Main

In a variety of animal species (1–3), and in the human (4–7) during late gestation, the fetal DA undergoes morphologic changes leading to the formation of "intimal cushions." These structures partially occlude the DA lumen and assure that the vessel closes completely when it constricts postnatally (8,9). We previously investigated the mechanism of intimal cushion formation in the fetal lamb, a process that takes place largely between 100 and 138 d of a 145-d gestation period and showed that cultured 100-d DA EC produce increased amounts of hyaluronan compared with 100-d Ao or 138-d DA EC (10).

The increased production of DA endothelial hyaluronan appears to be regulated by TGF-β in that neutralizing antibodies to TGF-β reduce hyaluronan synthesis to levels observed in the Ao (11). Thus, the increased production of endothelial hyaluronan in DA compared with Ao and its down-regulation between early and late gestation might also be accompanied by similar changes in the expression of TGF-β. Indeed, we have documented that there is increased intensity of immunostaining for TGF-β1 in 100-d DA compared with Ao tissues (11). However, the endothelial-specific expression of TGF-β1 had not been addressed. Increased expression of TGF-β1 in DA during its postnatal closure has also been reported by Tannenbaum et al. (12).

TGF-β is a homodimeric peptide with a molecular mass of 25 kD. TGF-β is synthesized and secreted by cells as a latent complex that requires activation to gain binding activity to its cell surface receptors. Of the three isoforms (TGF-β1, β2, and β3) that have been characterized in mammals, the expression of TGF-β1 has been shown to be tightly regulated during development in embryos (13,14) and in developing hearts (15). TGF-β1 is required for normal cardiogenesis (16) and vasculogenesis (17). Up-regulation of TGF-β1 expression is associated with vascular intimal thickening resulting from hypertension (18), atherosclerosis (19), and angioplasty (20,21). Both transcriptional (22–26) and posttranscriptional mechanisms are involved in the regulation of TGF-β1 (27–31).

In this study, we investigated, both in vivo and in vitro, whether there were changes in the production of endothelial TGF-β1 in the DA compared with the Ao in early and late gestation fetal lambs. We found, by immunohistochemical assessment, an increase in TGF-β in DA compared with Ao endothelium from 100-d gestation fetal lambs and a reduction to comparable levels of expression in tissue from 138-d gestation lambs. In keeping with these observations, we showed, by metabolic labeling and immunoprecipitation, that synthesis of TGF-β is increased significantly in cultured 100-d DA compared with Ao EC and decreased in 138-d DA EC to values similar to those in Ao EC. Western immunoblotting with a β1 isoform-specific antibody confirmed that these differences are related to TGF-β1. Northern blot analysis demonstrated that, despite a short half-life in 100-d DA compared with Ao cells, TGF-β1 mRNA levels are slightly increased in 100-d DA cells due to increased transcription of TGF-β1 mRNA, confirmed by nuclear run-on analysis. TGF-β1 mRNA levels were reduced in 138-d DA and Ao EC related to reduced mRNA transcription and reduced stability, respectively. Thus we have uncovered differences in the levels of regulation of DA endothelial TGF-β1 expression, which may determine its tissue-specific and developmental patterns of expression.

METHODS

Immunohistochemistry. The distribution and expression of TGF-β was evaluated by immunohistochemical staining of sections prepared from DA and Ao tissues from three different 100- and 138-d gestation fetal lambs. Tissues were fixed with 4% paraformaldehyde, and paraffin-embedded sections were dewaxed in xylene and rehydrated through standard graded ethanol solutions. Sections were stained with a neutralizing antibody against TGF-β (1:100 of a stock solution of 10 mg/mL; R&D Systems, Minneapolis, MN). This antibody is purified IgG and raised in the rabbit by injection of highly purified native porcine TGF-β1. The antibody is also cross-reactive with TGF-β2 with much less sensitivity. The sections were visualized using Vectastain ABC Kit (Vector Laboratories Inc., Burlingame, CA) with goat anti-rabbit antibody and hematoxylin blue counterstain. Control experiments were performed by replacing the primary antibody with normal rabbit IgG (Dako Corp., Carpinteria, CA). The stained tissue sections were assessed qualitatively, independently, and in a blinded fashion by the co-authors (B.Z., C.C., M.R.) and there was complete agreement.

