The correlation between sampling site and gene expression in the term human placenta

Placenta (2005), 26, 372e379 doi:10.1016/j.placenta.2004.07.003

The Correlation Between Sampling Site and Gene Expression in the Term Human Placenta* S. M. Wyatt, F. T. Kraus, C.-R. Roh, U. Elchalal, D. M. Nelson and Y. Sadovsky* Department of OBGYN, Washington University School of Medicine, St Louis, MO 63110, USA; Department of Cell Biology and Physiology, Washington University School of Medicine, St Louis, MO 63110, USA Paper accepted 16 July 2004

Using oligonucleotide microarrays we recently identified a set of transcripts that were up-regulated in hypoxic human trophoblasts. To test the hypothesis that expression of hypoxia-related placental transcripts depends on sampling site we analyzed nine different sites from term human placentas (n Z 6), obtained after uncomplicated pregnancies. These sites spanned the placental center to the lateral border and the basal to the chorionic plate. Relative gene expression at each site, determined using quantitative PCR, was correlated with villous histology. The expression of vascular endothelial growth factor (VEGF) and connective tissue growth factor (CTGF), the cytoskeleton proteins lamininA3 and a-tubulin, and the signal transduction protein Rad was enhanced in the subchorionic lateral border compared to medial basal site (1.6e2.9 fold, p ! 0.05). In contrast, the expression of NDRG1, adipophilin and human placental lactogen was unchanged. Enhanced villous maturation, syncytial knots and fibrin deposits were more frequent in the subchorionic placental lateral border, and correlated with up-regulation of hypoxiarelated transcripts (p ! 0.05). The association between sample site and expression level was not observed in placentas with marginal cord insertion. The expression of hypoxia-related genes in the term human placenta is dependent on sampling site within the placental disk, likely reflecting local differences in villous perfusion. Placenta (2005), 26, 372e379 Ó 2004 Elsevier Ltd. All rights reserved.

INTRODUCTION The maternal and fetal blood supply to the human placenta interface at the vasculosyncytial membrane and the supportive connective tissue. Maternal blood is supplied by uterine radial arteries, and percolates the placental intervillous space through the basal plate. Fetal blood is supplied by the two fetal umbilical arteries via the placental chorionic plate, and perfuses the villous core [1]. The trophoblasts at the villous surface regulate gas and nutrient exchange, and provide endocrine and immunological support to the growing fetus. Similar to most other tissues, the trophoblast is subject to ongoing injury and adaptation in response to diverse insults, such as hypoperfusion that results in cellular hypoxia. Whereas homeostasis is sustained in normal pregnancy, marked placental injury or inadequate repair may lead to villous damage, which is associated with fetal growth restriction [2]. To understand the molecular mechanisms underlying trophoblast response to placental hypoperfusion and cellular *

Presented, in part, at the 24th Meeting of the Society for Maternal-Fetal Medicine, February 2004. * Corresponding author. Washington University School of Medicine, OBGYN Campus Box 8064, 4566 Scott Avenue, St. Louis, MO 63110, USA. Tel.: C1 314 747 0937; fax: C1 314 747 1256. E-mail address: [email protected] (Y. Sadovsky). 0143e4004/$esee front matter

hypoxia we seek to analyze changes in gene expression in placental villi exposed to these insults. The advent of expression microarrays as high-throughput genetic screens, combined with detailed mapping of the human genome, provide for the first time an invaluable advance toward the goal of comprehensively defining the complex molecular mechanisms that underlie tissue response and adaptation to injury. Indeed, genomic and proteomic approaches were recently employed in order to gain insight into differentiation of human trophoblast and murine trophoblast stem cells [3e5]. Although microarray is an exceptionally robust technique, its use is hampered by technological and biological variability. Both sources of error are key determinants of microarray reliability, and therefore critical for assessing which genes are truly differentially expressed. Using a large replicate data set derived from human placental trophoblast we recently developed a novel methodology to analyze probe-specific contribution to technological variability of gene expression [6]. We also applied this methodology for the examination of altered gene expression in hypoxic trophoblasts [7]. Nevertheless, our approach does not address biological variability, which is largely contributed by tissue variations and experimental conditions. Placental architecture and blood flow are not uniform across the chorioallantoic human placental disk [8,9]. Proximity to Ó 2004 Elsevier Ltd. All rights reserved.

