar functions in other contexts, progress has been made toward understanding many components of placental development. For example, the fact that endovascular CTBs function as endothelial cells prompted investigators to study the role of vasculogenic/angiogenic molecules, including adhesion receptors, at the maternal-fetal interface (4, 5). As in many tumors, CTBs use matrix metalloproteinases for the purpose of invasion (6, 7). However, there are also numerous examples of seemingly novel mechanisms that are unique to placental development. For example, trophoblasts in all locations lack major histocompatibility class II expression, and upon allocation to the invasive pathway, CTBs up-regulate human leukocyte antigen-G, a nonclassical major histocompatibility class I molecule, in the absence of human leukocyte antigen-A and -B expression (8, 9). Accordingly, unbiased analyses, such as microarray approaches, are also crucial for obtaining new insights into the mechanisms that are required for normal basal plate formation and function during pregnancy.

As in many research areas, genome-wide expression profiling approaches are being used to understand trophoblast differentiation and placental development. Several types of experimental designs have been published. Aronow et al. (10) characterized human term CTB syncytialization and the role of the activator protein 2{alpha} transcription factor in this process (11). Kudo et al. (12) also studied syncytialization but focused on the BeWo line rather than on primary cells. Roh et al. (13) analyzed the effects of hypoxia on human term trophoblasts, which led to their most recent work showing a role for N-myc-down-regulated gene 1 (14). In addition, there are several published reports that describe the gene expression patterns of the placenta as a whole. Tanaka et al. (15) used a microarray approach to compare the genetic bases of murine placental and embryonic development at midgestation and subsequently validated and extended the results (16). Hemberger et al. (17) focused on the murine extraembryonic tissues together with the adjacent decidua at d 7.5, compared with d 17.5.

The Hemberger study design took into consideration that trophoblast-decidual interactions are critical determinants of pregnancy outcome. For this reason, we used the same basic strategy, profiling gene expression at the human maternal-fetal interface during five gestational age intervals: 14–16, 18–19, 21, 23–24, and 37–40 wk. RNA was isolated from the basal plate and processed to produce samples that were hybridized to high-density, short-oligonucleotide microarrays (GeneChips HG-U133A and HG-U133B; Affymetrix, Santa Clara, CA). Very few alterations in gene expression were observed during the midgestation period (14–24 wk). In contrast, hundreds of changes, including the expression of genes known to be up- or down-regulated over gestation, were modulated between midgestation and term (37–40 wk). These data allowed us to identify molecules that play potentially important roles in the formation and function of the maternal-fetal interface.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
Tissue collection
The University of California, San Francisco (UCSF), Committee on Human Research approved the tissue procurement protocol. Informed consent was obtained from each parturient before delivery. Basal plate biopsy specimens of the maternal-fetal interface from second- and third-trimester placentas [gestational ages 14–16 wk (n = 6), 18–19 wk (n = 9), 21 wk (n = 6), 23–24 wk (n = 6), and 37–40 wk (n = 9)] were collected. Second-trimester samples were obtained after elective terminations of singleton pregnancies. The products of terminations performed for fetal indications, infection, or maternal health complications were excluded. Term samples were collected after cesarean delivery at the conclusion of normal nonlabored singleton pregnancies. Pregnancies complicated by fetal anomalies, hypertension, diabetes, or other significant maternal health issues were excluded. Gestational age was determined during the second trimester by measuring foot length (18) and at term by using standard dating criteria (19). The basal plate was dissected from the placenta proper, rinsed in PBS and diced into approximately 3 x 3 mm3 pieces, which were snap frozen in liquid nitrogen and stored at –70 C. All samples were processed and frozen within 1 h of delivery. For immunohistochemistry, biopsy samples of the basal plate were fixed in 3% paraformaldehyde in PBS (wt/vol), passed through a sucrose gradient (5–15% in PBS), and frozen in optimal cutting temperature.

In addition, biopsies of several regions of the tissue were also fixed in 10% neutral-buffered formalin and embedded in paraffin. Tissue sections prepared from the blocks were stained with hematoxylin and eosin and examined by using a light microscope. In all cases, normal morphological features were noted; there were no histological signs of placental or decidual pathology.

Total RNA extraction
RNA was isolated from snap-frozen basal plate specimens using a modified Trizol method that was developed during the course of this work (20). Briefly, homogenization of 0.9–1 g of frozen basal plate specimens was carried out in 10 ml of cold Trizol reagent (Invitrogen, Frederick, MD) on wet ice (0–4 C). Cellular debris was pelleted by centrifugation at 12,000 x g for 10 min. Then the supernatant was transferred to Phase Lock Gel heavy tubes (Eppendorf, Netheler, Germany), and RNA was isolated according to the manufacturer’s instructions. The total RNA fraction was purified further by using an RNeasy mini kit (QIAGEN, Valencia, CA) according to the manufacturer’s instructions. Aliquots of the RNA isolated from the specimens were evaluated by using the Agilent RNA 6000 Nano LabChip kit (Agilent Technologies, Amstelveen, The Netherlands) on an Agilent Bioanalyzer 2100 system using the nano assay for eukaryote total RNA. Capillary electrophoresis data in commaseparated value files were analyzed by using the Degradometer version 1.41 software (available at www.dnaarrays.org) (21). Only RNA with a degradation factor of less than 11 was used in subsequent microarray experiments.

Microarray hybridization
The microarray platform was the high-density HG-U133A and HG-U133B GeneChips (Affymetrix) that use 45,000-oligomer probe sets representing 39,000 transcripts. Hybridization was accomplished by using the protocol devised by the UCSF Gladstone (National Heart, Lung, and Blood Institute) Genomics Core Facility (www.gladstone.ucsf.edu/gladstone/php/section). In brief, double-stranded cDNAs were generated from total RNA samples by using SuperScript II reverse transcriptase (Invitrogen) and a T7-oligo primer (QIAGEN). Biotin-labeled cRNA was synthesized by in vitro transcription using an Enzo Bioassay RNA labeling kit (Enzo Diagnostics, Farmingdale, NY). The labeled cRNA was purified with an RNeasy column (QIAGEN). Before hybridization, the quality of all in vitro transcription products was evaluated by using the Agilent Bioanalyzer 2100 system. Then the cRNA was fragmented at 94 C for 35 min in buffer [Tris-acetate 40 mmol/liter, potassium acetate 100 mmol/liter, magnesium acetate 30 mmol/liter (pH 8.1)]. Samples from individual basal plates were analyzed separately. Specifically, the HG-U133A and HG-U133B Affymetrix GeneChips were each hybridized with 15 µg of cRNA and then washed, stained, and imaged at the Gladstone Genomics Core Facility by using standard Affymetrix protocols. Data files were deposited in the Gene Expression Omnibus data repository with accession no. GSE5999.