Cell culture. Fetal Rambouillet lambs were delivered by cesarean section on d 100 or 138 of a 145-d timed gestation. Fetal lambs were maintained in a 100% nitrogen environment to prevent breathing and maintain patency of the DA. DA and Ao were removed en bloc as previously described (32), and the vessels were separated, opened, and rinsed in PBS containing 3% antibiotics/antimycotics (GIBCO, Burlington, ON). EC were harvested by scraping the luminal surface with a no. 11 scalpel blade (33) and maintained in Medium 199 (GIBCO) containing 20% heat-inactivated fetal bovine serum (GIBCO) and 1% antibiotics/antimycotics. EC were characterized by a contact-inhibited "cobblestone" morphology and positive staining for factor VIII (34), and used at passages 2 or 3. The animal experiments involved in this study have been approved by the Research Institute of The Hospital for Sick Children in Toronto, Ontario.

Immunoprecipitation. Subconfluent cultures of either DA or Ao EC were labeled with 100 µCi/mL [35S]cysteine and methionine (Amersham, Oakville, ON) for 24 h in medium containing 25% of the normal concentration of cysteine and methionine and 5% fetal bovine serum. Culture medium was removed in the presence of proteinase inhibitors (aprotinin, leupeptin, and pepstatin), added in a final concentration of 1 µg/mL of each. To activate TGF-β, aliquots of 1 mL of medium were transiently acidified by adding 25 µL of 5 M HCI for 2 h at room temperature, followed by neutralization with 35 µL of 0.7 M HEPES (pH 7.0) supplemented with 1.4 M NaOH (34). The samples were then precleared two rounds on ice for 30 min with 100 µg/mL normal rabbit IgG. The precleared supernatants were then incubated with the TGF-β-neutralizing antibodies (1:1000 of the stock solution, 10 mg/mL) at 4°C overnight, followed by addition of 100 µL of protein A-agarose (Sigma Chemical Co., St. Louis, MO) for 1 h. The samples were washed four times with immunoprecipitation buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, and 0.005% thimerosal) (35). The final pellets were resuspended in 50 µL of Laemmli (36) sample buffer, heated to 90°C for 10 min, and then resolved on a 12.5% SDS-PAGE under nonreducing conditions. The gel was treated with En3Hance (Dupont, Boston, MA) for 30 min, dried, and exposed to Kodak X-Omat AR-5 film for 1 wk at -70°C. Using the autoradiograph as a template, the prominent band at 25 kD, corresponding to TGF-β, was cut from the gel and counted by liquid scintillation spectrometry. To standardize values for cell number, DNA content was determined as the mean of duplicate cell suspension aliquots, using the fluorescent dye, bisbenzamide (Hoescht Reagent H3313, Calbiochem, San Diego, CA) as previously described (37). Results are expressed as counts/min/100 ng of DNA. Total protein was also assessed in these series of experiments by total trichloracetic acid-precipitated proteins found in the conditioned medium. Statistical analysis was performed using the computerized software program, SuperANOVA, with Duncan's new multiple range test to compare differences among the groups. Data reported in the figures are mean ± SD. p < 0.05 denotes a significant difference. The number of animals compared in each group is indicated in the corresponding figure legend.

