|Home | About | Journals | Submit | Contact Us | Français|
Activating mutations in the KRAS oncogene are associated with three related human syndromes, which vary in hair and skin phenotypes depending on the involved allele. How variations in RAS signals are interpreted during hair and skin development is unknown. In this study, we investigated the developmental and transcriptional response of skin and hair to changes in RAS activity, using mouse genetic models and microarray analysis. While activation of Kras (KrasG12D) in the skin had strong effects on hair growth and hair shape, steady state changes in downstream RAS/MAPK effectors were subtle and detected only by transcriptional responses. To model the transcriptional response of multiple developmental pathways to active RAS, the effects of growth factor stimulation were studied in skin explants. Here FGF acutely suppressed Shh transcription within 90 minutes but had significantly less effect on Eda, WNT, Notch or BMP pathways. Furthermore, in vivo Fgfr2 loss-of-function in the ectoderm caused derepression of Shh, revealing a role for FGF in Shh regulation in the hair follicle. These studies define both dosage sensitive effects of RAS signaling on hair morphogenesis and reveal acute mechanisms for fine-tuning Shh levels in the hair follicle.
RAS is a common downstream component of multiple signaling pathways, which include receptor tyrosine kinases (RTKS), associated growth factors, cytokines, integrin cross-linking, G-protein coupled receptors activation, and other stimuli (Reuther and Der, 2000). In its active state, RAS stimulates several parallel effector pathways, and depending on the experimental setting, a variety of cellular responses may ensue including migration, proliferation, survival, differentiation, and senescence (Malumbres and Barbacid, 2003; Rodriguez-Viciana et al., 1997). In the setting of cancer, mutations in RAS maintain RAS in its activated state and drive persistent proliferation and survival of transformed cells via one or more effector pathways. As a central mediator of multiple signaling pathways, pharmacologic inhibition of RAS, its post-translational modification, or its downstream effector proteins has been widely pursued as a potential treatment for cancer (Bollag et al., 2010; Downward, 2003; Rodriguez-Viciana et al., 2005).
Mutations in RAS paralogs are also found in a growing number of genetic disorders, including Costello, Cardiofaciocutaneous and Noonan syndromes (Rauen et al., 2010; Tidyman and Rauen, 2009). Collectively, these syndromes, termed RAS/MAPK syndromes or RASopathies, manifest in patients with several developmental abnormalities including craniofacial, cardiac, neural, cognitive and ectodermal defects. Cardiac defects include valvular malformations, cardiac hypertrophy, and arrhythmias (Lin et al., 2011), and in the nervous system, patients are affected by delayed development, cerebellar enlargement and cognitive defects (Axelrad et al., 2009; Gripp et al., 2010). In the integument, epidermal, hair, and nail defects are more divergent among RAS/MAPK syndromes. In Costello syndrome, redundant skin, papillomas, and palmar defects are highly characteristic, while sparse hair and perifollicular changes are common and characteristic in CFC patients (Siegel et al., 2012). Other phenotypes, including curly hair and heat intolerance, are common to both CFC and Costello syndrome. Differences in skin phenotypes provide important criteria for delineating between the three distinct syndromic forms (Siegel et al., 2011).
Cutaneous phenotypes among RAS/MAPK syndromes suggest that different RAS paralogs, alleles and effectors are capable of inducing distinct developmental responses. In Costello syndrome (CS), patients (>95%) carry activating HRAS mutations (Aoki et al., 2005); CFC patients (~60%) show predominantly BRAF, MEK1, or MEK2 mutations (Rodriguez-Viciana et al., 2006); 50% of Noonan syndrome patients carry PTPN11 loss-of-function mutations (Noonan, 2006). Uncommonly, mutations in the paralog Kras occur in RAS/MAPK syndromes. Mutations of this paralog have been reported in all three syndromes, including KRAS[K5E/K5N/F156L] in CS (Bertola et al., 2007; Zenker et al., 2007) and KRAS[P34R/V14I/T58I/D153V] in CFC/Noonan (Schubbert et al., 2006). The pleiotropic effects of KRAS mutations indicate that that paralog differences alone may not be sufficient to explain the phenotypic spectrum of RAS/MAPK syndromes. Moreover, in mouse models, when KrasG12D is activated in the skin, KRAS causes redundant skin, hair loss and papillomas, which are characteristics features of HRAS-associated Costello syndrome in humans (Mukhopadhyay et al., 2011). These studies suggest that phenotypic variation in the skin may be caused by both allelic differences and paralog differences in RAS/MAPK syndromes.