Wyatt et al.: Correlation between Sampling Site and Gene Expression

the umbilical cord, basal plate or chorionic plate may influence perfusion. Consistent with this notion, syncytial knots and villous fibrin are more common near the chorionic surface as well as near the placental margin compared to other locations in the placenta [10]. These regional differences, which are found in placentas derived from uncomplicated pregnancies, are also characteristic of under-perfused villi [11]. Whereas Doppler flow studies of the human placenta advance our understanding of gestational age dependent perfusion [12], detailed mapping of regional perfusion has not been performed. Despite these limitations it seems plausible that differences in sampling site contribute to variability in gene expression across the placental disk. Here we tested the hypothesis that the expression of hypoxia-related placental transcripts is dependent on placental sampling site. To test this hypothesis we analyzed regional differences in expression of several hypoxia-induced placental genes within term human placentas derived from uncomplicated pregnancies. These genes, identified using our microarray screen [6,7], included the growth factors vascular endothelial growth factor (VEGF) and connective tissue growth factor (CTGF), the cytoskeleton proteins lamininA3 and a-tubulin, the signaling proteins Rad (Ras associated with diabetes) and NDRG1, and the fatty acid droplet-associated protein adipophilin. MATERIAL AND METHODS Placental tissue acquisition Fresh samples of human placentas derived from uncomplicated pregnancies were obtained immediately following uneventful spontaneous vaginal deliveries. All pregnancies ended between 37 and 40 weeks of gestation, defined by standard clinical criteria. Placentas were excluded if the pregnancy was complicated by maternal diabetes, hypertensive disorder, tobacco or substance abuse, infection, prolonged rupture of membranes, multi-fetal gestation, meconium-stained amniotic fluid, placental abruption, previa, accreta or placentas harboring visible abnormalities, known fetal chromosomal abnormalities, structural anomaly or abnormal growth. Infant weight was between 3250 and 3865 g. Umbilical cord insertion was defined as central if located within the inner third of the placental disk radius, whereas marginal cord was defined when insertion was located within 1 cm from the lateral edge of the placental disk. Placentas with velamentous cord insertion were excluded. The study was approved by the Institutional Review Board of Washington University School of Medicine. Placental biopsy All samples were obtained within 10 min after placental delivery. Each placenta was placed in a sterile tray, and carefully inspected for any visible abnormalities as well as for location of the umbilical cord. Using a sterile scalpel we excised a triangular segment of the placenta, with its convex base (approximately 4 cm) at the lateral edge of the placenta and its apex at the placental center near the cord insertion site.

373

The dissected section was re-inspected for any visible abnormalities and both the decidual layer along the basal plate as well as the chorionic surface and membranes were removed by sharp dissection. The remaining tissue was then divided along the long axis into three portions, approximately 3 cm in length, from the medial to the lateral edge. Each of these portions was then divided into three horizontal segments from basal towards chorionic surface, as illustrated in Figure 1A. This resulted in nine samples, each approximately 0.75 cm3, that were labeled from the medial region (1) to lateral edge (3) and from the basal plate (A) to chorionic plate (C) as shown in Figure 1A. All samples were obtained 2 cm away from the lateral placental margin, in order to avoid areas rich in intervillous fibrin [13,14]. Samples from placentas with marginal cord insertion were collected in a similar manner, with the wedge shaped portion dissected from the region directly opposite to umbilical cord insertion. Six samples, corresponding to sites 1AeC and 3AeC in the placentas with central cord, were collected as described. Three additional samples, representing the three segments along the placental thickness from basal towards chorionic surface, were obtained from the area immediately adjacent to the marginal cord insertion. Each sample was rinsed thoroughly in PBS, and then divided with a scalpel into two identical specimens. One specimen was snap frozen in liquid nitrogen and stored at ÿ80 (C for RNA analysis while the other was placed in freshly prepared 4% paraformaldehyde in phosphate buffer and processed for paraffin embedding, sectioning and hematoxylin and eosin staining. RNA purification, reverse transcription and quantitative PCR Placental biopsies were homogenized in TriReagent (Molecular Research Center, Inc., Cincinnati, OH) as recommended by manufacturer. RNA was digested with DNase I (Roche Applied Sciences, Indianapolis, IN) and quantified by absorbance at 260 nm. Electrophoresis on a 1% agarose gel was performed to confirm the presence and quality of RNA. Reverse transcription was done with 2 mg of total RNA added to 10! RT buffer, 5.5 mmol/L MgCl2, 2.5 mmol/L random hexamers, 500 mmol/L dNTP mixture, 20 U RNase enzyme and 50 U MultiScribe reverse transcriptase for a final volume of 50 mL. Reverse transcription reactions were performed by incubation for 10 min at 25 (C, reverse transcription for 30 min at 48 (C and inactivation for 5 min at 95 (C. Relative expression levels of RNA per sample were quantified by SYBR Green I assay on GeneAmp 5700 Sequence Detection System (PE Applied Biosystems, Foster City, CA). For each transcript, PCR was performed in duplicates with 50 mL reaction volumes of 3 mL cDNA, 25 mL SYBR Green PCR master mix and 0.3 mmol/L of each primer. PCR was conducted using the following cycle parameters: 2 min at 50 (C, 10 min at 95 (C, and 40 two-step cycles of 15 s at 95 (C and 1 min at 60 (C. The sequences of forward and reverse primers were selected using Primer Express Software

