Distinct and nonoverlapping roles for pRB and cyclin D:cyclin-dependent kinases 4/6 activity in melanocyte survival

Abstract

Deregulation of the p16^INK4a–cyclin D:cyclin-dependent kinases (cdk) 4/6 –retinoblastoma (pRB) pathway is a common paradigm in the oncogenic transformation of human cells and suggests that this pathway functions linearly in malignant transformation. However, it is not understood why p16^INK4a and cyclin D:cdk4/6 mutations are disproportionately more common than the rare genetic event of RB inactivation in human malignancies such as melanoma. To better understand how these complexes contribute to altered tissue homeostasis, we blocked cdk4/6 activation and acutely inactivated Rb by conditional mutagenesis during mouse hair follicle cycling. Inhibition of cdk4/6 in the skin by subcutaneous administration of a membrane-transducible TAT-p16^INK4a protein completely blocked hair follicle growth and differentiation. In contrast, acute disruption of Rb in the skin of homozygous Rb^LoxP/LoxP mice via subcutaneous administration of TAT-Cre recombinase failed to affect hair growth. However, loss of Rb resulted in severe depigmentation of hair follicles. Further analysis of follicular melanocytes in vivo and in primary cell culture demonstrated that pRB plays a cell-autonomous role in melanocyte survival. Moreover, functional inactivation of all three Rb family members (Rb, p107, and p130) in primary melanocytes by treatment with a transducible TAT-E1A protein did not rescue the apoptotic phenotype. These findings suggest that deregulated cyclin D:cdk4/6 complexes and pRB perform nonoverlapping functions in vivo and provide a cellular mechanism that accounts for the low incidence of RB inactivation in cancers such as melanoma.

Transition of cells through the cell cycle is regulated by sequential activation of multiple cyclins and cyclin-dependent kinases (cdks). One of the major pathologies that occurs during oncogenesis is the deregulation of cdks governing the G₁ checkpoint (1–4). The G₁ phase of the cell cycle can be subdivided into an early and late phase controlled by cdk4/6 and cdk2, respectively. Remarkably, although the late G₁ phase is irreversible and signifies commitment to entry into S phase, it is the early phase of G₁ that is universally targeted by cancers of either sporadic or inherited origin (5, 6). Specifically, inappropriate activation of cyclin D:cdk4/6 complexes appears to be the common event in human malignancies. Chromosomal translocations and gene amplifications in lymphomas, breast cancer, and melanomas result in the overexpression of cyclin D1 and resulting deregulated cyclin D:cdk4/6 activity (7). Defective cdk4 inhibition by p16^INK4a loss of heterozygosity (LOH) or gene silencing is also a common finding in human cancer and contributes to the hereditary predisposition for pancreatic cancer and malignant melanoma. Mouse models deficient in p16^INK4a with or without loss of p19^ARF further confirm the critical role of early G₁ cell cycle machinery in malignant transformation (8).

The major target of cyclin D:cdk4/6 kinase activity is widely believed to be the retinoblastoma tumor suppressor protein (pRB), an important transcriptional regulator during the G₁ phase of the cell cycle (3). Consequently, this has led to the inference that cdk4-mediated cellular transformation promotes tumorigenesis in the same way as does deletion or inactivation of RB. This hypothesis is supported by findings that in certain cell lines p16^INK4a-mediated cell cycle inhibition depends on the presence of Rb (9, 10). However, direct evidence for the in vivo role of genetic inactivation of RB is limited to a subset of tumors including retinoblastoma, osteosarcomas, and small-cell lung carcinoma (11). In contrast, p16 loss, cyclin D amplification/overexpression leading to deregulated cyclin D:cdk4/6 activation is omnipotent. Furthermore, mutations in cyclin E:cdk2 complex, which inactivates pRB by hyperphosphorylation, are virtually nonexistent in human cancer. These observations point to complex and potentially divergent outcomes from either cdk4/6 or cdk2 activation. Insights into the different roles of cdk4 and cdk2 during cell cycle progression and pRB regulation from primary cell culture experiments reveal that phosphorylation by cdk4 and cdk2 impart different protein binding properties of pRB to the E2F family proteins (12). In early G₁, hypophosphorylation of pRB by cyclin D:cdk4/6 complexes facilitates binding of pRB to E2F, whereas hyperphosphorylation of pRB by cyclin E:cdk2 complexes in late G₁ leads to dissociation of pRB from E2F. Recruitment of pRB to E2F-regulated sites has important transcriptional consequences because pRB is involved in gene silencing and recruitment of chromatin-modifying complexes including histone deacetylase (HDAC) and Polycomb-group members (2, 4).