Western Immunoblot Analysis. Flasks (T-75) of subconfluent DA or Ao EC were incubated in 10 mL of serum-free Medium 199 for 24 h. Conditioned medium was prepared by dialyzing against 1 M acetic acid for 72 h with two changes, followed by dialysis against water and lyophilization. Equal amounts of protein (50 µg) were solubilized by boiling in 40 µL of nonreducing Laemmli sample buffer for 5 min and resolved on a 12.5% polyacrylamide gel. After electrophoresis, proteins were transferred from the polyacrylamide gel to a nitrocellulose membrane. The membrane was blocked by 5% dried milk in PBS containing 0.5% Tween 20 for 1 h and incubated overnight at 4°C with a TGF-β1 isoform-specific chicken IgG (R&D) at a concentration of 100 ng/mL, followed by incubation for 1 h with horseradish peroxidase-conjugated rabbit anti-chicken IgG diluted at 1:3000 (Bio-Rad, Mississaga, ON). Blots were visualized by an enhance chemiluminescence kit (ECL, Amersham), scanned, and analyzed using National Institutes of Health Image program.

Northern blot analysis. Total RNA was prepared from subconfluent DA and Ao EC harvested from 100- and 138-d gestation lambs by the phenol-chloroform extraction method (38). Samples of 30 µg of total RNA were then separated on a 1% agarose gel containing formaldehyde and transferred to Hybond nylon membranes (Amersham) by capillary transfer for 16 h and fixed by exposure to short wave UV irradiation. A 0.9-kb HindIII and SmaI porcine TGF-β1 cDNA fragment excised from pTGFB-Ch119 (a gift from Dr. M. B. Sporn, National Institutes of Health, Bethesda, MD) and an 0.8-kb PstI and XbaI human GAPDH cDNA fragment from pHcGAP purchased from ATCC (Rockville, MD) were labeled with [32P]dCTP using a random primer kit (Amersham) and purified with NucTrap™ push column (Stratagene, La Jolla, CA). Prehybridization and hybridization were performed for 16 h at 42°C. Blots were washed twice at 60°C in 2 × SSC, 0.1% SDS for 1 h and once in 0.1 × SSC, 0.1% SDS for 30 min. Autoradiographs of RNA blot hybridizations were analyzed by relative densitometric scanning. Values were standardized to mRNA levels of GAPDH to correct for any differences in total RNA loaded onto gels.

To study the stability of TGF-β1 mRNA, DA and Ao EC were cultured to subconfluence in the 100-mm dishes in Medium 199 and 10% fetal bovine serum. Actinomycin D was then added to the cultures at a concentration of 2.5 µg/mL, and cells were incubated for various times from 4 to 12 h before extraction of total RNA. Northern blot analyses were then carried out on each sample. After correction for any differences in total RNA loading by standardizing to mRNA levels of GAPDH, TGF-β1 mRNA signals were graphically plotted as a function of time. Simple linear regression was used to determine mRNA half-life. The differences in TGF-β1 mRNA half-life between cultured DA and Ao cells isolated from both 100- and 138-d gestation lambs were assessed by two-factor analysis of covariance. The number of animals compared in each group is indicated in the corresponding figure legend.

Nuclear run-on analysis. Nuclei isolation, in vitro transcription and purification of radiolabeled total RNA, and hybridization were performed as previously described (39). Briefly, 200 µL of nuclei isolated from each culture were added to an equal volume of reaction buffer [10 mM Tris-HCl, pH 8.3, 5 mM MgCl2, 300 mM KCl, 0.5 mM ATT, CTP, GTP, and 250 µCi of [32P]UTP (800 µCi/mmol)], and incubated for 30 min at 30°C. Purified nascent transcripts (5 × 106 to 7 × 107 cpm/mL) were then used for hybridization. Target cDNA samples (500 ng) used in slot blot and hybridization included a 0.9-kb HindIII and SmaI TGF-β1 cDNA fragment, a 0.8-kb PstI and XbaI GAPDH cDNA fragment, and linearized pGEM4. After a 3-d hybridization at 42°C, blots were washed three times, 30 min each, with 1 × SSC and 0.1% SDS at 42°C and exposed to film with two intensive screens.

