Id Proteins Are Dynamically Expressed in Normal Epidermis and Dysregulated in Squamous Cell Carcinoma¹

The development and maintenance of the epidermis is a complex,dynamic process whereby immature keratinocytes from the basal cell layer of the epidermis undergo upward migration and accumulation of proteins associated with barrier formation. Differentiation is regulated by signals from the dermis and by gradients of inductive epidermal growth signals, particularly calcium ions and cytokines such as EGF³and insulin-like growth factor I (1, 2). However, the molecular events responsive to these stimuli are less well understood. Recent studies have identified HLH transcription factors as key regulators of development [reviewed in Massari and Murre(3)]. Widely expressed class A factors (or E-proteins)dimerize with the temporally and spatially restricted class B factors to initiate tissue-specific gene expression. The formation of transcriptionally active complexes is blocked by the Id proteins, the expression of which is generally associated with naïve or proliferating cells [reviewed in Norton et al.(4)]. Increasing evidence implicates Id dysregulation in a range of tumors; indeed, ectopic Id expression was shown to immortalize primary human keratinocytes (5, 6). Nonmelanoma skin cancers are among the most common tumors, although few but the most aggressive forms are lethal. SCC is characterized by tongues of invasive keratinocytes in the epidermis, although spontaneous differentiation is apparent in subsets of cells, and foci of keratinized cells characterize more indolent tumors(7). We sought to investigate the pattern of HLH gene expression in human epidermis both in situ and in vitro and to assess their role in keratinocyte cell cycle progression. This led us to investigate HLH protein expression profiles in tumors derived from squamous epithelia.

Skin from face-lift or breast reduction operations was used for tissue sections and for primary human keratinocyte preparation(8). Biopsy sections from SCC of the head and neck were kindly provided by Dr. Nick Coleman (Department of Histopathology,Addenbrooke’s Hospital). All tissues were used with Local Research Ethics Committee approval.

SCC9 cells were purchased from American Type Culture Collection(9). HaCaT keratinocytes were a kind gift from Dr. R. Fusenig (University of Ulm, Ulm, Germany; Ref. 10). Primary keratinocytes were propagated in keratinocyte serum-free medium(Life Technologies, Inc., Paisley, United Kingdom). Cell lines were maintained in either DMEM (HaCaT) or DME:F-12 (1:1; SCC) containing 10% FCS, 1 mg/ml each of penicillin and streptomycin, 1 mg/ml glutamine, and 200 ng/ml hydrocortisone (all obtained from Sigma Chemical Co., Poole, United Kingdom).

These were performed essentially as described by Prunieras et al. (11). Briefly, epidermis was removed from human skin after 7–10 days incubation at 37°C in Ca²⁺- and Mg²⁺-free PBS. The dermis was acellularized by 10 cycles of freeze thawing and seeded with early-passage primary human keratinocytes and grown for 10–14 days.

cDNA was oligo dT primed from 1 μg of total human keratinocyte RNA and extended with SuperRT (HT Biotech, Cambridge, United Kingdom) as standard. PCR was performed using SuperTaq polymerase (HT Biotech) with the following primer pairs (all listed 5′- 3′-): Id-1,CGCGAATTCGCCAAGAATCATGAAAG,CGCTCTAGAGGCGCTTCAGCGACACA (EcoRI and XbaI sites are underlined); Id-2,ACAGCCTGTCGGACCACAGC, GCCTGCAAGGACAGGATGCT; Id-3,GACGACATGAACCACTGCTA, TTGGAGATGACAAGTTCCGGAG; E2A,ATCTACTCCCCGGATCACTC, TTCTCCTCCCGCTTGATCTC; p21^cip1waf1, AAGGTCAGTTCCTTGTGGAG,ATTAGGGCTTCCTCTTGGAG; involucrin, CTCTGCCTCAGCCTTACTGT,ATTCCCAGTTGCTCATCTCT; and β-actin,TACCTCATGAAGATCCTCAC, TTCGTGGATGCCACAGGACT (Eurogentec, Abingdon,United Kingdom). PCR products were excised from 2% TAE agarose gels and purified using GeneClean II (Bio101, St. Louis, MO) according to the manufacturer’s recommendations. Probes were subsequently labeled by multipriming (Amersham International, Buckinghamshire, United Kingdom).