Placenta (2005), Vol. 26

374

Table 1. Sequences used for real-time quantitative PCR Name

GenBank accession no.

Direction and site

Sequence

VEGF

AF024710 gi:2565322 X78947 gi:474933 D87953 gi:1596166 L34155 gi:551596 L24564 gi:439602 X06956 gi:32014 X97324 gi:1806039 J03071 gi:183148

F(681e699) R(754e735) F(639e657) R(717e698) F(755e773) R(822e801) F(232e256) R(322e303) F(795e819) R(870e852) F(4557e4576) R(4642e4621) F(147e167) R(211e193) F(51557e51579) R(51636e51614)

CTGGCGCTGAGCCTCTCTA CCGGTGTCCTCATCCCTGTA TGTGTGACGAGCCCAAGGA TCTGGGCCAAACGTGTCTTC CCGCCAGCACATTGTGAAT GGCTGTTGTAGGCATTGATGAA GATGGCTCAGGCATATGTGTTAACT GGACGGCGTTGCCATAGTAG CTTTGACTGCAAGTTCATTGAGACA GCGCACGACACCTTCAAAC TCATGGTGGACAACGAAGCA GAGGCGATTGAGGTTGGTGTAG GGCAGAGAACGGTGTGAAGAC TCTGGATGATGGGCAGAGC TCCTTCTCTTCCTTCACTTTGCA TTGCTGTAGGTCTGCTTGAGGAT

CTGF NDRG1 LamininA3 Rad a-Tubulin Adipophilin HPL

(PE Applied Biosystems, Foster City, CA), and are listed in Table 1. Analysis of transcript level was carried out using the sequence detection software supplied with the GeneAmp 5700. This software calculates the threshold cycle (Ct) for each reaction and this Ct is used to quantify the relative amount of starting template in the reaction. First, a difference in Ct values (DCt) is calculated by subtracting the mean Ct of the reference 18S from mean Ct of each transcript, examined in duplicates. Then, DDCt is calculated by subtracting the mean Ct of the calibrator from each value of DCt for each gene. The amount of target normalized to an endogenous reference and relative to a calibrator is computed by 2ÿDDCt [15]. Histological analysis Sections (5 mm) from each original tissue biopsy were stained with hematoxylin and eosin. The stained slides were then examined by an experienced placental pathologist (FTK) who was unaware of the sampling site within the placenta. The histology of each sample was graded based on the following parameters: (a) villous size (large, medium or small) (b) amount of fibrin deposits (sporadic, medium or increased) and (c) prevalence of syncytial knots (sparse, intermediate or common). The slides were then classified into grade IeIII, with grade I characterized by larger villi, sporadic fibrin deposits and sparse syncytial knots, whereas grade III characterized by small villi, increased amounts of fibrin deposits and common syncytial knots. Grade II was used for those samples that did not fulfill the criteria for either grade I or grade III. Statistics We utilized Spearman Rank correlation to analyze trends in our non-parametric data, computed using Analyze-It software (Analyze-It Software Ltd., Leeds, England). We also used Student’s t-test, where appropriate. Statistical significance was defined as p ! 0.05.