Differential cellular responses to RTK signal strength has been implicated as a mechanism to specify multiple cell fates and developmental responses using a single signaling pathway (Schweitzer and Shilo, 1997). In experimental models, amplitude and duration of RTK activation contribute to differential cellular and discrete transcriptional responses (Marshall, 1995; Murphy et al., 2002). The spectrum of phenotypes that differentiate RAS/MAPK syndromes suggest developmental dose-sensitive RTK responses may also exist in the hair and skin of humans and other mammals. To identify developmental and signaling mechanisms that sense changes in RAS activity, we sought to identify RAS-responsive molecular and developmental pathways in the skin using a KrasG12D gain-of-function model. In this model, KrasG12D allele induced a curly hair phenotype, altered hair shaft features important for the production of hair subtypes, and altered the transcriptional program of several developmental pathways of the hair follicle. To identify developmental pathways that are sensitive to RAS signal strength, the transcriptional response of skin explants was examined in response to varying doses of growth factors and inhibitors of the RAS signaling pathway. This approach revealed that growth factor stimulation acutely downregulates Shh in the hair follicle. These studies provide evidence for the sensitivity of developmental pathways of the skin in response to RAS signal variation.
Mice have been generated and genotyped as previously described for Krastm4Tyj/J (Tuveson et al., 2004) and Fgfr2flox/flox alleles (Yu et al., 2003). Lines have been maintained in mixed genetic backgrounds. All experiments were performed according to the institutional guidelines established by the University of California, San Diego, Institutional Animal Care and Use Committees.
Hair shafts were plucked from three P14 and three P47 animals from each genotype: wildtype and Msx2-cre; KrasG12D mice. The numbers of bends and medullar spaces were tabulated as independent features. Statistical analysis was performed with Prizm 5 (GraphPad Software).
Hair length was determined using Image (Wayne Rasband, NIH) and the distributions of hair lengths were compared using the Kendall-Sherman test. For analysis of hair growth from P4 to P7, fixed skin was dehydrated in ethanol and cleared in methyl salicylate. Thick sections (2–3 hair follicles wide) were cut, imaged, and analyzed by NeuronJ (Meijering et al., 2004). Cuticle casts were generated from imprints of hair shafts, embedded in Permount and air-dried overnight at room temperature.
Dorsal skin from postnatal (P) day 7 wildtype and Msx2-cre; KrasG12D mice was collected, minced, frozen, and pulverized in liquid nitrogen. Protein was isolated in lysis buffer containing 50 mM TrisHCl (pH 7.5), 150 mM NaCl, 1% NP-40, 0.25% sodium deoxycholate, 1 mM EDTA, 5 mM NaF, 2 mM sodium vanadate, and 1x Complete protease inhibitor (Roche). Antibodies used in Western blotting are listed in Supplementary Methods. Western analysis was performed with goat anti-mouse IgG (IRDye 800CW, Li-COR, USA) and goat anti-rabbit IgG (IRDye 680CW, Li-COR, USA), visualized using an Odyssey Infrared Imager (Li-COR, USA).
