INTRODUCTION

Insulin-like growth factor-I (IGF-I)¹ regulates a variety of actions in many somatic tissues (1). In skin, IGF-I induces keratinocyte replication and skin matrix protein synthesis (2). However, keratinocytes do not express detectable amounts of IGF-I and may rely on the circulation, or perhaps expression by other local skin cells, for a supply of this factor (3, 4). In contrast, keratinocytes express another growth regulator, parathyroid hormone-related protein (PTHrp), which shares amino-terminal sequence homology with PTH and acts at least in part through conventional PTH receptors (5, 6). PTHrp was first purified from squamous cell tumors where it was implicated in the paraneoplastic syndrome, humoral hypercalcemia of malignancy. It is a highly conserved and ubiquitous protein with many effects in a broad range of tissues (reviewed in Refs. 6 and 7).

Our earlier studies first demonstrated that IGF-I expression is enhanced by PTH, PTHrp, prostaglandin E₂, and other inducers of cyclic adenosine monophosphate (cAMP) in osteoblasts (8-10). Despite the assumption that skin, and dermis in particular, is not a primary target organ for PTH, receptors shared by PTH and PTHrp are clearly present on dermal fibroblasts (11-13) in which fibronectin synthesis increases after exposure to PTHrp (14). Therefore, we postulated that keratinocyte-derived PTHrp might act as a local regulator in skin in a manner similar to that by circulating PTH in the skeleton. In this study we asked specifically if PTHrp could increase IGF-I production by fetal dermal fibroblasts through a cAMP-dependent pathway. In this way, keratinocytes might support dermal fibroblast function through the action of PTHrp, and as a result dermal fibroblasts could support keratinocyte activity through an increase in IGF-I expression.

EXPERIMENTAL PROCEDURES

Fibroblasts were isolated from the forehead skin of 22-day-old Sprague-Dawley rat fetuses (Charles River Breeding Laboratories) using procedures approved by the Yale University Animal Care and Use Committee. As in earlier studies (15), skin flaps were incubated for 30 min with collagenase (600 µg/ml preincubated with tosylchloromethyl ketone to inhibit residual clostripain activity (Worthington Biochemical Co.)). Cells were collected by centrifugation and plated in Dulbecco's modified culture medium supplemented with 20 mM Hepes buffer (pH 7.2), 100 µg/ml ascorbic acid, penicillin, and streptomycin (Life Technologies, Inc.) for 3 days. Stock cultures were dispersed with trypsin for 1-2 passages, and plated in 9.6-cm² cultures where virtually all cells displayed typical fibroblast morphology. Cultures were serum-deprived for 24 h before treatment.

PTHrp (1-34) was obtained from Bachem Bioscience, Inc. Isobutylmethylxanthine, phorbol 12-myristate 13-acetate (PMA), forskolin, isoproterenol, prostaglandin E₂, 5,6-dichloro-1--ribofuranosylbenzimidazole (DRB) were obtained from Sigma. Radioisotopes were from NEN Life Science Products.

Serum-deprived cultures were incubated for 5 min with 0.5 mM isobutylmethylxanthine to inhibit endogenous phosphodiesterase activity and then treated with vehicle or test agent in medium supplemented with the inhibitor. Medium was aspirated and cell layers were extracted with 90% n-propanol. Extracts were dried and resolubilized in 0.05 N sodium acetate, and cAMP levels were measured by radioimmunoassay (RIA) with a commercial kit (Biomedical Technologies, Inc., Stoughton, MA) using data from the linear portion of the standard curve (5, 9).

Serum-deprived cultures were incubated with vehicle or test agent, and conditioned medium was collected and extracted with 2.5 M acetic acid and 95% ethanol to release IGF-I from IGF-binding proteins, as described previously (8, 9). Relative IGF-I levels were measured by RIA with a commercial kit (Nichols Institute) using data from the linear portion of a curve generated with human recombinant IGF-I as standard.

