Connective tissue growth factor (CTGF) is a recently described heparin-binding mitogen for fibroblasts and smooth muscle cells. The aim of this study was to investigate the production of CTGF by human uterine tissues using immunohistochemical and Northern blotting analyses. For immunohistochemistry, formalin-fixed human proliferative (n = 5), early secretory (n = 5; days 15–19), mid-secretory (n = 5; days 20–23), late secretory (n = 5; days 24–28) endometrial, and decidual (n = 5) tissues were stained using a highly specific affinity-purified polyclonal antibody raised against residues 81–94 of human CTGF. Myometrial (n = 5) and leiomyoma (n = 5) tissues were also used for CTGF immunochemistry. In proliferative endometrium, epithelial and vascular endothelial cells showed strong CTGF immunoreactivity, whereas stromal cells were negative or only weakly positive for the CTGF protein. Throughout the entire secretory stage, CTGF was detected in epithelial and endothelial cells of endometrium. Stromal cells showed strong immunoreactivity to CTGF only in oedematous areas for early and mid-secretory endometrium, and in decidualized regions of late secretory endometrium. During pregnancy, the decidual, epithelial and endothelial cells of the endometrium were all immunoreactive to CTGF. In myometrial and leiomyoma samples, CTGF immunoreactivity was found only in the endothelial cells. Northern blotting of mRNA from normal uterus (n = 2) or leiomyoma (n = 6) using a 320 bp human CTGF cDNA probe revealed a single 2.4 kb transcript. This study is the first to demonstrate CTGF gene expression and localization of its encoded protein in human uterine tissues. The cell- and cycle-specific localization of CTGF support a role for this molecule in regulating aspects of uterine cell growth, migration, and/or matrix production during the menstrual cycle and pregnancy.
Human connective tissue growth factor (hCTGF) was first described in 1991 as a 38 kDa secretory product of human vascular endothelial cells maintained in vitro (Bradham et al., 1991). CTGF expression is up-regulated by serum, several growth factors including transforming growth factor-β (TGF-β) and TGF-α/epidermal growth factor (EGF), and dexamethasone (Brunner et al., 1991; Ryseck et al., 1991; Igarashi et al., 1993; Grotendorst et al., 1996; Kothapalli et al., 1997; Dammeier et al., 1998; Wenger et al., 1999). Since its discovery, numerous studies have demonstrated high levels of CTGF expression in pathological states, e.g. scleroderma (Igarashi et al, 1995, 1996), desmoplasia (Frazier and Grotendorst, 1997; Wenger et al., 1999), atherosclerosis (Oemar et al., 1997; Hishikawa et al., 1999), benign mesenchymal tissue growths (Igarashi et al., 1998), pulmonary fibrosis (Allen et al., 1999) and renal fibrosis (Ito et al., 1998). CTGF is also produced as a response to mechanical tissue injury (Igarashi et al., 1993; Pawar et al., 1995; Dammeier et al., 1998), suggesting that this growth factor is involved in tissue repair and regeneration. CTGF-responsive cell types include fibroblasts (Bradham et al., 1991; Igarashi et al., 1993; Brigstock et al., 1997; Kireeva et al., 1997; Ball et al., 1998; Steffen et al., 1998), vascular smooth muscle cells (Brigstock et al., 1997), endothelial cells (Kireeva et al., 1997; Shimo et al., 1998) chondrocytes (Nakanishi et al., 2000) and epithelial cells (Kireeva et al., 1997). Biological effects elicited by CTGF include stimulation of mitosis, cell proliferation, extracellular matrix (ECM) production, cell adhesion, fibroplasia, and augmentation of the activity of platelet-derived growth factor (PDGF) or basic fibroblast growth factor (bFGF) (Bradham et al., 1991; Igarashi et al., 1993; Frazier et al., 1996; Brigstock et al., 1997; Kireeva et al., 1997; Ball et al., 1998; Steffen et al., 1998). Involvement of CTGF in apoptosis has also been reported (Hishikawa et al., 1999).
