The polarity protein Scrib mediates epidermal development and exerts a tumor suppressive function during skin carcinogenesis

To determine the role of Scrib during epidermal embryonic development, we initially examined the flank epidermis from viable wild-type (Wt) and Scrib ^+/− (Het) embryos compared to neonatal lethal Scrib ^-/- knockout (KO) embryos at E16.5, E17.5 and E18.5 (n = 8–29 per genotype and time point). Germline Cre-deleter driven recombination results in the systemic excision of a LoxP site flanking exons 4–13 of the Scrib gene, resulting in a frame-shift mutation and premature truncation of the protein (Fig. 1a). The remaining N-terminal product (approximately 100aa) has no known function. Quantitative real-time PCR (qRT-PCR) analysis confirmed a significant reduction in Scrib mRNA transcript expression levels in Het and KO embryonic flank skin compared to Wt littermates at E16.5 (Fig. 1b). Immunofluorescence (IF) to detect Scrib protein revealed uniform membranous staining in Wt and Het embryonic epidermal cells that is absent in KO littermates at E18.5 (Fig. 1c).

To establish whether Scrib is required for EPB acquisition, we assessed the penetration of toluidine blue dye in Scrib-deficient embryonic epidermis at several stages of development (Fig. 1d, n = 4–10 per genotype/time point). At E16.5, Wt and Scrib-deficient embryos displayed global dye penetration, consistent with previous work demonstrating that permeability barrier formation begins at this time point, but is not yet fully established [34]. By E17.5, Wt and Het embryos are resistant to toluidine blue permeation, reflecting SC formation and a functional EPB. In contrast, Scrib KO embryos displayed intense blue staining at E17.5, suggesting impaired barrier formation. Nevertheless, by E18.5 Scrib KO embryos did not allow dye penetration (except in exposed neural tube and eye) (Fig. 1d). Taken together, our findings indicate that Scrib deletion causes a transient delay in epidermal barrier acquisition at E17.5. Consistent with this, histological analysis revealed that Scrib KO flank skin was markedly thinner (Fig. 1e), and quantitation revealed a significant reduction in epidermal thickness at E16.5 and E17.5 compared to Wt and Het embryos that was recovered at E18.5 (Fig. 1f).

To investigate whether the delay in skin barrier formation we observe in Scrib KO embryos was simply due to a generalised delay in developmental timing, we analysed formation of the forelimb. Limb morphogenesis is highly stereotyped and as such, is commonly used as an indicator of developmental stage. Staining of E17.5 forelimbs with alcian blue and alizarin red, to detect cartilage and bone respectively, revealed no obvious difference in bone size, morphology or pattern of ossification in KO forelimbs compared to Wt or Het littermates (Fig. 2a; n = 3–6). At early stages of forelimb development, characteristic digit morphology and the dynamic pattern of interdigital apoptosis were unchanged in E13.5 KO embryos when compared to Wt or Het littermate embryos (Fig. 2b; n = 3). Together, these data indicate that a global delay in developmental timing is not responsible for the observed defects we see in skin barrier formation, and suggest that Scrib plays a specific role in regulating epidermal development. Nevertheless, although these data strongly suggest that the transient delay in EPB formation observed in Scrib KO embryos is most likely not a consequence of systemic abnormal development, we cannot completely exclude this possibility.

Interestingly, staining for bone and cartilage in Scrib KO embryos at E17.5 revealed widespread dysregulation of skeletal morphogenesis outside of the limb that to our knowledge has not been reported previously (Fig. 2c; n = 3–6). Scrib KO embryos were not only smaller, but displayed loss of dorsal cranial bones, hemivertebra and widespread proximal and distal rib fusions (Fig. 2d-f) resembling Chuzhoi mutant mouse embryos that harbour a mutation in the polarity gene Ptk7 [35] and Dvl2^−/− mouse embryos [36], further illustrating the importance of this network during development.