To clarify the roles of cyclin D:cdk4/6 activation and RB inactivation in vivo, we studied their involvement in developing skin by using both dominant-negative and loss-of-function genetic approaches. We used the skin and regenerating hair follicle model for cell cycle regulation by stimulating resting G₀ quiescent stem cell populations to enter the cell cycle in vivo by depilation (13–16). Using protein transduction technology (17), we demonstrate that inhibition of cdk4 and cdk2 activation in the skin by local injection of transducible TAT-p16 and TAT-p27 proteins completely blocked activation of the hair cycle. We further address the possible role of Rb in the hair cycle by acute inactivation of the Rb gene with TAT-Cre-mediated loxP mutagenesis (18) and show that Rb is dispensable for hair cycle activation. Unexpectedly, we find that melanocytes require functional pRB, as evidenced by severe loss of hair pigmentation, loss of melanocytes in this conditional knockout model, and the cell culture pRB-dependency of melanocytes. These findings suggest that deregulated cyclin D:cdk4/6 complexes and pRB perform nonoverlapping functions in vivo and provide a cellular mechanism that accounts for the low incidence of RB inactivation in cancers such as melanoma.

Materials and Methods

TAT Protein Construction and Isolation. The Cre recombinase coding region was PCR-amplified from pBS500 (19) and directionally cloned into XhoI–EcoRI of pTAT-HA (17). The fusion construct DNA sequence was confirmed and then transformed into BL-21(DE3)pLysS (Novagen). TAT-Cre protein was isolated from overnight cultures, sonicated in 20 mM Hepes/100 mM NaCl (pH 7.2)/1 μg/ml leupeptin/1 μg/ml aprotinin, purified on Ni-NTA column (Qiagen), and equilibrated in 10% glycerol in PBS for storage at –80°C or for injection. TAT-β-galactosidase (TAT-β-gal), TAT-GFP, and TAT-E1A proteins were produced as described (12, 20) and stored in 10% vol/vol glycerol.

Animal Housing, Maintenance, and Injections. Mice were maintained at the animal facilities of the Washington University School of Medicine under specific pathogen-free conditions according to Department of Comparative Medicine guidelines. Rb^LoxP mice have been described (21) and were bred to generate homozygous, heterozygous, and mixed litters. Mice were screened by PCR on tail DNA by using published primers Rb18 and Rb19 (21). Recombination of Rb^LoxP loci generates a 321-bp exon 19-deleted PCR fragment, and the wild-type exon 19 is 680 bp and the floxed exon 19 is 748 bp. Rb^LoxP mice were also bred to C57BL/6J-Gtrosa26^tm1Sor (ROSA26R) mice (22). Before depilation, mice were anesthetized with Avertin. TAT-fusion proteins (100–250 μl) in PBS were injected into depilated areas, equivalent to 2–5 μg total protein, along with 2 μg/ml leupeptin.

TAT-Cre Recombination. CATZ-17 cells (23) plated at 5 × 10⁴ cells per well in a 24-well dish and Histopaque-isolated (Sigma) Rb^LoxP splenocytes were treated with 2 μM TAT-Cre in media, 10% heat inactivated calf-serum, and 2 μg/ml leupeptin at 2 h intervals. Cells were assayed for recombination on the following day either by β-gal staining (20) or PCR analysis on the following day. For in vivo studies, day 5–10 postnatal mice were injected with 25–100 μl of TAT-Cre (1.25–5 μg total) and equivalent amounts of TAT-β-gal in PBS, 10% glycerol, and 2 μg/ml leupeptin. Intradermal injections in the dorsal skin were performed at the superior edge of the hindlimb. The opposite side was injected with TAT-control protein (TAT-β-gal or TAT-GFP). Injections were well tolerated, and no adverse effects were observed. Recombination in vivo was assayed by β-gal staining or by PCR of DNA isolated from whole tissue or frozen sections.

