Introduction

The TGF superfamily consists of more than 40 structurally related proteins that can be divided into several subfamilies, including TGF, bone morphogenetic proteins (BMPs) and activins/inhibins (Massague, 1998; Derynck et al., 2001; Wakefield and Roberts, 2002). TGF signaling plays many important functions in numerous biological processes. Some of the areas that are subjected to extensive investigation include skin development and neoplasia. Many members of TGF superfamily and their receptors are expressed in skin and developing hair follicles (Li et al., 2003a). It has been shown that alterations in BMP signaling pathway resulted in abnormal hair follicle growth (Kobielak et al., 2003; Li et al., 2003a; Andl et al., 2004; Ming Kwan et al., 2004; Yuhki et al., 2004). In addition, alterations in TGF signaling result in skin tumorigenesis (Wang, 2001). For instance, targeted disruption of TGF1 resulted in tumorigenesis of v-rasHa oncogene-transfected keratinocytes after they were grafted into nude mice (Glick et al., 1994). Consistently, transgenic mice expressing a dominant-negative type II TGF receptor exhibited epidermal hyperproliferation (Wang et al., 1997) and were more sensitive to chemically induced carcinogenesis (Amendt et al., 1998; Go et al., 1999). Benign hair follicle tumors were also observed in mice carrying a targeted disruption of BMP receptor 1A (BMPR1A) (Andl et al., 2004; Ming Kwan et al., 2004).

SMAD proteins, which constitute a family of eight members (SMAD1–8), serve as intracellular mediators for TGF signaling (Heldin et al., 1997; Massague, 1998; ten Dijke and Hill, 2004). SMAD2 and 3 respond to TGF and activins, SMAD1, 5 and 8 function in BMP signaling pathway, while SMAD4 is a common mediator for both signaling pathways (Heldin et al., 1997; Massague, 1998; ten Dijke and Hill, 2004). Loss of function mutations of SMAD4 has also been detected in pancreas cancer, colon cancer, gastric polyposis and adenocarcinomas (Hahn et al., 1996; Friedl et al., 1999; Howe et al., 1998). In mouse, loss of Smad4 results in lethality at mouse embryonic E6–7 due to impaired extraembroynic membrane formation and decreased epiblast proliferation (Sirard et al., 1998; Yang et al., 1998). Using the Cre-loxP-mediated tissue specific knockout approach to overcome embryonic lethality, we showed that mice lacking Smad4 in their mammary glands displayed abnormal differentiation of mammary epithelium and tumorigenesis (Li et al., 2003b). Meanwhile, Smad4 heterozygous mice developed gastric polyposis and cancer (Redman et al., 2005; Takaku et al., 1999; Xu et al., 2000), and one out of 16 Smad4+/- mice also developed squamous cell carcinomas (SCCs) of the skin (Redman et al., 2005), suggesting that Smad4 also functions as a tumor suppressor in the mouse.

We have previously studied expression of Smad1–7 in mouse skin. Our data indicated that Smad1–5 were expressed in all layers of the epidermis, including both the basal proliferative keratinocytes and suprabasal differentiated keratinocytes, whereas Smad6 and 7 were barely detectable in the epidermis (He et al., 2001). In the hair follicle, expression of Smad2 and 3 was quite weak, while Smad1, 4 and 5 were detected with intensities similar to those in the epidermis. Among these Smads, Smad4 expression was strongest in both epidermis and hair follicles, suggesting that Smad4 plays important roles in skin development and homeostasis. When we bred MMTV-Cre transgenic mice (Wagner et al., 1997) with Smad4-conditional mutant (MT) mice to study the role of Smad4 in mammary glands (Li et al., 2003b), we found that mice with Smad4 deletion also developed abnormalities in skin. In the current study, we further confirmed that MMTV-Cre is also expressed in the epidermis and hair follicles. Therefore, we investigated the role of Smad4 in skin development and differentiation. Our data showed that skin-specific disruption of Smad4 impaired differentiation of hair follicles, and increased proliferation of keratinocytes of the epidermis, consequently leading to malignant tumor formation. Some of these phenotypes are significantly different from the BMPR1A skin-null mice, where the abnormalities were mainly restricted to hair follicle development and benign tumor formation (Kobielak et al., 2003; Li et al., 2003a; Andl et al., 2004; Ming Kwan et al., 2004; Yuhki et al., 2004). Most notably, we found that the tumorigenesis is accompanied by inactivation of phosphatase and tensin homolog deleted on chromosome 10 (Pten) tumor suppressor, leading to the activation of PI-3/AKT pathway.