RESULTS Using a novel approach to analysis of microarray data we have previously identified a set of transcripts that exhibited marked up-regulation in hypoxic term villous trophoblasts [6,7]. These transcripts include VEGF, CTGF, NDRG1, lamininA3, Rad, a-tubulin, and adipophilin. Human placental lactogen (hPL), a marker of trophoblast differentiation, is down-regulated in hypoxia, and served as a control. We examined the expression of these transcripts in the different placental regions, depicted in Figure 1A and described in Material and methods. As shown in Figure 1B, we found that the expression of most hypoxia-related placental transcripts was significantly enhanced when samples from the lateral-chorionic sites (region 3C) were compared to samples from the medio-basal sites (region 1A). In contrast, the change in expression of NDRG1 and adipophilin was statistically insignificant. Similarly, the lower hPL expression in the chorionic plate compared to the basal plate was not statistically significant (Figure 1A). To buttress our findings we compared the fold-change in transcript expression at the lateral-chorionic site (region 3C) to the corresponding expression change of hPL at the same site. Using this analysis we found that the expression of VEGF (1.9-fold), CTGF (2.7-fold), NDRG1 (1.6-fold), lamininA3 (2.5-fold), Rad (2.2-fold) and a-tubulin (3.3-fold) was significantly elevated at the lateral-chorionic site, when compared to the medio-basal site (p ! 0.05, Student’s t-test). Together, these results demonstrate the non-uniform pattern of placental gene expression, with enhanced expression of several transcripts that are known to be up-regulated by hypoxia near the lateral placental margin at the chorionic plate compared to the central site. As previously noted, histological changes characteristic of hypoperfused villi are more common at the placental lateralsubchorionic region. We therefore surmised that the expression level of hypoxia-related transcripts might correlate with tissue histological changes. Using the grading scale defined in

Wyatt et al.: Correlation between Sampling Site and Gene Expression

375

Figure 1. The correlation between gene expression level and location of placental sampling site. (A) Location of sampling sites within the placental disk. (B) The level of transcript expression according to sample location (n Z 6). Each value is mean G SD of fold expression, normalized to transcript expression level at site 1A (defined as 1). Significance was determined by analysis of transcript level at positions 1A, 2B and 3C using Spearman Rank correlation test. * denotes p ! 0.05.

Material and methods we assigned a histological grade to each villous sample, irrespective of sampling site. Figure 2A demonstrates representative histology for grade designation I, II and III. As expected, histological grade I was most prevalent at site 1A, which represents the medio-basal region. Characteristic histological changes for grade IA included large villi, sporadic fibrin deposits and sparse syncytial knots. In contrast, histological grade III was the most prevalent at site

Figure 2. The correlation between villous histology and sampling site. (A) The three panels represent villous histology that was graded as I, II and III based on villous size, fibrin deposits and syncytial knots, as defined in Material and methods. (Bar Z 100 mM). (B) A stacking bar graph, representing the frequency of histological grades by selected placental sampling sites (1A, 2B and 3C, as defined above, and see Figure 1A). Each site was sampled in six different placentas. The differences in prevalence of the histological grades were significant (p ! 0.01, Spearman Rank correlation test).

376

3C, which represents the lateral-chorionic region (Figure 2B). Histological features at this site included small villi, prevalent fibrin deposits and frequent syncytial knots. Grade II was characterized by medium sized villi as well as intermediate numbers of syncytial knots and fibrin deposits. We next examined the correlation between gene expression level and villous histological grade, and found that the expression level of most hypoxia-related transcripts was up-regulated in grade III specimens, compared to grade I specimens (Figure 3). The corresponding results for VEGF and adipophilin were not statistically significant. The expression of hPL was unchanged

Figure 3. Gene expression level is associated with villous histology. The level of transcript expression according to histology, determined using six independent placentas. Each value is mean G SD of fold expression, normalized to transcript expression level for grade I (defined as 1). Significance was determined using Spearman Rank correlation test. * denotes p ! 0.05.