RESULTS

Increased expression of TGF-β in 100-d gestation DA tissue. Tissue sections prepared from DA and Ao tissue from three different 100- and 138-d gestation fetal lambs were evaluated by immunohistochemical staining with a rabbit anti-TGF-β neutralizing antibody, which mainly recognizes the TGF-β1 isoform, visualized using a Vectastain ABC Kit, and counterstained by hematoxylin blue. Control experiments were performed by replacing the primary antibody with normal rabbit IgG. The findings were consistent and reflected similar tissue-specific and developmentally regulated changes in all samples examined. As shown in Figure 1, positive staining for TGF-β was associated with EC in 100-d DA tissues (A), but not appreciated in the endothelial lining in tissues from the 100-d gestation Ao or 138-d gestation DA and Ao (B, C, and D, respectively). A negative control shows only background staining (E). Thus, there appears to be a tissue-specific increased expression of TGF-β in 100-d DA EC, which is developmentally regulated. In addition to its endothelial expression, TGF-β was also observed in the medial SMC in all the tissues examined (Fig. 2). However, SMC in 100-d DA tissues exhibited the highest intensity of staining (A) compared with 100-d Ao (B) or 138-d DA and Ao tissues (C and D, respectively). Control tissues using normal rabbit IgG showed no staining (E).

Figure 1
figure 1

Representative immunostaining of tissues from DA and Ao showing increased expression of endothelial TGF-β in 100-d gestation DA tissues. Positive staining for TGF-β was seen EC in 100-d DA tissues (A), but not in the endothelial lining in tissues from the 100-d Ao (B) or 138-d DA and Ao (C and D, respectively). A negative control shows only background staining (E). The comparison is representative of vessels from three different animals at each time point.

Full size image
Figure 2
figure 2

Immunostaining of tissues from DA and Ao showing increased TGF-β expression in medial SMC in 100-d gestation DA tissues (A) with less intense staining found in 100-d Ao (B) or 138-d DA and Ao SMC (C and D, respectively). Control tissues using normal rabbit IgG are negative for immunostaining (E). The comparison is representative of vessels from three different animals at each time point.

Full size image

Increased TGF-β1 synthesis in cultured 100-d gestation DA EC. Primary cultures of EC from both 100- and 138-d gestation lambs were established, and biosynthesis of TGF-β1 was assessed using the same antibody described above in the immunohistochemistry studies. Values were normalized for total DNA content and expressed as counts/min/100 ng of DNA. Figure 3 summarizes the immunoprecipitation studies. Figure 3A shows a representative autoradiograph of these studies, and Figure 3B presents quantitative analysis of data from all harvests. TGF-β was synthesized and secreted by both DA and Ao EC cultured from 100- and 138-d gestation fetal sheep. The majority of newly synthesized TGF-β was secreted in the culture medium, whereas small amounts remained cell-associated (data not shown). Thus, analysis of biosynthesis of TGF-β could be achieved by measuring TGF-β secreted in the culture medium. As shown in Figure 3, a significant 1-fold increase in newly synthesized TGF-β is secreted by 100-d DA EC compared with Ao EC (lane 1 versus 2, p < 0.05), whereas no significant increase in TGF-β production is found in 138-d DA EC compared with 138-d Ao EC (lane 3 versus 4). TGF-β synthesis thus decreases significantly in DA EC from 138-d compared with 100-d gestation lambs (lane 1 versus 3, p < 0.05), whereas comparable amounts of TGF-β were found in 100- and 138-d Ao EC. By assessing total trichloroacetic acid-precipitated proteins in the conditioned medium and normalizing for DNA content in the monolayer, we also established that changes in TGF-β synthesis did not reflect differences in total protein synthesis or all number (data not shown).

Figure 3
figure 3

Increased TGF-β synthesis in cultured 100-d gestation DA EC. (A) A representative autoradiograph of immunoprecipitation studies showing that TGF-β was synthesized and secreted by both DA and Ao EC cultured from 100- and 138-d gestation fetal lambs. The arrow indicates the 25-kD TGF-β protein. (B) Statistical analysis of immunoprecipitation studies indicates a significant 1-fold increase in newly synthesized TGF-β secreted by 100-d DA EC compared with Ao EC. Compared with 100-d DA EC, TGF-β synthesis also decreases in 138-d DA cells, but no differences were found between 100- and 138-d Ao EC. The bar reflects the mean ± SD; *p < 0.05; n = 7, for number of comparisons from different lambs at 100 d, and n = 5, for comparisons from different lambs at 138 d.

