Probing the Regulation of Snail Gene Expression Reveals an Essential Link to TGF-β2 Signaling in the Developing Hair Bud
The temporal spike of Snail mRNA expression in the hair bud prompted us to consider what factor(s) may be regulating the Snail gene. A variety of extracellular signals have an impact on the cell type-specific expression of different Snail family members, and many of them, including Wnts, BMPs, FGFs, and TGF-βs, also affect hair bud development [2,16,28]. Since Snail is not expressed in cultured skin keratinocytes that secrete active BMPs and FGFs (see Figure 1B), we focused our attention on Wnt and TGF-β signaling as more likely candidates for Snail induction in this cell type.
Previously, we showed that effective transmission of a Wnt-3a signal in cultured keratinocytes can be achieved through their exposure to the BMP inhibitor noggin, which induces LEF-1 expression [4]. In vitro, these conditions down-regulated the E-cadherin promoter and induced a LEF-1/β-catenin-sensitive reporter gene, TOPFLASH [4]. In contrast, Snail expression was not induced by these conditions (Figure 5A). Thus, despite essential roles for Wnts and noggin in hair follicle specification [4,29,30], our studies did not support an essential role for these signals in governing Snail expression in keratinocytes.
Figure 5  TGF-β2, but Not Wnt/noggin, Transiently Induces Snail Expression In Vitro
(A) Failure of Wnt and noggin signaling to induce Snail in cultured keratinocytes. Primary mouse keratinocytes were treated with Wnt- and/or noggin-conditioned medium (+) or the corresponding control medium (–). These conditions are known to activate the LEF-1/β-catenin reporter TOPGal and down-regulate the E-cadherin promoter (see [4] for details). Using Western blot analyses, cellular proteins were then analyzed for Snail, LEF-1, β-catenin, and tubulin. Proteins from keratinocytes transfected with K14-Snail were used as a positive control for Snail expression.
(B) TGF-β2 can induce Snail protein. Primary keratinocytes were treated for the indicated times with recombinant TGF-β2 (+) or heat inactivated TGF-β2 (–).Total cellular proteins were then isolated and analyzed by Western blot for Snail, pSMAD2 (reflective of activated TGF- signaling), and tubulin. Note the activation of Snail expression, peaking at 2 h post-TGF-β2 treatment and then disappearing thereafter.
(C) Snail mRNA expression is transiently induced by TGF-β2. The experiment in (B) was repeated, and this time, total RNAs were isolated from keratinocytes treated with TGF-β2 for the indicated times. RT-PCR was then used with (+) or without (–) reverse transcriptase (RT) and with primer sets specific for Snail and GAPDH mRNAs. Note that Snail mRNA expression also peaked at 2 h, paralleling Snail protein.
(D) TGF-β2 treatment results in enhanced activity of a Snail promoter-β-galactosidase reporter. Keratinocytes were transfected with a β-galactosidase reporter driven by a Snail promoter that is either WT (wt prom) or harbors a mutation in a putative pSMAD2/pSMAD4 binding site (mt prom). At 2 d posttransfection, cells were treated with either TGF-β or heat-inactivated TGF-β2 (inact) for the times indicated. β-galactosidase assays were then conducted, and results are reported as fold increase over a basal level of activity of 1. The experiment was repeated three times in triplicate, and error bars reflect variations in the results. TGF-β1 has been shown to induce Snail family members in hepatocytes and heart [15, 31]. In keratinocytes, however, TGF-β1 inhibits keratinocyte growth and seems to be involved in triggering the destructive phase of the cycling hair follicle [32]. Of the loss-of-function mutations generated in each of the TGF-β genes, only the TGF-β2 null state blocked follicle development at the hair bud stage [32]. Thus, we turned towards addressing whether TGF-β2 might be involved in regulating Snail expression in keratinocytes isolated from the basal layer of the epidermis. Though there is no cell culture system available to specifically study placodal cells, these keratinocytes are their progenitors and are the closest approximation available to study the behavior of epithelial cells of the placode.
Interestingly, treatment of cultured keratinocytes with as little as 5 ng/ml of TGF-β2 caused a rapid and transient induction of Snail (Figure 5B). Following this treatment, Snail protein was detected within 30 min, peaked at 2 h, and then declined thereafter. The induction of Snail appeared to be specific for the active form of the growth factor, as pretreatment of TGF-β2 for 10 min at 100 °C obliterated the response [Figure 5B, lanes marked (–)]. By contrast, although TGF-β receptor activation remained elevated during the duration of the experiment (as measured by the sustained phosphorylation of the downstream effector SMAD2) Snail expression could not be maintained (Figure 5B). Thus, although Snail expression correlated with phosphorylated SMAD2 (pSMAD2) induction, its decline seemed to rely on secondary downstream events.
The rapid kinetics of Snail expression were reflected at the mRNA level, suggesting that Snail promoter activity in keratinocytes might be sensitive to TGF-β2 signaling (Figure 5C). To test this possibility, we engineered a transgene driving the β-galactosidase reporter under the control of approximately 2.2 kb of promoter sequence located 5′ from the transcription initiation site of the mouse Snail gene. At 2 d after transient transfection, keratinocytes were treated with TGF-β2 (t = 0) and then assayed for transgene activity over the same time course in which we had observed Snail protein induction. The results of this experiment are presented in Figure 5D.