Dorsal skin where Msx2-cre is most active was used for RNA studies. Whole skin was minced, transferred into TRIzol (Life Technologies Corp.), and homogenized by mechanical bead disruption (BioSpec Products). Total RNA was isolated according to manufacturer’s protocol. Quantitative RT-PCR was performed using Maxima qPCR reagents (Fermentas Inc.) and a LightCycler 480 (Roche). Taqman reagents and primers for SYBR reactions are listed in Supplementary Methods. Relative mRNA levels were normalized to beta-actin and expressed using a baseline-independent method was used to determine the fractional cycle number and the cycle number was used to calculate the amount of each product using the delta CP calculations. Levels of PCR product were expressed as a function of beta-actin.
Microarray analysis was performed in triplicate from total RNA of control and Msx2-cre; KrasG12D P7 mice by the UCSD BIOGEM Microarray core for quality control, reverse transcription, labeling, and hybridization to Mouse WG-6 V2 Beadchips (Illumina, Inc.). Differential gene expression p-values were calculated from multiple hypothesis correction (Yekutieli and Benjamini, 2001). For comparison of microarray data to hair follicle lineages, we used the standards described in GEO Dataset GDS1323 (Rendl et al., 2005) to determine cell-type specific genes, as defined by ≥ 2-fold in one cell population as compared to all others. Additional statistical methods are detailed in Supplemental methods. Wildtype and Msx2-cre; KrasG12D mouse skin microarray data can be accessed at the ArrayExpress database, accession: E-MTAB-700.
Histology was performed as previously described (Mukhopadhyay et al., 2011). In situ hybridization was performed as previously described (Etchevers et al., 2001). Anti-sense digoxygenin riboprobes were generated, according to manufacturer’s instructions (Roche). K17 riboprobe was generated from a PCR template using forward (5’-TGGCAGTGGTTATGGAGGCAAC-3’) and reverse primers (5’-CCAGCAATCCTACCTTGTTCTTCAG-3’) containing T7 promoter. Shh riboprobe was previously described (Lewis et al., 2001). Image capture was performed with a DP71 (Olympus) camera mounted on a BX51 stereomicroscope (Olympus).
Immunofluorescence staining was performed on paraformaldehyde-fixed tissue in conjunction with citrate antigen retrieval. Primary antibodies used were: phospho-H3 (Ser10) (1:250, Cell Signaling), AE15 (1:500, Santa Cruz Biotechnologies), AE13 (1:500, Santa Cruz Biotechnologies).
For cell counts, positive phospho-H3 and Shh staining in hair follicles were counted from at least three mice and subjected to a Mann-Whitney statistical test from Prism 5 (GraphPad Software, Inc.)
Wildtype 1.5 mm punch biopsies were obtained from the dorsal skin of euthanized CF1 outbred and C576Bl/6 inbred P7 animals and placed in DMEM supplemented with 10% fetal calf serum (Mediatech, Inc.) and antibiotics. Growth factors and their sources are provided on Supplemental Methods. Explants were treated with vehicle, media and serum for 90 minutes were used as controls. RNA was extracted from biopsies as above and analyzed by quantitative RT-PCR.
In order to activate KrasG12D from its endogenous locus, we utilized a loxP permissive allele, LSL-KrasG12D, in which a loxP-STOP-loxP (LSL) cassette prevents gene expression until excised by Cre recombinase. To perform Cre-loxP recombination in the hair follicle, we bred LSL-KrasG12D mouse to the Msx2-cre transgenic line, which is active in the matrix of the postnatal hair follicle and in the dorsal skin during embryonic development (Pan et al., 2004; Sun et al., 2000). In other embryonic tissues, Msx2-cre is expressed in the apical ectoderm ridge of the early limb bud. Offspring carrying both Msx2-cre and LSL-KrasG12D (hereafter referred to Msx2-cre; KrasG12D) were viable and lacked obvious external changes, e.g. normal limbs and facies. Thus, Msx2-cre overcomes the neonatal lethality associated with previously reported activation of KrasG12D by keratinocyte-specific Keratin 14 (K14)-cre (Tuveson et al., 2004).
