Preterm birth is a multifactorial disease, and its prevention involves understanding the causes of uterine contractions, cervical dilatation, and preterm premature rupture of the fetal membranes (PROM).1,2 It is often difficult to study these in isolation because contractions and preterm PROM usually occur concurrently.
This laboratory has focused previously on the hormone relaxin and its involvement in preterm PROM. It is produced and acts within the fetal membranes to cause accelerated collagen degradation by matrix metalloproteinases.3–5 The classic cytokines, produced as a consequence of infection, are also capable of stimulating production and activation of the matrix metalloproteinases, and we have been able to distinguish the relaxin-mediated pathway from an infection-stimulated pathway to preterm PROM.6 We identified the need to define the cellular events in common and unique to these pathways that result in preterm PROM and pre-term delivery.
A number of new molecular techniques for identifying differentially expressed genes could provide important new insights and broaden our approach to the study of the complex problem of preterm PROM. These new techniques include differential display,7 representational difference analysis,8 suppression subtractive hybridization,9 and serial analysis of gene expression.10 This laboratory extensively evaluated the suppression subtractive hybridization method in a homogeneous cell line11 before using it on a complex tissue such as the fetal membranes. The use of this method in the present study is an attempt to determine important changes in gene expression and to identify genes not previously known to be related to preterm PROM or preterm labor.
Materials and Methods
The tissues were from patients who agreed to participate in this study between March 1996 and December 1997 at Kapiolani Medical Center for Women and Children, Honolulu, Hawaii. Fetal membranes including amnion, chorion, and decidua were cut from the placenta, frozen in liquid nitrogen, and stored at −80 C until use. All patients delivered vaginally except for those having cesarean delivery. The University Committee on Human Experimentation and the Hospital Institutional Review Board approved the protocols of this study.
The control tissue for the suppression subtractive hybridization experiment was from a 21-year-old woman at 28 weeks and 5 days' gestation who underwent emergency cesarean delivery after presenting with decreased fetal movement and a fetal biophysical profile of 2. The experimental tissues for this experiment were from three women who presented with preterm PROM and delivered vaginally at 25,33, and 35 weeks' gestation. The 25-week tissue was subsequently shown to have histologic evidence of chorioamnionitis. The suppression subtractive hybridization patients were part of a larger group of patients, which included those who had preterm cesarean deliveries and those who had preterm PROM, labor, and vaginal deliveries. The demographic data from the 13 uninfected patients are shown in Table 1. Quantitative expression of the candidate genes was determined on Northern blots of these 13 patients with preterm cesarean delivery and PROM.
Interleukin-8 and regulatory-signal G-protein gene expression were quantitated in additional membranes obtained from four patients with preterm PROM and from five patients with preterm labor and intact membranes throughout the latent and early active phases of labor. These tissues were collected after vaginal delivery and had been used previously as a part of a study on infection.6 None of the membranes had histologic evidence of chorioamnionitis. Seven of nine patients were exposed to antibiotic therapy during labor, and three of nine patients were exposed to magnesium sulfate. Regulatory-signal G-protein and interleukin-8 gene expression were also measured in term fetal membranes collected from women with uncomplicated pregnancies experiencing normal term labor (n = 5) or undergoing elective cesarean delivery for abnormal lie or previous cesarean (n = 8). These tissues had been used in a study of measurements of intrauterine surface area. None had chorioamnionitis, PROM, or exposure to Celestone (Schering Corp., Kenilworth, NJ) or magnesium sulfate. Two patients in each group had received antibiotics before delivery.
For the suppression subtractive hybridization experiment, 1.2 g of frozen tissue from each of the three patients with preterm PROM was pooled and processed for messenger RNA (mRNA) as one sample. In all other cases, each sample was from a single patient. Total and poly(A)+ RNA were isolated by the Chomczynski and Sacchi12 and oligo(dT)13 methods, respectively. The suppression subtractive hybridization method is designed to selectively amplify differentially expressed genes and followed the protocol for the PCR-Select cDNA Subtraction Kit (Clontech, Palo Alto, CA).9 Briefly, mRNAs from the preterm PROM membranes were designated as experimental, or the population whose upregulated genes were to be identified; and the mRNAs from a membrane of a preterm cesarean delivery were designated as control, or the population whose genes served as a reference for subtraction or control. An outline of the method is shown in Figure 1.