Data analysis
The raw image data were analyzed by using GeneChip Expression Analysis software (Affymetrix) to produce perfect match and mismatch values. Subsequently quality control, preprocessing, and linear modeling were performed using Bioconductor (22), an open-source and open-development software project based on the R statistical package (www.r-project.org). Clustering analysis was performed using Acuity software (Molecular Devices Corp., Sunnyvale, CA). Initial hybridization quality was assessed by using Bioconductor package affyPLM, and the slight variations in quality were compensated for during the preprocessing stage, which was performed in two steps. First, we used a Probe Level robust linear model (23) to obtain separate normalized log intensities for each chip (i.e. background subtraction, quantile normalization, and probe set summarization). Second, we applied a global median normalization at the probe set level to all A and B GeneChips (n = 72) and then combined these data into a matrix of log2-based gene expression measures, in which columns corresponded to different cRNA samples, and rows corresponded to the different probe sets.

Initial analyses showed that gene expression during the second-trimester intervals was stable. Therefore, subsequent analyses were performed by comparing the gene expression data from the midgestation samples (14–24 wk; n = 27) with those obtained at term (37–40 wk; n = 9). Estimated log ratios (M value) between term and midgestation were determined by using the limma software package in R (24). Then differentially expressed genes were selected by determining the moderated t statistic-adjusted P values (<0.05 using Bonferroni correction). The results showed that the expression of 418 genes (505 probe sets) was significantly modulated. Then the normalized intensity values for this data set were centered to the median intensity value for each probe set, after which the probe sets were ranked according to their M values (representing fold change) and depicted as a gene expression color map.

The gap statistic with Euclidean distance was used to select the cluster number (k = 11) for subsequent application of the K-means algorithm (25). Then differentially expressed genes from each cluster were presented as a hierarchical dendogram of the normalized log intensity data based on the Euclidean squared metric and average linkage. This analysis enabled us to visually evaluate both transcript levels and patterns of coregulation.

Pathway and network analysis
Initially, gene ontogeny (GO) annotations were determined (www.genetools.microarray.ntnu.no) and used to categorize the differentially expressed genes according to the biological processes in which they were involved (level 2). When biological process information was lacking, genes were annotated according to molecular function. To determine whether there was a significant overrepresentation of differentially expressed genes in particular functions or physiologic processes, the data set was analyzed by using Ingenuity Pathway Analysis 3.1 software (www.ingenuity.com). The data set containing gene identifiers and their corresponding expression values was uploaded as an Excel spreadsheet using the template provided in the application. Each gene identifier was mapped to its corresponding gene object in the Ingenuity Pathways Knowledge Base. Differentially regulated genes, identified by using an adjusted P < 0.05 as the cut-off, were then used as the starting point for generating biological networks.

Specifically, all the differentially expressed genes as well as the subset that exhibited more than a 2-fold change in expression were evaluated according to their molecular and cellular functions and the physiological processes in which they participated. In addition, their participation in metabolic and signaling pathways was assessed. Finally, the differentially expressed genes were subjected to network analysis.

Quantitative PCR
RT of basal plate (total) RNA was carried out by using the TaqMan Gold RT-PCR kit (Applied Biosystems, Foster City, CA) as described by the manufacturer, followed by real-time PCR, performed in triplicate by using the Applied Biosystems 9700HT sequence detection system. All templates were amplified by using Assay-on-Demand kits (Applied Biosystems) or primer/probe sets designed by the UCSF Biomolecular Research Center (see supplemental Table 1, published as supplemental data on The Endocrine Society’s Journals Online web site at http://endo.endojournals.org). Briefly, 5 µl of cDNA was added to 20 µl of 1x TaqMan Universal PCR master mix containing AmpErase UNG and 1 µl of a primer/probe. Negative controls contained either RNA that was not reverse transcribed or lacked template inputs. Reactions were incubated at 50 C for 2 min and then 95 C for 10 min, followed by 40 cycles of 95 C for 15 sec and 60 C for 1 min. Relative quantification was determined by using the standard curve method (see Applied Biosystems user bulletin no. 2; www.appliedbiosystems.com). In preliminary experiments, we investigated the utility of 11 potential targets as endogenous controls (endogenous control plate; Applied Biosystems). The results showed that the 18S rRNA did not vary with gestational age. Accordingly, the levels of this transcript were used to obtain normalized values for the target amplicons. Then these values were calibrated to a 14-wk sample, the earliest gestational age included in our analysis. Results were reported as the relative fold mRNA levels ± SD for each basal plate specimen. The means of the term and midgestation samples were compared using a two-tailed Student’s t test (P < 0.05).

Immunohistochemistry
Frozen sections (5 µM) cut from optimal cutting temperature-embedded tissues were washed in PBS and nonspecific reactivity was blocked with 3% BSA, 0.1% Triton X-100, and 0.5% Tween 20 in PBS for 30 min. Then the experimental sections were incubated with mouse antihuman lipoprotein lipase (LPL) antibody (1:100; 5D2; the kind gift of John Brunzell, University of Washington, Seattle, WA) for 1 h, after which they were washed in PBS three times for 5 min. Negative controls were incubated in the absence of the primary antibody. Then both experimental and control sections were incubated in rat antihuman cytokeratin antibody (CK) [1:100; 7D3 (26)]) for 1 h and washed in PBS as described above. To localize the bound primary antibodies, the sections were incubated with Alexa Fluor 594-conjugated goat antimouse IgG (1:1000; Molecular Probes Inc., Eugene, OR) and fluorescein isothiocyanate-labeled donkey antirat IgG (1:200; Jackson ImmunoResearch Laboratories, West Grove, PA) antibodies for 30 min and washed again in PBS. Tissue sections were mounted in Vectashield containing 4'-6-diamidino-2-phenylindol (Vector Laboratories, Burlingame, CA), which allowed visualization of the nuclei. Immunoreactivity was imaged using a Leica DM 5000B fluorescent microscope equipped with a Leica DFC 350FX digital camera (Leica Instruments, San Jose, CA). Expression of fatty acid binding protein 4 (FABP4) in the basal plate was evaluated by using a rabbit antimurine polyclonal antibody that specifically reacted with this protein (27). The immunostaining protocol was identical with that described for LPL except that the primary antibody was at a 1:1000 dilution and the secondary antibody was Alexa Fluor 594-conjugated goat antirabbit IgG (1:1000; Molecular Probes).

Protein function annotation by sequence homology and structural similarity
To determine the function of the differentially expressed genes that lacked annotations (www.genetools.microarray.ntnu.no; July 2005), we used protein sequence homology searches along with protein structure modeling. Briefly, protein sequences for the differentially expressed genes were extracted by using their UniGene identifiers (28). Homology searches were done using PSI-BLAST (29). Five iterations of the PSI-BLAST were run using an e-value cutoff of 10–5 for sequences to be included in the profile. For protein sequences with detectable homology to other proteins of known structure, comparative structure models were built through the MODWEB server (30), which uses the program MODELLER (31). The resulting models were deposited in the model database MODBASE (32). Proteins that could not be assigned a function based on homology searches were subjected to threading using the mGenTHREADER software (33).