Total RNA was extracted with acid guanidine-monothiocyanate (16), precipitated with isopropanol, and heat denatured in 2.2 M formaldehyde, 12.5 M formamide at 60 °C. Equal amounts of RNA (20 µg/lane) were fractionated on 1.5% agarose, 2.2 M formaldehyde gels and blotted onto charged modified nylon (NEN Life Science Products), and loading was visualized by ethidium staining. Membranes were hybridized with 3 × 10⁷ cpm of rat IGF-I cDNA probe that was gel purified and radiolabeled with [-³²P]dCTP and [-³²P]dTTP by the random hexanucleotide-primed second strand synthesis method (17). Filters were washed at 55 °C in 0.2 × SSC, 0.2% SDS, and bound material was visualized by autoradiography (8, 9). Alternately, RNase protection assays to examine IGF-I transcripts by differential promoter utilization were performed as described earlier (18). Briefly, total RNA (10 µg) was combined with [-³²P]UTP-labeled antisense RNA probes that can distinguish the utilization of either promoter 1 or promoter 2 of the rat IGF-I gene. Protected fragments were separated on 6% polyacrylamide, 8.3 M urea gels, and visualized by autoradiography.

Cultures were pulsed with 5 µCi/ml [methyl-³H]thymidine (80 Ci/mmol) for the last 2 h of culturing and lysed in 0.1 M SDS, 0.1 N sodium hydroxide. The insoluble material formed by precipitation with 10% trichloroacetic acid was collected and assessed by scintillation counting (15).

Cultures were pulsed with 5 µCi/ml [2,3-³H]proline (2.5 Ci/mmol) for the last 2 h of culturing. Cell layers were lysed by freeze-thawing and extracted in 0.5% Triton X-100 (Sigma). Samples were precipitated with 10% trichloroacetic acid and chilled, and the insoluble material was collected by centrifugation. Precipitates were acetone-extracted, dried, redissolved in 0.5 N acetic acid, and neutralized with sodium hydroxide. [³H]Proline incorporation into collagen (collagenase-digestible protein) and noncollagen protein (all other proteins) was measured using bacterial collagenase free of nonspecific protease activity (15, 19).

Cultures at 50-60% confluence were rinsed in serum-free medium and transfected with plasmids containing portions of rat IGF-I promoter 1 fused upstream of the reporter gene luciferase (total DNA 1.5 µg/culture) using Lipofectin (Life Technologies, Inc.) for 3 h. The solutions were then replaced with medium containing 5% serum and incubated for 48 h. Cultures were treated with vehicle or test agents in serum-free medium, rinsed, and then extracted with cell lysis buffer (Promega). In some experiments cells were co-transfected with pSV--galactosidase control vector (Promega) at 1.0 µg/culture to normalize for transfection efficiency, but -galactosidase activity never varied by more than 6% (S.D.) within an experiment. Commercial kits were used to measure luciferase (Promega) and -galactosidase (Tropix) and corrected for protein content by the Bradford method (20, 21).

Data were analyzed in multiple samples after multiple determinations and where appropriate are expressed as means ± S.E. In experiments that compared more than one variable or treatment group, statistical differences were assessed by analysis of variance with limits set by the Student-Newman-Keuls' test. In experiments where a single group was compared with control, analysis defaulted to the Student's t test. Comparisons were performed with a commercial statistical software package (SigmaStat®). Differences among groups were considered significant when p values were <0.05.

RESULTS

PTHrp caused rapid, time-, and dose-dependent increases in cAMP accumulation in the dermal fibroblast cultures. At 10-30 nM PTHrp, there was a maximal 8-fold to 10-fold increase in cAMP that occurred within 1-5 min of treatment and regressed by 15 min. Basal cAMP levels were reduced in the absence of isobutylmethylxanthine, suggesting an active conventional phosphodiesterase system for normal cAMP turnover in these cells. Treatment with the phorbol ester PMA had no effect on cAMP levels, whereas forskolin was stimulatory (Fig. 1). In parallel cultures, cAMP was similarly increased by treatment with two other positive control reagents, isoproterenol (7.5 ± 0.5-fold) and prostaglandin E₂ (12.4 ± 3.4-fold).

