Distribution of acidic and basic fibroblast growth factors in ovine skin during follicle morphogenesis

The fibroblast growth factor (FGF) family is composed of a number of heparin-binding proteins characterized by their potent mitogenic effects on cells, their ability to promote migration and differentiation, and to induce angiogenesis (reviewed by Folkman and Klagsbrun, 1987; Gospodarowicz, 1988). FGF has also been implicated in the regulation of vertebrate development. Mesoderm-inducing activity by basic FGF (bFGF) has been demonstrated in early Xenopus embryos, and FGF mRNA transcripts, FGF-like proteins and FGF-receptors, were described in this system (Kimelman and Kirschner, 1987; Slack et al., 1987; Kimelman et al., 1988; Gillespie et al., 1989; Slack and Isaacs, 1989). Other members of the FGF family have also been identified in association with developmental processes. Represa et al. (1991) implicated the int-2 proto-oncogene, which encodes several FGF-like proteins, in the induction of the otic vesicle of the chick. In mammals, evidence of a role for FGF in pattern specification is less direct, although the differential expression of bFGF, acidic FGF (aFGF) and FGF-5 genes during the latter half of embryonic development in the mouse certainly suggests such a function (Hébert et al., 1990).

A role for FGF in the growth and maintenance of skin cells was indicated when epidermal keratinocytes and dermal fibroblasts were found to proliferate in response to both aFGF and bFGF in vitro (O’Keefe et al., 1988; Ristow and Messmer, 1988; Shipley et al., 1989; Pisansarakit et al., 1991). In the sheep, aFGF was found to be more mitogenic than bFGF for skin keratinocytes (Pisansarakit et al., 1990). Basic FGF has also been detected in cultured fibroblasts (Story, 1989; Pisansarakit et al., 1991) and human keratinocytes (Halaban et al., 1988), and a bFGF-like pep-tide has been detected in immunoblots of ovine epidermis (Pisansarakit et al., 1990). However, relatively little is known about the distribution of aFGF or bFGF in the skin in vivo. A recent study by Gonzalez et al. (1990) identified bFGF predominantly as a component of the basement membrane in the developing whisker follicle of the 18-day rat fetus. The growth factor was also present in the mesenchymal cells that were adjacent to the basement membrane. Of equal interest was the identification of FGF-receptor gene transcripts (including flg (FGFR1) and bek (FGFR2)) in the epidermis and dermis, respectively, and in their derivatives in hair follicles of rat and mouse embryos (Orr-Urtreger et al., 1991; Wanaka et al., 1991; Peters et al., 1992).

The activities of FGF on cultured cells derived from the skin and the highly specific distributions of members of the FGF family and their receptors suggest that they participate in critical events during the development of mammalian skin and the morphogenesis of the cutaneous appendages. Here, we report the results of a study to identify aFGF and bFGF in ovine skin during fetal life with particular attention to their localizations during wool follicle morphogenesis.

Rams were introduced to a flock of Border Leicester × Merino ewes in the Autumn. The day on which ewes bore a mating mark (Radford et al., 1960) was regarded as the first day of pregnancy. A real-time ultrasound scanner (ADR model 2130) was used to confirm pregnancy and obtain an approximate estimation of the stage of gestation of each fetus. A final determination was subsequently made from crown-to-rump length measurements of the fetuses (Cloete, 1939). Skin was excised with a 1 cm trephine from the right or left flank of 40, 69, 76, 90 and 140-day-old fetuses and from the ewes, and the tissues fixed in a solution of 95% ethanol, formalin and glacial acetic acid (6:3:1, by vol.), prior to processing for histology. Tissue sections were cut at a thickness of 8 μm at right angles to the skin surface; they were transferred to gelatinized slides and deparaffinized.

Antiserum to aFGF was prepared by immunizing rabbits with a synthetic peptide (residues 1-11 of bovine aFGF; Bachem, California) conjugated to bovine serum albumin (BSA) (Goodfriend et al., 1964). Polyclonal antibodies to bFGF were similarly prepared by conjugating a synthetic peptide of bovine bFGF (residues 1-24 of bovine bFGF) to keyhole limpet hemocyanin (Auspep, Melbourne). Antisera were purified firstly by Protein A-affinity chromatography, followed by either aFGF- or bFGF-affinity chromatography. Sera for control incubations were obtained from the rabbits prior to immunization and purified using Protein A-affinity chromatography. The protein contents of the antibody preparations were estimated using the procedure of Bradford (1976).