Histology and Immunohistochemistry. Longitudinal skin sections were obtained to maintain proper anteroposterior orientation and placed in 10% formalin. For immunohistochemical stains, deparaffinized sections (5–7 μm) were microwaved in water for 10 min for antigen retrieval and preincubated in 3% H₂O₂ and avidin/biotin solutions (Vector Laboratories). Sections were incubated overnight with primary antibodies at 4°C including rabbit anti-tyrosinase-related protein 2 (TRP-2) (hPEP8) (24) and rabbit anti-proliferating cell nuclear antigen (PCNA) (Santa Cruz Biotechnology) antibodies diluted 1:200. Sections were further stained with biotinylated anti-rabbit antibodies and the ABC detection kit (Vector Laboratories) and developed with 3-amino-9-ethylcarbazole (AEC) substrate (Biomeda, Foster City, CA). Heavily pigmented sections were further bleached in 10% H₂O₂ overnight at 4°C to enhance visualization. Sections were counterstained with hematoxylin and mounted. Immunofluorescent stains were performed on acetone-fixed 7 μM frozen sections by using 1:10 AE13 (25), 1:10 AE15 (26), and 1:200 anti-mouse FITC (Sigma). Modified citrulline staining was performed in 0.5% p-dimethylaminocinnamaldehyde (Sigma) in 0.5 M HCl and 2 M NaCl (27).

Whole-Mount and β-Gal Assays. Specimens as procured above were placed in 4% paraformaldehyde and incubated in X-gal staining buffer overnight. Focal areas of β-gal activity were processed as above or examined as whole tissue. To isolate epidermis and hair components, the sections were initially incubated in 2 M NaBr to retain activity of β-gal and spare antigens for immunohistochemical staining. Sections were stored in 10% glycerol or mounted.

Melanocyte Isolation and Culture. Murine epidermis was isolated from dispase-treated dorsal skin from killed P0 and P1 pups and treated with trypsin for 5 min. Dispersed cells were cultured in keratinocyte growth media (Clonetics, San Diego) for 2–3 days, then grown in melanocyte selective media (Cascade Biologics, Portland, OR). Each well of a 24-well dish containing cells from a single animal was grown for 2 weeks until <10% fibroblasts were present. Melanocytes were counted based on their morphology and pigmentation for 4 days before administration of 500 nM TAT-Cre for 2 h. Cells were counted for an additional 2 days posttreatment and viability was ultimately confirmed by trypan blue exclusion. Melanocytes were treated with TAT-E1A or TAT-p16 protein (200 nM) daily for 3 days and quantified daily. Recombination and melanocyte identity was further determined by rabbit anti-β-gal and TRP-2 antibody immunohistochemical staining diluted at 1:200. Results from melanocyte survival assays were combined from two experiments. TdT-mediated UTP nick end labeling (TUNEL) was performed by using the In Situ Cell Death Detection Kit, Fluorescein, according to manufacturer's instructions (Roche). Three independent homozygous melanocyte lines were used to calculate percent TUNEL-positive (n = 51, 40, and 72 total homozygous cells; n = 62 and 96 total heterozygous cells examined).

Results

Inactivation of cdk4/6 and cdk2 in Skin. To study the requirements for cyclin D:cdk4/6 and cyclin E:cdk2 activity in vivo, we injected transducible cdk inhibitors, TAT-p16, TAT-p27, and control TAT-β-gal proteins (13, 19, 20), into wild-type mouse skin postdepilation day 0 (Fig. 1). Analysis of hair growth on day 3 postdepilation of control TAT-β-gal-treated areas showed the expected stage of anagen hair growth (Fig. 1 A; see also 6A, which is published as supporting information on the PNAS web site). In contrast, depilated areas treated with daily and single administrations of TAT-p16 or TAT-p27 proteins demonstrated severe impairment of hair growth in all follicles (n = 10–15 follicles per low-power field). Proliferation as detected by S phase-specific expression of the PCNA and early markers of hair follicle differentiation, including trichohyalin and hair keratin, were notably absent in TAT-p16- and TAT-p27-treated regions, confirming impaired activation and differentiation, whereas regions treated with control TAT proteins showed proper expression of markers indicating IRS and hair shaft differentiation (Figs. 1 B and C and 6B). There were no noted differences morphologically between TAT-p16- and TAT-p27-treated hair follicles. Interestingly, when administered 24 h postdepilation, TAT-p16 protein was unable to prevent normal hair follicle development, as indicated by the production of the citrulline-rich medulla and IRS and by the presence of PCNA-positive matrix cells (Fig. 1C). These observations suggest that activation of the hair cycle is cyclin D-dependent during a critical temporal window.