Although we observe multiple developmental and skeletal abnormalities in Scrib KO embryos, our analysis of interdigital apoptosis and limb morphogenesis strongly indicates that developmental timing is not broadly disrupted. Thus, we reasoned that the transient delay in epidermal barrier acquisition might reflect impaired proliferation, Tight Junction (TJ) formation and/or keratinocyte maturation. To test this, we first analysed proliferation by performing immunohistochemistry (IHC) to detect the proliferation marker PCNA at E16.5, E17.5 and E18.5 (Fig. 2 g and data not shown, n = 3 per genotype/time-point). Remarkably, Wt and Scrib-deficient embryonic epidermis displayed a similar number of PCNA-positive cells, indicating that Scrib loss does not cause a transient delay in EPB acquisition by reducing proliferation.

IF to detect the TJ protein ZO-1 revealed that Scrib Wt, Het and KO embryonic epidermis display comparable, uniform membranous ZO-1 staining (Fig. 2 h, n = 3), indicating that impaired TJ formation is not responsible for the observed delay in EPB formation in Scrib KO embryos. In support, ultrastructure analysis of Scrib KO embryonic flank epidermis revealed normal TJ morphology (e.g. kissing points) that were comparable to Wt and Het littermates at E17.5 (Fig. 2i, n = 3–6). Furthermore, we did not observe any changes in the distribution of the apical-basal polarity protein Dlg or the Adheren’s Junction (AJ) component and polarity protein E-cadherin (Additional file 1: Figure S1, n = 3). Collectively, these data suggest that delayed EPB formation in Scrib KO embryos is not the result of aberrant polarity or cell-cell adhesions.

To assess whether Scrib loss disrupts keratinocyte maturation, we measured the thickness of both the SC and the SG in Scrib Wt, Het and KO epidermis at E16.5, E17.5 and E18.5. We show that Scrib KO embryos display a significant reduction in SC and SG thickness at E17.5 compared to Wt and Het littermates (Fig. 3a-c, n = 3). Since SG and SC thickness is restored and partially recovered respectively in Scrib KO epidermis at E18.5, these data suggest that Scrib loss perturbs keratinocyte maturation to delay EPB formation. In support, co-staining for the spinous layer cytoskeletal protein cytokeratin 10 (K10) and the SG component loricrin, revealed that at E16.5 and E17.5, Scrib KO epidermis displays a diminished SC and limited SG formation compared to Wt and Het littermates, which recovered at E18.5 (Fig. 3d, n = 3).

Given that Scrib facilitates EPB acquisition during development, we next analysed whether Scrib functions to maintain epidermal homeostasis in adult mice. As Scrib KO mice are neonatal lethal, we crossed K14Cre(Δneo) mice [37] with Scrib floxed mice [7] to specifically delete Scrib in the entire epidermis after birth. Cohorts of K14Cre ^+/− ;Scrib ^+/+ , K14Cre ^+/− ;Scrib ^+/fl or K14Cre ^+/− ;Scrib ^fl/fl mice (hereafter denoted Scrib ^+/+ , Scrib ^+/fl , Scrib ^fl/fl respectively) were generated and aged to 100 days (n = 10). PCR analysis of genomic DNA isolated from dorsal skin confirmed recombination of the Scrib ^+/fl and Scrib ^fl/fl alleles in the epidermis, while only the Wt allele was present in control animals (Fig. 4a). QRT-PCR to detect Scrib confirmed that mRNA transcript levels were significantly decreased by 56.3 % (±1.92 SD) and 89.0 % (±1.84 SD) in Scrib ^+/fl and Scrib ^fl/fl mice respectively (Fig. 4b). Histological analysis of adult epidermis at 100 days revealed no obvious phenotypic difference in Scrib ^+/+ , Scrib ^+/fl , Scrib ^fl/fl epidermis, despite the absence of Scrib protein in Scrib ^fl/fl epidermal cells, determined by IF staining (Fig. 4c). Scrib ^+/+ and Scrib ^+/fl mice showed strong, continuous Scrib staining within the membrane cortex of epidermal cells. Together, these findings indicate that Scrib is dispensable for the maintenance of adult epidermal tissue homeostasis.