Deparaffinized tissue sections were washed in Tris-buffered saline containing Tween-20 (TBST: 10 mM Tris, 150 mM NaCl, 0.05% Tween-20) at pH 8.0 for 10 min. This was followed by incubation at 4°C in ethanol and acetone (1:1, v/v) to permeabilize the tissues, and three 10-min rinses in TBST buffer. Non-specific staining was blocked with 3% BSA in TBST at room temperature for 1.5 h. The sections were subsequently incubated at 4°C for 18 h with the primary antisera diluted in 1% BSA/TBST solution (1:20 for both aFGF and bFGF) and washed (3×10 min) in TBST. The secondary antiserum was then applied for 1.5 h; this comprised swine anti-rabbit IgG complexed to fluorescein isothiocyanate (FITC; Dako Corp.) in TBST buffer containing 70 μg/ml Evans Blue dye as a counterstain (Mauger et al., 1982). The sections were rinsed in distilled water and mounted in buffered glycerol. Control incubations included substitution of the primary antisera with either pre-immune rabbit serum at a similar protein content in TBST buffer, or Protein A-purified antisera pretreated with either aFGF or bFGF peptides. To investigate epitope masking, some sections were digested for 2 h at 37°C with heparinase and heparitinase (Seikagaku America), 1 unit/ml in 0.1 M sodium acetate (pH 7.0) with 1 mM calcium acetate (McCarthy and Couchman, 1990) prior to antibody incubation. Representative sections were photographed using a Leitz fluorescence microscope fitted with a filter of band-pass range 450-490 nm.

Groups of approximately 50 freshly isolated wool follicles were dispersed in 0.0625 M Tris-HCl buffer (pH 6.8) containing 4% sodium dodecyl sulfate (SDS) and 2.5% 2-mercaptoethanol, and the proteins extracted by sonication at room temperature. The protein extract, bovine aFGF (UBI, Inc.), bovine bFGF (R & D Systems), epidermal growth factor (EGF; isolated from male mouse submaxillary glands, according to the method of Savage and Cohen, 1972) and pre-stained low molecular mass markers (Rainbow markers; Amersham Corp.) were subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE) using the discontinuous system of Laemmli (1970). The proteins were separated on a 15% to 20% acrylamide gradient gel with a 3% stacking gel, then electrophoretically transferred to Immobilon-P membrane (Millipore Corp.). The membrane was incubated with either aFGF or bFGF antisera. Antibody binding was detected using a biotinylated goat anti-rabbit IgG secondary antibody (Vector Laboratories, Inc.) followed by a streptavidin-peroxidase complex (Zymed Laboratories, Inc.) and then visualized with 4-chloro-1-naphthol (Sigma) as substrate.

Antisera raised against aFGF- and bFGF-conjugated peptides were specific for aFGF and bFGF proteins, respectively. They did not cross-react with other growth factors such as TGF-α, IGF or PDGF (data not shown) or with EGF, which is known to be present in developing skin and in mature follicles (du Cros et al., 1992) (Fig. 1A,D). Skin sections incubated with pre-immune serum showed very low non-specific binding. Similarly, binding of aFGF and bFGF antibodies to skin sections was reduced when they were pretreated with aFGF or bFGF peptides, respectively. Treatment of sections with heparinase and heparitinase did not alter the distribution of immunoreactive sites.

We dissected follicles from mature sheep skin to determine the molecular masses of aFGF and bFGF. Following SDS-PAGE and immunoblotting, a band showing strong aFGF reactivity was detected with the antibody probe. This band had a molecular mass of approximately 27 kDa (Fig. 1A).