Fig. 1.

TAT-p16 and TAT-p27 proteins block hair follicle remodeling. (A) Histology (hematoxylin/eosin staining). (B) Hair keratin and trichohyalin expression on day 3 postdepilation mice treated daily with control TAT-β-gal (Left), TAT-p16^INK4a (Center), or TAT-p27^KIP1 (Right) proteins. A Left shows follicular down-growth (dotted outlines) and envelopment of the dermal papilla by the hair bulb (arrows) consistent with anagen II as are the expression of hair keratin and trichohyalin, markers of matrix and inner root sheath (IRS) differentiation. (C) PCNA immunostaining of day 2 mice and citrulline stain of IRS and medulla in day 3 postdepilation mice treated with single-injection control TAT-β-gal or TAT-p16^INK4a (Left and Center) proteins beginning on day 0 or TAT-p16^INK4a proteins on day 1 (Right). (Scale bar, 50 μm.)

TAT-Cre Recombination. pRB is believed to be a major target of cyclin D:cdk4/6 kinase activity and therefore, predictably, pRB inactivation should mimic activation of cyclin D:cdk4/6 kinases. Although the inactivation of cyclin D complexes in skin could be achieved by treatment with transducible sequestering proteins, pRB sequestering proteins, such as E1A, T Ag, and E7, also bind other pRB family members, namely p107 and p130 (1–4), and would thereby potentially obscure the role of pRB. Therefore, to specifically examine pRB's role in hair follicle cycling, we used a genetic mouse model of Rb where exon 19 of Rb is flanked by LoxP recombination sites, Rb^LoxP/LoxP (21). To recombine Rb^LoxP/LoxP, we generated a transducible TAT-Cre recombinase (Fig. 2A). TAT-Cre protein treatment of murine fibroblasts in culture and subcutaneous injection into ROSA26R reporter mice harboring Cre-activated LacZ transgenes (Fig. 2 A Lower) resulted in significant levels of recombination (Fig. 2B). Importantly, persistent lacZ expression was also detected in the epidermis, hair, dermis, muscle, and adipose tissue of ROSA26R mice 4–6 weeks postinjection of TAT-Cre, indicating the stable propagation of the recombined allele in both proliferating and terminally differentiated cellular compartments (Fig. 2C). Thus, TAT-Cre allows for recombination of LoxP-flanked DNA segments in multiple cell types in culture and in vivo.

Fig. 2.

Activation of lacZ^LoxP reporter gene in cells and tissues by transducible TAT-Cre recombinase. (A) Diagram of TAT-Cre fusion protein (Upper), containing His-purification tag, TAT protein transduction domain, HA-epitope tag, and Cre coding sequence. (Lower) Diagram of Cre-activated lacZ^LoxP reporter gene. Recombination of the LoxP sites results in excision of LoxP-Stop-LoxP DNA segment and expression of lacZ (β-gal). (B) β-gal staining of fibroblast lacZ^LoxP reporter cells (Left)or ROSA26R lacZ^LoxP mice (Right) treated with control or TAT-Cre protein. (C) Stable expression of lacZ reporter gene in ROSA26R mice 6 weeks after single TAT-Cre treatment was observed in basal keratinocytes, fibroblasts, and the hair follicle. (Scale bar, 20 μm.)