Hair growth in Msx2-cre; KrasG12D became visibly abnormal by 1–2 weeks (Fig. 1A). Mice produce four different hair types, guard, awl, auchene and zigzags, which can be identified based on differences in their length, relative frequency, number of constrictions (bends), and internal columns (medulla) (Mustonen et al., 2003; Sundberg, 1994). Microscopic analysis of Msx2-cre; KrasG12D hair revealed multiple abnormalities in hair shaft formation (Fig. 1B). Hair bends were absent in Msx2-cre; KrasG12D mice. In addition, the number of medulla per hair (P = 0.004; ANOVA) and length (P=1.7 x 10−12; ANOVA) were reduced in Msx2-cre; KrasG12D mice (Fig. 1C). To characterize the short hair phenotype, we examined hair length at P4 and P7 of cleared whole skin (Fig. 2A). Image tracings of hair from wildtype skin demonstrated an average of 1.81 mm growth over the three-day period, while Msx2-cre; KrasG12D littermates grew only 0.82 mm within the same interval (Fig. 2B–D). These findings indicate that the reduced hair length of Msx2-cre; KrasG12D mice is caused by slower hair growth.
To investigate other possible causes of short hair, we examined the timing of hair development in Msx2-cre; KrasG12D mice. The earliest hair follicles of the dorsal skin initiate development at embryonic (E) day 14.5 and are responsible for the production of the longest hairs (guard hairs). We assessed that presence of hair follicles at E15.5 by the expression of keratin 17 (K17) and found that placode maturation and density was normal in Msx2-cre; KrasG12D embryos (Fig. 3A). At neonatal and early postnatal stages, when all hair types are present, hair follicles were identified at normal density and maturation (Fig. 3B, C; Suppl. Fig. S2). In addition, differentiation into hair shaft lineages, medulla, cortex and cuticle, was readily detected (Fig. 3D, E). Further analysis of the external morphology of Msx2-cre; KrasG12D hair shafts showed normal cuticle morphology (Suppl. Fig. S2). To investigate the reduced hair production of Msx2-cre; KrasG12D mice, proliferation of the hair matrix was examined (Fig. 3F, G). The number of phospho-histone H3-positive cells in the hair follicle bulb region, a molecular change equivalent to mitotic index, was significantly reduced in Msx2-cre; KrasG12D mice (P=0.03). These findings indicate the activated RAS does not specifically block hair differentiation but alters the rate of hair growth.
The KrasG12D allele, while acting as a gain-of-function, has been shown to be hyperactive rather than persistent and constitutive in other models (Tuveson et al., 2004). To determine if the steady-state level of RAS activation was normal in the skin of Msx2-cre; KrasG12D mice, we examined the overall activity of the RAS/MAPK pathway at P7 by a quantitative, non-enzymatic Western approach (Fig. 4A, B). In addition, whole skin was utilized as extensive dissection of tissue would likely lead to non-physiologic activation of the RAS/MAPK pathway. Levels of phosphorylated MEK, ERK and AKT were found to be normal in the Msx2-cre; KrasG12D mice.
Although KrasG12D is active and sufficient to induce phenotypic changes in the hair, epigenetic events such as the upregulation of negative feedback networks may reset the steady state levels of phosphorylated downstream proteins in vivo (Courtois-Cox et al., 2006). Because multiple RAS/MAPK antagonists are transcriptionally-induced in response to growth factor stimulation, their expression was used as a transcriptional signature of elevated RAS/MAPK signaling (Kawakami et al., 2003; Verheyden et al., 2005). Extensive dissection of hair follicles and injury potentially activate RAS/MAPK ectopically (Corson et al., 2003). To reduce the likelihood of ectopic activation, whole skin of P7 animals was used as a source for RNA with the caveat that some changes in hair follicle gene expression may be diluted by surrounding tissue. Extracted RNA was subjected to microarray hybridization analysis, interrogating 45,281 unique, expressed transcripts of 18,122 genes (Fig. 5). Microarray analysis revealed 1,891 significantly altered genes (≥25% change from wildtype, with P<0.05) and 144 genes altered by greater than 2-fold (Fig. 5A). Affected genes in Msx2-cre; KrasG12D mice included RAS/MAPK antagonists and other biomarkers of RAS/MAPK activation, e.g. Ets variant transcription factors, Etv1, and Etv4, Etv5, Dusp6 and others (Fig. 5B). For several genes, real-time PCR was performed to confirm the altered quantities of these transcripts (Fig. 5C). These studies confirmed that the Msx2-cre; KrasG12D skin represents a heightened level of RAS/MAPK signaling and suggested that in a developmental steady state, the developmental and transcriptional response to RAS gain-of-function may be more sensitive than biochemical changes.