Results

Cre-mediated inactivation of Smad4 in skin resulted in progressive hair loss

Previous studies indicated that MMTV-Cre is expressed in the skin (Wagner et al., 1997, 2001). Using a Rosa-26 reporter strain (Soriano, 1999), we confirmed this observation and also showed that Cre activity was gradually increased in the skin of postnatal mice. At the early postnatal (P) skin, -galactocidase-positive keratinocytes were mainly detected in the differentiated layers of the epidermis and hair follicles (Supplementary Figure 1a). When examined at later stages (from P16 to P30), -galactocidase-positive keratinocytes expanded to the proliferative layers of the epidermis and hair follicles (Supplementary Figure 1b). Cre-mediated excision of Smad4 exon 8 was also detected in skin of Smad4^Co/CoMMTV-Cre mice, but not in Smad4^Co/+and Smad4^Co/Co mice by PCR (Figure 1a and b). Western blot analysis revealed significantly decreased Smad4 protein in MT skin compared with control skin (Figure 1c).

Figure 1.

Targeted disruption of Smad4 in skin resulted in progressive hair loss. (a) Diagram of Smad4 conditional allele and deletion allele. (b) MMTV-Cre promotes deletion of Smad in skin revealed by PCR. The age and genotype of the smapls were as indicated. Primers a/c amplify about 500 bp from the recombined allele. Primers a/b amplify about 450 bp from conditional allele. The sequences of the primers are: a, 5'-GACCCAAACGTCACCTTCAG-3'; b, 5'-GGGCAGCGTAGCATATAAGA-3'; and c, 5'-AAGAGCCACAGGTCAAGCAG-3'. Cre primers are: Cre-1 (5'-CCT GTT TTG CAC GTT CAC CG-3') and Cre-2 (5'-ATG CTT CTG TCC GTT TGC CG-3'). (c) Western blot analysis. (d–g) Progressive hair loss found in Smad4 MT (e–g), but not in control (d) mice. Ages of the mice are P21 (d, e), 4 month (f), and 6 month (g).

Full figure and legend (269K)

Inspection of Smad4^Co/CoMMTV-Cre mice before P16 did not reveal obvious abnormalities. However, the MT mice gradually displayed rough fur and decreased hair density (Figure 1e) compared with control mice (Figure 1d). This phenotype became progressively severe when animals were getting older (Figure 1f). MT mice of both genders exhibited similar abnormalities, while all control mice (Smad4^Co/Co, Smad4^Co/+ and wild type (WT)) were normal. We have also followed hair growth of MT mice that carried two copies of MMTV-Cre and found that these mice displayed more extensive hair loss and, in more severe cases, became completely bald (Figure 1g), perhaps due to more efficient deletion of Smad4 in hair follicles.

Loss of Smad4 resulted in hyperproliferation of keratinocytes and epidermal hyperplasia, and aberrant hair follicle cycling

To identify the causes for hair loss, we studied skin of Smad4^Co/CoMMTV-Cre mice at different stages of hair cycle. Histologic analysis of P7 and younger Smad4 MT mice did not detect observable defects (data not shown). However, examination of skin isolated from P10 and older Smad4^Co/CoMMTV-Cre mice revealed progressive epidermal hyperplasia and abnormal hair follicle differentiation. At P10, both control and MT hair follicles were in anagen phase, with the MT hair follicles slightly bigger than controls (Figure 2a and b). At this stage, although majority of BrdU-positive cells in both control and Smad4 MT hair follicles were in the bulb region, more BrdU-positive cells were found in the outer root sheath (ORS) (arrowheads, Figure 2b). Another notable change was that many Smad4 MT hair follicles did not contain hair shafts. The epidermis of MT skin was significantly thicker, with about five-fold more BrdU-positive cells, on the average, than control epidermis (Figure 2a and b; Student's t-test, P<0.0001), indicating increased proliferation of epidermal keratinocytes. At P17, hair follicles in the dorsal skin of WT mice entered the catagen stage, as characterized by diminished BrdU-positive staining (Figure 2c). However, the epidermis of MT skin was significantly thicker, and contained more BrdU-positive cells than that of control (Figure 2d), suggesting that Smad4^Co/CoMMTV-Cre skin did not enter into the catagen phase. The MT skin also contained abnormally enlarged hair follicles (arrow), and sebaceous glands (arrowheads). At P20, WT skin was very thin and had entered the telogen phase (Figure 2e). MT skin still contained abnormal hair follicles even with some BrdU-positive cells (Figure 2f). This observation indicates that MT hair follicles were only partially regressed. At P25, WT skin had finished the first hair cycle and re-entered into the anagen phase (Figure 2g). In contrast, MT skin did not enter the cycle of regeneration (Figure 2h), but exhibited many features observed in earlier stages, such as abnormal hair follicles (i.e. empty cysts without hair, arrowhead), enlarged sebaceous glands (arrow), missing the hair shaft and epidermal hyperplasia.

Figure 2.