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
Gene expression patterns at the maternal-fetal interface change dramatically between second trimester and term
First, we determined the gene expression profiles of 36 human basal plate samples that were collected between the gestational ages of 14 and 40 wk. The relevant clinical data pertaining to each specimen are presented as supplemental data (supplemental Table 2). Pair-wise comparisons between each of the five gestational age intervals showed remarkably stable patterns of gene expression among the second trimester arrays (14–24 wk). In contrast, numerous differences were observed between midgestation (14–24 wk) and term (37–40 wk). The differentially expressed transcripts, a total of 418 genes/expressed sequence tags (505 probe sets), were identified. Statistical analysis of maternal age and parity showed that there were no significant differences in these parameters between the RNA samples that were prepared from midgestation and term basal plate biopsies.

An MA scatter-plot that depicts the fold change and signal intensity for every probe set is presented as supplemental data (supplemental Fig. 1). The differentially expressed genes normalized to the median value and ordered by their fold change are shown as a heat map in Fig. 2AGo. The areas that contain the 35 most highly up- or down-regulated probe sets have been enlarged and annotated (Fig. 2BGo, upper and lower panels, respectively). Annotation of the complete heat map is provided as supplemental data (supplemental Fig. 2). The entire set of genes that were up-regulated at term is summarized in Table 1GoGoGo, where they are also categorized according to the biological processes in which they participate or their molecular functions based on the relevant GO annotation (www.genetools.microarray.ntnu.no). The probe set identifier, gene symbol, GO annotation, and fold change are also included in Table 1GoGoGo. Table 2GoGoGoGo presents analogous information concerning the genes that were down-regulated at term. As expected, genes with known expression patterns, either up-regulated [e.g. CRH (34) and inhibin-ßA subunit (35)] or down-regulated [e.g. C-X-C chemokine ligand 14 (36) and angiopoietin-2 (ANGPT-2) (5)] at term, were present in the data set.


Figure 2
View larger version (86K):
[in this window]
[in a new window]

  		FIG. 2. Heat map of differentially expressed genes. A, The normalized log intensity values for all 505 differentially expressed probe sets were centered to the median value of each probe set and colored on a range of –2 to +2. Red denotes up-regulated, yellow denotes intermediate, and blue denotes down-regulated expression levels as compared with the median value. Columns contain data from a single basal plate specimen, and rows correspond to a single probe set. Samples are arranged from left to right, ordered by increasing gestational age. Rows are ranked by fold change [mean term value (n = 9) divided by mean midgestation value (n = 27)]. B, The most highly up-regulated (upper panel) and down-regulated (lower panel) probe sets at term were enlarged and annotated with the gene name and the fold change value. An annotated heat map of all 505 probe sets is available as supplemental data (supplemental Fig. 2).

 

View this table:
[in this window]
[in a new window]

  		TABLE 1. Differentially expressed genes upregulated at term, compared with midgestation

 

View this table:
[in this window]
[in a new window]

  		TABLE 1A. Continued

 

View this table:
[in this window]
[in a new window]

  		TABLE 1B. Continued

 

View this table:
[in this window]
[in a new window]

  		TABLE 2. Differentially expressed genes down-regulated at term, compared with midgestation

 

View this table:
[in this window]
[in a new window]

  		TABLE 2A. Continued

 

View this table:
[in this window]
[in a new window]

  		TABLE 2B. Continued

 

View this table:
[in this window]
[in a new window]

  		TABLE 2C. Continued

 
Coregulation of differentially expressed genes
To examine patterns of coexpression and possibly coregulation, we performed K-means cluster analysis. Cluster sizes ranged from 4 to 121 probe sets. Five clusters contained up-regulated genes and six clusters contained down-regulated genes. Examples of the clusters (n = 4) created are shown as heat maps that were constructed by hierarchical clustering of the log intensity data (Fig. 3Go). The remaining clusters (n = 7) are available as supplemental data (supplemental Fig. 3). Cluster A is composed of four probe sets for two genes, ANGPT-2 and microcephalin (MCPH1; formerly hypothetical protein FLJ12847). The Pearson correlation coefficients among these four probe sets were extremely high (0.92–0.97), suggesting very tight coregulation. On further examination, the chromosomal location for both genes was found to be 8p23.1, in which they are transcribed from the same DNA sequence in opposite directions. Cluster B contains the most highly up-regulated genes: LPL, FABP4, CRH, HBB, and ALPP. Cluster C includes a significant proportion of genes with immunological functions that are regulated in a similar manner. Finally, cluster D contains numerous genes that encode proteins with unknown functions.


Figure 3
View larger version (58K):
[in this window]
[in a new window]

  		FIG. 3. Selected K-means clusters of differentially expressed genes that are coexpressed. Differentially expressed probe sets were clustered into 11 groups using K-means (Euclidean). Clusters A, B, C, and D are presented as heat maps of the normalized log intensity data, which allows visual assessment of the signal intensity and the degree of change over gestation. Arrays (columns) are arranged from left to right, ordered by increasing gestational age. Each probe set is annotated with the probe set ID, gene symbol, chromosomal location, and fold change. The seven additional clusters that were generated are included as supplemental data (supplemental Fig. 3).

 
Differentially expressed genes: functions, pathways, and networks
The GO annotations suggested that the differentially expressed genes were involved in a variety of biological processes. At least one sixth were expressed sequence tags or hypothetical proteins and thus were not annotated. Of the known differentially expressed genes, 17 were related to lipid metabolism, 10 were involved with formation or regulation of the extracellular matrix, 21 were immune effectors or modulators, 24 were transcription factors, and six had angiogenic/vasculogenic functions. Interestingly, of the 39 genes that were involved in signal transduction, five functioned in the Wnt-ß-catenin pathway (FRAT1, CTNNBIP1, DKK1, SFRP1, and KREMEN1).

The GO annotations indicated that the differentially expressed genes within the basal plate region were involved in a variety of specific functions and biological processes. Thus, we used Ingenuity Pathway Analysis (IPA) to determine the significance of these observations. IPA showed that 23 molecular functions and 22 physiologic processes included more differentially expressed genes than would be expected by chance (–log [significance] of > 1.25, which corresponds to a P < 0.05; see www.ingenuity.com). These results, summarized by functional categories, are shown as supplementary data (supplemental Fig. 4). Genes that had at least a 2-fold change in expression were most significantly represented in the following categories: cell movement, cell-cell signaling, cell death, lipid metabolism, small molecule biochemistry, and gene expression.

Next, we used the IPA software to further evaluate the participation of the differentially expressed genes of the human basal plate in metabolic and signaling pathways (supplemental Table 3). Analysis of genes with at least a 2-fold change highlighted two metabolic pathways: folate biosynthesis and N-glycan degradation involving mannose-containing structures. With regard to signaling, the differentially expressed genes were significantly overexpressed in the Wnt-ß-catenin (supplemental Fig. 5) and TGF-ß pathways. Finally, several other pathways were just below the 1.25 threshold, namely peroxisome proliferative activated receptor and IL-6.