Within 0.5 h of TGF-β2 treatment, Snail promoter activity had increased 3-fold, and by 2 h, it peaked to approximately 10-fold over control levels (Figure 5D). Thereafter, Snail promoter activity rapidly returned to the basal levels seen in unstimulated keratinocytes. The kinetics of Snail promoter activity closely paralleled those observed for Snail protein induction. Moreover, the stimulatory effects appeared to be specific to TGF-β2, since they were abrogated either by heat inactivation of the TGF-β2 protein or by mutation of a putative SMAD binding element located about 1.8 kb 5′ from the Snail transcription start site (Figure 5D). Taken together, these results suggested that in keratinocytes, TGF-β2 signaling results in a pSMAD2-dependent transient activation of the Snail gene, and that maintenance of Snail protein relies, in part, upon sustained promoter activity.
The brevity of Snail gene and protein induction in TGF-β2 treated cultured keratinocytes resembled the temporal appearance of Snail mRNA and protein at the initiation of hair follicle morphogenesis in embryonic mouse skin. To test whether TGF-β2 might be required for Snail induction in hair bud formation in vivo, we first analyzed whether TGF-β2 was expressed in or around the hair bud. Consistent with previous observations [33], an anti-TGF-β2 antibody labeled developing hair buds (Figure 6A). This labeling appeared to be specific as judged by the lack of staining in follicle buds from mice homozygous for a TGF-β2 null mutation (Figure 6A; [34]). Moreover, the downstream effector of TGF-β2 signaling, pSMAD2, was also expressed in WT, but not TGF-β2-null, hair buds (Figure 6B). Together, these data underscore the importance of the TGF-β2 isoform despite expression of both TGF-β1 and TGF-β2 in developing hair buds at this stage.
Figure 6  TGF-β2 Is Necessary to Induce Snail Expression and Regulate Proliferation and E-Cadherin in the Hair Bud
(A–D) Skins from TGF-β2 WT or KO E17.5 embryos were analyzed for expression of TGF-β2 protein (A), which is present in the epidermis and dermis as previously described [33] and in the hair bud, pSMAD2 (B), Snail (C), and Snail mRNA (D). Arrows point to the hair buds.
(E) Western blot analyses of Snail expression in the skins of 2-wk-old K14-Smad2 transgenic (SMAD2 TG) and WT littermate (WT) mice. Antibody to tubulin was used as a control for equal protein loadings. The K14-Smad2 Tg mouse was previously shown to possess activated TGF-β signaling [35].
(F–G) Proliferation markers Ki67 (F) and pMAPK (G) are diminished in TGF-β2-null hair relative to its WT counterpart.
(H–J) TGF-β2-null hair fails to down-regulate E-cadherin (H). Wnt and noggin signaling pathways are still intact in the TGF-β2 null hair as nuclear LEF-1 (I) and nuclear β-catenin (J) are still expressed. To further explore the possible relation between Snail and TGF-β2, we examined the status of Snail expression in TGF-β2-null hair buds. As judged by immunohistochemistry, Snail protein was absent from E17.5 skin of TGF-β2-null embryos but not from that of control littermates (Figure 6C). This effect appeared to be exerted at the transcriptional level, since Snail mRNAs were also not found in TGF-β2 null hair buds under conditions in which the signal was readily detected in the hair buds of littermate skin (Figure 6D).
Conversely, in 2-wk-old K14-Smad2 Tg mice, which display elevated TGF-β signaling in skin [35], Snail protein was readily detected by Western blot analyses, where it was not found in postnatal skin (Figure 6E). Taken together, these results provide compelling evidence that TGF-β2 is functionally important for inducing Snail gene expression in a pSMAD-dependent manner in developing hair buds. Whether pMARK activity also contributes to Snail induction was not addressed in the present study [15].
Although some hair buds still formed in TGF-β2 null skin, their number was reduced by approximately 50% [32]. Thus, although the pathway mediated by TGF-β2 signaling impacts the earliest step of epithelial invagination, it does not appear to be essential for bud morphogenesis. Consistent with this notion, basement membrane remodeling still took place in the TGF-β2-null buds, as judged by immunofluorescence with antibodies against β4 integrin, an integral component of keratinocyte-mediated adhesion to its underlying basement membrane (Figure 6F). In contrast, TGF-β2 signaling appeared to be an important factor for the early proliferation that occurs in the developing hair buds, as judged by anti-Ki67 and anti-pMAPK immunofluorescence (Figure 6F and 6G).
If TGF-β2 stimulates Snail expression in developing buds, loss of this morphogen would be expected to affect the expression of genes that are typically repressed by Snail. Since a major target for Snail-mediated repression is the E-cadherin gene [12,13], we investigated the status of E-cadherin in TGF-β2-null buds. As shown in Figure 6H, hair buds in TGF-β2 null skin displayed elevated immunofluorescence staining relative to their WT counterparts.
Previously we demonstrated that the concerted action of the extracellular signals Wnt and noggin are required for the generation of a LEF-1/β-catenin transcription complex to repress E-cadherin transcription at the onset of hair fate specification. As shown in Figure 6I and 6J, both WT and TGF-β2 null buds exhibited nuclear LEF-1 and β-catenin localization, signs that the Wnt-noggin signaling pathway was intact. These data suggest that during hair follicle morphogenesis, TGF-β2 functions subsequently to Wnt/noggin-mediated determination of hair fate. Moreover, through activation of Snail gene expression, TGF-β2 appears to work in tandem with these other morphogens to down-regulate E-cadherin levels, which contributes to the activation of proliferative circuitries.