Histology and BrdU labeling of control (a, c, e, g) and MT (b, d, f, h) skin. Ages of samples are P10 (a, b), P17 (c, d), P21 (e, f) and P25 (g, h). Arrows point to the epidermis (b), and to the enlarged sebaceous glands (d, h). Arrowheads in (b) point to the ORS, and in (d, f, h) point to the abnormal hair follicles. Ep: epidermis; hf: hair follicle; hs: hair shaft. BrdU-positive cells in five arbitrary areas in the epidermis were counted using a microscope and subjected to Student's t-test analysis. The average number of BrdU-positive cells of MT vs control was: P10: 371.33 vs 71.67 (P<0.0001), and P17: 242.33 vs 40.33 (P<0.0001).

Full figure and legend (955K)

Smad4-deficient hair follicles did not exhibit normal differentiation

The above analyses indicated that hair cycling of the MT hair follicles was blocked at the end of the first hair cycle. The morphology of MT hair follicles suggests that they suffered abnormal differentiation. To determine if this is the case, we assessed the differentiation status of hair follicles using molecular markers. We first examined expression of K14 that labels basal layer keratinocytes of the epidermis and ORS. Our data showed that the MT hair follicles displayed significantly increased layer of K14-positive cells than controls (Figure 3a and b). K14 staining also revealed increased thickness of the epidermis (Figure 3b). We next examined expression of K6, which is an early differentiation marker for the hair follicles. K6 staining was restricted to the companion layer of the ORS in control follicles (Figure 3c), but was aberrantly expressed in all layers of the ORS in the MT hair follicles as well as interfollicular epidermis (Figure 3d). This staining pattern is consistent with previous reports that aberrant K6 expression is associated with hyperproliferation. In contrast, K10, a differentiation-specific keratinocyte marker that is expressed only in cells committed to terminal differentiation of the interfollicular epidermis, was expressed with a normal pattern in the MT epidermis (not shown).

Figure 3.

Targeted disruption of Smad4 blocks hair follicle differentiation. Antibodies used are as indicated. (a, c, e, g) are control and (b, d, f, h) are Smad4 MT mice. Ages of samples are P21 (a, b), and P17 (a, b, c–f). The inset in (b) shows that K14 expression has expanded into the superbasal layer in the MT epidermis.

Full figure and legend (815K)

Next, we examined expression of a number of markers that are expressed during hair follicle differentiation to further characterize phenotypes of the MT hair follicles. We first checked expression of AE13 and AE15. In control hair follicles, AE15 normally stains the cortical and medulla cells of the hair shaft, whereas AE13 stains the inner root sheath (IRS) (Figure 3e and g). However, the MT hair follicles were negative for these markers (Figure 3f and h). Thus, the Smad4-deficient hair follicles were missing the hair shaft, the IRS and the cortical and medulla cells. We then determined expression of Lef1, an HMG box transcription factor that is expressed in the hair matrix and the precursor cells of the hair shaft (Figure 4a), and is essential for hair growth (Kratochwil et al., 1996; DasGupta and Fuchs, 1999). In MT mice, we found that Lef1 staining was negative, suggesting the lack of such progenitors in the MT hair follicles (Figure 4b).

Figure 4.

Alteration of gene expression due to targeted disruption of Smad4. (a, b) Expression of Lef1 in P17 control (a) and Smad4 MT (b) skin. (c, d) Analysis of gene expression in WT, MT skin and tumors by quantitative PCR using primers as indicated. All qRT–PCR results shown are averaged from 3–5 samples in each group.

Full figure and legend (265K)

Transcriptional alterations resulted from a block of Smad4-mediated signaling pathways

It has been shown recently that targeted disruption of BMPR1A in the hair follicle results in upregulation (Kobielak et al., 2003), or no significant alteration of Lef1 (Andl et al., 2004; Ming Kwan et al., 2004). However, here we show that Lef1 expression is significantly downregulated in the Smad4 MT hair follicles (Figure 4b). We suspect that it is because the absence of Smad4, a common mediator for TGF superfamily, has a broader effect than the absence of BMPR1A, which only mediates signals of the BMP subfamily. To investigate this, we performed quantitative PCR analysis on samples from Smad4 MT and control skin. We first used primers for Smad4 in order to evaluate the extent of Smad4 deletion, and detected markedly reduced Smd4 transcription in Smad4^Co/CoMMTV-Cre skin (Figure 4c). Next, we evaluated transcripts of Smad6 and 7, which are immediate downstream targets of BMP and TGF signaling, respectively (Nagarajan et al., 1999; Ishida et al., 2000). Our data revealed significant downregulation of both genes (Figure 4d), confirming that Smad4 deficiency impaired both TGF and BMP signals. Consistently, our further analysis on several additional downstream targets of TGF (p21) and BMP (Id1) indicated that their expression levels were significantly decreased (Figure 4d). Altogether, these data indicate that the targeted ablation of Smad4 resulted in a block of both TGF and BMP signaling pathways.