TAT-Cre-Mediated Acute Inactivation of Rb in Skin. We next tested the ability of transducible TAT-Cre protein to faithfully recombine Rb in cells from Rb^LoxP/LoxP mice (Fig. 3A). Treatment of splenocytes and skin in vivo from homozygous Rb^LoxP/LoxP and heterozygous Rb^LoxP/+ mice with TAT-Cre protein resulted in excision of exon 19 and detection of the 321-bp recombined fragment as early as 12 h postadministration (Fig. 3B). Treatment of cells and tissues from Rb^+/+ mice with TAT-Cre protein showed no alteration of exon 19, as determined by PCR analysis. Sequence analysis confirmed the retention of a single, complete LoxP recombination site in TAT-Cre-treated Rb^LoxP/LoxP cells (data not shown). Thus, the Rb^LoxP/LoxP locus is recombinable by TAT-Cre in vitro and in vivo.

Fig. 3.

In vivo recombination of Rb^LoxP/LoxP leads to localized depigmentation. (A) Schematic of Rb^LoxP locus. (B) PCR of Rb^LoxP and wild-type splenocytes and skin treated with TAT-Cre or control protein. TAT-Cre-specific recombination generates a 321 base pair fragment in Rb^LoxP mice. The wild -type locus is 680 bp, and the unrecombined Rb^LoxP locus is 748 bp. (C) Dorsal skin (Upper) and hair shafts (Lower)of Rb^LoxP/LoxP mice treated with control TAT-β-gal (Left) or TAT-Cre (Right) proteins showed normal morphology, but hair shafts from TAT-Cre-treated mice were deficient in melanin (Lower Left). (D) Hair pigmentation in control (Left) and TAT-Cre-treated homozygous Rb^LoxP/LoxP (Center) mice and heterozygous Rb^loxP/+ mice (Right).

Subcutaneous administration of a single dose of TAT-Cre or control TAT-β-gal proteins into small, demarked areas of Rb^LoxP/LoxP P10 (postnatal day 10) mice resulted in normal timing of hair growth and hair follicle activation during the postnatal cycle in both control and experimental regions (Fig. 3C). In addition, hair length and diameter were similar in both control and Rb deficient regions, indicating that normal growth had occurred (data not shown). However, despite showing normal hair morphology, the hair shafts of TAT-Cre-treated Rb^LoxP/LoxP mice were severely depigmented in 11 of 11 mice (Fig. 3C Right). In contrast, hair pigmentation was normal in the skin of Rb^LoxP/LoxP mice treated with control TAT-β-gal protein (Fig. 3C Left). Although the Rb^LoxP allele in heterozygous Rb^LoxP/+ mice is susceptible to recombination (Fig. 3B), four of four Rb^LoxP/+ mice treated with TAT-Cre protein developed normally pigmented hair shafts (Fig. 3D Right), confirming the specificity of the phenotype to homozygous deletion of Rb. Importantly, this observation excluded the potential contribution of recombination-mediated activation of a DNA damage checkpoint or recombination at cryptic LoxP sites (28, 29).

pRB-Dependency of Melanocyte Survival. Hair pigmentation occurs as the result of melanin transfer from follicular melanocytes into the matrical keratinocytes and subsequently into the hair shaft. Thus, abnormalities in hair pigmentation could arise from a melanocyte defect or defective hair morphogenesis. Although we observed sporadic abnormalities in the skin, including the presence of calcified hair remnants, keratinous cysts, and dilated follicular infundibulum (data not shown), these findings were infrequent and could not account for the highly penetrant hair depigmentation phenotype. Because of the nature of the Rb exon 19 deletion, we were unable to directly test the in situ loss of pRb. To aid in the identification of recombinant tissue, we crossed Rb^LoxP/LoxP mice into ROSA26R mice (22) to introduce a lacZ reporter for TAT-Cre-mediated recombination. Consistent with the observations above, morphological examination of hair shafts from TAT-Cre-treated homozygous Rb^LoxP/LoxP/ROSA26R and heterozygous Rb^LoxP/+/ROSA26R mice by histological examination and whole-mount preparations demonstrated normal anagen morphology and positivity for lacZ expression (Fig. 4A). These observations demonstrate that there is no negative selection against Rb-deficient cells and that, excluding the absence of pigmentation, hair shaft morphogenesis occurs in a pRB-independent and cyclin D:cdk4/6-dependent fashion.

Fig. 4.