RNA was prepared with TriReagent (Sigma) according to the manufacturer’s recommendations. Five μg of total RNA were resolved in 1.2% agarose gels in 0.5× TBE (45 mm Tris-borate, 1 mm EDTA) and transferred to Hybond N+ membranes(Amersham) by capillary blotting. Nucleic acids were fixed by UV, and hybridization was performed in Rapid Hybridization Buffer (Amersham)for 2.5 h at 65°C prior to washes for 30 min at 65°C in each of the following solutions: in 2× SSC (twice), 1× SSC, and 0.5× SSC(all containing 0.1% SDS).

Human face-lift skin was fixed overnight at 4°C in neutral buffered formalin and embedded in paraffin wax. Five-μm sections were deparaffinized, rehydrated, and then subjected to 10 min of high temperature antigen retrieval by pressure cooking in 80 mmsodium citrate. Sections were incubated in 3% hydrogen peroxide solution for 30 min, followed by a 10% normal goat serum block. Sections were incubated for 48 h at 4°C with one of the following antibodies: polyclonal Id-1 (rabbit antimouse Id-1, residues 129–148; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), used at 5μg/ml; polyclonal Id-2 and Id-3 (rabbit antimouse, full-length, a generous gift from Dr. Edward Prochownik, Department of Hematology/Oncology, University of Pittsburgh, Pittsburgh, PA), each used at 5 μg/ml after an overnight preblock with BSA; monoclonal E2A (antihuman E47, residues 195–208; Santa Cruz) at 1μg/ml; calcitonin (anti human; Europath, Bude, United Kingdom) at 5μg/ml or involucrin (antihuman; Novocastra, Newcastle, United Kingdom) at 1 μg/ml. Detection was with a biotinylated IgG (diluted 1:500 in 1% BSA/Tris buffered saline), streptavidin-conjugated horseradish peroxidase and diaminobenzidine precipitation, followed by hematoxylin counterstaining. Blocking of Id signals was performed using a 10-fold excess by mass of the appropriate Id GST fusion peptides (a kind gift from Edward Prochownik; data not shown). Immunoprecipitation studies demonstrated low cross-reactivity between Id antibodies (data not shown). Five hundred cells from three patients were counted per SCC stage to quantify Id staining. ISH was performed according to the method of Thomas et al. (12) with the following 5′- and 3′-digoxygenin-labeled primers (listed 5′- 3′): Id-1sense, CTCTACGACATGAACGGCTGCTACTC; Id-1, antisense,GAGTAGCAGCCGTTCATGTCGTAGAG.

Expression of Id-1, Id-2, and Id-3mRNA was observed in subconfluent proliferating human primary keratinocytes maintained in low calcium (0.05 mm)media (Fig. 1,A, day 0) by Northern blotting. Differentiation in response to raised calcium (2.0 mm) resulted in down-regulation of Id mRNA after 24 h (Fig. 1,A, day 1), and by 5 days, little message remained (Fig. 1,A, day 5). This correlated inversely with the increased expression of the cyclin-dependent kinase inhibitor p21^cip1waf1 and the differentiation marker involucrin (13). E2A messages persisted throughout differentiation. Furthermore, ISH localized Id-1 message to the basal and spinous layers of the epidermis (Fig. 2,B).Id mRNAs were highly expressed in cell lines derived from squamous epithelia (SCC9 and HaCaT keratinocytes; Fig. 1,B). Such cells do not undergo terminal differentiation in response to elevated calcium ions in monolayer culture and display little contact inhibition, consistent with maintained Idexpression (compare Fig. 1 B, pre- and postconfluence; Refs.9 and 10).