Absence of Smad4 resulted in SCC formation

We have followed more than 30 Smad4^Co/CoMMTV-Cre mice for the long-term effect of Smad4 mutation and found that they all developed various numbers of nodules on their skin. In severe cases, these nodules covered the entire skin (Figure 5a). Histological analysis on MT skin revealed that the nodules invariably contained abnormally folding, hair canels and cysts (Figure 5c) and increased thickness (Figure 5d) of the epidermis compared with control (Figure 5b). MT skin of some mice also exhibited aberrant defective dilated hair follicles containing keratinaceous debris (Figure 5e). Additionally, some nodules also contained increased size and number of sebaceous glands (Figure 5d).

Figure 5.

Targeted disruption of Smad4 results in skin malformation. (a) A 1-year-old Smad4 MT mouse. The boxed area is amplified and placed on the right panel. (b–e) Histology of the skin of 6-month-old control (b), and MT (c–e) mice.

Full figure and legend (409K)

All MT mice developed visible skin tumors at ages ranging from 3 to 13 months (Figure 6a). Histological analysis on 20 tumors revealed that the majority were SCCs (Figure 6b). Other types of tumors, such as sebaceous adenoma (Figure 6c), basal cell carcinomas, and tricoepithelomas, appeared infrequently (not shown).

Figure 6.

Targeted disruption of Smad4 results in skin tumor formation. (a) A curve showing the percent of tumor-free Smad4^Co/CoMMTV-Cre mice. Of 21 MT mice shown, only three were female because majority of them also develop mammary tumors and therefore were not used in this study. No skin tumors were observed in over 20 WT control mice. (b, c) Histology of an SCC (b) and a sebaceous adenoma (c). (d–h) Staining of SCCs using antibodies as indicated. K14 was in red; K8 or K13 was in green in (g, h).

Full figure and legend (580K)

Our gene expression study indicated that tumor cells only expressed Smad4 at a background level (Figure 4c) and exhibited impaired expression of a number of genes downstream of TGF/Smad4 (Figure 4d). To further characterize the tumors, we performed immunohistochemical analysis using several markers for keratinocytes. Our data revealed that most tumors were positive for K14 (Figure 6d) and K6 (Figure 6e), but negative for K10 (Figure 6f) and K1 (data not shown), suggesting that these tumors have passed the benign papilloma stage. Staining with K8, which is a marker for late-stage poorly differentiated SCCs (Santos et al., 1997), revealed a widespread expression pattern that largely overlaps with K14 (Figure 6g), while expression of K13, a marker for early-stage SCC progression that is lost at late-stage SCCs (Santos et al., 1997), was detected with a patchy pattern (Figure 6h).

Increased cell proliferation and activation of AKT observed in tumor cells

Concomitant to the block in the differentiation, tumor cells exhibited significantly increased proliferation as revealed by BrdU labeling (Figure 7a) and cyclin D1 (Figure 7b), which is known to promoter cells from G1 to enter and complete the S phase. It has been shown that tumor suppressor Pten induces G1 arrest by decreasing the level and nuclear localization of cyclin D1 (Weng et al., 2001; Radu et al., 2003). Next, we investigated whether Smad4 deficiency could affect the signaling pathway mediated by Pten. Using an antibody that detects phosphorylated Pten (pPten), an inactivated form of the protein, we detected significantly increased level of Pten phosphorylation in all the tumors (n=10) (Figure 7c). Comparable elevated levels of phosphorylated form of AKT (pAKT) were also observed in all of the tumors examined (Figure 7d). Of note, our Western blot analysis revealed that expression of both the pPten and unphosphorylated form of Pten was also increased in tumors, but not in WT and normal MT skin (Figure 7f). Thus, the tumorigenesis in Smad4-deficient skin, at one hand, requires inactivation of Pten, but, on the other hand, increases total levels of Pten proportionally. While the genetic interaction between Smad4 and Pten remains elusive, the Western blot confirmed the increased level of pAKT in extracts from tumors but not from WT and normal MT skin (Figure 7f). This observation suggests that Smad4 deficiency by itself is not sufficient to activate PI-3/Akt pathway, but facilitates activation of this pathway if there are additional genetic/epigenetic alterations. The synergistic effect of these alterations with Smad4 deficiency may play a causal role in skin tumor formation through increasing nuclear accumulation of cyclin D1 and other unidentified factors in Smad4^Co/CoMMTV-Cre keratinocytes.

Figure 7.

Increased cell proliferation and activation of AKT in the tumors. (a–e) Immunohistochemical staining of skin tumors using antibodies as indicated, except for panel (e), where first antibody was not used. (f) Western blot analysis on extracts from tumors, WT and MT skin using antibodies as indicated.