Complementary DNAs (cDNAs) from the experimental and control mRNAs were digested with Rsa I. The experimental cDNA was divided in half and each was ligated to a different cDNA adaptor. This was followed by a first hybridization in which excess control cDNA was added to each of the two adaptor-ligated experimental cDNAs. In this step, molecules corresponding to high- and low-abundance sequences became equalized, while other molecules became enriched for differentially expressed cDNAs. In the second hybridization, the two samples were mixed together and additional control cDNA was added to further enrich the differentially expressed sequence mixture. Two rounds of polymerase chain reaction (PCR) followed to amplify the differentially expressed sequences. To decrease the number of false-positive clones or cDNAs common to both experimental and control samples that had not been successfully subtracted, we used the PCR-Select Differential Screening Kit (Clontech).
The PCR products, which represented the differentially expressed sequences after suppression subtractive hybridization subtraction and two rounds of PCR amplification, were cloned and transformed into cells (TA Cloning Kit; Invitrogen, Carlsbad, CA). Random recombinant colonies were picked and transferred onto the master plates. Individual colonies were grown, and an aliquot of each sample then served as a template for PCR amplification using nested primers. The PCR products were dot-blotted onto two duplicate membranes. One was hybridized with forward-subtracted and the other with reverse-subtracted cDNA probes. The forward-subtracted probe was made from the same subtracted cDNA used to construct the original cDNA library. For the reverse-subtracted probe, another suppression subtractive hybridization was performed using mRNA from preterm cesarean as the experimental sample and mRNA from preterm PROM as control. Probes were labeled and hybridized to duplicate filters as described previously.6 When clones hybridized only with the forward-subtracted probe or when the intensity of the hybridization signal was at least fivefold greater in the forward-subtracted compared with the reverse-subtracted probe, the results were considered to represent truly differentially expressed mRNAs (Figure 2). Each positive clone was purified with the QIAGEN Plasmid Mini Kit (QIAGEN, Chatsworth, CA) and sequenced using the PRISM dyeDeoxy Terminator Cycle Sequencing kit (Applied Biosystems, Foster City, CA). The sequences were analyzed and sequence similarities were checked using the Basic Local Alignment Search Tool (BLAST) program at the National Center for Biotechnology Information (Bethesda, MD). DNAs of interest were digested out, separated on gel agarose, extracted (QIAquick Gel Extraction Kit; QIAGEN), labeled, and used for Northern analysis. The procedures for Northern analysis were published previously.6
Statistical analysis was performed using the two-tailed, nonparametric Mann-Whitney U test to compare groups. P < .05 was considered statistically significant. Linear regression analysis was performed to determine correlation coefficients.
Eight genes were identified by suppression subtractive hybridization when the mRNAs from a control fetal membrane from preterm cesarean delivery were subtracted from mRNAs obtained from three patients with preterm PROM. None of the three preterm PROM patients had any clinical evidence of infection, and all had gone through labor and delivery. The eight genes identified are shown in Table 2. Four genes (E–H) are probably independent of inflammation or infection. However, four genes (A–D) have products that are known to be involved in the response to inflammation or infection: complement factor-B, ferritin H-chain, interleukin-8, and the invariant gamma-chain of class II histocompatibility antigen. This raises the question of whether there was any infection present in the preterm PROM tissues that was not recognized clinically. Individual aliquots of these tissues had been stored at −80 C, and pieces were then fixed in Bouin's fixative, embedded in paraffin, and stained with hematoxylin.
Figure 3 shows the three tissues from the patients with preterm PROM (1–3), together with the control tissue (C). The first and second preterm PROM tissues were normal (Figure 3; no. 1 and 2). In the second tissue, the amnion had separated from the choriodecidua. However, the third preterm PROM tissue showed extensive infiltration of neutrophils through the extracellular matrix of the amnion and chorion (Figure 3; no. 3), strongly suggesting chorioamnionitis. The control tissue was normal (Figure 3C).
