Unique domain functions of p63 isotypes that differentially regulate distinct aspects of epidermal homeostasis.

p63 is critical for squamous development and exists as multiple isotypes of two subclasses, TA and DeltaN. DeltaNp63 isotypes can antagonize transcription by TAp63 and p53, and are highly expressed in squamous cell cancers. Using mouse keratinocytes as a biological model of squamous epithelium, we show that multiple p63 isotypes, DeltaN- and TA-containing, are expressed and differentially modulated during in vitro murine keratinocyte differentiation. DeltaNp63alpha declines with Ca2+-induced differentiation, while a smaller DeltaN-form, DeltaNp63s, persists, suggesting unique functions of the two DeltaN-forms. To investigate the impact of dysregulated p63 expression that is observed in cancers and to define the biological contribution of the different domains of the p63 isotypes, DeltaNp63alpha, DeltaNp63p40, TAp63alpha, TAp63gamma or beta-galactosidase were overexpressed in primary murine keratinocytes. Microarray, RT-PCR and western blot analyses revealed that overexpression of DeltaNp63p40, which lacks the entire alpha-tail present in DeltaNp63alpha, permits expression of a full panel of differentiation markers. This is in contrast to overexpression of the full-length DeltaNp63alpha, which blocks induction of keratin 10, loricrin and filaggrin. These findings support a role for the alpha-tail of DeltaNp63alpha in blocking differentiation-specific gene expression. Overexpression of either TAp63 isotype permits keratin 10 and loricrin expression, thus the alpha-terminus requires the cooperation of the DeltaN domain in blocking early differentiation. However, both TA isotypes block filaggrin induction. The DeltaN-terminus is sufficient to maintain keratinocytes in a proliferative state, as both DeltaN forms block Ca2+-mediated p21WAF1 induction and S-phase arrest, while sustaining elevated PCNA levels. No alteration in cell cycle regulation was observed in keratinocytes overexpressing TAp63alpha or TAp63gamma. Clarifying the functional distinctions between p63 isotypes and domains will help to elucidate how their dysregulation impacts tumor biology and may suggest novel therapeutic strategies for modulating behavior of tumor cells with altered expression of p53 family members.

p53 and tumor necrosis factor alpha regulate the expression of a mitochondrial chloride channel protein.

A novel chloride intracellular channel (CLIC) gene, clone mc3s5/mtCLIC, has been identified from differential display analysis of differentiating mouse keratinocytes from p53+/+ and p53-/- mice. The 4.2-kilobase pair cDNA contains an open reading frame of 762 base pairs encoding a 253-amino acid protein with two putative transmembrane domains. mc3s5/mtCLIC protein shares extensive homology with a family of intracellular organelle chloride channels but is the first shown to be differentially regulated. mc3s5/mtCLIC mRNA is expressed to the greatest extent in vivo in heart, lung, liver, kidney, and skin, with reduced levels in some organs from p53-/- mice. mc3s5/mtCLIC mRNA and protein are higher in p53+/+ compared with p53-/- basal keratinocytes in culture, and both increase in differentiating keratinocytes independent of genotype. Overexpression of p53 in keratinocytes induces mc3s5/mtCLIC mRNA and protein. Exogenous human recombinant tumor necrosis factor alpha also up-regulates mc3s5/mtCLIC mRNA and protein in keratinocytes. Subcellular fractionation of keratinocytes indicates that both the green fluorescent protein-mc3s5 fusion protein and the endogenous mc3s5/mtCLIC are localized to the cytoplasm and mitochondria. Similarly, mc3s5/mtCLIC was localized to mitochondria and cytoplasmic fractions of rat liver homogenates. Furthermore, mc3s5/mtCLIC colocalized with cytochrome oxidase in keratinocyte mitochondria by immunofluorescence and was also detected in the cytoplasmic compartment. Sucrose gradient-purified mitochondria from rat liver confirmed this mitochondrial localization. This represents the first report of localization of a CLIC type chloride channel in mitochondria and the first indication that expression of an organellular chloride channel can be regulated by p53 and tumor necrosis factor alpha.