We also used the IPA software to map networks of the differentially expressed genes at the human basal plate. The largest network contained genes that were involved in cell motility, cell-to-cell signaling/interaction, and tissue development. Diagrams of these networks are shown as supplementary data (supplemental Fig. 6).

Confirmation of the microarray results
To further validate the microarray data, we used two approaches: quantitative PCR (Q-PCR) to assess relative mRNA levels and immunolocalization to confirm differential expression at the protein level. With regard to Q-PCR, our analysis focused primarily on the genes that exhibited the greatest fold change and those that encoded hypothetical proteins with potentially novel functions. The sequences of the primer/probe sets are shown as supplemental data (supplemental Table 2). The 14 genes that we subjected to Q-PCR analyses all showed the expected expression patterns (Fig. 4GoGo). Six were up-regulated: CRH, hypothetical protein LOC284561, mucin 15 , trophoblast-derived noncoding RNA, hypothetical protein FLJ11292, and platelet/endothelial cell adhesion molecule-1 (Fig. 4AGoGo). Eight were down-regulated: MCPH1, target of Nesh-SH3, chemokine ligand 14, hypothetical protein FLJ11539, phospholipase A2, group VII, transcription factor-like 2, adipocyte-specific adhesion molecule, and vascular cell adhesion molecule 1 (Fig. 4BGoGo). Additionally, the Q-PCR data and log intensity values for each basal plate specimen showed a high degree of cross-correlation (Fig. 4GoGo, A and B).


Figure 4
View larger version (33K):
[in this window]
[in a new window]

  		FIG. 4. Q-PCR confirmation of a subset of differentially expressed genes. Samples of total RNA isolated from basal plate biopsy specimens that were obtained at midgestation (14–24 wk; n = 8 or 9) or term (37–40 wk; n = 4 or 5) were analyzed using TaqMan primer/probe sets. A, Q-PCR data for transcripts up-regulated at term: CRH (P = 0.0003), hypothetical protein LOC284561 (P = 0.0006), mucin 15 (MUC15; P = 0.00002), trophoblast-derived noncoding RNA (TncRNA; P = 0.01), hypothetical protein FLJ11292 (P = 0.04), and platelet/endothelial cell adhesion molecule-1 (PECAM1; P = 0.03). Relative RNA levels were normalized to 18S values and then divided by a calibrator, in this case a 14-wk sample. Each bar represents the mean ± SD of triplicate determinations (midgestation, light gray; term, dark gray). Dashed lines are the mean values for the midgestation or term samples. Significance was determined by using Student’s t test (P < 0.05). For comparison, the insets show the corresponding microarray log intensity data for the same samples (log2).

 

Figure 4A
View larger version (34K):
[in this window]
[in a new window]

  		FIG. 4A. Continued. B, Q-PCR data for transcripts down-regulated at term: MCPH1 (P = 0.054), target of Nesh-SH3 (TARSH; P = 0.008), chemokine (C-X-C motif) ligand 14 (CXCL14; P = 0.02), hypothetical protein FLJ11539 (P = 0.008), phospholipase A2, group VII (PLA2G7; P = 0.002), transcription factor-like 2 (TCF7L2; P = 0.12), adipocyte-specific adhesion molecule (ASAM; P = 0.01), and vascular cell adhesion molecule 1 (VCAM1; P = 0.34).

 
With regard to the immunolocalization experiments, we focused on LPL, one of the lipid metabolizing enzymes that were highly up-regulated (3.7-fold) at the mRNA level over gestation. The staining pattern of LPL, which was primarily cytoplasmic, changed with advancing gestational age. Specifically, at 16 wk, the vast majority of cells in the basal plate region, including CK-positive CTBs within the uterine wall, failed to react with anti-LPL (Fig. 5AGo, b and c). At 23 wk, a slightly higher level of immunoreactivity was detected, primarily in association with CTBs; intense staining was occasionally observed (Fig. 5AGo, e and f). In contrast, many CTBs exhibited strong immunoreactivity at 39 wk (Fig. 5AGo, h and i), and minimal staining was detected in association with the decidua.


Figure 5
View larger version (83K):
[in this window]
[in a new window]

  		FIG. 5. Invasive CTBs up-regulate expression of LPL and FABP4 at term. A, Tissue sections of basal plate biopsy specimens from 16-wk (a–c), 23-wk (d–f), and 39-wk (g–i) placentas were double stained with anti-CK (1:50; a, d, and g) to identify CTBs and an anti-LPL antibody, 5D2 (1:100; b, e, and h; the kind gift of J. D. Brunzell, University of Washington). Binding was detected with the appropriate species-specific secondary antibodies. Nuclei were labeled with Hoechst (1:1000; blue). Merging of the green (CK) and red (LPL) images showed that CTBs stained for LPL, and expression increased dramatically with advancing gestational age (c, f, and i). Each inset is a x63 magnification of the region contained within the white dashed lines (d–i). Photos are representative of the basal plate staining at 16 wk (n = 3), 23–24 wk (n = 5), and 37–40 wk (n = 4). B, Expression of FABP4, which is localized to the nuclei of human basal plate CTBs, increased expression at term. Human basal plate biopsy specimens from 19-wk (a–c) or 37-wk (d–f) placentas were stained with anti-CK (1:50) to identify the invasive CTBs (a and d) and Hoechst (1:1000) to identify nuclei. Sections were costained with anti-FABP4 antibody (1:1000; b and e). Strong FABP4 immunoreactivity was detected in association with cells of the basal plate at term, compared with the second trimester (compare b and e). CK-positive CTBs stained primarily in a nuclear pattern, whereas staining in maternal cells was largely in the cytoplasmic compartment (c and f). Photos are representative of the basal plate staining at midgestation (15–24 wk; n = 11) and term (n = 3).

 
We also used an immunolocalization approach to validate up-regulation of FABP4 expression at term. In accord with the microarray data, faint immunoreactivity was detected in association with cells of the basal plate during the second trimester. Maternal cells (CK negative) expressed FABP4 in the cytoplasmic compartment (Fig. 5BGo, c), although occasional nuclear staining was also observed in some of the earlier gestation samples (data not shown). Fetal cells (CK-positive trophoblasts) showed essentially no FABP4 staining. In contrast, strong FABP4 immunoreactivity was detected in association with cells of the basal plate at term. In general, the fetal cells had intense nuclear and minimal cytoplasmic staining. Maternal cells had a similar staining pattern, but at much lower levels (Fig. 5BGo, f).

Protein modeling to gain insight into possible functions of the nonannotated differentially expressed genes
Given the large number of hypothetical proteins and nonannotated sequences among the differentially expressed genes, we used modeling to gain insight into their potential functions. Of the 78 hypothetical proteins or nonannotated sequences encoded by the differentially expressed genes that were evaluated, 31 were reliably annotated for potential functions. These results are shown in Table 3Go. Subsets of potential functions included myosin related molecules, enzymes, and ribosomal-like proteins. The complete analysis with links to the MODBASE models is provided as supplemental data (supplemental Table 4).