Placenta (2005), Vol. 26

among the three histological grades. Taken together, these data support the notion that the expression of hypoxia-related transcripts correlates with placental regions relative to mediobasal region and with histological grade. The placental region we defined as ‘‘medial’’ is located at the center of the placental disk, and near the centrally inserted cord. To determine if proximity to the cord insertion site influences gene expression we repeated our analysis for a selected group of hypoxia-related transcripts using samples derived from placentas with marginal cord insertion, where the umbilical cord was inserted on the chorionic plate within 1 cm from the placenta margin. We postulated that if proximity to the cord was a key determinant of gene expression level, samples obtained near the marginally inserted cord would have lower expression than those obtained near the placental center or from the margin opposite from the cord insertion. As shown in Figure 4, we found no correlation between expression level of several hypoxia-related transcripts and sampling site in these placentas. Importantly, unlike our data using placentas with centrally inserted cord, in placentas with marginal cord insertion the expression of our set of hypoxia-related transcripts in the lateral region (opposite from the cord insertion site) was not increased. These data may suggest that cord insertion site exerts some influence, at least in part, on placental transcript expression.

Figure 4. The level of transcript expression according to sample location in placentas with marginal umbilical cord insertion (n Z 5). All medial samples were obtained from the basal region, and all lateral samples were obtained from the subchorionic region. Each value is mean G SD of fold expression, normalized to transcript expression level at the medial site (defined as 1). All differences were statistically insignificant.

Wyatt et al.: Correlation between Sampling Site and Gene Expression

DISCUSSION In this work we correlated the sampling site with the level of gene expression in term human placentas, obtained from women following an uncomplicated pregnancy, labor and delivery. Our results indicate that the expression of hypoxiarelated transcripts in placental villous biopsies is dependent on the sampling site within the placental disk, with a lower expression level near the placental center at the basal plate and in proximity to the cord insertion, and a higher expression level at the placental periphery at the chorionic plate. The changes in gene expression were relatively modest (up to three-fold), reflecting the fact that all assays were performed using placentas from term uncomplicated pregnancies. We have also confirmed that histological changes characteristic of underperfused placenta [10], including villous maturation, fibrin deposition and syncytial knots [16], are also more commonly observed at the periphery of the placental disk near the chorionic plate. Furthermore, the expression of hypoxiarelated transcripts in placental villous biopsies independently correlated with the histological changes. Although our study was not designed to prove a cause and effect, this information suggests that diminished blood flow at the placental margin, particularly at the subchorionic region, predisposes the histological changes characteristic of hypoperfusion as well as the up-regulation of hypoxia-related transcripts. Our data are also consistent with previously published work on altered cotyledonal perfusion and oxygenation, linking regional oxygen gradients with altered gene expression [17] or with histological changes that reflect hypoxia at the lateral-subchorial sites within the entire placenta or within each cotyledon [8,9,18e20]. Regional perfusion differences might also be influenced by placental implantation site, a variable that was not addressed in our study. Lastly, we chose our sites based on an unbiased systematic sampling of different placental regions. Previous studies utilized other approaches to spatial analysis of placental histomorphology or function (for example, see Refs. [21e24]), and it is likely that different methodologies might influence the extent of gene expression changes. Transcript expression at the medio-basal region might be influenced by the location at the center of the placenta or by the proximity to the umbilical cord insertion. Marginal insertion of the umbilical cord may be considered a normal variant [10]. Analysis of these placentas revealed no correlation between expression level of hypoxia-related transcripts and sampling site. The fact that transcript expression was not relatively lower near the marginally inserted cord compared to other placental regions suggests that cord insertion site is not the only determinant of transcript expression. Interestingly, unlike our observation for placentas with centrally inserted cord, the expression level of hypoxia-related transcripts at the marginal site opposite from the marginally inserted cord was not enhanced, implying at least a limited influence of cord insertion on regional perfusion and gene expression. In this study we focused on expression of genes that we previously identified by their up-regulation in hypoxic