Antibody probing of different batches of freshly prepared wool follicle extracts consistently revealed the same band although sometimes the intensity varied and other bands were recognized (Fig. 1B). These other bands were at approximately 10 kDa and at the expected 16 and 18 kDa. The relative intensities of the 10, 18 and 27 kDa bands varied such that when the 27 kDa band diminished in intensity, the other two became more prominent. Highly degraded follicle extracts contained the smaller proteins but not the 27 kDa isoform (Fig. 1C). The aFGF antibody did not recognise a 27 kDa band in sheep pituitary extracts (fresh or degraded) although several others were revealed including the 10, 16 and 18 kDa forms (Fig. 1B,C).

As with aFGF, we attempted to determine the molecular mass of bFGF in the wool follicles. However, presumably because of the low abundance of bFGF in the follicles, the antibody was unable to detect the growth factor in wool follicle extracts (data not shown).

At 40 days of gestation, the epidermis consists of a basal layer, a discontinuous intermediate layer and a flattened periderm; the dermis is composed of a matrix that contains scattered fibroblasts (Fig. 2A). No aFGF was detected in the skin at this stage, which is prior to follicle initiation. Initiation of wool follicles is first observed in midside skin at approximately 55-60 days of gestation (Hardy and Lyne, 1956), although the initiation period continues over a number of weeks. The process begins with the aggregation of epidermal and dermal cells at distinct sites adjacent to the dermal-epidermal junction (DEJ). Immunoreactivity to aFGF was first evident in the epidermis at this stage of follicle development, being associated with the basal cells; more diffuse staining was associated with the distal layers (Fig. 2B). In follicle primordia, immunoreactivity was associated with the borders of the cells forming the epidermal aggregates. Some cells of the dermal condensations were also stained. During the next stage, at which the epidermal cells began to proliferate and penetrate the dermis, cells at the core of the follicle were also stained (Fig. 2C).

A variety of growth stages is exhibited by the follicle population at 90 days of gestation, the most advanced follicles appearing as elongated columns of cells penetrating the dermis (Fig. 2D,E). As this process occurred, aFGF fluorescence extended to the distal regions of the epidermal column. A strong reaction associated with the mesenchymal cells of the dermal condensations was also observed at this stage of development, but little or no reaction was seen in the surrounding dermis (Fig. 2D).

In adult skin, the epidermis showed strong immunofluorescence. This extended into the proximal regions of the pilary canals, which are continuous with the outer root sheath (ORS) of the wool follicles (Fig. 2F). However, the cells comprising the distal zone of the ORS showed no aFGF immunofluorescence (Fig. 2G). The suprabulbar cells of the wool follicle and those in the keratogenous zone displayed strong aFGF reactivity (Fig. 2G). The aFGF immunoreactivity seen in the dermal condensation at 90 days of gestation was not apparent in the dermal papilla of the mature follicle. Equivalent sections of tissue exposed to control (pre-immune) serum or antibodies pre-incubated with aFGF peptide showed reduced or no fluorescence (see Fig. 2E, for example).

The distribution of bFGF differed from that of aFGF and was first detected at day 40 of gestation. Basic FGF immunoreactivity was predominantly localized to the cytoplasm of cells of the intermediate layer and periderm of the epidermis; faint punctate labelling was also visible along the DEJ (Fig. 3A). The cells of the basal layer appeared otherwise unstained. The distribution of bFGF fluorescence was similar at 69 days of gestation (Fig. 3B). It did not change with the appearance of the follicle primordia, although the intensity of immunoreaction at the DEJ increased during early follicle development (Fig. 3C). Basic FGF was also localized at the periphery of the epidermal plug adjacent to the dermis and dermal condensation. At 90 days of gestation, bFGF immunoreactivity remained associated with the intermediate layer of the epidermis. In the follicles, staining was located in the basement membrane zone although core cells of the elongating epithelial column were also immunostained (Fig. 3D). The immunoreactivity increased as follicle development progressed. Equivalent sections exposed to control (preimmune) serum or antibodies pre-incubated with bFGF peptide showed reduced or no fluorescence (Fig. 3E).