View this table:
[in this window]
[in a new window]

  		TABLE 3. Functional annotation determined from protein modeling

 

   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
We performed a global analysis of gene expression profiles at the human maternal-fetal interface between midgestation (14–24 wk) and term (37–40 wk). The 418 differentially expressed genes included molecules with well-characterized expression patterns, specifically up-regulation of CRH, and the inhibin-ßA subunit and down-regulation of C-X-C chemokine ligand 14 and angiopoietin-2, which served as positive controls. The analysis also revealed a number of novel observations, such as the large proportion of nonannotated or hypothetical proteins. Independent analyses of the expression patterns of 16 of the most highly regulated genes gave the predicted results, suggesting the validity of the data set as a whole. The differentially expressed genes encompassed numerous biological processes, including angiogenesis, cell motility, extracellular matrix modulation, gene transcription, signal transduction, immune response, protein biosynthesis, and lipid metabolism. The significance of a portion of these data are discussed in greater detail below. K-means cluster analysis showed that some genes were tightly coexpressed, suggesting the possibility of common regulatory mechanisms. Pathway analysis demonstrated significant alterations in several metabolic and signaling processes. Most notably, a number of differentially expressed genes were associated with folate biosynthesis, N-glycan biosynthesis, and Wnt-ß-catenin signaling. Modeling of the hypothetical proteins and nonannotated sequences suggested possible functions for a subset of these putative molecules. Together, these data provide new insights into the dynamics of gene expression at the human maternal-fetal interface during pregnancy and numerous directions for additional functional analyses.

To our knowledge, this is the first application of a microarray approach for profiling the gene expression patterns over gestation of the conjoined areas of the human placenta and decidua that compose the maternal-fetal interface. Given the revised estimate that the human genome contains 29,000–36,000 protein-encoding genes (37), our analysis of 39,000 transcripts generated a comprehensive data set. We note that other investigators have used microarray technologies to evaluate gene expression patterns during human trophoblast or placental development. Cheng et al. (11) used a cDNA platform with 9600 cloned sequences to separately evaluate the gene expression profiles of first-trimester decidua and chorionic villi. In their analysis, the maternal-fetal interface was deliberately excluded. Recently Wyatt et al. (38) compared gene expression patterns of samples taken from the medial and lateral portions of the placenta. In many instances, no spatial variations were noted, but in a few cases, up to a 3-fold difference was observed. Although that paper was published after the samples for our study had been collected, we took this potentially confounding factor into account by pooling samples from both locations. Most recently Sood et al. (39) reported the gene expression profiles of different anatomical locations of human placentas that were obtained after normal term deliveries.

In this study, our goal was to begin to analyze the molecular components of the dialogue that takes place between maternal and fetal cells in the human basal plate region over gestation. In interpreting the microarray data obtained from analysis of this complex tissue, we were cognizant of the fact that differential gene expression could reflect changes that involve one or multiple cell types, with alterations in the cellular composition of the basal plate region yet another possibility. Further investigation of the expression patterns and biological functions of the up- and down-regulated molecules will shed additional light on this issue. Another potential confounder could be maternal blood contamination. The maternal-fetal interface is richly vascularized, and blood flow to this region increases at term. Furthermore, it is impossible to completely remove blood cells from this tissue, even by multiple washes in large volumes of PBS, the procedure we used. Therefore, the expression levels of hemoglobin-ß, a transcript that is highly expressed in red blood cells, were evaluated in relationship to the expression patterns of the differentially expressed genes. This analysis showed that only hemoglobin-{delta} expression correlated tightly with that of hemoglobin-ß (Pearson’s coefficient r2 = 0.92), followed by Kruppel-like factor 2 (r2= 0.757), which was below the level of significance. Therefore, these results suggest that none of the expression patterns of the other differentially expressed genes can be solely explained by varying amounts of blood contamination. Finally, drug exposure could also be a possible confounder because we did not have access to the clinical data regarding the type of sedation that was used in each case. In general, the midgestation procedures were performed under conscious sedation (fentenyl and versed), and patients who underwent cesarean sections had regional anesthesia (epidural or spinal).

Because the basal plate region is comprised of a mixture of maternal and fetal cells, we also determined whether changes in the proportion of cytotrophoblasts could account for any of the differentially expressed genes. To address this possibility, we exploited the fact that approximately 50% of the placentas were from pregnancies in which the fetus was male. We thus estimated the contribution of fetal cells by examining the expression patterns of two Y-linked genes, ribosomal protein S4–1 and (DEAD) box polypeptide 3. Analysis of the data showed that the log intensity values of these probe sets had less variability than those of the differentially expressed genes. Furthermore, there was no significant correlation between the expression patterns of these transcripts and genes that were either up- or down-regulated over gestation (Pearson correlation coefficients < 0.8). Additionally none of the differentially expressed genes showed significant correlations with cytokeratin-7, which is specifically expressed by all the trophoblast populations in this region (Pearson > 0.08). Together, these data suggest that the observed changes in gene expression were unlikely to be attributable to variable contributions of fetal cells to the basal plate samples we analyzed.

With regard to specific categories of the differentially expressed genes, we were particularly interested in those that we had not expected to show dramatic patterns of regulation, e.g. molecules that play a role in lipid metabolism. As a result, we included a number of these gene products in the validation studies. Immunolocalization of LPL and FABP4 illustrated the fact that the differentially expressed genes that were identified in the microarray analysis can reflect changes that occur in either a single cell type or both maternal and fetal cells. Specifically, our analysis showed that invasive CTBs dramatically up-regulated expression of cytoplasmic LPL levels at term with minimal expression noted in maternal cells. This pattern of regulation is in contrast to maternal plasma LPL levels, which decrease during human pregnancy (40). It is possible that the increase of CTB LPL expression at term occurs in response to an increased demand for fatty lipid precursors that are used in the synthesis of prostaglandins required for labor. This theory is supported by the analysis of the LPL immunoreactivity that is associated with the cells that compose the human fetal membranes at term (i.e. the location of known prostaglandin production for parturition). Specifically, anti-LPL localized to the basolateral surface of amniotic epithelia, the chorionic CTB layer exhibited plasma membrane immunoreactivity, and the extracellular compartment stained as well (41). It is interesting to note that invasive CTBs and cells of the amnion/chorion appear to have different subcellular patterns of LPL distribution. In this regard, the fetal membranes studied by Huter et al. (41) were obtained after vaginal delivery, whereas our samples were obtained from patients who did not experience labor. Together, these findings suggest that LPL might be stored within the cell before initiation of parturition, subsequently localizing to the plasma membrane and extracellular compartment during the delivery process.