Genetic deletion of p21WAF1 enhances papilloma formation but not malignant conversion in experimental mouse skin carcinogenesis.

Tumor suppression by p53 is believed to reside in its ability to regulate gene transcription, including up-regulation of p21WAF1. In p53(-/-) mice, chemical- or oncogene-induced skin tumors undergo accelerated malignant conversion. To determine the contribution of the p21WAF1 gene product to epidermal carcinogenesis, animals +/+, +/-, and -/- for a null mutation in the p21WAF1 gene were treated once with 25 nmol 7,12-dimethylbenz[a]anthracene, followed by 5 microg of TPA two times/week for 20 weeks. Papilloma frequency was higher in the p21WAF1-deficient mice. However, the frequency of malignant conversion was similar among all three genotypes. After TPA treatment, all genotypes developed epidermal hyperplasia, although the labeling index was lower in p21WAF1 (-/-) epidermis compared with p21WAF1 (+/+). Furthermore, the expression of differentiation markers was the same across genotypes in untreated or TPA-treated epidermis. Similar frequencies of malignant conversion were also observed in an in vitro assay. Thus, p21WAF1 suppresses early stages of papilloma formation but not malignant progression in mouse skin carcinogenesis, and decreased levels of p21WAF1 do not account for the enhanced malignant conversion of p53 null epidermal tumors.

Cooperation of p53 loss of function and v-Ha-ras in transformation of mouse keratinocyte cell lines.

We previously demonstrated that after transduction with the v-Ha-ras oncogene and grafting onto nude mouse hosts, primary epidermal keratinocytes with a null mutation in the p53 gene form tumors with increased growth rates and predisposition to malignant conversion relative to p53 wild-type keratinocytes (Weinberg WC, et al., Cancer Res 54:5584-5592, 1994). To further explore the cooperation between p53 loss of function and activation of the ras oncogene, cell lines were established from the normal epidermises of newborn and adult p53-null mice, and parallel subclones were reconstituted with the p53val135 temperature-sensitive mutant. Reconstituted lines C, G, N, and V demonstrated functional p53 transcriptional activator activity at the wild-type-permissive temperature of 32 degrees C, compared with the hygromycin-selected control line X and parental p53-null lines NHK4 and AK1b. Hygromycin-selected subclones, but not the parental lines, made normal skin in vivo; all cell lines made carcinomas after introduction of v-Ha-ras, independent of p53 status. These cell lines were compared in vitro at 32 degrees C to maximize the amount of p53val135 in the wild-type conformation. Expression of v-Ha-ras did not consistently alter p53-mediated transcriptional activity, suggesting tat ras acts downstream or independently of p53. No correlation was observed between p53-mediated transcriptional activity and in vitro growth rates, colony formation after exposure to ultraviolet light, or suppression by normal neighboring keratinocytes. However, keratinocyte cell lines devoid of p53 and expressing v-Ha-ras formed colonies in soft agar; this was blocked at 32 degrees C in all cell lines reconstituted with p53val135. These keratinocyte lines provide a model for exploring the role of p53 and the interaction of p53 and ras in keratinocyte transformation.

Loss of p21CIP1/WAF1 does not recapitulate accelerated malignant conversion caused by p53 loss in experimental skin carcinogenesis.