In fully formed wool follicles, bFGF staining was present in the ORS cells where fluorescence was particularly enhanced in the basement membrane zone; the staining extended distally to include the region of the basement membrane enveloping the bulb matrix (Fig. 3F). Immunofluorescence was also seen occasionally in the basement membane zone between the dermal papilla and the bulb matrix. With sloughing of the periderm at approximately 115 days of gestation and maturation of the epidermis, bFGF was found predominantly in the suprabasal cells. This epidermal distribution was also seen in adult skin (Fig. 3G).

Very little bFGF immunofluorescence was detected at any stage in the dermal components of the developing skin or follicles. Similarly, no activity was found in the dermis of mature skin, or particularly associated with the dermal papilla of the follicle.

In an immunohistochemical study of bFGF antibody-binding sites in the 18-day-old rat fetus, Gonzalez et al. (1990) found a widespread distribution of immunoreactive material among different tissues, and also at extracellular sites associated with basement membranes. FGFs have strong affinities for heparan sulfate proteoglycan (Vlodavsky et al., 1987) and, as this is a component of basement membranes, it is possible that bFGF binds to it and remains in the environment to serve as a cell trophic factor (Gospodarowicz and Cheng, 1986; Sommer and Rifkin, 1989). Indeed, recent studies suggest that binding to heparin-like molecules is a prerequisite for the interaction of bFGF with its high-affinity receptor (Yayon et al., 1991; Klagsbrun and Baird, 1991). Furthermore, it is likely that the cell-surface proteoglycan syndecan, which binds FGF by its heparan sulfate chains, performs this function (Ruoslahti and Yamaguchi, 1991). In the present work, the detection of bFGF in the cytoplasm of the epidermal cells and at the DEJ of young fetal skin raises the distinct possibility that the growth factor is synthesized locally and deposited extracellularly in the basement membranes.

The association of aFGF and bFGF with the DEJ, during hair follicle initiation in particular, clearly positions these molecules at the interface of epidermal-mesenchymal interactions, which are crucial for early morphogenesis. The signal that triggers the aggregation of cells at follicle initiation sites and later epithelial penetration of the dermis during follicle development has not been identified. However, it may be significant that transcripts of bone morphogenetic protein-4 (BMP-4), a member of the transforming growth factor-beta (TGF-β) gene family, are found in the dermal condensations of early whisker follicle primordia (Jones et al., 1991). This, together with the present evidence of FGF at the DEJ, implicates both molecules in a synergistic relationship during follicle morphogenesis, perhaps by redirecting the developmental fates of local mesenchymal cells into a pre-papilla cell population (Moore et al., 1989) prior to their condensation adjacent to the epidermal aggregation. The influence of both TGF-β and bFGF on the developmental fates of cells has already been demonstrated during mesoderm induction in the early Xenopus embryo (Kimelman and Kirschner, 1987). The expression of BMP-4 is transient whereas FGF immunoreactivity persists throughout follicle development.

Syndecan has also been identified in the condensing mesenchyme of vibrissa follicles (Trautman et al., 1991), perhaps induced by the overlying epidermis (Vainio et al., 1989). The localization of this low-affinity cell surface receptor for FGF further confirms a role for the growth factor in the immediate post-induction processes of morphogenesis. Moreover, it has recently been shown that, when administered to newborn mice, bFGF inhibits hair follicle initiation and development, and subsequently affects the hair cycle (du Cros, 1993). It is known that bFGF stimulates cell migration, the synthesis of collagen IV and collagenase activity (Rifkin and Moscatelli, 1989). It is likely that these activities contribute, at least in part, to the mechanism by which the epidermal plugs penetrate the dermis. In this context, it is of interest that the migratory activity and invasiveness of capillary endothelial cells is correlated with their FGF content (Tsuboi et al., 1990). Furthermore, the presence of aFGF, transcripts of FGF receptor genes (Orr-Urtreger et al., 1991; Peters et al., 1992) and syndecan (Trautman et al., 1991) in the developing dermal papilla suggests an autocrine function for aFGF in regulating the morphogenesis of this cell population. Certainly, the papilla is necessary for normal follicle development and activity (Moore et al., 1992) and, in the mature follicle, its size is related to the dimensions of the growing fiber (Rudall, 1956; Ibrahim and Wright, 1982).