RAS gain-of-function also had significant effects on hair signaling and gene expression. A comparison of KrasG12D-affected genes with hair lineage gene expression (Rendl et al., 2005) revealed repression of hair matrix lineage genes, including many involved with ectodermal development and proliferation (Fig. 5D). In adjacent tissues, genes expressed by the adjacent dermal papilla were not significantly altered and upregulated in the outer root sheath. Lineage gene expression changes indicate that RAS has selective effects on hair follicle lineages. The downregulation of matrix-specific genes and positive regulators of proliferation led us to study the effect of KrasG12D on lineage differentiation and cell proliferation of the hair follicle (Fig. 5E).
One of the goals of investigating RAS gain-of-function in the hair and skin was to identify the dose-sensitive relationship between growth factor signaling and developmental pathways. The downregulation of hair matrix gene expression could reflect one or more signaling pathways that are sensitive to RAS activation and that are known to participate in hair morphogenesis, including Bone morphogenetic protein (BMP), Ectodysplasin (EDA), FGF, Hedgehog (HH), WNT, and Notch (Fig. 6A)(Schneider et al., 2009). Microarray analysis and real-time PCR of 19 genes representative of transcriptional targets or critical ligands of six pathways using microarray data and real-time PCR results suggested a number of RAS-sensitive pathways (Fig. 6B, C). However, whether these changes in Msx2-cre; KrasG12D mice reflect secondary change to due the altered development of the hair follicle could not be determined. To address these issues, we utilized a skin explant assay to model physiologic RAS activation, e.g. growth factor treatment, and to identify the acute transcriptional response of the skin (Suppl. Fig. S3). Among several candidate pathways, the level and statistical significance of Shh RNA modulation by FGF2 was most prominent. Within 90 minutes of FGF2 stimulation, Shh RNA levels were reduced up to 2.8 fold, P=2x10−5 (Suppl. Fig. S2). The acute response in wildtype explants suggested that the relationship between RTK signaling and Shh regulation may be direct.
Microarray data also corroborated the significance of Shh changes in vivo as several Hedgehog-related targets, including Shh, Ptch1, Ptch2 and Foxe1 were also downregulated in Msx2-cre; KrasG12D mice (Fig. 7A) (Brancaccio et al., 2004; Tabata and Kornberg, 1994). Confirmation by quantitative real-time RT-PCR revealed a 3.78-fold reduction in Shh mRNA and a 2.17-fold reduction in its target gene, Ptch1 in Msx2-cre; KrasG12D skin (Fig. 7B, Suppl. Fig. S4). The reduced expression of transcriptional targets of the Hedgehog pathway indicated that in addition to reduced ligand expression, activation of the Hedgehog pathway was also reduced. To determine if reduced Shh levels affect the expression pattern of Shh in the Msx2-cre; KrasG12D hair follicles, in situ hybridization was performed (Fig. 7C). The domain of Shh expression was essentially normal in Msx2-cre; KrasG12D hair follicles but had reduced staining compared to wildtype hair follicles. The reduction of Hedgehog target genes suggested the reduction or inhibition of SHH ligand. Western analysis revealed ~25% reduction of mature SHH protein (Fig. 7D). Thus, ex vivo RTK stimulation and RAS gain-of-function in vivo repress cutaneous Shh levels.