The p21(CIP1/WAF1) protein is considered a downstream effector of tumor suppression by p53. We have previously demonstrated that p53 null keratinocytes have lower basal p21(CIP1/WAF1) mRNA levels and that tumors derived from these cells following transduction with the v-ras(Ha) oncogene grow faster than wildtype keratinocytes and rapidly progress to undifferentiated carcinomas (Cancer Res 54: 5584-5592, 1994). In this study, primary keratinocytes differing in p21(CIP1/WAF1) gene dose were transduced with v-ras(Ha) encoding retrovirus and grafted to nude mouse hosts to test whether the p53 null phenotype is mediated through p21(CIP1/WAF1). Resulting tumors from all genotypes were well differentiated papillomas; focal carcinomas were observed in 43, 30 and 44% of papillomas derived from +/+, +/- and -/- keratinocytes, respectively. p21(CIP1/WAF1) deficient keratinocytes expressing v-ras(Ha) do not display the degree of increased growth observed in p53 deficient tumors in vivo or the decreased responsiveness to negative growth regulation by Ca2+ in vitro. These results suggest that p21(CIP1/WAF1) does not regulate the differentiated phenotype or malignant progression of v-ras(Ha) initiated keratinocytes and that additional functions of the p53 protein other than transcriptional regulation of the p21(CIP1/WAF1) gene are required for p53 mediated tumor suppression.

Functions of the POU domain genes Skn-1a/i and Tst-1/Oct-6/SCIP in epidermal differentiation.

Here we report on investigation of the role of the POU domain genes Skin-1a/i (Skn-1a/i/Epoc/Oct-11) and Testes-1 (Tst-1/Oct-6/SCIP) in epidermis where proliferating basal keratinocytes withdraw from the cell cycle, migrate suprabasally, and terminally differentiate to form a multilayered, stratified epithelium. The expression of the Skn-1a/i and Tst-1 genes is linked to keratinocyte differentiation in vivo and in vitro, whereas the ubiquitous POU domain factor Oct-1 is expressed highly in both proliferating and post-mitotic keratinocytes. Analysis of Skn-1a/i gene-deleted mice reveals that the Skn-1a/i gene modulates the pattern of expression of the terminal differentiation marker loricrin and inhibits expression of genes encoding markers of the epidermal keratinocyte wounding response. Although epidermis from Tst-1 gene-deleted mice develops normally, epidermis from mice deleted for both Skn-1a/i and Tst-1 is hyperplastic and fails to suppress expression of K14 and Spr-1 in suprabasal cells when transplanted onto athymic mice. This suggests that Skn-1a/i and Tst-1 serve redundant functions in epidermis. Therefore, at least two POU domain genes, Skn-1a/i and Tst-1, serve both distinct and overlapping functions to regulate differentiation of epidermal keratinocytes during normal development and wound healing.

Transforming growth factor beta 1 suppresses genomic instability independent of a G1 arrest, p53, and Rb.

Alterations in expression of or responsiveness to transforming growth factor beta (TGF-beta) are frequently found in human and animal epithelial cancers and are though to be important for loss of growth control in the neoplastic cell. We show here that keratinocyte cell lines from mice with a targeted deletion of the TGF-beta 1 gene have significantly increased frequencies of gene amplification in response to the drug N-phosphonoacetyl-L-aspartate (PALA) compared to TGF-beta 1-expressing control keratinocyte cell lines. In contrast to the control lines, the PALA-mediated G1 arrest did not occur in the TGF-beta 1 null keratinocytes despite the presence of wild-type p53 in both genotypes. Exogenous TGF-beta 1 suppresses gene amplification in the null keratinocytes at concentrations that do not cause a G1 growth arrest and in human tumor cell lines that are insensitive to TGF-beta 1-mediated growth inhibition. The pathway of TGF-beta 1 suppression is independent of the p53 and Rb genes, but requires an intact TGF-beta type II receptor. These studies reveal a novel TGF-beta-mediated pathway regulating genomic stability and suggest that defects in TGF-beta signaling may have profound effects on tumor progression independent of cell proliferation.

Nitric oxide-induced p53 accumulation and regulation of inducible nitric oxide synthase expression by wild-type p53.