377

cultured trophoblasts and in placentas from pregnancies complicated by hypoperfusion and intrauterine growth restriction. The two growth factors, VEGF and CTGF, are regulated by hypoxia and play a role in embryonic development, cell differentiation and angiogenesis [25e29]. VEGF has been previously shown to be up-regulated in hypoxic trophoblasts [30,31]. CTGF is a member if the heparin-binding connective tissue growth factor/cysteine-rich 61/nephroblastoma over-expressed (CCN) family [32]. While the expression of CTGF transcript in the placenta has been documented, the function of CTGF in the placenta remains unknown. The signaling proteins NDRG1 and Rad have been associated with cell differentiation [33,34], and therefore are implicated in adaptive cellular response to hypoxia. NDRG1 is up-regulated in hypoxic immortalized first trimester human trophoblasts (HTR-8/Svneo) as well as human MDA-MB231 breast cancer line [35,36]. Rad is a member of the Ras guanosine triphosphatase family of proteins, which has been implicated in cellular growth, vesicle transport, signal transduction and glucose transporter function [34]. LamininA3 and a-tubulin are both cytoskeleton related proteins. LamininA3 is the alpha-3 chain of the glycoprotein laminin 5 (epiligrin, kalinin), and is important in wound healing and maintenance of skin dermal epidermal junction [37]. a-Tubulin is expressed at a lower level than normal in cytotrophoblast and syncytiotrophoblasts from fetuses with trisomy 21 [38]. The lipid droplet-associated adipophilin plays a role in uptake and turnover of cellular lipids [39]. We have recently shown that adipophilin is expressed in term human placental trophoblasts, and that its expression is increased during trophoblast differentiation and by the ligand-activated PPARg/RXR complex [40]. Unlike these hypoxia-related transcripts, we and others have previously demonstrated that hPL, a member of the hGH/hPL gene family that regulates maternal and fetal metabolism as well as fetal growth [41e44], was downregulated by hypoxia in trophoblasts [7,45]. Thus, hPL served as a negative control in our studies. It should be pointed out that our analysis focused on transcripts that were selected based on marked expression change in hypoxic trophoblasts [7]. The level of hypoxia inducible factor-1a (HIF-1a) mRNA did not exhibit a significant change in our transcriptome analysis [7]. This likely represents the fact that although hypoxia induces HIF-1a mRNA in some cells, the level of HIF-1a protein is primarily regulated post-translationally, with additional regulation at the level of cellular localization binding to promoter targets and gene activation function [46e49]. At the present time it is unclear whether changes in gene expression in response to hypoxia contribute to placental dysfunction, or may represent an adaptive response, designed to attenuate injury. Nevertheless, our findings suggest that perfusion and tissue architecture within the placental sampling site influence transcript expression. Our results also underscore the need to stratify the analysis of placental gene expression by sampling multiple sites within the chorioallantoic placental disk.

378

Placenta (2005), Vol. 26

ACKNOWLEDGMENTS We thank Elena Sadovsky and Lori Rideout for technical support. We also thank Dr. Michael Province from the Division of Biostatistics at Washington University for reviewing our analysis. This research was supported by NIH grant R01-HD29190 (to DMN) and R01-ES11597 (to YS).