Recent observations have demonstrated production of CTGF by the developing embryo and uterus of pregnant and non-pregnant mammals, highlighting a potential role for this factor in embryogenesis, uterine function, and pregnancy (Brigstock et al., 1997; Kireeva et al., 1997; Ball et al., 1998; Harding et al., 1998; Surveyor et al., 1998; Surveyor and Brigstock, 1999). Synthesis of porcine CTGF (pCTGF) by uterine explants has been documented (Harding et al., 1998), endometrial expression of pCTGF mRNA has been confirmed (Brigstock et al., 1997; Ball et al., 1998; Harding et al., 1998), and various low mass (10–20 kDa) pCTGF isoforms have been detected in uterine luminal fluids throughout the oestrous cycle and early pregnancy (Brigstock et al., 1997; Ball et al., 1998). In pregnant pigs, CTGF concentrations in uterine fluids peak around the time of blastocyst elongation and oestrogen production (Ball et al., 1998; Brigstock, 1999a), the latter of which is the molecular cue for pregnancy recognition in this species (Flint et al., 1979). In the mouse uterus, CTGF is produced mainly by glandular and luminal epithelial cells during the oestrous cycle and early pregnancy (Surveyor et al., 1998). This pattern of CTGF localization persists until the day of implantation, at which time epithelial localization of CTGF is strongly diminished (Surveyor et al., 1998). Thereafter, CTGF localization is pronounced in the decidua and in the developing embryo. The objective of this study was to demonstrate production of CTGF by human uterine tissues and to establish the immunohistochemical localization of the CTGF protein as a function of reproductive status. In addition, since CTGF is associated with abnormal tissue growth, we also examine the presence and distribution of CTGF in human uterine leiomyoma.
Materials and methods
For immunostaining, proliferative (n = 5), early (n = 5; days 15–19), mid- (n = 5; days 20–23) and late secretory (n = 5; days 24–28) endometrial samples were obtained from patients undergoing diagnostic laparotomy or laparoscopy who had: (i) a normal uterus with no adhesions; (ii) a history of regular menstrual cycles; (iii) not received steroids or intrauterine contraceptive devices for at least 6 months prior to collection of biopsies; and (iv) an age range of 26–43 years. Decidual tissues (n = 5) were obtained from abortion cases of patients who were 28–44 years old. Myometrial samples were taken from patients (33–40 years old) who were undergoing hysterectomy for medical reasons, other than endometrial cancer or leiomyoma. Leiomyoma samples (n = 5) were from patients (36–41 years old) with a preoperative diagnosis of uterine fibroids. Pathological examination of the uterus confirmed the presence of a leiomyoma.
For Northern blotting, a Northern Territory Real™ Human Uterus Tumor Blot (Invitrogen Corporation, Carlsbad, CA, USA) that comprised 2 μg per lane of mRNA from individual donors of normal human uterus (n = 1; 36 years old) or leiomyomata (n = 3; 36, 43, and 51 years old) was used. In addition, a Northern Territory Real™ Human Tumor Panel Blot II (Invitrogen) that comprised 2 μg per lane of individual mRNA samples from donors of normal uterus (n = 1; 45 years old) or breast (n = 1; 50 years old), or of pooled mRNA samples from donors of leiomyoma (n = 3; 39, 41 and 49 years old) or moderately well-differentiated invasive ductal breast carcinoma (n = 4; 42, 49, 60 and 69 years old) was used.
Formalin-fixed paraffin-embedded tissues were cut into 4 μm sections, mounted on aminoalkylsilane-treated slides, deparaffinized with xylene and rehydrated through graded alcohols into distilled water. The sections were microwaved in 10 mmol/l citrate buffer (pH 6) for 3 min on the `high' setting until the solution began to boil and then for 15 min on the `medium–low' setting. Slides were then washed in distilled water and endogenous peroxidase activity was quenched using a 3% hydrogen peroxide solution over 15 min. Slides were then incubated for 10 min in PBS containing 0.3% BSA and then overnight at room temperature with 1 μg/ml CTGF antibody. This antibody was raised in rabbits against residues 81–94 of hCTGF, a domain which is non-conserved among other CTGF-related molecules. The antibody was affinity-purified and has been previously validated for immunohistochemistry (Steffen et al., 1998; Surveyor et al., 1998). Negative controls were obtained by substituting the primary antibody with 1 μg/ml normal rabbit immunoglobulin G (IgG). Immunoreactivity was visualized using a commercial streptavidin–biotin horseradish peroxidase (HRP) system (LSAB+ Kit; Dako Corporation, Carpinteria, CA, USA) in which slides were successively incubated with biotinylated anti-rabbit IgG, streptavidin–HRP, and 3-3′ diaminobenzidine (DAB). Sections were counterstained with Harris' haemotoxylin.