Northern analyses were performed to quantify each of the eight candidate genes in each of the tissues (1,2, 3, and C) used for the original suppression subtractive hybridization experiment. The expression of each gene was quantified as a ratio with the housekeeping gene, glyceraldehyde-3-phosphate dehydrogenase, in each sample. The results are shown in Figure 4. Four genes previously implicated in the process of inflammation or infection were clearly upregulated in preterm PROM tissue 3, which subsequently showed histologic infection (Figure 3). The genes for F-actin capping protein and chitinase precursor (E, F) also showed the highest expression in this tissue (Figure 4). Although these genes have not been associated previously with the response to inflammation or infection, they may be a part of this response, or upregulated as a consequence of the infection. Two genes (G, H), a regulatory G-protein signaling protein and osteopontin, were not expressed at the highest level in the tissue with chorioamnionitis (Figure 4). Indeed, the expression of the regulatory G-protein signaling gene (G) was highest in preterm PROM tissue number 1, whereas osteopontin (H) showed slightly higher expression in the control tissue (C) than in the three preterm PROM tissues (Figure 4). These results suggest that regulatory G-protein signaling is upregulated in the PROM tissue independent of infection and that osteopontin was a methodologic false positive.
Additional Northern analyses for the expression of the eight candidate genes were performed on a larger group of patients that included preterm PROM with labor (n = 7) and preterm cesarean deliveries (n = 6). These two groups also included the two patients with preterm PROM and one preterm cesarean patient used for the original suppression subtractive hybridization experiment. In the absence of infection, there was no significant increase in the expression of ferritin, complement factor-B, or histocompatibility antigen in the preterm PROM tissues, whereas interleukin-8 was significantly increased (P < .02) in preterm PROM (Table 3). F-actin capping protein and chitinase precursor, possibly involved in inflammation or infection (Table 3), showed no significant difference with PROM. The regulatory G-protein signaling gene showed a marginal increase in PROM. Osteopontin (H) was confirmed to be a methodologic false positive.
Because the demographic composition of the two groups (preterm cesarean and PROM) differed (Table 1), we attempted to assess for some confounding variables. This was difficult because of the small numbers of samples in each group. Two patients in each group were exposed to celestone and magnesium sulfate. There were no significant differences in regulatory G-protein signaling expression with or without celestone exposure (0.73 ± 0.44 and 0.38 ± 0.24, respectively) or with or without exposure to magnesium sulfate (0.69 ± 0.48 and 0.40 ± 0.24, respectively). Not surprisingly, there were more patients with preeclampsia in the preterm cesarean group compared with the PROM group (33% compared with 0, respectively), and also more of the neonates were small for gestational age (SGA) at birth in the cesarean group (50% compared with 0). There was no difference in regulatory G-protein signaling gene expression in patients with preeclampsia compared with normotensive women (0.38 ± 0.13 and 0.51 ± 0.36, respectively). Regulatory G-protein signaling expression in the SGA group was similar to that of the group with normal growth (0.27 ± 0.03 and 0.56 ± 0.36, respectively).
There was no significant difference in interleukin-8 expression in tissues with or without exposure to celestone (1.07 ± 1.10 and 1.12 ± 1.78, respectively) or to magnesium sulfate (1.05 ± 1.12 and 1.13 ± 1.77, respectively). There was no difference in interleukin-8 gene expression in patients with preeclampsia compared with normotensive women (0.14 ± 0.02 and 1.28 ± 1.64, respectively). Interleukin-8 expression in the SGA group was similar to that of the group experiencing normal growth (0.23 ± 0.26 and 1.31 ± 1.70, respectively).
As would be expected, patients with preterm PROM were exposed to more frequent and longer antibiotic therapy than patients with preterm cesarean delivery. Additionally, all patients in the preterm PROM group labored, but only one in the preterm cesarean group had uterine contractions before surgery. Therefore, it was not possible from these results to determine whether interleukin-8 and regulatory G-protein signaling gene expression were upregulated as a result of PROM, labor, or antibiotic therapy. Therefore, two further experiments were conducted. In the first experiment, patients delivering vaginally preterm were divided into two groups, with and without preterm PROM, to determine whether interleukin-8 and regulatory G-protein signaling genes were upregulated by preterm PROM. The second experiment compared term patients with and without labor to show any effect of labor on interleukin-8 and regulatory G-protein signaling gene expression.