Full figure and legend (669K)

Discussion

Our analysis indicated that Smad4-deficient hair follicles did not differentiate after the first hair cycle. The hair shaft progenitor cells are derived from stem cells in the bulge (Alonso and Fuchs, 2003). These cells cease proliferation when they rise up in the follicle beyond the level of the dermal papilla, and differentiated into the hair shaft and the IRS. Analysis of MT mice generated by gene targeting has revealed the involvement of many molecules in the formation of hair follicle, including members or downstream mediators of Wnt and TGF families (Fuchs et al., 2001; Stenn and Paus, 2001; Li et al., 2003a). We show here that expression of Lef1 in the MT hair follicles was reduced, suggesting that Smad4 signaling is required for maintaining expression of this gene in the hair shaft progenitor cells. Previous investigations indicated that Lef1 plays an essential role for hair follicle morphogenesis and targeted disruption of Lef1 blocked hair formation (Kratochwil et al., 1996; DasGupta and Fuchs, 1999). We found that, at the end of the first cycle, Smad4^Co/CoMMTV-Cre hair follicle did not contain matrix and precortex cells, leading to the absence of the hair shaft and the IRS in the MT hair follicles. Thus, the regeneration of Smad4-deficient hair follicle is blocked after the first hair cycle. Our further analysis suggests that the block to differentiation may cause abnormal expansion of the basal layer of keratinocytes or the ORS of hair follicles (Figure 3), which eventually results in the poorly differentiated SCC formation (Figure 6).

Previous investigations indicated that targeted disruption of BMPR1A also results in hair follicle defects and benign hair follicle tumors (Kobielak et al., 2003; Andl et al., 2004; Ming Kwan et al., 2004; Yuhki et al., 2004). A comparison between Smad4 MT and BMPR1A MT skin revealed some significant differences. First, in Smad4-deficient hair follicles, expression of Lef1 is diminished (Figure 4h), whereas, in the BMPR1A-null hair follicles, Lef1 is maintained or even increased (Kobielak et al., 2003; Andl et al., 2004; Ming Kwan et al., 2004; Yuhki et al., 2004). This observation suggests that Smad4-, but not BMPR1A-mediated signaling is critical for Lef1 expression. Second, Smad4 MT skin exhibited expanded proliferative compartment in the epidermis and ORS of the hair follicle, while BMPR1A deficiency results in defects in the IRS and the hair shaft, with relatively normal ORS and epidermis (Kobielak et al., 2003; Andl et al., 2004; Ming Kwan et al., 2004; Yuhki et al., 2004). It is known Smad4 interacts with Smad1, 5 and 8 to mediate BMP signaling, and with Smad2 and 3 to mediate signaling of TGF subfamily members, while BMPR1A only mediates signals of BMP subfamily (Heldin et al., 1997; Massague, 1998; ten Dijke and Hill, 2004). We have shown previously that all these Smads, except for Smad8, whose expression in skin has not been studied, are expressed in hair follicles and epidermis (He et al., 2001). Thus, it is plausible that the increased cell proliferation in the epidermis and ORS of the hair follicle is largely due to a block of signals of the TGF subfamily.

Furthermore, tumors developed in BMPR1A-/- skin are exclusively benign hair follicle tumors (Andl et al., 2004; Ming Kwan et al., 2004), while tumors in Smad4^Co/CoMMTV-Cre mice have multiple origin, although the majority is epidermal-derived SCC. Previous studies showed that transgenic mice expressing a dominant-negative type II TGF receptor, which blocks signaling of TGF subfamily, exhibited epidermal hyperproliferation (Wang et al., 1997). These mice do not develop spontaneous skin tumors; however, they are more sensitive to chemically induced carcinogenesis than control mice (Amendt et al., 1998; Go et al., 1999). Thus, the spontaneous skin tumor formation in Smad4^Co/CoMMTV-Cre mice could be due to the fact that Smad4 is a common mediator of both TGF and BMP signaling. Consistent with this observation, our data indicated that the absence of Smad4 blocked both TGF and BMP signaling (Figure 4f).

Significantly, we found that all tumors displayed high levels of pPten and unphosphorylated form of Pten. Pten is a potent tumor suppressor gene whose mutations have been found in a wide range of human cancers (reviewed in Sansal and Sellers, 2004). The increased levels of both forms of Pten suggest an essential role of this tumor suppressor in repressing skin tumor formation. The cause for the increased unphosphorylated form of Pten in Smad4-deficient skin tumors is unclear, but is currently under investigation. Our preliminary data indicate that Smad4 negatively regulates Pten expression under some conditions (Xu and Deng, unpublished observation), which may account for the increased Pten levels in the tumors. However, the details of this regulation are not currently clear and remain to be investigated. It has been shown that the absence of functional Pten leads to constitutive activation of downstream components of the PI-3 kinase pathway, including the AKT/PKB, a survival factor that protects various cell types against apoptosis (Downward, 1998; Maehama and Dixon, 1998). Our observation that Smad4 MT tumors exhibited high levels of AKT phosphorylation suggests that the overall functions of Pten were reduced in the Smad4 tumor cells, which allowed tumor cells to survive and proliferate.