PTHrp also induced time- and dose-dependent increases in IGF-I expression. A 1.7-2-fold increase in immunoreactive IGF-I (iIGF-I) polypeptide was secreted within 24 h of treatment with 30 nM PTHrp (Fig. 2a) and persisted for 48-72 h (data not shown). Similar to their effects on cAMP accumulation, treatment with forskolin enhanced iIGF-I secretion, whereas PMA did not (Fig. 2b). As shown in Fig. 3, dermal fibroblasts exhibit a complex IGF-I transcript pattern with predominate bands at 6.5, 4.1, 1.7, and 0.9 kb, analogous to that previously found in fetal rat bone cells (8-11). 6 h of exposure of dermal fibroblasts to PTHrp caused a 2-fold increase in steady state IGF-I transcript levels (Fig. 3, left panel), which remained elevated for 12 h and rescinded by 24 h (data not shown). Consistent with their effects on cAMP and iIGF-I polypeptide, forskolin increased steady state IGF-I transcripts, and PMA had no effect. Next, the cultures were treated for 6 h with control medium or with PTHrp to induce IGF-I mRNA and then supplemented with DRB for the next 24 h to suppress new mRNA transcription (18, 22). IGF-I transcripts decreased throughout the 24 h period of DRB treatment in control cultures. In contrast, the stimulatory effect of PTHrp persisted for the next 6 h and was still moderately elevated even 24 h after DRB exposure, indicating that PTHrp enhanced IGF-I mRNA stability.

Unlike their effects on IGF-I expression, PTHrp and forskolin did not enhance DNA synthesis, collagen synthesis, or general noncollagen protein synthesis by dermal fibroblasts, whereas PMA increased each of these biochemical activities (Fig. 4). Therefore, the effects of the cAMP stimulators on IGF-I expression appeared selective, and the fibroblasts were able to respond to PMA in other ways.

Studies with a probe that distinguishes the use of two separate gene promoters for IGF-I transcription (18) revealed that only promoter 1 drove IGF-I expression in fetal rat dermal fibroblasts. Several larger protected bands of 202-243 nucleotides in length, corresponding to the use of promoter 2 in rat liver, were not evident with RNA from control or treated dermal fibroblasts. A prominent smaller protected band of 143 nucleotides, corresponding to transcripts initiated from promoter 1, was enhanced by PTHrp and forskolin, but not by PMA (Fig. 5). To locate the minimal region of promoter 1 required for IGF-I expression in the dermal fibroblast cultures, they were transfected with various IGF-I promoter/reporter plasmid constructs truncated from the 5 or 3 ends (21). Negligible reporter gene was expressed using the 4.3-kb fragment of IGF-I promoter DNA, and maximal activity occurred with a 5 truncated fragment of 1.7 kb (Fig. 6a). An important enhancer element occurs within exon 1 at nucleotides +202 to +209 (23). In agreement with this, when 3 sequences from exon 1 encompassing this site were removed from the active 1.7 kb upstream promoter sequence, basal promoter activity in dermal fibroblasts was reduced (Fig. 6b). Therefore, basal IGF-I promoter activity in fetal rat dermal fibroblast cultures is fully consistent with its expression in fetal rat osteoblasts (21). However, in striking contrast to the changes that occur in osteoblasts, no stimulator of cAMP enhanced the activity of any promoter construct that we tested. In this context, we examined the maximally active 1.7-kb constructs truncated from the 5 and 3 ends to eliminate possible suppressor sequences and compared effects by PTHrp with dibutyryl cAMP, isoproterenol, and PMA, and invariably found no significant variations (Fig. 6, panels a-c).