The tumor suppressor gene product p53 plays an important role in the cellular response to DNA damage from exogenous chemical and physical mutagens. Therefore, we hypothesized that p53 performs a similar role in response to putative endogenous mutagens, such as nitric oxide (NO). We report here that exposure of human cells to NO generated from an NO donor or from overexpression of inducible nitric oxide synthase (NOS2) results in p53 protein accumulation. In addition, expression of wild-type (WT) p53 in a variety of human tumor cell lines, as well as murine fibroblasts, results in down-regulation of NOS2 expression through inhibition of the NOS2 promoter. These data are consistent with the hypothesis of a negative feedback loop in which endogenous NO-induced DNA damage results in WT p53 accumulation and provides a novel mechanism by which p53 safeguards against DNA damage through p53-mediated transrepression of NOS2 gene expression, thus reducing the potential for NO-induced DNA damage.

p53-mediated transcriptional activity increases in differentiating epidermal keratinocytes in association with decreased p53 protein.

The regulation of p53 protein synthesis and p53-mediated gene transactivation were evaluated in cultured mouse keratinocytes maintained as basal cells or induced to differentiate by Ca2+ > 0.1 mM. p53 protein half-life, p53 protein synthesis and the level of p53 mRNA decreased during terminal differentiation, as detected by immunoprecipitation with a panel of p53-specific antibodies and Northern blotting. Thus differentiating keratinocytes have lower levels of p53 protein. This decline is not observed following growth arrest alone, or in papilloma cell lines which do not terminally differentiate in response to Ca2+. In contrast, the ability of endogenous p53 to transactivate transcription from the PG13 CAT plasmid increased during differentiation in vitro. This change in activity cannot be explained by changes in p53 conformation or nuclear localization. Consistent with these findings, mRNA for the p53-mediated genes WAF1 and mdm-2 increased with Ca(2+)-induced differentiation in a time dependent manner, suggesting activation of p53 contributes to the differentiated phenotype. However, p53-null mice exhibit histologically normal skin and epidermal keratinocytes from these mice express the appropriate markers of differentiation and suppression of DNA synthesis in vitro when the [Ca2+] is > 0.1 mM. The observation that proliferating cells have higher levels of p53 protein which is less active for its function than differentiated cell types could have a consequence for the selection of p53 gene mutations during carcinogenesis, depending upon the stage of differentiation of the tumor cell type.

Role of oncogenes and tumor suppressor genes in multistage carcinogenesis.

The introduction of the techniques of molecular biology as tools to study skin carcinogenesis has provided more precise localization of biochemical pathways that regulate the tumor phenotype. This approach has identified genetic changes that are characteristic of each of the specific stages of squamous cancer pathogenesis: initiation, exogenous promotion, premalignant progression, and malignant conversion. Initiation can result from mutations in a single gene, and the Harvey allele of the ras gene family has been identified as a frequent site for initiating mutations. Heterozygous activating mutations in c-rasHa are dominant, and affected keratinocytes hyperproliferate and are resistant to signals for terminal differentiation. An important pathway impacted by c-rasHa activation is the protein kinase C (PKC) pathway, a major regulator of keratinocyte differentiation. Increased activity of PKC alpha and suppression of PKC delta by tyrosine phosphorylation contribute to the phenotypic consequences of rasHa gene activation in keratinocytes. Tumor promoters disturb epidermal homeostasis and cause selective clonal expansion of initiated cells to produce multiple benign squamous papillomas. Resistance to differentiation and enhanced growth rate of initiated cells impart a growth advantage when the epidermis is exposed to promoters. The frequency of premalignant progression varies among papillomas, and subpopulations at high risk for progression have been identified. These high-risk papillomas overexpress the alpha 6 beta 4 integrin and are deficient in transforming growth factor beta 1 and beta 2 peptides, two changes associated with a very high proliferation rate in this subset of tumors. The introduction of an oncogenic rasHa gene into epidermal cells derived from transgenic mice with a null mutation in the TGF beta 1 gene have an accelerated rate of malignant progression when examined in vivo. Thus members of the TGF beta gene family contribute a tumor-suppressor function in carcinogenesis. Accelerated malignant progression is also found with v-rasHa transduced keratinocytes from skin of mice with a null mutation in the p53 gene. The similarities in risk for malignant conversion by initiated keratinocytes from TG beta 1 and p53 null geneotypes suggest that a common, growth-related pathway may underly the tumor-suppressive functions of these proteins in the skin carcinogenesis model.