The expression of FABP4, or adipocyte P2, which plays a role in lipid signal transduction, also increased dramatically at term predominantly in the invasive CTBs; decidual cell expression was detected from midgestation onward. This molecule, which is highly expressed in adipose tissue, serves as a marker of adipocyte maturation (42). Like other FABPs, its functions include fatty acid uptake and intracellular shuttling (43). Consistent with this role, FABP4 physically interacts with the hormone-sensitive lipase of adipose tissue, binding a fatty acid product, thereby facilitating lipolysis (44) and intracellular delivery of ligands to regulatory proteins. Indeed, FABP4 specifically provides lipid ligands to the nuclear receptor peroxisome proliferative activated receptor-{gamma}, which in turn regulates gene transcription (45). To our knowledge, this is the first report of FABP4 expression in the human basal plate. Of note is the fact that FABP4, LPL, and CRH (cluster B) had very similar expression patterns, with r2 values of 0.84 to 0.85. The coexpression of these molecules in the basal plate region suggests that enhanced availability of fatty acids may be part of the molecular preparation for parturition. The coregulation of these molecules also raises the possibility that CRH may govern LPL and FABP4 expression. In this regard, there is a recent report that CRH influences the functions of other lipases (46). Alternatively, the coexpression of all three molecules might be governed by a higher-order mechanism, the basis of which could provide important new clues to the signals that trigger labor.

Examination of the highly regulated immune molecules highlighted the possibility that changes in gene expression could reflect alterations in the cellular composition of the basal plate. The expression of granulysin, which localizes to the cytolytic granules of T cells, natural killer (NK) cells (47), and certain dendritic cells (48), is down-regulated at term. We speculate that the decreased expression we observed reflects the known parallel decrease in the number of T and NK cells at the maternal-fetal interface at term (49). The down-regulation of the expression of Ly96, another NK cell-specific molecule, provides further support for this concept. Although the mechanisms that lead to the eventual disappearance of decidual leukocytes from the maternal-fetal interface are not known, the decrease in expression noted for the chemotactic molecules, such as chemokine-like factor superfamily 6 and secreted phosphoprotein 1 (Table 2GoGoGoGo), could be a related phenomenon.

Our group has also been interested in the functions of the myriad of angiogenic factors that are produced at the maternal-fetal interface (3, 50). Consistent with our previously published work, we found that the down-regulated genes included ANGPT-2 (51). This microarray analysis also revealed a striking co-down-regulation at term of ANGPT-2 and MCPH1. MCPH1 controls brain size in humans by regulating the proliferative and hence differentiative capacity of neuroblasts, ultimately exerting its affects through cell cycle regulators such as checkpoint kinase 1 and breast cancer 1 (38, 52). MCPH1 was previously reported to be expressed by (fetal) brain, liver and kidney. Thus, it will be interesting to determine which cell type(s) at the maternal-fetal interface expresses this molecule. Additional analysis of the ANGPT-2 and MCPH1 genes revealed they are transcribed from opposite strands of the same region (chromosome 8p23.1). Their tight coexpression suggests that transcription from this area could be silenced at term, perhaps by local chromatin modifications or the recruitment of inhibitory protein complexes to the same promoter element. It will be interesting to determine whether the pattern of coexpression of ANGPT-2 and MCPH1 occurs in other tissues or is placental specific. Finally, there is evidence that strong genetic selection has been exerted on MCPH1 during recent human evolution (53). Whereas the most obvious selection pressure may be on brain size, another interesting possibility is that the placenta is coevolving.

In summary, the maternal-fetal interface is a remarkable chimeric tissue that holds the answers to many interesting biological questions regarding invasive behavior, vasculogenesis/angiogenesis and immunotolerance. Our study provides a global analysis of the expression patterns of genes that are involved in these and other processes. The fact that molecules with known expression patterns were correctly regulated bolsters our confidence in the novel data we obtained. Finally, the results of this study provide reference data sets of gene expression profiles against which changes that occur in a variety of pregnancy complications can be measured. For example, impaired CTB invasion has been associated with recurrent miscarriages (54, 55), preeclampsia (56), intrauterine growth restriction (56), and a subset of preterm labor cases (57). Now we are in a position to differentiate changes in gene expression that occur in these conditions from those that are attributable to advancing gestational age.


   Acknowledgments
 
We thank Ms. Jean Perry, the study research nurse coordinator, who assisted in tissue collection, as did the nurses, residents, and faculty at San Francisco General Hospital Women’s Options Center and the UCSF Birth Center. Dr. Chris Barker, Dr. Chandi Griffin, and Ms. Jennifer Gregg, members of the UCSF Gladstone (National Heart, Lung, and Blood Institute) Shared Microarray Facility, contributed invaluable technical and intellectual expertise. We also thank Drs. David Erle and Michael Salazar, members of the Sandler Genomics Core Facility, for helpful discussions and assistance with data deposition. We are grateful to Mr. Evan Messenger and Dr. Kathy Ivanetich, UCSF Biomolecular Research Center, for technical assistance in performing the Q-PCR experiments. We thank Dr. John Brunzell (University of Washington) for the kind gift of the LPL-specific antibody that was used in the immunolocalization studies. Finally, we are grateful to Ms. Mary McKenney and Dr. Leonard Dragone for critical review of the manuscript.


   Footnotes
 
This work was supported by National Institutes of Health Grants R01 HL 64597 (to S.J.F.), R21 AI53782 (to L.P.), DK 053189 (to D.A.B.), 5-MO1-RR 00083 (General Clinical Research Center, San Francisco General Hospital), and HL 072301 (University of California, San Francisco, National Heart, Lung, and Blood Institute Shared Microarray Facility). V.D.W. is a March of Dimes Reproductive Scientists Development Program scholar (National Institute of Child Health and Human Development 5K12HD00849) and an American Board of Obstetrics and Gynecology/American Association of Obstetricians and Gynecologists Foundation scholar.

Author Disclosure Summary: V.D.W., R.H.-K., A.C.P., Y.J.Y., M.S.M., M.G., K.-T.V.F., S.M., L.P., A.S., and S.J.F. have nothing to declare. D.A.B. has received lecture fees from Merck and Pfizer.

First Published Online December 14, 2006

Abbreviations: ANGPT-2, Angiopoietin-2; CK, cytokeratin; CTB, cytotrophoblast; FABP4, fatty acid binding protein 4; GO, gene ontogeny; IPA, Ingenuity Pathway Analysis; LPL, lipoprotein lipase; MCPH1, microcephalin; NK, natural killer; Q-PCR, quantitative PCR.

Received May 19, 2006.