The above findings suggest that use of FGF in explant studies recapitulate some aspects of Shh dysregulation in Msx2-cre; KrasG12D hair follicles. To further localize the response of Shh to a component of RAS/MAPK signaling pathway, we examined the effect of inhibition of RAS effector pathways. Treatment of explants with inhibitors against downstream components of the RAS/MAPK pathway, c-RAF1 (GW5074; Lackey et al., 2000) or MEK (U0126; Favata et al., 1998), caused an increase in Shh mRNA levels (Fig. 8A). Treatment with a PI-3-kinase inhibitor (LY294002; Vlahos et al., 1994) also resulted in increased Shh mRNA levels. These studies provide further evidence that downstream effectors of RAS, i.e. MEK, RAF and PI-3-kinase, participate in the regulation of Shh.
We also reasoned that if Shh downregulation is RAS-mediated then the Shh response should be broadly induced by multiple receptor tyrosine kinase (RTK) signals. Thus, the response of Shh to epidermal growth factor (EGF), platelet-derived growth factor (PDGF), and additional FGFs (FGF7, FGF10) was also examined (Fig. 8B). Each growth factor demonstrated a similar downregulation of Shh mRNA. Thus, in a wildtype hair follicle, RTK activation or inhibition of RAS downstream effector pathways both elicit rapid changes in Shh levels.
The rapid loss of Shh mRNA in the presence of FGF2 could be explained by a short intrinsic half-life of Shh mRNA or by FGF-induced destabilization of Shh mRNA. Since the half-life of Shh mRNA is unknown, we characterized the kinetics of Shh mRNA loss after pharmacologic inhibition of global transcription (Fig. 8C). Treatment of explants with actinomycin resulted in rapid decay of Shh mRNA, revealing an intrinsically short half-life. Furthermore, non-linear best-fit modeling showed no significant change in the Shh RNA half-life of FGF2 vs. actinomycin-treated explants (4.6 vs. 4.3 minutes, respectively, R2>0.98). These findings suggest that FGF represses Shh mRNA levels by inhibiting transcription. To determine if transcriptional repression is responsible for decreased Shh mRNA in Msx2-cre; KrasG12D skin, the levels of primary transcript of Shh RNA was evaluated (Fig. 8D). Since primary transcripts are unspliced, primers against introns 1–2, 2–3, and proximal regions of the first exon of Shh were generated to detect the primary transcripts of Shh RNA. Like its mature Shh mRNA, the primary transcripts of Shh were decreased in Msx2-cre; KrasG12D mice. These findings indicate that the Shh mRNA is short-lived and transcriptionally repressed by growth factor signaling.
The above studies suggested that RAS plays a role in Shh regulation. However, because RAS activation responds to many types of cellular signals, whether endogenous growth factor ligands participate in Shh transcriptional regulation in the hair follicle is unknown. In explant studies, we observed that several growth factor ligands were capable of repressing Shh RNA levels. FGF ligands play multiple roles in hair development, including regulation of hair growth, length determination, and medulla formation, and were ideal candidates for normal regulators of hair follicle Shh levels (Schlake, 2007). FGF ligands bind to one of four FGF receptors (FGFR) and activate their receptors in complex with heparan sulfate proteoglycans (Yayon et al., 1991). All four FGFRs are expressed in the hair follicle (Rosenquist and Martin, 1996). Mouse knockout studies reveal that Fgfr3 and Fgfr4 are not required for normal morphogenesis of the hair follicle (Deng et al., 1996; Weinstein et al., 1998). Roles of Fgfr1 and Fgfr2 in the hair follicle have been tested through tissue-specific inactivation and have demonstrated their roles in early formation of the hair and later in the maintenance of the hair follicle unit (Grose et al., 2007; Yang et al., 2010), but their role in ectodermal Shh regulation is unknown.