p53 gene dosage modifies growth and malignant progression of keratinocytes expressing the v-rasHa oncogene.

Epidermal keratinocyte cultures were established from newborn mice expressing a null mutation in the p53 gene to explore the contribution of p53 to epidermal growth regulation and neoplasia. Keratinocytes were initiated by transduction with a replication-defective retrovirus encoding the v-rasHa oncogene and grafted onto nude mouse hosts. Tumors arising from keratinocytes heterozygous or null for functional p53 in the presence of v-rasHa have growth rates approximately 5-fold higher than those derived from p53(+/+) controls and rapidly form carcinomas, in contrast to the benign phenotype observed in p53(+/+)/v-rasHa grafts. In vitro, p53-deficient keratinocytes with and without v-rasHa expression display decreased responsiveness to the negative growth regulators transforming growth factors beta 1 and beta 2. In combination with v-rasHa, p53-deficient keratinocytes also exhibit decreased responsiveness to elevated Ca2+. These differences between genotypes cannot be attributed to changes in transforming growth factor beta receptor types present or altered levels of epidermal growth factor receptor and are independent of c-myc transcript levels. mRNA expression for the p-53 inducible protein WAF1 correlates with p53 gene dosage, but low levels are still detectable in p53(-/-) keratinocytes. The altered responsiveness of p53 deficient keratinocytes to negative growth regulators may provide a growth advantage to such cells in vivo and render them more susceptible to genetic alterations and malignant conversion.

In vivo regulation of murine hair growth: insights from grafting defined cell populations onto nude mice.

The nude mouse graft model for testing the hair-forming ability of selected cell populations has considerable potential for providing insights into factors that are important for hair follicle development and proper hair formation. We have developed a minimal component system consisting of immature hair follicle buds from newborn pigmented C57BL/6 mice and adenovirus E1A-immortalized rat vibrissa dermal papilla cells. Hair follicle buds contribute to formation of hairless skin when grafted alone or with Swiss 3T3 cells, but produce densely haired skin when grafted with a fresh dermal cell preparation. The fresh dermal cell preparation represents the single cell fraction after hair follicles have been removed from a collagenase digest of newborn mouse dermis. It provides dermal papilla cells, fibroblasts, and possibly other important growth factor-producing cell types. Rat vibrissa dermal papilla cells supported dense hair growth at early passage in culture but progressively lost this potential during repeated passage in culture. Of 19 E1A-immortalized, clonally derived rat vibrissa dermal papilla cell lines, the four most positive clones supported hair growth to the extent of approximately 200 to 300 hairs per 1-2 cm2 graft area. The remaining clones were moderately positive (five clones), weakly positive (three clones), or negative (seven clones). Swiss 3T3 cells prevented contraction of the graft area but did not appear to affect the number of hairs in the graft site produced by dermal papilla cells plus hair follicle buds alone. The relatively low hair density (estimated 1-5% of normal) resulting from grafts of hair follicle buds with the most positive of the immortalized dermal papilla cell clones compared to fresh dermal cells suggests that optimal reconstitution of hair growth requires some function of dermal papilla cells partially lost during the immortalization process and possibly the contribution of other cell types present in the fresh dermal cell preparation, which is not supplied by the Swiss 3T3 cells. The current graft system, comprising hair follicle buds and immortalized dermal papilla cell clones, provides an assay for positive or negative influences on hair growth exerted by added selected cell types, growth factors, or other substances. Characterization of the phenotype of the dermal papilla cell lines, which differ in their ability to support hair growth when grafted with hair follicle buds, may provide insight into specific dermal papilla cell properties important for their function in this system.