Full size image

To confirm that the results of the immunoprecipitation studies reflected synthesis of TGF-β1, we carried out western immunoblot analysis using a β1 isoform-specific antibody. A 25-kD immunoreactive band corresponding in molecular mass to TGF-β1 protein was observed (Fig. 4). Consistent with the immunoprecipitation data, there was a greater than 1-fold increase in TGF-β1 protein produced by 100-d DA compared with 100-d Ao EC. In 138-d DA, the TGF-β1 protein was reduced, as was the case with the 138-d Ao cells, although there appeared to be an increase in 138-d DA compared with 138-d Ao EC, which was not as striking as the difference at 100 d.

Figure 4
figure 4

A Western immunoblot analysis of TGF-β1 using a β1 isoform-specific antibody. Increased production and secretion of TGF-β1 in DA compared with Ao EC cultured from both 100- and 138-d lambs were found by Western blot analysis (A). There is a developmentally related decrease in both DA and Ao TGF-β1 at 138 d. The arrow indicates the 25-kD TGF-β1 protein. (B) Graphic presentation of the measurements of the blot using National Institutes of Health image program.

Full size image

Increased levels of TGF-β1 mRNA in cultured 100-d DA EC. We next measured the steady-state levels of TGF-β1 mRNA to determine whether there were changes corresponding to differences in protein synthesis. Using Northern blot analyses, expression of TGF-β1 mRNA was detected in both cultured DA and Ao EC from 100- and 138-d gestation fetal lambs, and this appears as a single mRNA of 2.4 kb (Fig. 5). In contrast to the protein synthesis, TGF-β1 mRNA levels in 100-d DA compared with Ao EC were slightly, but not significantly, increased. There is, however, a significant fall in TGF-β1 mRNA levels in 138-d DA EC (p < 0.05). In Ao EC steady-state mRNA levels were also decreased in cells from 138-d compared with 100-d cells (p < 0.05).

Figure 5
figure 5

Northern blot analysis of steady-state TGF-β1 mRNA levels. (A) A representative Northern blot analysis. Arrows indicate TGF-β1 mRNA or GAPDH mRNA. (B) Quantitative studies show that TGF-β1 mRNA levels in 100-d DA compared with Ao EC are not significantly increased, whereas there is a significant fall in TGF-β1 mRNA levels in both DA and Ao EC in a comparison of cells from 100- and 138-d gestation lambs. The bar represents the mean ± SD from assays from four different animals at each time point; *p < 0.05.

Full size image

Analysis of stability of TGF-β1 mRNA. We then addressed whether the differences in TGF-β1 mRNA state levels could be attributed to alterations in mRNA stability. In these studies, transcription was inhibited by actinomycin D, and the remaining synthesized TGF-β1 mRNA was chased for various times ranging from 4 to 12 h. Compared with 100-d Ao EC, the levels of TGF-β1 in 100-d DA EC decreased relatively more rapidly (Fig. 6A). The estimated mRNA half-life is approximately 9.5 h for 100-d DA EC and 20 h for 100 d Ao EC TGF-β1. TGF-β1 mRNA decay was comparable in late gestation DA and Ao EC, where the half-life of the mRNA was approximately 13 and 9 h, respectively (Fig. 6B). Thus, although the developmentally down-regulated TGF-β1 mRNA level in Ao cells is likely due to the relative decrease in mRNA stability; TGF-β1 developmental down-regulation in DA cells without a marked change in mRNA stability suggests that a decrease in transcription may occur. The comparable amount of steady-state TGF-β1 mRNA in 100-d DA versus 100-d Ao EC with reduced stability in DA cells also suggests an increased transcription of TGF-β1 mRNA in 100-d DA EC.