FGF7 and FGF10 selectively bind to ectodermal isoforms of FGFR1 and FGFR2 and are sufficient to downregulate Shh levels in explants (cf. Fig. 8B). Specifically, we wished to study the transcriptional response of Shh in the absence of Fgfr2 in the same tissue domains as Msx2-cre; KrasG12D mice (Fig. 9). Msx2-cre; Fgfr2flox/− mice (Msx2-cre; Fgfr2cKO) were thus generated. These mice displayed limb defects as previously reported (Yu and Ornitz, 2008) but had normal skin and hair until two weeks of age. After two weeks, the hair from Msx2-cre; Fgfr2cKO mice became long, shiny and brittle, but the overall density was normal (see below). Thus, in order to study the response of Shh RNA to the loss of Fgfr2, we assayed Shh mRNA levels at P7 prior to the onset of these abnormalities. We detected Shh at more than 2-fold higher levels in Msx2-cre; Fgfr2cKO mice relative to littermate controls, which included three genotypes, Msx2-cre; Fgfr2flox/+, Fgfr2flox/−, and Fgfr2flox/+ (Fig. 9A, Suppl. Fig S4 and not shown). Increased levels of Shh mRNA were also accompanied by increased expression of downstream genes, Ptch1, Ptch2, Hhip1, and Gli1, indicating that Hedgehog signaling was increased in the Msx2-cre; Fgfr2cKO skin.
During normal hair follicle growth, Shh is expressed by a transient migrating population of cells, which begins near the hair matrix. As these cells move distally, they abruptly lose Shh expression and eventually acquire an IRS or cuticle cell fate (Greco et al., 2009). Ectopic or persistent expression of Shh might also result in increased Shh levels in the whole skin of Msx2-cre; Fgfr2cKO mice. The negative feedback loop involving KRAS and Shh transcription suggest that growth factor signals might regulate the pattern or magnitude of Shh expression in the hair follicle. RNA in situ hybridization was performed on Msx2-cre; Fgfr2cKO skin to distinguish between these possibilities (Fig. 9B). In situ hybridization studies revealed that the overall expression pattern of Shh was unchanged in the absence of Fgfr2. A comparison of the number of Shh-positive cells in wildtype, Msx2-cre; Fgfr2cKO, and Msx2-cre; KrasG12D hair follicles revealed no significant difference in their numbers (P= 0.4187 and P=0.4355) (Fig. 9C). To verify whether loss of ectodermal Fgfr2 was sufficient to block acute Shh downregulation ex vivo, Msx2-cre; Fgfr2cKO explants were treated with FGF2 (Fig. 9D). There was a small but significant response of Shh transcription to FGF2 in the Msx2-cre; Fgfr2cKO skin, which indicating the presence of some residual functional FGFR activity. Thus, increased Shh mRNA levels in Msx2-cre; Fgfr2cKO skin arise independently of ectopic Shh expression (Fig. 9E).
In this study, hair morphogenesis was used as a model to study the developmental response to RAS gain-of-function. We found that in addition to the induction of a growth factor-responsive pathways, the KrasG12D allele also causes the downregulation of Shh. Downregulation of Shh appears to be a normal response to growth factor signaling in the hair follicle as verified through in vitro studies and genetic loss of FGFR2 signaling.
The hair follicle is sensitive to gain-of-function RAS mutations. In humans, hair shape and growth changes are prevalent in patients with germline HRAS, KRAS, BRAF, MEK1, or MEK2 (Roberts et al., 2006). In the mouse, we find that Kras gain-of-function affects hair formation in multiple ways. KrasG12D alters hair length, features of the hair shaft, and as described previously, hair cycling (Mukhopadhyay et al., 2011). Altered proliferation in the Msx2-cre; KrasG12D hair follicle may contribute to many of the observed hair defects. Proliferation in the hair matrix is linked to the growth rate of the hair shaft. In addition, changes in proliferation are associated with altered appearance of the hair shaft including effects on hair bends in mice (Weger and Schlake, 2005). which is consistent with reduced hair growth and may further contribute to reduced numbers of medullary column and bend formation. KrasG12D may also have independent effects of the above characteristics as phenotypic changes in length, bends and medulla can independently affected in mouse genetic mutants (Sundberg, 1994; Schlake, 2007).