DISCUSSION

Interactions between epithelial and mesenchymal cells are common during development, remodeling, repair, and perhaps the aging of many tissues. Often these interactions are dependent on the effects of local or systemic growth factors. Paracrine factors appear to influence skin cell metabolism (24), and of these IGF-I is thought to have an integral role (2, 25-27). Initial studies demonstrated that an epithelial component of skin, keratinocytes, do not express detectable levels of IGFs, but can respond to the IGFs released from co-cultured mesenchymal fibroblasts (3). Although IGF-I increases keratinocyte activity in vitro (2, 3, 28) and overexpression or targeted disruption of the genes encoding IGF-I or IGF type 1 receptor (29, 30) potently affect the dermis, little is still known about mechanisms that control production of IGF-I in this tissue.

Previous studies revealed an increase in PTHrp expression by keratinocytes in response to serum factors or to fibroblast-derived culture medium (31, 32), predicting interactions between these cell populations through soluble mediators. Furthermore, in cultures of BEN squamous carcinoma cells, PTHrp production is regulated by several factors including IGF-I and IGF-II (33). Skin fibroblasts express a common receptor shared by PTH and PTHrp and produce cAMP in response to these hormones (11-13). In contrast, keratinocytes fail to respond to these hormones and accordingly exhibit undetectable levels of mRNA for the conventional PTH/PTHrp receptor that is coupled to cAMP generation (34, 35). We initially reported that all inducers of cAMP, including PTH and PTHrp, increase IGF-I expression by differentiated osteoblasts (8-10). For these reasons, we postulated that keratinocyte-derived PTHrp might promote IGF-I expression by skin fibroblasts and in this way complete a paracrine loop within skin and support keratinocyte metabolism. A model for these interactions is shown in Fig. 7.

Our current studies confirm the potent effect of PTHrp on cAMP production in dermal fibroblasts and now demonstrate its ability to enhance IGF-I secretion and steady state mRNA levels by these cells. In addition, IGF-I expression was also enhanced by other cAMP stimulators but not by PMA, indicating protein kinase A-dependent mechanisms. This differs from the protein kinase C-dependent permissive effect of growth hormone in liver cells that may respond similarly to cAMP (36). We found an early and transient increase in IGF-I mRNA in PTHrp-induced dermal fibroblasts. IGF-I mRNA levels remained high in PTHrp treated cultures in the presence of DRB, consistent with mRNA stabilization. However, we saw no change in IGF-I promoter activation, which directly contrasts with our results in osteoblasts (18, 21). By gel shift analysis with nuclear extracts from cAMP-stimulated osteoblasts, we recently detected large molecular mass complexes of nuclear factors that associate with specific sequences of the IGF-I promoter (23), and find that a subset of these complexes is missing in control and treated dermal fibroblasts.² This suggests that some endogenous or cAMP-activated nuclear factors expressed by osteoblasts may be lacking in dermal fibroblasts. Therefore, although steady state IGF-I mRNA is enhanced by cAMP stimulators in both dermal and skeletal cells, it appears to be through different mechanisms in each tissue.

An increase in IGF-I in response to PTHrp may stimulate additional keratinocyte growth or subsequent matrix synthesis. Interactions of this sort between the epithelium and mesenchymal cells is consistent with a role for PTHrp in many tissues (7). Furthermore, immunohistochemical and in situ hybridization studies in embryologic tissue localize PTHrp and its receptors to the interface between these two tissue types (37). The importance of PTHrp as a local paracrine factor in skin is consistent with results from the transgenic mouse targeted to overexpress PTHrp in this tissue. In these animals the dermis and the epidermis are disrupted and there is abnormal development of hair follicles (38). Our current studies suggest that changes in local IGF-I expression may account in part for these effects, and that imbalances may occur when normal regulatory processes in skin are subverted by local overexpression of PTHrp. Analogous to other events that occur in skeletal cells, increases in cAMP may also alter fibroblast IGF-binding protein expression. In so doing, these changes could enhance or inhibit IGF activity (39). In this regard, our initial results suggest that activators of the cAMP and PKC pathways induce different effects in cultured skin and bone cells (22, 40).³ Further studies to determine the mechanisms by which these factors are expressed and necessarily suppressed under normal conditions may enhance our ability to focus on molecular events that can more efficiently control skin growth and repair.