Accepted for publication December 4, 2006.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Damsky CH, Librach C, Lim KH, Fitzgerald ML, McMaster MT, Janatpour M, Zhou Y, Logan SK, Fisher SJ 1994 Integrin switching regulates normal trophoblast invasion. Development 120:3657–3666[Abstract]
  2. Zhou Y, Fisher SJ, Janatpour M, Genbacev O, Dejana E, Wheelock M, Damsky CH 1997 Human cytotrophoblasts adopt a vascular phenotype as they differentiate. A strategy for successful endovascular invasion? J Clin Invest 99:2139–2151[Medline]
  3. Red-Horse K, Zhou Y, Genbacev O, Prakobphol A, Foulk R, McMaster M, Fisher SJ 2004 Trophoblast differentiation during embryo implantation and formation of the maternal-fetal interface. J Clin Invest 114:744–754[CrossRef][Medline]
  4. Clark DE, Smith SK, Licence D, Evans AL, Charnock-Jones DS 1998 Comparison of expression patterns for placenta growth factor, vascular endothelial growth factor (VEGF), VEGF-B and VEGF-C in the human placenta throughout gestation. J Endocrinol 159:459–467[Abstract]
  5. Zhou Y, McMaster M, Woo K, Janatpour M, Perry J, Karpanen T, Alitalo K, Damsky C, Fisher SJ 2002 Vascular endothelial growth factor ligands and receptors that regulate human cytotrophoblast survival are dysregulated in severe preeclampsia and hemolysis, elevated liver enzymes, and low platelets syndrome. Am J Pathol 160:1405–1423[Abstract/Free Full Text]
  6. Staun-Ram E, Goldman S, Gabarin D, Shalev E 2004 Expression and importance of matrix metalloproteinase 2 and 9 (MMP-2 and -9) in human trophoblast invasion. Reprod Biol Endocrinol 2:59[CrossRef][Medline]
  7. Librach CL, Feigenbaum SL, Bass KE, Cui TY, Verastas N, Sadovsky Y, Quigley JP, French DL, Fisher SJ 1994 Interleukin-1ß regulates human cytotrophoblast metalloproteinase activity and invasion in vitro. J Biol Chem 269:17125–17131[Abstract/Free Full Text]
  8. McMaster MT, Librach CL, Zhou Y, Lim KH, Janatpour MJ, DeMars R, Kovats S, Damsky C, Fisher SJ 1995 Human placental HLA-G expression is restricted to differentiated cytotrophoblasts. J Immunol 154:3771–3778[Abstract]
  9. Mattsson R 1998 The non-expression of MHC class II in trophoblast cells. Am J Reprod Immunol 40:383–384
  10. Aronow BJ, Richardson BD, Handwerger S 2001 Microarray analysis of trophoblast differentiation: gene expression reprogramming in key gene function categories. Physiol Genomics 6:105–116[Abstract/Free Full Text]
  11. Cheng YH, Aronow BJ, Hossain S, Trapnell B, Kong S, Handwerger S 2004 Critical role for transcription factor AP-2{alpha} in human trophoblast differentiation. Physiol Genomics 18:99–107[Abstract/Free Full Text]
  12. Kudo Y, Boyd CA, Sargent IL, Redman CW, Lee JM, Freeman TC 2004 An analysis using DNA microarray of the time course of gene expression during syncytialization of a human placental cell line (BeWo). Placenta 25:479–488[CrossRef][Medline]
  13. Roh CR, Budhraja V, Kim HS, Nelson DM, Sadovsky Y 2005 Microarray-based identification of differentially expressed genes in hypoxic term human trophoblasts and in placental villi of pregnancies with growth restricted fetuses. Placenta 26:319–328[CrossRef][Medline]
  14. Chen B, Nelson DM, Sadovsky Y 2006 N-myc down-regulated gene 1 modulates the response of term human trophoblasts to hypoxic injury. J Biol Chem 281:2764–2772[Abstract/Free Full Text]
  15. Tanaka TS, Jaradat SA, Lim MK, Kargul GJ, Wang X, Grahovac MJ, Pantano S, Sano Y, Piao Y, Nagaraja R, Doi H, Wood 3rd WH, Becker KG, Ko MS 2000 Genome-wide expression profiling of mid-gestation placenta and embryo using a 15,000 mouse developmental cDNA microarray. Proc Natl Acad Sci USA 97:9127–9132[Abstract/Free Full Text]
  16. Galaviz-Hernandez C, Stagg C, de Ridder G, Tanaka TS, Ko MS, Schlessinger D, Nagaraja R 2003 Plac8 and Plac9, novel placental-enriched genes identified through microarray analysis. Gene 309:81–89[CrossRef][Medline]
  17. Hemberger M, Cross JC, Ropers HH, Lehrach H, Fundele R, Himmelbauer H 2001 UniGene cDNA array-based monitoring of transcriptome changes during mouse placental development. Proc Natl Acad Sci USA 98:13126–13131[Abstract/Free Full Text]
  18. Drey EA, Kang MS, McFarland W, Darney PD 2005 Improving the accuracy of fetal foot length to confirm gestational duration. Obstet Gynecol 105:773–778[Abstract/Free Full Text]
  19. American College of Obstetricians and Gynecologists 1999 Practice bulletin 10. Induction of labor. Washington, DC: American College of Obstetricians and Gynecologists
  20. Haimov-Kochman R, Fisher SJ, Winn VD 2006 Modification of the standard Trizol-based technique improves the integrity of RNA isolated from RNase-rich placental tissue. Clin Chem 52:159–160[Free Full Text]
  21. Auer H, Lyianarachchi S, Newsom D, Klisovic MI, Marcucci G, Kornacker K 2003 Chipping away at the chip bias: RNA degradation in microarray analysis. Nat Genet 35:292–293[CrossRef][Medline]
  22. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B, Gautier L, Ge Y, Gentry J, Hornik K, Hothorn T, Huber W, Iacus S, Irizarry R, Leisch F, Li C, Maechler M, Rossini AJ, Sawitzki G, Smith C, Smyth G, Tierney L, Yang JY, Zhang J 2004 Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 5:R80
  23. Bolstad BM 2004 Low Level analysis of high-density oligonucleotide array data: background, normalization and summarization. Dissertation, University of California, Berkeley
  24. Smyth GK 2005 Limma: linear models for microarray data. In: Gentleman RC, Carey V, Duboit R, Irizarry W, Huber, ed. Bioinformatics and computational biology solutions using R and Bioconductor. New York: Springer; 397–420
  25. Tibshirani R, Walther G, Hastie T 2001 Estimating the number of clusters in a data set via the gap statistic. J R Stat Soc Series B 63:411–423[CrossRef]
  26. Damsky CH, Fitzgerald ML, Fisher SJ 1992 Distribution patterns of extracellular matrix components and adhesion receptors are intricately modulated during first trimester cytotrophoblast differentiation along the invasive pathway, in vivo. J Clin Invest 89:210–222[Medline]