Interestingly, Pten phosphorylation and AKT activation have been found in mice carrying targeted ablation of BMPR1A in the intestine epithelium (He et al., 2004). This resulted in the activation of -catenin, leading to the expansion of stem cells populations and formation of intestinal polyposis (He et al., 2004). Our previous study in Smad4-deficient mammary epithelial cells also revealed that inactivation of Smad4 in these cells resulted in the increased expression of -catenin. In the cultured condition, we showed that TGF treatment inhibited transcription of -catenin in the WT mammary epithelial cells but not in Smad4-/- cells, establishing that TGF treatment inhibited transcription of -catenin through Smad4. However, our analysis indicated that majority of Smad4-deficient skin tumors did not show increased expression, nor the nuclear localization of -catenin (data not shown), suggesting a distinct mechanism, rather than -catenin activation, may be involved in this model. To this end, we observed a significant increase in cyclin D1 (Figure 7b) and decrease in p21 (Figure 4d), which may have a significant contribution of tumor progression in the Smad4 MT skin. While both cyclin D1 and p21 are known downstream targets of TGF/Smad4 signaling, they are also subjected to regulation of multiple other factors. It has been demonstrated that activated AKT can result in the nuclear accumulation of cyclin D1 (Segrelles et al., 2002; Radu et al., 2003). Our analysis revealed spontaneous inactivation of Pten, and increased levels of AKT phosphorylation, which could be one of the causes for the nuclear accumulation of cyclin D1 in the Smad4-deficient skin tumor. Thus, different from mammary and intestinal epithelial cells, where targeted ablation of Smad4 (Li et al., 2003b) or BMPR1A (He et al., 2004) results in activation of -catenin, the absence of Smad4 in skin causes activation of cyclin D1 in a -catenin-independent manner.

In summary, we have studied the role of Smad4 in skin development and neoplasia using the Cre-Loxp approach. We have observed three major phenotypes displayed by Smad4-deficient skin. First, we showed that Smad4-deficient skin gradually lost hair due to a block of hair formation at the end of the first hair cycle, suggesting a critical role for Smad4 in hair generation and growth. Second, our analysis indicated that Smad4-deficient epidermis was significantly hyperproliferative, suggesting that Smad4 plays an essential role in repressing proliferation of epidermal keratinocytes. Third, we showed that the absence of Smad4 triggered skin tumor formation. Most notably, our data indicated that tumorigenesis is accompanied by inactivation of Pten, which may lead to the activation of AKT-cyclin D1 signaling pathway. This observation reveals a genetic interplay between tumor suppressors Smad4 and Pten during skin tumor formation, and our current study warrants future study for the underlying mechanisms of these interactions.

Materials and methods

Mice and mating

MMTV-Cre transgenic mice (Wagner et al., 1997) were crossed with Smad4 conditional MT (Smad4^Co/Co) mice (Yang et al., 2002) to generate Smad4^Co/+;MMTV-Cre mice. This strain of mice was interbred to generate Smad4 conditional MT mice (Smad4^Co/Co) mice carrying one copy or two copies of MMTV-Cre, as well as control mice with various genotypes (such as Smad4^Co/+, Smad4^Co/Co and Smad4^+/+). As Smad4^Co/Co mice carrying two copies of MMTV-Cre exhibited severer and earlier onset of skin abnormalities than those carrying one copy of the MMTV-Cre, we kept a colony of Smad4 and MMTV-Cre double-homozygous mice. Genotypes of mice and detection of Cre-mediated deletion of Smad4 were determined as described (Li et al., 2003b). Once the mice were confirmed to be homozygous for the Cre (i.e. all offspring contain Cre), they were interbred to maintain their homozygous state. The protocol for animal studies was approved by the 'Animal Care and User Committee' of the NIDDK.

Histology, immunohistochemical staining and in situ hybridization

For histology, dorsal skin and tumors were fixed in 10% formalin, blocked in paraffin, sectioned, stained with hematoxylin and eosin, and examined by light microscopy. Detection of primary antibodies was performed using the ZYMED Histomouse™ SP Kit according to the manufacturer's instructions. Western analysis was performed using standard procedures. Cyclin D1 and Smad4 antibodies were purchased from Santa Cruz Biotechnology. Antibodies for K5, K6, K14, K10 and BrdU were purchased from Covance Inc.; and -catenin was from BD Pharmingen. Antibodies to AE13 and AE15 were kindly provided by Dr TT Sun at NYU. Antibody to pPten and pAKT were from Cell Signaling, Inc. Expression of Lef1 was detected by in situ hybridization using a probe containing a fragment of cDNA corresponding to nt 1155–1500 (NM016269) kindly provided by Dr RL Johnson at the University of Texas MD Anderson Cancer Center.