TAT-Cre-mediated inactivation of Rb results in depletion of melanocytes. (A) Hair shafts from TAT-Cre-treated Rb^LoxP/LoxP; ROSA26R and Rb^LoxP/+; ROSA26R mice show activation of lacZ reporter, but pigment granule loss is specific to homozygous mice. (B) TRP-2-positive melanocytes were absent from TAT-Cre-treated Rb^LoxP/LoxP mice but present in control TAT-β-gal-treated homozygous Rb^LoxP/LoxP mice and TAT-Cre-treated heterozygous Rb^LoxP/+ mice. PCNA-positive bulb cells confirms proliferative anagen hair follicles in control and TAT-Cre-treated homozygous and heterozygous mice. (Scale bar, 20 μm.)

To determine the presence of melanocytes, we examined the expression of a melanocyte-specific marker, TRP-2 (29). TAT-Cre treatment of heterozygous Rb^LoxP/+ mice retained TRP-2-positive melanocytes in the hair follicle (Fig. 4B Upper Right), suggesting that retention of a single Rb allele is sufficient for maintenance of follicular melanocytes during development. In contrast, TRP-2-positive melanocytes were absent from hair follicles of Rb^LoxP/LoxP mice treated with TAT-Cre protein (Fig. 4B Upper Center). However, TRP-2-positive melanocytes were readily detected in skin sections from the same Rb^LoxP/LoxP mice treated with control TAT-β-gal protein (Fig. 4B Upper Left). Activation of melanogenesis and recruitment to the hair follicle is tightly associated with the proliferative phase of the hair cycle. Therefore, to include/exclude this developmental cue as mechanistic, we analyzed skin sections for the presence of PCNA expression in the hair matrix. Regardless of the genetic composition or the type of TAT-fusion protein treatment, all hair follicles were PCNA-positive (Fig. 4B Lower). These observations demonstrated that pRB was essential for the specific maintenance of follicular melanocytes in vivo.

β-gal positivity appeared to underestimate the affected area of depigmentation. Although consistent with previous observations related to locus and tissue-specific recombination (30), we could not rule out a non-cell-autonomous effect of melanocyte survival. Melanocyte patterning and survival is highly dependent on cues from the epidermis and hair follicle, including those of the stem cell factor (SCF)/c-kit and hepatocyte growth factor (HGF)/c-met receptor signaling pathways (31). To determine whether melanocyte survival was due to a cell-autonomous requirement for pRB versus nonautonomous effects of stromal cells and/or environmental changes, primary murine melanocytes were isolated from homozygous Rb^LoxP/LoxP and heterozygous Rb^LoxP/+ mice, cultured and treated with a single dose of TAT-Cre protein on day 4 (Fig. 5A). Homozygous deletion of Rb in Rb^LoxP/LoxP melanocytes resulted in a dramatic decrease in melanocyte viability, whereas loss of a single Rb allele from heterozygous Rb^LoxP/+ melanocytes had little to no effect on the cultures. The fate of the Rb^LoxP/LoxP melanocytes was confirmed by TUNEL DNA fragmentation assay that indicated loss of viability due to programmed cell death (Fig. 5B). Consistent with the observations above and those of Deng et al. (32) in melanoma cells, functional inactivation of all three pocket proteins (Rb, p107, and p130) in primary murine melanocytes by treatment with a transducible TAT-E1A protein did not rescue the apoptotic phenotype (Fig. 7 A and B, which is published as supporting information on the PNAS web site). In contrast, functional inactivation of cyclin D:cdk4/6 complexes by treatment with TAT-p16 protein resulted in cell cycle arrest and increased cell size. Taken together, these observations argue that pRB function is required for melanocyte autonomous survival.

Fig. 5.

Cell autonomous requirement for pRB. (A) Primary murine melanocytes isolated from Rb^LoxP/+ and Rb^{LoxP /LoxP} mice were quantified daily before and after a single TAT-Cre treatment on day 4. Nonviable melanocytes were confirmed by trypan blue exclusion and by detection of DNA fragmentation by TUNEL assay (B). Average percent TUNEL-positive cells of three homozygous primary melanocyte lines and two heterozygous melanocyte lines treated with TAT-Cre are detailed in A.