To confirm Scrib is not required for normal adult skin homeostasis, we first performed IHC for PCNA, to visualise proliferating cells, and Cleaved Caspase-3 (CC3) to visualise apoptotic cells. Enumeration of these IHC identified that there was no significant difference in the number of proliferating or apoptotic cells in the epidermis of Scrib ^+/+, Scrib ^{+/f l} and Scrib ^fl/fl mice (Fig. 4c-e) at 100 days (n = 3). In support, qRT-PCR to detect mRNA expression of the stem cell gene Lgr5 revealed no significant difference in transcript levels in Scrib ^+/+, Scrib ^{+/f l} and Scrib ^fl/fl epidermis (Additional file 2: Figure S2A, n = 3). Cytokeratin 6 (K6) is detected in the interfollicular epidermis only during hyperproliferative conditions, including psoriasis [38] and wound healing [39], and thus can be used as a surrogate marker of aberrant homeostasis in the epidermis. IHC to detect cytokeratin 6 (K6) confirmed that K6 was restricted to the hair follicles in Scrib ^+/+ and Scrib-deficient epidermis (Additional file 2: Figure S2B, n = 3), indicative of normal homeostatic conditions [38]. Finally, we investigated differentiation in the epidermis by performing co-IF staining for basal (K5) and suprabasal (K10) differentiation markers, which revealed no difference between Scrib ^+/+, Scrib ^{+/f l} and Scrib ^fl/fl adult epidermis (Additional file 2: Figure S2C, n = 3). Importantly, skin pathologies were not observed in aged Scrib ^+/+, Scrib ^{+/f l} and Scrib ^fl/fl mice (400 d, n = 10, data not shown). Collectively, these data indicate that Scrib is not essential for murine adult epidermal homeostasis.

We have previously shown that conditional homozygous deletion of Scrib mediated by Probasin- and MMTV-driven Cre-recombinase expression causes low-grade prostate intraepithelial neoplasia and mammary gland dysplasia respectively in mice, establishing that Scrib is a mammalian tumor suppressor gene [7, 30]. To examine the role of Scrib during epidermal tumor initiation/progression, we used the well-characterised DMBA/TPA two-step chemical skin carcinogenesis model [40, 41]. DMBA invariably induces HRas activating mutations, while TPA promotes tumor growth by deregulating several signaling networks, including the PKC-Ras/MAPK cascade that mediates proliferation, differentiation and inflammation [42‐44].

Cohorts of Scrib ^+/+ , Scrib ^+/fl and Scrib ^fl/fl mice were generated (n = 12–16; mixed gender) and underwent long-term DMBA/TPA treatment at 8 weeks of age (Fig. 5a). Weekly examination of dorsal epidermis revealed that papillomagenesis initiated 7 weeks post-DMBA in Scrib ^+/+ , Scrib ^+/fl and Scrib ^fl/fl mice (Fig. 5b and c), indicating that Scrib loss does not disturb the latency of DMBA/TPA-induced epidermal lesions. Notably, the entire Scrib ^fl/fl cohort had developed epidermal lesions at 14 weeks post-DMBA, while 43.7 % and 33.3 % of Scrib ^+/+ and Scrib ^+/fl cohorts respectively were tumor free (Fig. 5b). Indeed, 30.0 % of Scrib ^+/+ and 13.3 % of Scrib ^+/fl mice remained tumor-free during the course of the experiment (Fig. 5b), however overall survival was not significantly different between genotypes (Additional file 3: Figure S3A). qRT-PCR of RNA isolated from epidermal lesions and IF to detect Scrib confirmed Scrib mRNA is significantly reduced in Scrib ^+/fl and Scrib ^fl/fl lesions compared to Scrib ^+/+ controls, and that Scrib protein is uniformly distributed at the membrane cortex in Scrib ^+/+ and Scrib ^+/fl DMBA/TPA lesions, but negligible in Scrib ^fl/fl mice (Additional file 3: Figure S3B and C; n = 3). These data demonstrate that Scrib confers a tumor suppressive role during DMBA/TPA-induced skin carcinogenesis.