RNA extraction and quantitative RT–PCR (qRT–PCR)

Total RNA was isolated from control skin, MT skin and tumors arising on MT skin using RNA-Stat-60 from Tel-Test, Inc. The qRT–PCR was achieved by combining in vitro reverse transcription with quantitative PCR, which was performed in a Stratagene^® Mx3000P™ thermal cycler (Stratagene, Lo Jolla, CA). Briefly, 5 g of purified total RNA from each sample was treated with DNase (Ambion, Austin, TX) and was then subjected to a reverse transcription reaction using AMV reverse transcriptase (Roche, Indianapolis, MN). The resultant cDNA products were used as templates for quantitative PCR to examine the levels of transcripts of mouse Smad4, 6, 7, p21, inhibitor of differentiation (Id)-1, and Lef1, using corresponding Taqman^® Assays-on-Demand^TM probes (Applied Biosystems Inc.) (catalogue numbers are: Smad4: Mm00484724; Smad6: Mm00484738; Smad7: Mm00484741; Id-1: Mm00775963; Lef-1: Mm00550265; and p21: Mm00432448). A 2-microglobulin (2M) probe (Mm00432448) was used as an internal control and the data (C_T values) were analysed using the Stratagene^® Mx3000P™ Comparative Quantitation software (Stratagene, La Jolla, CA).