Regulation of hair follicle development: an in vitro model for hair follicle invasion of dermis and associated connective tissue remodeling.

During embryonic development presumptive hair follicle cells of epithelial and mesenchymal origin are determined in defined body locations. This is followed by rapid proliferation of epithelial cells and associated penetration into the dermis in response to as yet undetermined signals. A collagen matrix culture system, which maintains the three-dimensional relationships of hair follicle cells to each other, was developed to study the regulation of the enlargement of immature hair follicles and the accompanying remodeling of the dermis. In studies with a heterogeneous dermis-derived preparation of murine hair follicles, ranging in size from the earliest down-growing budding cell mass to hair-forming follicles, we had previously shown that cell proliferation was stimulated by cholera toxin and epidermal growth factor, but only the epidermal growth factor-stimulated proliferation was accompanied by digestion of the collagen matrix due to release of collagenolytic enzymes. Further studies revealed that transforming growth factor-alpha also stimulated hair follicle cell proliferation and collagenase release. However, although transforming growth factor-beta inhibited the transforming growth factor-alpha-stimulated proliferation, it enhanced the release and activation of collagenases and other gelatin-degrading enzymes detectable by gelatin zymography. Stimulation of collagenolytic activity depended on the three-dimensional hair follicle structure and did not occur in monolayer cultures of hair follicle cells. Comparison of hair follicle buds with more developed dermis-derived hair follicles, plated at the same cell density (based on DNA content), suggested that a greater fraction of cells in the bud-stage follicle responded to the growth factors by release of collagenases. Possibly only the cells in the advancing portion of growing hair follicles that are closest to the dermal papilla cell cluster produce the collagenases in response to growth factors. To examine the participation of dermal papilla cells in collagenase release and activation, several immortalized rat whisker dermal papilla cell lines were co-cultured with mouse hair follicle buds. Co-culture resulted in a marked enlargement of follicles as well as activation of the 92-kDa type IV collagenase, produced by hair follicle buds, that correlated with ability of the dermal papilla cells to stimulate hair formation in grafts of hair follicle buds on nude mice. Dermal papilla cells cultured alone produced the 72-kDa type IV collagenase, which was also activated during co-culture with hair follicle buds. Thus, two activities, both relevant for hair follicle development, namely, cell proliferation and release and activation of collagenases, have been stimulated in immature hair follicle buds by either growth-factor supplementation or interaction with dermal papilla cells.(ABSTRACT TRUNCATED AT 400 WORDS)

Reconstitution of hair follicle development in vivo: determination of follicle formation, hair growth, and hair quality by dermal cells.

Combinations of cultured and uncultured epidermal and dermal cell preparations from newborn and perinatal mice were grafted onto the backs of athymic nude mouse hosts to elucidate the cellular requirements for skin appendage formation. All epidermal populations studied, including a total epidermal keratinocyte preparation from trypsin-split skin, developing hair follicle buds isolated from epidermis, and preformed hair follicles isolated from dermis, make haired skin when grafted with fresh dermal cells. Only pre-formed hair follicles produce haired skin on grafts without an additional dermal component. Hair follicle buds grafted alone or with cultured dermal cells will reconstitute skin but without appendage formation. Thus, cells or factors present in fresh, but not cultured, dermal cells are essential for supporting hair growth from budding follicles, whereas more developed (pre-formed) follicles appear to contain all the necessary components for hair formation. Dissociation of isolated hair follicles by trypsin/ethylenediaminetetraacetic acid prior to grafting is permissive for hair growth, suggesting that follicle cells can be re-induced or reassociate in vivo. Dermal papilla cells, microdissected from rat vibrissal follicles and cultured for up to 14 passages, stimulate hair growth from follicle buds and influence the quality of hair growth from pre-formed hair follicles. Thus, dermal papilla cells maintain inductive capacity in culture and contribute to the reconstituted skin. This reconstitution model should be useful for identifying cell populations within the hair follicle compartment necessary for hair growth and for examining the effects of specific gene products on hair follicle growth and development in vivo.