REFERENCES [1] Wigglesworth JS. Perinatal pathology. In: Livolsi VA, editor. Major problems in pathology, vol. 15. London: W.B. Saunders Company Ltd; 1996. [2] Mayhew TM. Villous trophoblast of human placenta: a coherent view of its turnover, repair and contributions to villous development and maturation. Histol Histopathol 2001;16:1213e24. [3] Aronow BJ, Richardson BD, Handwerger S. Microarray analysis of trophoblast differentiation: gene expression reprogramming in key gene function categories. Physiol Genomics 2001;6:105e16. [4] Tanaka TS, Kunath T, Kimber WL, Jaradat SA, Stagg CA, Usuda M, et al. Gene expression profiling of embryo-derived stem cells reveals candidate genes associated with pluripotency and lineage specificity. Genome Res 2002;12:1921e8. [5] Hoang VM, Foulk R, Clauser K, Burlingame A, Gibson BW, Fisher SJ. Functional proteomics: examining the effects of hypoxia on the cytotrophoblast protein repertoire. Biochemistry 2001;40:4077e86. [6] Budhraja V, Spitznagel E, Schaiff WT, Sadovsky Y. Incorporation of gene-specific variability improves expression analysis using high-density DNA microarrays. BMC Biol 2003;1:1. [7] Roh CR, Budhraja V, Kim HS, Nelson DM, Sadovsky Y. Microarraybased identification of differentially expressed genes in hypoxic term human trophoblasts and in placental villi of pregnancies with growth restricted fetuses. Placenta; in press, doi:10.1016/j.placenta.2004.06.013. [8] Matijevic R, Meekins JW, Walkinshaw SA, Neilson JP, McFadyen IR. Spiral artery blood flow in the central and peripheral areas of the placental bed in the second trimester. Obstet Gynecol 1995;86:289e92. [9] Ramsey EM, Chez RA, Doppman JL. Radioangiographic measurement of the internal diameters of the uteroplacental arteries in rhesus monkeys. Am J Obstet Gynecol 1979;135:247e51. [10] Fox H. Pathology of the placenta. In: Livolsi VA, editor. Major Problems in Pathology, vol. 7. London: W.B. Saunders Company, Ltd; 1997. [11] Benirschke K, Kaufmann P. Pathology of the Human Placenta. New York: Springer; 2000. [12] Konje JC, Huppertz B, Bell SC, Taylor DJ, Kaufmann P. 3-Dimensional colour power angiography for staging human placental development. Lancet 2003;362:1199e201. [13] Boyd PA, Brown RA, Stewart WJ. Quantitative structural differences within the normal term human placenta: a pilot study. Placenta 1980;1: 337e44. [14] Fox H. The histopathology of placental insufficiency. J Clin Pathol Suppl (R Coll Pathol) 1976:1e8. [15] Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(ÿDDC(T)) method. Methods 2001; 25:402e8. [16] Fox H. The pattern of villous variability in the normal placenta. J Obstet Gynaecol Br Commonw 1964;71:749e58. [17] Hempstock J, Bao YP, Bar-Issac M, Segaren N, Watson AL, CharnockJones DS, et al. Intralobular differences in antioxidant enzyme expression and activity reflect the pattern of maternal arterial bloodflow within the human placenta. Placenta 2003;24:517e23. [18] Alvarez H, Benedetti WL, Morel RL, Scavarelli M. Trophoblast development gradient and its relationship to placental hemodynamics. Am J Obstet Gynecol 1970;106:416e20. [19] Schuhmann R, Stoz F, Maier M. Histometric studies of placentones of the human placenta. Z Geburtshilfe Perinatol 1986;190:196e203. [20] Sala MA, Matheus M, Valeri V. A stereologic study on the thickness of the trophoblastic layer and its regional variation in the human term placenta. Anat Anz 1984;157:257e60. [21] Kosanke G, Castellucci M, Kaufmann P, Mironov VA. Branching patterns of human placental villous trees: perspectives of topological analysis. Placenta 1993;14:591e604. [22] Mayhew TM. Patterns of villous and intervillous space growth in human placentas from normal and abnormal pregnancies. Eur J Obstet Gynecol Reprod Biol 1996;68:75e82.