The use of an explant system facilitated the assessment of various growth factor signals that are normally present during hair follicle morphogenesis. The observation that several different growth factors are capable of Shh repression implies the activity of a common downstream signaling component. We also find that two parallel RAS downstream pathways, RAF/MAPK and PI-3-kinase, participate in Shh repression; further supporting the concept that a common component of growth factor signaling is a key regulator of Shh. We propose that RAS normally functions in the hair follicle as a mediator of FGFR2 activation to suppress Shh. SHH plays a major role in hair follicle growth (Chiang et al., 1999; Mill et al., 2003; St-Jacques et al., 1998), and thus, it seems feasible that hair growth defects in RAS gain-of-function mutations may result in part from decreased SHH levels. RAS gain-of-function affected additional signaling pathways in the hair follicle. How these additional signals cooperate with the suppression of Shh in the hair follicle and other organs in RAS/MAPK syndromes requires further investigation.
Much of the knowledge on Shh transcriptional regulation has focused on the complex cis-regulatory elements regulating its restrictive pattern during organogenesis (Amano et al., 2009; Epstein et al., 1999). However, little is known regarding the mechanisms that control its level of expression. In this study, we identify a rapid transcriptional response of the Shh promoter in response to RAS activation and inhibition. The half-life of Shh mRNA has not been previously reported. We find that unlike developmental signaling molecules, e.g. fgf8 (Dubrulle and Pourquié, 2004), Shh mRNA has a very short half-life. The 3' untranslated region of Shh is AU-rich like many short-lived mRNAs but lacks classic AU-rich elements (Hau et al., 2007; Zubiaga et al., 1995). The short half-life of Shh mRNA allows for highly regulated patterns of Shh expression in developing organs, where cells transiently occupy signaling centers. The short half-life of Shh places significant reliance on transcriptional mechanisms to achieve high steady state levels of Shh mRNA. We find evidence that regulation of Shh by FGF and RAS signaling play important role in regulating the amplitude of Shh expression but appears to be distinct from regulatory mechanisms that regulate Shh pattern (cf. Fig 9E). Increased expression of morphogens by expansion of the expression domain often have deleterious effects on the morphogenesis of an organ (Oro et al., 1997). Modulation of morphogen level or sensitivity without disturbing pattern is a possible mechanism for increasing organ growth and size without affecting its shape.
The observations from the hair follicle may not be readily generalized to other organs. For example, in the limb bud, FGFR2 signaling plays both a role in the initiation and maintenance of normal domains of Shh expression (Lewandoski et al., 2000) and a second role in preventing ectopic Shh expression in the anterior limb bud (Mao et al., 2009; Zhang et al., 2009). It is also possible that FGFR2 function changes during the dynamic period of the hair cycle, and thus we cannot rule out a later function of FGF signaling in the regulation of Shh pattern during other stages of hair growth. The variation in hair follicle organ size at different body sites may well depend on the scalability of Shh expression via FGF modulation.
The above studies reveal that hair formation, shape and growth are highly sensitive to RAS/MAPK signals and serve as an even more sensitive measure than traditional biochemical markers of RAS/MAPK activation. Further studies of hair response to RAS/MAPK gain and loss-of-function may find useful application in the clinic not only in the evaluation of congenital syndromes but also as a biomarker in cancer or response to RAS/MAPK inhibition in cancer therapy.
We are grateful to the following: T. Jacks (LSL-KrasG12D), D. Ornitz (Fgfr2fl/fl), and G. Martin and R. Maxson (Msx2-cre), the UCSD BIOGEM Microarray Core (microarray hybridization and scanning), C. Jamora and members of the laboratory for reagents, technical assistance and discussion. This work was supported by the NIH/NIAMS to B.Y. (AR056667), CIRM to B.Y. (RN2-00908), NIH/Ruth Kirschstein NRSA to J.B. (HL07491).