Myometrial and leiomyoma samples were also immunostained for CD34, which is a positive marker for endothelial cells. Following incubation with DAB, the samples were washed in distilled water and then Tris buffer. The samples were incubated with a CD34 antibody (1:200; Novocastra, Newcastle upon Tyne, UK) at room temperature overnight. Immunoreactivity to CD34 was visualized using a commercial streptavidin–biotin–alkaline phosphatase system (LSAB+ Kit; Dako Corp.) in which slides were successively incubated with biotinylated link antibody, streptavidin–alkaline phosphatase, and New Fuchsin substrate system. Sections were then counterstained with Harris' haemotoxylin.
Each Northern blot was probed with a previously described 320 bp hCTGF cDNA probe (Steffen et al., 1998). The probe was labelled to a specific activity of >108 cpm/μg with [32P]-ATP using a RadPrime kit (Life Technologies, Grand Island, NY, USA). Membranes were incubated in 6 ml prehybridization solution [6× sodium chloride/sodium citrate (SSC), 2× Denhardt's, 1 μg/ml denatured salmon sperm DNA, 50% deionized formamide, 2% sodium dodecyl sulphate (SDS)] at 42°C for 2 h. The probe was boiled, quick-chilled, and added to 6 ml of fresh prehybridization solution, then incubated with the membrane at 42°C for 18 h. The membranes were washed three times in 10 ml 2× SSC, 0.05% SDS for 10–15 min at 42°C followed by two washes in 20 ml of 0.1× SSC, 0.1% SDS at 50°C. Blots were exposed to X-ray film for 4–6 h at –70°C. Each blot was stripped using 5× SSC for 1–2 min at room temperature followed by three washes in 20 ml of boiling 0.1% SDS solution. They were then incubated with a [32P]-labelled 724 bp β-actin probe (Invitrogen) essentially as described above. Blots were exposed to X-ray film for up to 30 h at room temperature.
Histopathological examination of the endometrial sections demonstrated no malignancy or granulomatous inflammation such as tuberculosis. In proliferative endometrium, CTGF was immunolocalized to luminal (not shown) and glandular epithelial and vascular endothelial cells while stromal cells showed no or weak immunoreactivity (Figure 1A). In early- and mid-secretory endometrium, CTGF immunoreactivity was observed in glandular epithelial, luminal epithelial and vascular endothelial cells. Stromal cells also showed CTGF immunoreactivity in oedematous areas (Figure 1B,C). Since both early and mid-secretory stage endometrium showed very similar pattern for CTGF immunolocalization, a representative sample from early-secretory stage is shown (Figure 1B,C). In late secretory endometrium, glandular epithelial, luminal epithelial and vascular endothelial cells showed strong immunoreactivity (Figure 1D,E). CTGF immunoreactivity was also observed in decidualized stromal cells around arterioles (Figure 1E). In endometrium of pregnancy, decidual, epithelial and vascular endothelial cells demonstrated CTGF immunoreactivity whereas polymorphonuclear cells did not (Figure 1F,G). Tissue sections stained with normal rabbit IgG were negative at all stages, for which a typical example is shown in Figure 1L. In both normal myometrium (Figure 1H,I) and leiomyoma (Figure 1J,K), immunoreactivity to CTGF and CD34 were co-localized to vascular endothelial cells which accounted for the majority of the staining seen in these tissues (Figure 1H–K).
Northern blotting demonstrated a single 2.4 kb CTGF transcript in two (out of two) samples of normal uterus, two (out of three) individual leiomyomata, one (out of one) leiomyomata pooled from three patients, one (out of one) normal breast, and one (out of one) breast cancer pooled from four patients (Figure 2). This transcript size for hCTGF is consistent with previous reports (Bradham et al., 1991; Ryseck et al., 1991; Steffen et al., 1998) as well as uterine pCTGF (Ball et al., 1998; Harding et al., 1998). When probed for β-actin, a single 2 kb transcript was present in all samples and was of uniform intensity (data not shown), confirming that the mRNA was intact and that equivalent amounts had been loaded in all lanes.
CTGF belongs to the expanding `CCN' family which also includes cyr61, nov, elm1 and HICP (O'Brien et al., 1990; Joliot et al., 1992; Delmolino et al., 1997; Hashimoto et al., 1998). Although CCN family members share a significant structural similarity, especially conserved cysteine residues (Bork, 1993), the biological properties of these molecules are extremely varied (Brigstock, 1999b). CTGF and cyr61 stimulate cell proliferation, chemotaxis, adhesion, and production of ECM components (Bradham et al., 1991; Frazier et al., 1996; Kireeva et al., 1996, 1997; Brigstock et al., 1997; Steffen et al., 1998). Effects of cyr61 and fisp-12/mCTGF on cell adhesion and migration have been attributed to its binding to integrin αVβ3 or integrin αIIbβ3,which represent the only molecularly-defined cell surface receptors for any family member to date (Kireeva et al., 1998; Babic et al., 1999; Jedsadayanmata et al., 1999). It has been proposed that most, if not all, of the activities of CCN proteins may be attributed to their interaction with integrins (Lau and Lam, 1999).