References

1. Alonso L & Fuchs E. (2003) Proc Natl Acad Sci USA 100: 11830–11835. | Article | PubMed | ChemPort |
2. Amendt C, Schirmacher P, Weber H & Blessing M. (1998) Oncogene 17: 25–34. | Article | PubMed | ISI | ChemPort |
3. Andl T, Ahn K, Kairo A, Chu EY, Wine-Lee L & Reddy ST et al.. (2004) Development 131: 2257–2268. | Article | PubMed | ISI | ChemPort |
4. DasGupta R & Fuchs E. (1999) Development 126: 4557–4568. | PubMed | ISI | ChemPort |
5. Derynck R, Akhurst RJ & Balmain A. (2001) Nat Genet 29: 117–129. | Article | PubMed | ISI | ChemPort |
6. Downward J. (1998) Curr Opin Cell Biol 10: 262–267. | Article | PubMed | ISI | ChemPort |
7. Friedl W, Kruse R, Uhlhaas S, Stolte M, Schartmann B & Keller KM et al.. (1999) Genes Chromosomes Cancer 25: 403–406. | Article | PubMed | ISI | ChemPort |
8. Fuchs E, Merrill BJ, Jamora C & DasGupta R. (2001) Dev Cell 1: 13–25. | Article | PubMed | ISI | ChemPort |
9. Glick AB, Lee MM, Darwiche N, Kulkarni AB, Karlsson S & Yuspa SH. (1994) Genes Dev 8: 2429–2440. | PubMed | ISI | ChemPort |
10. Go C, Li P & Wang XJ. (1999) Cancer Res 59: 2861–2868. | PubMed | ISI | ChemPort |
11. Hahn SA, Hoque AT, Moskaluk CA, da Costa LT, Schutte M & Rozenblum E et al.. (1996) Cancer Res 56: 490–494. | PubMed | ISI | ChemPort |
12. He W, Cao T, Smith DA, Myers TE & Wang XJ. (2001) Oncogene 20: 471–483. | Article | PubMed | ISI | ChemPort |
13. He XC, Zhang J, Tong WG, Tawfik O, Ross J & Scoville DH et al.. (2004) Nat Genet 36: 1117–1121. | Article | PubMed | ISI | ChemPort |
14. Heldin CH, Miyazono K & ten Dijke P. (1997) Nature 390: 465–471. | Article | PubMed | ISI | ChemPort |
15. Howe JR, Roth S, Ringold JC, Summers RW, Jarvinen HJ & Sistonen P et al.. (1998) Science 280: 1086–1088. | Article | PubMed | ISI | ChemPort |
16. Ishida W, Hamamoto T, Kusanagi K, Yagi K, Kawabata M & Takehara K et al.. (2000) J Biol Chem 275: 6075–6079. | Article | PubMed | ISI | ChemPort |
17. Kobielak K, Pasolli HA, Alonso L, Polak L & Fuchs E. (2003) J Cell Biol 163: 609–623. | Article | PubMed | ISI | ChemPort |
18. Kratochwil K, Dull M, Farinas I, Galceran J & Grosschedl R. (1996) Genes Dev 10: 1382–1394. | PubMed | ISI | ChemPort |
19. Li AG, Koster MI & Wang XJ. (2003a) Cytokine Growth Factor Rev 14: 99–111. | Article | PubMed | ISI | ChemPort |
20. Li W, Qiao W, Chen L, Xu X, Yang X & Li D et al.. (2003b) Development 130: 6143–6153. | Article | PubMed | ISI | ChemPort |
21. Maehama T & Dixon JE. (1998) J Biol Chem 273: 13375–13378. | Article | PubMed | ISI | ChemPort |
22. Massague J. (1998) Ann Rev Biochem 67: 753–791. | Article | PubMed | ISI | ChemPort |
23. Ming Kwan K, Li AG, Wang XJ, Wurst W & Behringer RR. (2004) Genesis 39: 10–25. | Article | PubMed | ChemPort |
24. Nagarajan RP, Zhang J, Li W & Chen Y. (1999) J Biol Chem 274: 33412–33418. | Article | PubMed | ISI | ChemPort |
25. Radu A, Neubauer V, Akagi T, Hanafusa H & Georgescu MM. (2003) Mol Cell Biol 23: 6139–6149. | Article | PubMed | ISI | ChemPort |
26. Redman RS, Katuri V, Tang Y, Dillner A, Mishra B & Mishra L. (2005) J Oral Pathol Med 34: 23–29. | Article | PubMed | ISI |
27. Sansal I & Sellers WR. (2004) J Clin Oncol 22: 2954–2963. | Article | PubMed | ISI | ChemPort |
28. Santos M, Ballestin C, Garcia-Martin R & Jorcano JL. (1997) Mol Carcinogen 20: 3–9. | Article | ISI | ChemPort |
29. Segrelles C, Ruiz S, Perez P, Murga C, Santos M & Budunova IV et al.. (2002) Oncogene 21: 53–64. | Article | PubMed | ISI | ChemPort |
30. Sirard C, de la Pompa JL, Elia A, Itie A, Mirtsos C & Cheung A et al.. (1998) Genes Dev 12: 107–119. | PubMed | ISI | ChemPort |
31. Soriano P. (1999) Nat Genet 21: 70–71. | Article | PubMed | ISI | ChemPort |
32. Stenn KS & Paus R. (2001) Physiol Rev 81: 449–494. | PubMed | ISI | ChemPort |
33. Takaku K, Miyoshi H, Matsunaga A, Oshima M, Sasaki N & Taketo MM. (1999) Cancer Res 59: 6113–6117. | PubMed | ISI | ChemPort |
34. ten Dijke P & Hill CS. (2004) Trends Biochem Sci 29: 265–273. | PubMed | ISI | ChemPort |
35. Wagner KU, McAllister K, Ward T, Davis B, Wiseman R & Hennighausen L. (2001) Transgenic Res 10: 545–553. | Article | PubMed | ISI | ChemPort |
36. Wagner KU, Wall RJ, St-Onge L, Gruss P, Wynshaw-Boris A & Garrett L et al.. (1997) Nucleic Acids Res 25: 4323–4330. | Article | PubMed | ISI | ChemPort |
37. Wakefield LM & Roberts AB. (2002) Curr Opin Genet Dev 12: 22–29. | Article | PubMed | ISI | ChemPort |
38. Wang XJ. (2001) Microsc Res Tech 52: 420–429. | Article | PubMed | ISI | ChemPort |
39. Wang XJ, Greenhalgh DA, Bickenbach JR, Jiang A, Bundman DS & Krieg T et al.. (1997) Proc Natl Acad Sci USA 94: 2386–2391. | Article | PubMed | ChemPort |
40. Weng L, Brown J & Eng C. (2001) Hum Mol Genet 10: 237–242. | Article | PubMed | ISI | ChemPort |
41. Xu X, Brodie SG, Yang X, Im YH, Parks WT & Chen L et al.. (2000) Oncogene 19: 1868–1874. | Article | PubMed | ISI | ChemPort |
42. Yang X, Li C, Herrera PL & Deng CX. (2002) Genesis 32: 80–81. | Article | PubMed | ISI | ChemPort |
43. Yang X, Li C, Xu X & Deng C. (1998) Proc Natl Acad Sci USA 95: 3667–3672. | Article | PubMed | ChemPort |
44. Yuhki M, Yamada M, Kawano M, Iwasato T, Itohara S & Yoshida H et al.. (2004) Development 131: 1825–1833. | Article | PubMed | ISI | ChemPort |

Acknowledgements

We thank Dr TT Sun for providing antibodies for AE13 and AE15, and members of the Deng laboratory for helpful discussions and critical reading of the manuscript. This work was supported by the intramural support of NIDDK to C-XD, and NIH grants CA87849 and AR47898 to X-JW.

Supplementary Information accompanies the paper on Oncogene website (http://www.nature.com/onc)