Discussion

The current dogma of G₁ phase cell cycle regulation is that alteration of cyclin D:cdk4/6 activity in cancer leads to deregulated proliferation by inactivation of pRB. Thus, genetic events that lead to the deregulated activation of cyclin D:cdk4/6 should theoretically lead to a phenotype similar to genetic inactivation of RB. Here we examined this question in an in vivo hair follicle cycling mouse model and find that cyclin D:cdk4/6 and Rb have distinct and nonoverlapping functions.

Regeneration of the hair during the hair cycle is a complex process involving multiple cell–cell interactions, homeostatic mechanisms controlling cell death, and recruitment of additional cell lineages including melanocytes and endothelial cells (33). Activation of cdk4 and cdk2 proved to be essential for the initiation of the hair cycle. Injection of either TAT-p16 or p27 was sufficient to block regeneration of the hair after depilation. Only a single injection of TAT-p16 was required to block initiation of the hair cycle when given on day 0 of depilation. However, if injected 24 h later, TAT-p16 was not able inhibit the hair cycle. This may reflect the in vivo avidity of p16^INK4a for nascent and unbound cdk4/6 in the absence of cyclin D expression has been shown in previous biochemical studies compared with the more difficult problem of overcoming cdk4/6 activation in cycling cells that are continuously synthesizing cyclin D (34). However, inappropriate activation of cdk4/6 alone is probably not sufficient to circumvent the many signals required for initiation of the hair cycle. In transgenic studies using the keratin 5 promoter, overexpression of cyclin D1–3 or cdk4 did not appear to promote hair growth or activation despite having profound effects on the epidermis (35). In addition, other costimulatory signals necessary for mobilization of follicular stem cells may also be affected by cdk4 inhibition and thereby lead to an absence of hair follicle cycling.

pRB is thought to be principal target of cyclin D:cdk4/6 complexes and, therefore, we studied the requirements of Rb in the hair follicle cycling. However, because of embryonic lethality associated with Rb deficiency (36, 37), we used a local, subcutaneous injection of TAT-Cre protein into in Rb^LoxP/LoxP mice. Surprisingly, Rb loss did not affect hair growth or hair follicle cycling. Similarly, transgenic mice expressing E1a, which sequesters and functionally inactivates all Rb-family members (pRb, p107, and p130), in the epidermis had little effect on cell proliferation (38). Recently, Ruiz et al. (39) reported defects in doubly mutant p107/p130 mice that include defects in skin differentiation and hair morphogenesis and may play a role in the hair cycle. However, here we show that specific inactivation of Rb in the mouse skin revealed a previously unrecognized role for Rb in the survival of melanocytes.

Somatic deletion of Rb resulted in depigmentation of hair shafts as a consequence of melanocyte loss, and in culture, we found that Rb was required for melanocyte survival. Neither single or double mutant p107/p130 mice have been reported to have pigmentation defects. Similarly, mice deficient in p16^INK4a, cdk4, or D-type cyclins are viable but do not have hair follicle pigmentation defects (40–43). Thus, in melanocytes, inactivation of Rb and activation of cyclin D:cdk4/6 do not appear to have overlapping functions. Melanocyte defects have also not been reported in Rb-deficient chimeric analysis (44) or in animals derived by placental rescue (45). It is possible that during embryonic development, loss of Rb can be complemented for by altered gene expression, whereas postnatal melanocytes that are exquisitely sensitive to Rb deficiency lack the machinery to compensate for acute Rb loss.

The requirement for Rb in melanocyte survival has not previously been examined and provides a basis for understanding the infrequency of Rb mutations in melanoma. Consistent with the findings in sporadic and familial melanoma, where p16^INK4a mutations are frequent and RB inactivation is rare (5, 6), these observations further suggest that physiologic deregulation of cyclin D:cdk4/6 is not equivalent to genetic inactivation of RB and that cyclin D:cdk4/6 may have additional functions (12, 46, 47). Altered homeostasis of malignancies where cdk4 is activated may reflect epigenetic changes such as altered cellular senescence distinct from pRB function. Our observations demonstrate a critical role for pRB in the cell-autonomous survival of melanocytes and potentially account for the observed limited role of RB inactivation during oncogenesis. The role of RB in melanocyte survival may provide clues to a potential vulnerability of melanomas to targeted inactivation of pRB.