Although members of the CCN family have been implicated in processes related to tissue pathology, e.g. wound healing, fibrosis and tumour growth, some family members may also play roles in physiological processes. For example, strong changes in levels of CTGF mRNA occur during follicular and luteal growth, and ovulation which involve processes such as angiogenesis and tissue repair (Wandji et al., 2000). Recent evidence has suggested that CTGF and cyr61 play roles in processes such as implantation, placentation, embryogenesis, differentiation and development (Brigstock, 1999b). In addition, the production of CTGF by uterine tissues of the pig and mouse has been documented, as have clear spatio–temporal changes in its levels and localization during both the oestrous cycle and early pregnancy (Ball et al., 1998; Surveyor et al., 1998). These data suggest that CTGF is physiologically important in both the cycling and pregnant uterus. To extend these studies, we performed CTGF immunohistochemistry on human endometrium and found that epithelial and endothelial cells were strongly immunoreactive for CTGF at all phases examined. The observed epithelial immunoreactivity in this study is consistent with the localization of CTGF in luminal and glandular epithelial cells in the mouse uterus (Surveyor et al., 1998) and in a variety of other epithelial structures, especially those found in secretory tissues (Kireeva et al., 1997). Endothelial immunoreactivity for CTGF is consistent with its production by cultured endothelial cells (Bradham et al., 1991; Shimo et al., 1998) as well as its immunohistochemical detection in endothelia of the cardiovascular system of day 14–18 mouse embryos (Kireeva et al., 1997). Although to understand its exact function in uterine biology warrants additional studies, CTGF that is ubiquitously produced by epithelial and endothelial cells may function as a mitogenic stimulus for uterine growth and angiogenesis.
Our observed lack of immunoreactivity of human endometrial stromal cells during the proliferative phase is similar to that reported throughout the oestrous cycle and early pregnancy of the mouse (Surveyor et al., 1998). Although CTGF immunoreactivity was observed in stromal cells during the secretory phase, the strongest immunoreactivity was seen in areas of focal oedema during the early and mid-secretory stage, and in areas of localized decidualization during late secretory stage. Strong and consistent immunoreactivity for CTGF was also seen in the decidua. Since the tissues were from abortion cases, this finding may not be used to draw a definitive conclusion for the normal decidua. However, since immunoreactivity was also observed in decidualized stromal cells of the late secretory stage, it seems likely that CTGF is produced as a prelude to or consequence of the decidualization process. A very similar conclusion has been drawn from observations in the mouse (Surveyor et al., 1998). To understand whether CTGF plays any role in decidualization and exact nature of this role, if any, requires further investigaton.
The role of CTGF in tumour growth remains unclear. Levels of CTGF gene expression have been studied in mammary gland tumours (this study; Frazier and Grotendorst, 1997), sarcoma cells (Steffen et al., 1998), chondrosarcoma cells (Nakanishi et al., 1997), and various tumours of the nervous and vascular systems (Xin et al., 1996; Igarashi et al., 1998). Only one of these studies (Frazier and Grotendorst et al., 1997) has produced credible data supporting a possible functional relationship of CTGF in tumour development in that CTGF expression was correlated positively with the incidence of desmoplasia in mammary gland carcinomas. We found that leiomyoma was positive for CTGF expression in two out of three individual mRNA samples and one (out of one) pooled mRNA samples from three patients. However, CTGF immunoreactivity was restricted to vascular endothelial cells in both myometrium and leiomyoma samples. Therefore, this finding does not support the presence of a specific role for CTGF in leiomyoma. Nevertheless, recent evidence for a role of CTGF in endothelial cell function (Lin et al., 1998; Babic et al., 1999; Boes et al., 1999; Shimo et al., 1999), suggests that this molecule could act through an angiogenic pathway to stimulate tumour growth and development.
In conclusion, this study has shown that the CTGF gene is transcribed in human uterine tissues and that its encoded protein is localized within the endometrium, myometrium and decidua in a cell- and stage-specific manner. These findings are consistent with a potential role for CTGF in the regulation of a variety of normal cellular processes in the uterus including growth, locomotion, and ECM production.
We thank Hamit Ayberk for his expert technical help with the immunohistochemical staining. DRB was supported by HD 30334 from the National Institutes of Health.