Figure 6
figure 6

Northern blot analysis of stability of TGF-β1 mRNA in 100-d DA and Ao EC (A) and in 138-d DA and Ao EC (B). mRNA decay curves were generated by plotting the remaining mRNA in each of three experiments in the 100-d groups and in each of two experiments in 138-d groups according to a related time point. After correcting for any differences in total RNA loading by standardizing to mRNA levels of GAPDH, TGF-β1 mRNA signals were plotted semilogarithmically as a function of time. Simple linear regression was used to determine the mRNA half-life. The difference in TGF-β1 mRNA half-life between DA and Ao cells is represented by the difference in slopes of the curves as assessed by two-factor analysis of covariance. A significant difference in TGF-β1 mRNA half-life was found between 100-d DA and Ao EC (p = 0.0248).

Full size image

Analysis of newly transcribed TGF-β1 mRNA. To confirm whether there are development-related and tissue-specific differences in TGF-β1 mRNA transcription, we carried out nuclear run-on analyses. As shown in Figure 7, and consistent with the increased steady-state mRNA level, the level of newly transcribed TGF-β1 mRNA in DA cells was higher in 100-d cultures than in 138-d cultures, whereas the transcription rate of TGF-β1 was also increased in 100-d DA cells compared with 100-d Ao cells. In contrast, the transcription rate was comparable in 100- versus 138-d Ao cells. Taken together, these data suggest an increased transcription and translation of a relatively unstable TGF-β1 mRNA in 100-d DA compared with Ao EC, and a reduction in TGF-β1 mRNA transcription in the 138-d DA cells.

Figure 7
figure 7

Nuclear run-on analysis of newly transcribed TGF-β1 mRNA. (A) Autoradiograph shows an increase in nascent TGF-β1 transcript in 100-d DA EC compared with 100-d Ao cells, and a drop in nascent TGF-β1 transcript in DA 138-d cells, whereas no difference related to Ao cells was observed. (B) Graphic presentation of the measurements of the blot using National Institutes of Health Image program.

Full size image

DISCUSSION

Although the underlying mechanisms regulating intimal cushion formation and the signals that initiate this process are presently unknown, our previous studies demonstrated that TGF-β1 was responsible for the initial extracellular matrix changes, specifically the subendothelial deposition of hyaluronan (11). TGF-β1 is known to be the predominant TGF-β isoform up-regulated in remodeling vessels (18–21) and the most potent activator of extracellular matrix production (40–44). Infusion of TGF-β1 or direct transfer of the TGF-β1 gene into the iliofemoral artery will induce intimal hyperplasia, predominantly due to extracellular matrix production (41,42). Conversely, antibodies against TGF-β1 suppress intimal hyperplasia in acutely injured vessels, and this is associated with reduced extracellular matrix production (45).

In previous studies, we demonstrated by immunohistochemical staining that increased amounts of TGF-β1 are associated with 100-d gestation DA compared with Ao tissues (11). We also showed that the increase in DA endothelial hyaluronan is dependent on an active form of TGF-β1, because the addition of neutralizing antibodies to TGF-β in 100-d DA EC cultures decreased synthesis of hyaluronan to the level of that seen in Ao EC (11). We have extended those observations in this study, first by comparing the presence of TGF-β1 in the tissue at 100- and 138-d gestation time points and then by comparing the expression of TGF-β1 in DA and Ao cultured EC as indicated by protein synthesis and mRNA levels.