[23] Mayhew TM, Bowles C, Yucel F. Hypobaric hypoxia and villous trophoblast: evidence that human pregnancy at high altitude (3600 m) perturbs epithelial turnover and coagulation-fibrinolysis in the intervillous space. Placenta 2002;23:154e62. [24] Rajakumar A, Whitelock KA, Weissfeld LA, Daftary AR, Markovic N, Conrad KP. Selective overexpression of the hypoxia-inducible transcription factor, HIF-2a, in placentas from women with preeclampsia. Biol Reprod 2001;64:499e506. [25] Ferrara N, Henzel WJ. Pituitary follicular cells secrete a novel heparinbinding growth factor specific for vascular endothelial cells. Biochem Biophys Res Commun 1989;161:851e8. [26] Ferrara N, Houck K, Jakeman L, Leung DW. Molecular and biological properties of the vascular endothelial growth factor family of proteins. Endocr Rev 1992;13:18e32. [27] Lau LF, Lam SC. The CCN family of angiogenic regulators: the integrin connection. Exp Cell Res 1999;248:44e57. [28] Shimo T, Kubota S, Kondo S, Nakanishi T, Sasaki A, Mese H, et al. Connective tissue growth factor as a major angiogenic agent that is induced by hypoxia in a human breast cancer cell line. Cancer Lett 2001; 174:57e64. [29] Wulff C, Weigand M, Kreienberg R, Fraser HM. Angiogenesis during primate placentation in health and disease. Reproduction 2003;126: 569e77. [30] Taylor CM, Stevens H, Anthony FW, Wheeler T. Influence of hypoxia on vascular endothelial growth factor and chorionic gonadotrophin production in the trophoblast-derived cell lines: JEG, JAr and BeWo. Placenta 1997;18:451e8. [31] Shore VH, Wang TH, Wang CL, Torry RJ, Caudle MR, Torry DS. Vascular endothelial growth factor, placenta growth factor and their receptors in isolated human trophoblast. Placenta 1997;18:657e65. [32] Brigstock DR. The connective tissue growth factor/cysteine-rich 61/ nephroblastoma overexpressed (CCN) family. Endocr Rev 1999;20: 189e206. [33] Xu B, Lin L, Rote NS. Identification of a stress-induced protein during human trophoblast differentiation by differential display analysis. Biol Reprod 1999;61:681e6. [34] Reynet C, Kahn CR. Rad: a member of the Ras family overexpressed in muscle of type II diabetic humans. Science 1993;262:1441e4. [35] Kokame K, Kato H, Miyata T. Nonradioactive differential display cloning of genes induced by homocysteine in vascular endothelial cells. Methods 1998;16:434e43. [36] Park H, Adams MA, Lachat P, Bosman F, Pang SC, Graham CH. Hypoxia induces the expression of a 43-kDa protein (PROXY-1) in normal and malignant cells. Biochem Biophys Res Commun 2000;276:321e8. [37] Mizushima H, Miyagi Y, Kikkawa Y, Yamanaka N, Yasumitsu H, Misugi K, et al. Differential expression of laminin-5/ladsin subunits in human tissues and cancer cell lines and their induction by tumor promoter and growth factors. J Biochem (Tokyo) 1996;120:1196e202. [38] Massin N, Frendo JL, Guibourdenche J, Luton D, Giovangrandi Y, Muller F, et al. Defect of syncytiotrophoblast formation and human chorionic gonadotropin expression in Down’s syndrome. Placenta 2001; 22(Suppl A):S93e7. [39] Heid HW, Moll R, Schwetlick I, Rackwitz HR, Keenan TW. Adipophilin is a specific marker of lipid accumulation in diverse cell types and diseases. Cell Tissue Res 1998;294:309e21. [40] Bildirici I, Roh CR, Schaiff WT, Lewkowski BM, Nelson DM, Sadovsky Y. The lipid droplet-associated protein adipophilin is expressed in human trophoblasts and is regulated by peroxisomal proliferatoractivated receptor-g/retinoid X receptor. J Clin Endocrinol Metab 2003; 88:6056e62. [41] Walker WH, Fitzpatrick SL, Barrera-Saldana HA, Resendez-Perez D, Saunders GF. The human placental lactogen genes: structure, function, evolution and transcriptional regulation. Endocr Rev 1991;12:316e28. [42] Hoshina M, Hussa R, Pattillo R, Camel HM, Boime I. The role of trophoblast differentiation in the control of the hCG and hPL genes. Adv Exp Med Biol 1984;176:299e312.

Wyatt et al.: Correlation between Sampling Site and Gene Expression [43] Hoshina M, Hussa R, Pattillo R, Boime I. Cytological distribution of chorionic gonadotropin subunit and placental lactogen messenger RNA in neoplasms derived from human placenta. J Cell Biol 1983;97:1200e6. [44] Hoshina M, Boothby M, Hussa R, Pattillo R, Camel HM, Boime I. Linkage of human chorionic gonadotrophin and placental lactogen biosynthesis to trophoblast differentiation and tumorigenesis. Placenta 1985;6:163e72. [45] Alsat E, Wyplosz P, Malassine A, Guibourdenche J, Porquet D, Nessmann C, et al. Hypoxia impairs cell fusion and differentiation process in human cytotrophoblast, in vitro. J Cell Physiol 1996;168:346e53.

379 [46] Wenger RH. Cellular adaptation to hypoxia: O2-sensing protein hydroxylases, hypoxia-inducible transcription factors, and O2-regulated gene expression. Faseb J 2002;16:1151e62. [47] Bruick RK. Oxygen sensing in the hypoxic response pathway: regulation of the hypoxia-inducible transcription factor. Genes Dev 2003;17: 2614e23. [48] Kaelin Jr WG. How oxygen makes its presence felt. Genes Dev 2002;16: 1441e5. [49] Schofield CJ, Ratcliffe PJ. Oxygen sensing by HIF hydroxylases. Nat Rev Mol Cell Biol 2004;5:343e54.