Quantitation of lesion multiplicity revealed that the incidence of DMBA/TPA-induced lesions is increased in Scrib ^fl/fl mice compared to Scrib ^+/+ and Scrib ^+/fl mice (Fig. 5d). In addition, half of the Scrib ^fl/fl cohort (6/12) developed lesions that reached our ethical size limit and were sacrificed prior to 26 weeks post-DMBA, compared to only 20 % (3/15) and 25 % (4/16) of the Scrib ^+/+ and Scrib ^+/fl cohorts respectively (Additional file 3: Figure S3A). Scrib ^fl/fl mice also developed larger DMBA/TPA-induced lesions compared to Scrib ^+/+ and Scrib ^+/fl mice at comparable time-points (Fig. 5e; 7, 10, 12 and 14-weeks post DMBA treatment shown). Together, these data indicate that bi-allelic loss of Scrib sensitises the epidermis to papillomagenesis and facilitates their growth.

Histological analysis of Scrib ^+/+ , Scrib ^+/fl and Scrib ^fl/fl mice at sacrifice determined that long-term DMBA/TPA treatment causes epidermal hyperplasia, benign papillomas and invasive SCC in Scrib ^+/+ , Scrib ^+/fl and Scrib ^fl/fl mice (Fig. 6a). While only a single SCC was observed in the Scrib ^+/+ cohort (3.8 % incidence), Scrib ^+/fl and Scrib ^fl/fl mice displayed an increased frequency of benign papilloma progression to invasive SCC (14 % and 25 % incidence respectively) (Fig. 6b). Thus, epidermal tumor progression correlates directly with reduced Scrib expression.

Next, we assessed the proliferation status of volume-matched (40 – 80 mm³) early benign papillomas from all 3 genotypes, to ascertain whether Scrib loss can accelerate DMBA/TPA-induced papilloma growth. IHC to detect BrdU incorporation revealed that bi-allelic Scrib loss causes a small, yet significant increase in the number of mitotic cells (37.3 % ± 1.754 SD) compared to Scrib ^+/fl (32.3 % ± 1.742 SD) or Scrib ^+/+ (33.1 % ± 0.991 SD) papillomas (Fig. 6c and d). Thus, elevated proliferation contributes to the observed increased in DMBA/TPA epidermal tumor volume in Scrib ^fl/fl mice.

Given that DMBA initiation invariably causes activating Ras mutations in murine epidermis [45], these findings support previous work demonstrating that Scrib loss cooperates with Ras oncogenic activation to promote progression and/or invasion [7, 29, 30, 46]. However, we did not observe any significant changes in the number of p-ERK positive nuclei in DMBA/TPA-induced size-matched benign papillomas from Scrib ^+/+ , Scrib ^+/fl and Scrib ^fl/fl (Additional file 3: Figure S3D and E), suggesting that the increased tumour phenotype observed upon deletion of Scrib is unlikely to reflect changes in Ras/MAPK signalling.

Of note, we did not detect any alteration in the distribution of the tight-junction protein ZO-1, or the AJ component E-cadherin (Additional file 3: Figure S3F, n = 3), that has been previously shown to co-localise with Scrib to regulate E-cadherin-mediated cell-cell adhesion [47]. These data suggest that Scrib does not alter cell-cell adhesion to facilitate DMBA/TPA progression.