  27. Hertzel AV, Bennaars-Eiden A, Bernlohr DA 2002 Increased lipolysis in transgenic animals overexpressing the epithelial fatty acid binding protein in adipose cells. J Lipid Res 43:2105–2111[Abstract/Free Full Text]
  28. Wheeler DL, Barrett T, Benson DA, Bryant SH, Canese K, Chetvernin V, Church DM, DiCuccio M, Edgar R, Federhen S, Geer LY, Helmberg W, Kapustin Y, Kenton DL, Khovayko O, Lipman DJ, Madden TL, Maglott DR, Ostell J, Pruitt KD, Schuler GD, Schriml LM, Sequeira E, Sherry ST, Sirotkin K, Souvorov A, Starchenko G, Suzek TO, Tatusov R, Tatusova TA, Wagner L, Yaschenko E 2006 Database resources of the National Center for Biotechnology Information. Nucleic Acids Res 34:D173–180
  29. Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W, Lipman DJ 1997 Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25:3389–3402[Abstract/Free Full Text]
  30. Eswar N, John B, Mirkovic N, Fiser A, Ilyin VA, Pieper U, Stuart AC, Marti-Renom MA, Madhusudhan MS, Yerkovich B, Sali A 2003 Tools for comparative protein structure modeling and analysis. Nucleic Acids Res 31:3375–3380[Abstract/Free Full Text]
  31. Sali A, Blundell TL 1993 Comparative protein modelling by satisfaction of spatial restraints. J Mol Biol 234:779–815[CrossRef][Medline]
  32. Pieper U, Eswar N, Braberg H, Madhusudhan MS, Davis FP, Stuart AC, Mirkovic N, Rossi A, Marti-Renom MA, Fiser A, Webb B, Greenblatt D, Huang CC, Ferrin TE, Sali A 2004 MODBASE, a database of annotated comparative protein structure models, and associated resources. Nucleic Acids Res 32:D217–D222
  33. McGuffin LJ, Jones DT 2003 Improvement of the GenTHREADER method for genomic fold recognition. Bioinformatics 19:874–881[Abstract/Free Full Text]
  34. McLean M, Smith R 1999 Corticotropin-releasing hormone in human pregnancy and parturition. Trends Endocrinol Metab 10:174–178[CrossRef][Medline]
  35. Petraglia F, Calza L, Garuti GC, Abrate M, Giardino L, Genazzani AR, Vale W, Meunier H 1990 Presence and synthesis of inhibin subunits in human decidua. J Clin Endocrinol Metab 71:487–1492[Abstract]
  36. Drake PM, Red-Horse K, Fisher SJ 2004 Reciprocal chemokine receptor and ligand expression in the human placenta: implications for cytotrophoblast differentiation. Dev Dyn 229:877–885[CrossRef][Medline]
  37. Collins JE, Goward ME, Cole CG, Smink LJ, Huckle EJ, Knowles S, Bye JM, Beare DM, Dunham I 2003 Reevaluating human gene annotation: a second-generation analysis of chromosome 22. Genome Res 13:27–36[Abstract/Free Full Text]
  38. Wyatt SM, Kraus FT, Roh CR, Elchalal U, Nelson DM, Sadovsky Y 2005 The correlation between sampling site and gene expression in the term human placenta. Placenta 26:372–379[CrossRef][Medline]
  39. Sood R, Zehnder JL, Druzin ML, Brown PO 2006 Gene expression patterns in human placenta. Proc Natl Acad Sci USA 103:5478–5483[Abstract/Free Full Text]
  40. Kinoshita T, Shirai K, Itoh M 2003 The level of pre-heparin serum lipoprotein lipase mass at different stages of pregnancy. Clin Chim Acta 337:153–156[CrossRef][Medline]
  41. Huter O, Wolf HJ, Schnetzer A, Pfaller K 1997 Lipoprotein lipase, LDL receptors and apo-lipoproteins in human fetal membranes at term. Placenta 18:707–715[CrossRef][Medline]
  42. Reese-Wagoner A, Thompson J, Banaszak L 1999 Structural properties of the adipocyte lipid binding protein. Biochim Biophys Acta 1441:106–116[Medline]
  43. Zimmerman AW, Veerkamp JH 2002 New insights into the structure and function of fatty acid-binding proteins. Cell Mol Life Sci 59:1096–1116[CrossRef][Medline]
  44. Smith AJ, Sanders MA, Thompson BR, Londos C, Kraemer FB, Bernlohr DA 2004 Physical association between the adipocyte fatty acid-binding protein and hormone-sensitive lipase: a fluorescence resonance energy transfer analysis. J Biol Chem 279:52399–52405[Abstract/Free Full Text]
  45. Tan NS, Shaw NS, Vinckenbosch N, Liu P, Yasmin R, Desvergne B, Wahli W, Noy N 2002 Selective cooperation between fatty acid binding proteins and peroxisome proliferator-activated receptors in regulating transcription. Mol Cell Biol 22:5114–5127[Abstract/Free Full Text]
  46. Okada S, Shimizu T, Yokotani K 2003 Brain phospholipase C and diacylglycerol lipase are involved in corticotropin-releasing hormone-induced sympatho-adrenomedullary outflow in rats. Eur J Pharmacol 475:49–54[CrossRef][Medline]
  47. Krensky AM 2000 Granulysin: a novel antimicrobial peptide of cytolytic T lymphocytes and natural killer cells. Biochem Pharmacol 59:317–320[CrossRef][Medline]
  48. Raychaudhuri SP, Jiang WY, Raychaudhuri SK, Krensky AM 2004 Lesional T cells and dermal dendrocytes in psoriasis plaque express increased levels of granulysin. J Am Acad Dermatol 51:1006–1008[CrossRef][Medline]
  49. King A, Loke YW, Chaouat G 1997 NK cells and reproduction. Immunol Today 18:64–66[CrossRef][Medline]
  50. Levine RJ, Karumanchi SA 2005 Circulating angiogenic factors in preeclampsia. Clin Obstet Gynecol 48:372–386[CrossRef][Medline]
  51. Zhou Y, Bellingard V, Feng KT, McMaster M, Fisher SJ 2003 Human cytotrophoblasts promote endothelial survival and vascular remodeling through secretion of Ang2, PlGF, and VEGF-C. Dev Biol 263:114–125[CrossRef][Medline]
  52. Xu X, Lee J, Stern DF 2004 Microcephalin is a DNA damage response protein involved in regulation of CHK1 and BRCA1. J Biol Chem 279:34091–34094[Abstract/Free Full Text]
  53. Evans PD, Gilbert SL, Mekel-Bobrov N, Vallender EJ, Anderson JR, Vaez-Azizi LM, Tishkoff SA, Hudson RR, Lahn BT 2005 Microcephalin, a gene regulating brain size, continues to evolve adaptively in humans. Science 309:1717–1720[Abstract/Free Full Text]
  54. Khong TY, Liddell HS, Robertson WB 1987 Defective haemochorial placentation as a cause of miscarriage: a preliminary study. Br J Obstet Gynaecol 94:649–655[Medline]
  55. Michel MZ, Khong TY, Clark DA, Beard RW 1990 A morphological and immunological study of human placental bed biopsies in miscarriage. Br J Obstet Gynaecol 97:984–988[Medline]
  56. Kaufmann P, Black S, Huppertz B 2003 Endovascular trophoblast invasion: implications for the pathogenesis of intrauterine growth retardation and preeclampsia. Biol Reprod 69:1–7