Preterm patients with either PROM (n = 4) or preterm labor with intact fetal membranes throughout the latent phase (n = 5) were studied to determine whether there was any effect of rupture on the expression of interleukin-8 and regulatory G-protein signaling genes. All of these patients labored and delivered vaginally and none had histologic evidence of infection, although eight of nine were exposed to antibiotics. In patients with preterm PROM or preterm labor, there were no significant differences in interleukin-8 expression (0.03 ± 0.06 and 0.70 ± 0.86, respectively) or in the regulatory G-protein signaling protein, RGS2 (0.19 ± 0.16 and 0.26 ± 0.33, respectively). These results suggest that these two genes are upregulated by labor, independent of PROM. This was confirmed by quantitating their expression in women undergoing elective repeat cesarean delivery at term (n = 8) and women delivering vaginally after normal term pregnancy (n = 5). Two women in each group were exposed to antibiotic therapy, and none had PROM or chorioamnionitis. Interleukin-8 gene expression was significantly increased in the tissues after labor compared with nonlabor tissues (4.15 ± 1.77 and 0.16 ± 0.15, respectively; P < .003). Expression of the regulatory G-protein signaling protein RGS2 was also significantly increased in the tissues after labor compared with nonlabor tissues (6.91 ± 4.70 and 2.21 ± 1.54, respectively; P < .02).
When the expression of each of the eight candidate genes in the patients with preterm PROM (n = 7) was analyzed against the length of latency in these patients, only complement factor-B showed a positive correlation (r = .89) (Figure 5). There was no correlation between the expression of any of the genes and gestational age (data not shown).
This study shows that suppression subtractive hybridization can be used successfully on a complex tissue such as fetal membranes to study changes in gene expression in specific clinical situations. We have thus related the upregulation of six genes, four as expected and two novel, to chorioamnionitis. We have also shown significant upregulation of a novel gene, a regulatory G-protein signaling protein, with labor, while confirming that labor also significantly increases expression of interleukin-8. Finally, we have shown that complement factor-B expression is related to the duration of membrane rupture in preterm PROM.
The results described demonstrate the need to couple this method with both rigorous histologic assessment of the tissues and validation of candidate genes by quantitative Northern analysis. Further analyses are also mandated to assess patient variability in the expression of each candidate gene. This is important because of the many uncontrolled variables for each tissue collected before term, which could result in misleading data. The use of a pool of three preterm PROM tissues for our original experiment was an attempt to reduce such patient variability. However, one of the preterm PROM tissues had chorioamnionitis, detectable only on histologic examination. We were able to directly relate the expression of four candidate genes—complement factor-B, ferritin H-chain, interleukin-8, and histocompatibility antigen—quantitatively by Northern analysis to their expression in this single infected tissue in comparison with each of the noninfected tissues. This indeed showed the unique upregulation of each of these four genes with infection. These genes are all involved in the response to infection or inflammation. Complement factor-B has been shown to be synthesized by human amnion and is implicated in the local host-defense response.14 Ferritin production is increased in infiltrating macrophages at the amniochorionic-decidual interface after bacterial colonization,15 and elevated serum ferritin levels are predictive of early spontaneous preterm delivery.15 Interleukin-8 is a monocyte-derived neutrophil chemotactic factor that is stimulated by the inflammatory reaction.16 Its identification and upregulation in the tissue exhibiting chorioamnionitis are in keeping with its use as a marker of infection and predictor of preterm delivery.17 The histocompatibility antigen is also a macrophage product needed for immune-cell interaction,18 and its upregulation in the infected tissue could also be expected.