Our data demonstrate an increased synthesis of TGF-β1 in DA EC when compared with that in Ao EC at 100-d gestation. In addition, there is a developmentally associated down-regulation of TGF-β1 synthesis in DA cells between the 100- and 138-d gestation time points. Thus, the peak in TGF-β1 expression in the 100-d DA is associated with the onset of neointimal formation. Because there are similar expression levels of the different isoforms of TGF-β in the closing lamb DA (12), we studied the specific regulation of TGF-β1 isoform expression by assessing steady-state levels of TGF-β1 mRNA. There were, however, no significant differences in mRNA levels comparing DA with Ao EC at the both gestation time points but a developmental decrease in TGF-β1 mRNA level in both DA and Ao cells. mRNA stability studies demonstrated relatively rapid degradation of TGF-β1 mRNA in 100-d DA EC compared with Ao cells. Further analysis of transcription rate by nuclear run-on analysis indicated that there was an increase in 100-d DA compared with Ao cells and 138-d DA cells. Thus, the increase in TGF-β1 protein in 100-d DA compared with Ao cells is regulated by increased transcription and translation of a relatively unstable mRNA. This finding is compatible with previous observations that the reduced stability of growth factor or cytokine mRNAs are coupled with their increased translational efficiency (46–48). A reduced rate of transcription accounts for the decrease in TGF-β1 mRNA and protein in 138-d DA cells, whereas in Ao EC the developmental down-regulation in TGF-β1 mRNA level is due mainly to reduced mRNA stability.

Immunohistochemistry studies also revealed abundant TGF-β1 in the SMC of the DA but, in our previous studies, we were unable to correlate this observation definitively with a functional property in cell cultures specifically related to production of matrix (11). Comparisons also focused on DA and Ao, but, because our previous studies also showed a similar increase in hyaluronan when comparing DA with pulmonary artery cells (10), we might expect that there was a similar increase with respect to TGF-β1 expression.

Previous studies investigating the molecular mechanism regulating TGF-β1 expression have shown that the human TGF-β1 promoter contains two transcription start sites and several binding domains for known transcription factors (22,23). For example, phorbol ester-responsive elements have been identified, in both the upstream and downstream domains of the gene, and seem to be important in regulating TGF-β1 expression at the transcriptional level (23,49). These observations suggest that the initiating signal for stimulation of synthesis of TGF-β1 may occur via the activation of protein kinase C. Moreover, the transcription of TGF-β1 has been shown to be stimulated by the product of the retinoblastoma gene through specific retinoblastoma response elements (25).

There is also a growing body of evidence indicating that expression of TGF-β1 may be regulated posttranscriptionally (30,34,50,51). For example, nerve growth factor induces TGF-β1 expression, partly due to an increase in its mRNA half-life from 6 to 30 h in PC12 cells (50). The immunosuppressant cyclosporine has also been shown to prolong the half-life of TGF-β1 mRNA in activated human T lymphocytes (51). Although the mechanisms of posttranscriptional control of TGF-β1 expression are not well defined, several studies have demonstrated that the GC-rich sequence in the 5′-UTR may play a role in controlling TGF-β1 production from mRNA (27). Computer analysis showed that this region of the 5′-UTR contained a stable secondary stem-loop structure between sequences +49 to +76. This stem-loop region alone is sufficient to inhibit expression of the growth hormone gene, suggesting that it may contain a cis element that plays an important role in the posttranscriptional regulation of TGF-β1 gene expression. Moreover, Scotto and Assoian (29) have been able to show that a GC-rich domain in the 3′-UTR of TGF-β1 mRNA can have a bifunctional effect on overall protein expression; it may decrease the steady-state TGF-β1 mRNA while increasing its protein production by improving the efficiency of translation. The relative contributions of these inhibitory and stimulatory bifunctional elements to TGF-β1 protein production may result from temporal or cell-specific expression of certain trans-acting factors (29).

Whether or not these proposed mechanisms are responsible for the site-specific and developmental regulation of TGF-β1 expression in the DA and Ao EC is unknown. We recently identified an RNA-binding protein that binds to an AU-rich element in the 3′-UTR of fibronectin mRNA and appears to play a role in increasing the translational efficiency of fibronectin mRNA in DA SMC (52). Immunohistochemical analysis showed that this RNA-binding protein is also up-regulated in DA versus Ao EC (our unpublished data). Because there is an AU-rich-like element in the 3′-UTR of TGF-β1 mRNA, it would be interesting to investigate whether this protein or other trans-acting factors existing in DA EC could play a functional role in controlling expression of TGF-β1.