Papillomas developed significantly faster in Scrib ^fl/fl mice suggesting that loss of Scrib may provide a tumor promoting environment in the skin. To investigate if Scrib loss promotes the survival of transformed cells in the DMBA/TPA model, we performed IHC to detect CC3 in volume-matched early benign papillomas expressing either Scrib ^+/+ , Scrib ^+/fl or Scrib ^fl/fl alleles (Fig. 6c). Immunostaining quantitation revealed that the number of CC3 positive cells in Scrib ^fl/fl papillomas was significantly reduced compared to Scrib ^+/+ or Scrib ^+/fl papillomas (Fig. 6e). In support, we show that the number of cells positive for the histone variant H2AX phosphorylated at Ser139 (termed γH2AX), previously shown to be associated with ATM- and JNK-dependent apoptosis [48, 49], directly correlates with the number of CC3-positive apoptotic cells (Fig. 6c and f). Taken together, our findings indicate that Scrib deletion promotes DMBA/TPA-induced epidermal lesion growth by enhancing proliferation and survival of transformed cells.

As Scrib ^fl/fl mice were more predisposed to develop DMBA/TPA-induced lesions (Fig. 5b-e), this suggests that Scrib loss may sensitize the epidermis to cutaneous carcinogenesis. To investigate this, we performed a short-term DMBA/TPA study to analyse Scrib-deficient epidermis prior to papillomagenesis. To this end, Wt, Scrib ^+/fl and Scrib ^fl/fl mice received a single topical application of DMBA, followed by 4 weeks of TPA treatment (Fig. 7a; n = 4–7; mixed gender). Histological analysis of short-term cohorts revealed comparable hyperplastic epidermis associated with epidermal thickening was present in all 3 genotypes (Fig. 7b). We next analysed proliferation rates by quantitating IHC to detect BrdU-positive cells. Surprisingly, Scrib loss does not affect proliferation in DMBA/TPA-induced pre-neoplastic epidermis (Fig. 7b and c). In corroboration, neither DMBA nor TPA treatment alone was sufficient to induce a proliferative advantage in Scrib deficient skin (Fig. 7c and Additional file 4: Figure S4A). Taken together, our findings suggest that although Scrib can suppress proliferation to inhibit the growth of established DMBA/TPA-driven papillomas, Scrib does not affect DMBA/TPA-induced proliferation during the pre-neoplastic state.

The dermal infiltration of immune cells in response to TPA has previously been shown to augment proliferation and promote skin carcinogenesis (reviewed in [50]). Our analysis of mast cells (toluidine blue staining) and macrophages (IHC to detect F4/80) in Scrib ^+/+ and Scrib-deficient epidermis that has undergone short-term treatment with either acetone, DMBA, TPA or DMBA/TPA revealed no difference in the frequency of these immune cell types (Additional file 4: Figure S4B and C). These data indicate that Scrib deletion is not likely to facilitate tumour initiation/growth by mediating inflammatory responses.

Since we observed no changes in proliferation in pre-neoplastic Scrib-deficient epidermis, we next investigated the apoptotic index in these mice. Importantly, acetone controls confirmed that Scrib loss is not sufficient to alter CC3-mediated keratinocyte apoptosis. However, compared to Scrib ^+/+ and Scrib ^+/fl mice, the number of CC3-positive cells detected by IHC was significantly decreased in Scrib ^fl/fl epidermis treated with either DMBA, TPA or DMBA/TPA for 4 weeks (Fig.  7b, d and Additional file 5: Figure S5A). Likewise, DMBA treatment (+/− TPA) significantly reduced the number of γH2AX-positive cells in Scrib ^fl/fl epidermis (Fig.  7b, e and Additional file 5: Figure S5B), however the number of p53 and p21 positive nuclei remained unchanged (Additional file 5: Figure S5C-E, and data not shown). Thus, these data suggest that Scrib loss impairs the DMBA/TPA-induced apoptotic response via a p53-independent mechanism, presenting a direct mechanism whereby loss of Scrib causes a pre-neoplastic setting in the epidermis.