A comparison of interfollicular and hair follicle derived cells as targets for the v-rasHa oncogene in mouse skin carcinogenesis.

Methods to isolate and culture intact mouse hair follicles and interfollicular epidermal cells provide a model to test the potential of each to form tumors as a consequence of rasHa gene activation and to determine the risk for progression in the resultant tumors. The v-rasHa oncogene was introduced into intact or dissociated hair follicle cells or interfollicular epidermal cells from newborn mouse skin via a defective retroviral vector. Either immediately after infection or after an additional 6 days of culture, the v-rasHa cells were transferred to nude mice as a skin graft. Both cell populations formed squamous papillomas which were indistinguishable based on morphology and immunocytochemistry. All papillomas expressed epidermal specific markers whether derived from hair follicle or interfollicular cells, and many regressed. After 16 weeks in vivo, 20-30% of the benign skin tumors in all groups converted to malignancy. In addition to papillomas, hair follicle derived populations produced hemangiomas in many animals. None of the groups formed basal cell carcinomas, keratoacanthomas or tumors with characteristics of differentiating hair follicle cells. These studies indicate that ras gene activation can contribute to benign squamous neoplasia originating from several skin-derived cell types. The underlying factors which determine the variable risk for neoplastic progression of skin papillomas after ras gene activation is not simply the origin of the tumor cell from hair follicle or interfollicular epidermis. The activated ras oncogene can also transform skin endothelial cells but does not appear to directly contribute to transformation of the more differentiated cells of the hair follicle.

Growth factors specifically alter hair follicle cell proliferation and collagenolytic activity alone or in combination.

A three-dimensional culture model for isolated murine pelage hair follicles in a type I collagen gel has been utilized to study the effects of selected growth factors on follicle cell proliferation and release of collagenolytic factors. Cultured follicle organoids differentially express cytokeratins 6 and 14 in a pattern suggesting they contain cells of the outer root sheath, inner root sheath and follicle matrix. Using incorporation of [3H]thymidine as a measure of proliferation, follicle organoids show a peak of DNA synthesis between day 1 and 5 of culture, depending on plating density, and then have a low rate of DNA synthesis. Thymidine incorporation is stimulated by transforming growth factor-alpha (TGF-alpha) in a dose-dependent response. Only peripheral cells presumably of the outer root sheath, incorporate thymidine in basal or stimulated conditions. TGF-beta 1 and TGF-beta 2 inhibit constitutive cell proliferation and oppose growth stimulation by TGF-alpha. Hair follicles lyse the collagen gel matrix when exposed to certain cytokines. Epidermal growth factor (EGF) and TGF-alpha stimulate gel lysis, but TGF-beta 1, TGF-beta 2 and cholera toxin do not. Other skin-derived cells, such as interfollicular epidermal cells, dermal fibroblasts, or combinations thereof, do not lyse gels in this culture model even when exposed to growth factors. Combinations of EGF or TGF-alpha with TGF-beta 1 or TGF-beta 2 are synergistic for collagenase release. These cytokines stimulate release of multiple species of matrix metalloproteinases, but the 92-kDa and 72 kDa type IV procollagenases and their activated derivatives predominate on zymograms. In cytokine-stimulated follicles, both peripheral and centrally located cells in the organoids express the 72-kDa type IV collagenase and a similar immunostaining pattern is present in developing follicles in vivo. Thus growth factors appear to work in concert for certain hair follicle responses and in opposition for others. These combined actions may play a role in different phases of hair follicle development that require cell replication and invasion into the deeper dermis.