The F-actin capping protein and chitinase precursor genes upregulated in the infected fetal membrane have not been associated previously with inflammation or infection. However, the F-actin capping protein identified is involved in the reorganization of actin filaments within the cell.19 Experimental evidence suggests that the actin cytoskeleton is actively involved in hormone secretion. For example, the actin cytoskeleton of the lactotroph appears to integrate the multiple factors regulating prolactin secretion.20 Disruption of the actin filaments by depolymerizing agents such as fungal or bacterial toxins enhances the stimulated secretion of regulatory substances by a number of different cell types.20 It is therefore likely that the F-actin capping protein gene, identified and shown to be upregulated in the infected fetal membrane, is indeed activated in one or more cell types present in the tissue in response to the infection. The chitinase precursor (YKL-39) is also a product of macrophages and may be associated with the late stages of monocyte to macrophage maturation.21 Little is known about the functions of this protein, and although it was first described in the media of chondrocyte cells, it does not possess any enzymatic activity against chitinase substrates.22
We have shown that the expression of complement factor-B in the fetal membranes is directly related to the duration of membrane rupture in women with preterm PROM. Complement factor-B expression has been localized to the amnion and shown to be the source of complement proteins in amniotic fluid.14 Five of the seven patients with preterm PROM (71%) had received antibiotic therapy, and none of the tissues after delivery showed any histologic evidence of infection. These data suggest that complement factor-B expression is indeed an early immunologic response identified only at the subclinical level. It has been shown previously that both maternal and cord blood levels of complement activity are related to the duration of membrane rupture.23 However, in a prospective follow-up study, complement levels failed to correlate with either the duration of the latent period or maternal or neonatal infection.24 Because our study measured gene expression directly, rather than a decrease in complement as it is used up in the host-defense mechanism, we have successfully shown a relation between its expression and the duration of membrane rupture. Therefore, an increase in complement factor-B expression is likely to be a direct measure of an early immune response mechanism once the membranes are ruptured.
Another interesting finding was the identification of the regulatory G-protein signaling gene and its upregulation in term labor. This gene is a member of a newly identified family of regulators of G-protein signaling, which all contain an evolutionarily conserved RGS domain.25 These proteins act as inhibitors of G-protein signaling within the cell, and recent studies have shown that they drive guanosine triphosphate-bound G proteins back to the inactive guanosine diphosphate-bound form.26 This gene was expressed equally in the membranes of women with vaginal preterm delivery with and without preterm PROM, but was upregulated at term with labor. Expression of the regulatory G-protein signaling gene is therefore unrelated to preterm PROM but is upregulated in response to uterine contractions or some other component of the labor process. Further work is in progress to clarify this. The concurrent upregulation of interleukin-8 by labor at term has been well described27,28 and was confirmed in the tissues used here. Finally, although osteopontin, an extracellular-matrix glycosylated glycoprotein, is expressed by human cytotrophoblast,29 it was shown not to be upregulated in either the tissues used for the original experiment or the tissues used to study patient variability in expression and was therefore considered a methodologic false positive.
We have shown that the method of suppression subtractive hybridization, if used critically with rigorous follow-up experiments, can provide important leads to better understand the pathologic processes of preterm birth. The further development of such leads has potential clinical relevance in the future application of novel markers for diagnostic purposes and in our basic understanding of the mechanisms by which these pathologies develop.
1. The proceedings of an international conference on preterm birth: Etiology, mechanisms and prevention. Charleston, North Carolina, October 1997. Prenat Neonat Med 1998;3:1–185.
2. Parry S, Strauss JF III. Premature rupture of the fetal membranes. N Engl J Med 1998;338:663–70.
3. Koay ESC, Bryant-Greenwood GD, Yamamoto SY, Greenwood FC. The human fetal membranes: A target tissue for relaxin. J Clin Endocrinol Metab 1986;62:513–21.
4. Qin X, Garibay-Tupas J, Chua PK, Cachola L, Bryant-Greenwood GD. An autocrine/paracrine role of human decidual relaxin. I. Interstitial collagenase (matrix metalloproteinase-1) and tissue plasminogen activator. Biol Reprod 1997;56:800–11.
5. Qin X, Chua PK, Ohira RH, Bryant-Greenwood GD. An autocrine/paracrine role of human decidual relaxin. II. Stromely-sin (MMP-3) and tissue inhibitor of matrix metalloproteinase-1 (TIMP-1). Biol Reprod 1997;56:812–20.
6. Millar LK, Boesche MH, Yamamoto SY, Killeen J, DeBuque L, Chen R, et al. A relaxin-mediated pathway to the preterm premature rupture of the fetal membranes, independent of infection. Am J Obstet Gynecol 1998;179:126–34.
7. Liang P, Pardee AB. Differential display of eukaryotic messenger RNA by means of the polymerase chain reaction. Science 1992;257:967–71.