In addition to heparin-like low-affinity receptors, two of four high-affinity receptors for FGF (reviewed by Partanen et al., 1992) have been identified in the skin. Recent studies have shown that FGFR1 is expressed in mesodermal derivatives of the skin whereas FGFR2 is predominantly expressed in epithelial tissues (Orr-Urtreger et al., 1991; Peters et al., 1992). Each of these receptors has two isoforms, one of which binds both aFGF and bFGF, and another that binds only aFGF. However, the isoforms were not distinguished in either study and the functional significance of receptor distribution is not yet known.

Fig. 4 illustrates the current status of our knowledge of the distributions of aFGF, bFGF, FGFR1 and FGFR2. The spatio-temporal pattern of aFGF and bFGF distributions shown here in the skin at maturity suggests that these factors perform a number of related, and perhaps complementary, differentiation and proliferative functions in the adult animal. The two growth factors have different, possibly even exclusive, localizations. Basic FGF was confined to the suprabasal, differentiating layers of the epidermis, which suggests that it does not function as an autocrine mitogen for the germinative cells. This is consistent with its relatively low activity in cultured keratinocytes (Pisansarakit et al., 1990). Nevertheless, the presence of bFGF in the ORS of the distal region of the follicle and, more particularly, in the basement membrane adjacent to the proliferative zone of the follicle bulb, may indicate a specific growth-promoting function. Bound to the basement membrane in an active form, bFGF could provide a continuous proliferative stimulus to the matrix cells and thus directly influence the rate of fiber growth.

By contrast, the presence of aFGF in the basal cells of the epidermis and the proximal zone of the ORS, which is contiguous with the epidermis, suggests that it acts as an autocrine mitogen for skin keratinocytes (Pisansarakit et al., 1990). However, high concentrations of aFGF immunoreactivity in the suprabulbar cells of the follicle are more indicative of a role in differentiation rather than proliferation. The bulb matrix cells have a number of fates, forming the inner root sheath of the follicle as well as the fiber. It is possible that differing concentrations of aFGF in these cells may modulate their functions, directing them to follow alternative differentiation pathways.

It is noteworthy that the 27 kDa aFGF identified in mature wool follicles has an apparent molecular mass higher than is normally reported for this growth factor. We concluded that this form of aFGF is composed of a complex between the expected 16-18 kDa proteins and a smaller truncated form. These conclusions are based on the relative intensities of the three proteins revealed in many different fresh tissue preparations, and on the findings that highly degraded follicle extracts contain the smaller proteins but not the 27 kDa isoform. Higher molecular mass forms of aFGF have previously been identified by others: for example, Fu et al. (1991) identified aFGF-immunoreactive bands at 26-28 kDa in rat embryos. They concluded that these were naturally occurring forms of the growth factor, speculating that the 28 kDa band is composed of the expected 16-18 kDa or 24 kDa aFGF plus a small binding protein. Since the band was not found in adult tissue, the results implied that the high molecular mass aFGF was developmentally regulated. Basic FGF has also been identified in higher molecular mass forms in developing human tissues, possibly produced by initiation at alternative start sites during protein translation (Giordano et al., 1992). In addition, species specificity for different forms of bFGF has been suggested (Brigstock et al., 1990). However, on the basis of the immunoblotting results and the finding that the 27 kDa aFGF band was not present in sheep pituitary extracts, we conclude that rather than being specific for sheep, the protein must represent a form of aFGF found in the differentiating bulb cells of wool follicles.

In summary, the data presented in this study provide further evidence of the important physiological roles of aFGF and bFGF. The exclusive locations of aFGF and bFGF in the developing skin and hair follicles suggest distinct and synergistic roles, for aFGF in differentiation events and bFGF in cellular proliferation, both during morphogenesis and in fiber production at maturity.

This work was supported, in part, by the Australian Wool Corporation and the National Institutes of Health, USA (HD 17664). We thank Judy Bond for dissecting wool follicles from sheep skin for the electrophoretic analyses.

Aspects of this work were presented at the New York Academy of Sciences Conference on the Molecular and Structural Biology of Hair, January 1991.