8. Hubank M, Schatz DG. Identifying differences in mRNA expression by representational difference analysis of cDNA. Nucleic Acids Res 1994;22:5640–8.
9. Diatchenko L, Lau YF, Campbell AP, Chenchik A, Moqadam F, Huang B, et al. Suppression subtractive hybridization: A method for generating differentially regulated or tissue-specific cDNA probes and libraries. Proc Natl Acad Sci USA 1996;93:6025–30.
10. Velculescu VE, Zhang L, Vogelstein B, Kinzler KW. Serial analysis of gene expression. Science 1995;270:484–7.
11. Nemeth E, Bole-Feysot C, Tashima LS. Suppression subtractive hybridization (SSH) identifies prolactin stimulation of p38 MAP kinase gene expression in Nb2 T lymphoma cells. J Mol Endocrinol 1998;20:151–6.
12. Chomczynski P, Sacchi N. The single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 1988;162:156–9.
13. Aviv H, Leder P. Purification of biologically active globin messenger RNA by chromatography on oligothymidylic acid-cellulose. Proc Natl Acad Sci USA 1972;69:1408–12.
14. Katz Y, Gur S, Aladjem M, Strunk RC. Synthesis of complement proteins in amnion. J Clin Endocrinol Metab 1995;80:2027–32.
15. Tamura T, Goldenberg RL, Johnston KE, Cliver SP, Hickey CA. Serum ferritin: A predictor of early spontaneous preterm delivery. Obstet Gynecol 1996;87:360–5.
16. Fortunate SJ, Menon R, Swan KF. Amniochorion: A source of interleukin-8. Am J Reprod Immunol 1995;34:156–62.
17. Foulon W, Van Liedekerke D, Demanet C, Decatte L, Dewaele M, Naessens A. Markers of infection and their relationship to preterm delivery. Am J Perinatol 1995;12:208–11.
18. Omu AE, Makhseed M, al-Qattan F. The comparative value of interleukin-4 in sera of women with preeclampsia and cord sera. Nutrition 1995;11:688–91.
19. Fowler VM. Capping actin filament growth: Tropomodulin in muscle and nonmuscle cells. Soc Gen Physiol Ser 1997;52:79–89.
20. Carbajal ME, Vitale ML. The cortical actin cytoskeleton of lactotropes as an intracellular target for the control of prolactin secretion. Endocrinology 1997;138:5374–84.
21. Rehli M, Krause SW, Andreesen R. Molecular characterization of the gene for human cartilage gp-39 (CHI3L1), a member of the chitinase protein family and marker for late stages of macrophage differentiation. Genomics 1997;43:221–5.
22. Hakala BE, White C, Recklies AD. Human cartilage gp-39, a major secretory product of articular chondrocytes and synovial cells, is a mammalian member of a chitinase protein family. J Biol Chem 1993;268:25803–10.
23. Levy DL, Arquembourg PC. Maternal and cord blood complement activity: Relationship to premature rupture of the membranes. Am J Obstet Gynecol 1981;139:38–40.
24. Levy DL, Cox A, Leffell MS, Wilds PL. Serum complement activity in pre-term pregnancies: Relationship to duration of ruptured membranes and clinical infection. Am J Reprod Immunol 1982;2:142–7.
25. Koelle MR, Horvitz HR. EGL-10 regulates G protein signaling in the C elegans
nervous system and shares a conserved domain with many mammalian proteins. Cell 1996;84:115–25.
26. Dohlman HG, Thorner J. RGS proteins and signaling by heterotrimeric G proteins. J Biol Chem 1997;272:3871–4.
27. Osmers RG, Blaser J, Kuhn W, Tschesche H. Interleukin-8 synthesis and the onset of labor. Obstet Gynecol 1995;86:223–9.
28. Elliott CL, Kelly RW, Critchley HO, Riley SC, Calder AA. Regulation of interleukin 8 production in the term human placenta during labor and by antigestagens. Am J Obstet Gynecol 1998;179:215–20.
29. Daiter E, Omigbodun A, Wang S, Walinsky D, Strauss JF III, Hoyer JR, et al. Cell differentiation and endogenous cyclic adenosine 3′,5′-monophosphate regulate osteopontin expression in human trophoblasts. Endocrinology 1996;137:1785–90.