To evaluate the cytokine sensitivity of Id expression, HaCaT cells were arrested by serum withdrawal and reinduced by the addition of serum, EGF, or insulin (Fig. 1 C). Id levels were typically reduced 4-fold after 24 h of serum deprivation and further reduced by the presence of heparin, which blocks autocrine amphiregulin stimulation of the EGF receptor (14). Reinduction with each factor resulted in an increase of Id-3mRNA after 15 min, consistent with immediate early gene expression,peaking at 1 h. Similar kinetics were obtained for Id-1and Id-2 (data not shown).

To evaluate the pattern of HLH protein expression in human skin in situ, sections were stained with a range of antibodies(Fig. 2, C–J). We found cytoplasmic Id-1 localization in the basal layers of the epidermis (Fig. 2, E and F), but some suprabasal nuclear staining was apparent in the spinous and granular layers. Little staining was apparent in the corneocytes. Id-2 staining was nuclear, and the basal layer appeared to show the greatest levels of expression (Fig. 2,G), with down-regulation as cells transited to the stratum corneum. Subcellular Id-3 localization was also nuclear and predominantly basal, although some cells in the spinous layers also stained (Fig. 2,H). Interestingly, discrete columns of Id-2 and Id-3 staining cells persisted throughout all of the suprabasal layers (Fig. 2, G and H, arrows). Involucrin identifies the granular and cornified layers (Fig. 2,J), whereas nuclear expression of the products of the E2A gene (E12 and/or E47)was apparent throughout the epidermis (Fig. 2 I).

The stratified organization of the LSE model closely resembles that of human skin (Fig. 2, K–O), and the pattern of Id staining was qualitatively similar to that seen in normal skin.

Sections from poorly (Fig. 3, A–D), moderately (Fig. 3, E–H), and well-differentiated SCCs (Fig. 3, I–L) were stained for Id-1, Id-2, and Id-3. In each case, Id-1 showed the abundant cytoplasmic staining characteristic of basal keratinocytes, whereas expression of Id-2 and Id-3 maintained a nuclear pattern. The keratinized foci characteristic of well-differentiated SCCs displayed a lack of Id staining, (Fig. 3, I, K, and L, arrows). The majority of malignant keratinocytes in poorly differentiated SCC expressed Id-1 and Id-3(Table 1) with reduced staining in well-differentiated SCCs. Interestingly,intermediate values were obtained in moderately differentiated SCCs. Fewer cells expressed Id-2, and levels were similar in each stage.

Recent reports implicate Id proteins in cancers of the nervous system, pancreas, breast, and chondrocytes [15, 16;reviewed by Israel et al. (5)]. Furthermore,ectopic Id expression causes the formation of adenomas and leads to the transformation of primary human keratinocytes (6, 17). In this report, we show that Id proteins are expressed in normal human epidermis and are dysregulated in SCC.

Differentiation of primary keratinocytes in response to calcium in vitro was marked by rapid Id mRNA down-regulation, consistent with cell cycle withdrawal and the onset of differentiation. Furthermore, the response of Id expression to mitogenic stimulation in serum-deprived HaCaT keratinocytes is consistent with a growth-promoting role. HaCaT cells undergo partial growth arrest in response to serum withdrawal and can be induced to re-enter the cell cycle in response to insulin and EGF, although entry into S-phase is delayed with the latter cytokine (2). Insulin (acting via the insulin-like growth factor I receptor)stimulates similar Id induction to serum, whereas EGF alone shows a strong peak of Id message after 1 h but very little signal after 24 h, consistent with slower transit through the cell cycle. This suggests that EGF and insulin may, at least in part, exert their mitogenic effects through activation of Id expression.

HaCaT cells form a stratified epithelium in an organotypic coculture model (18). However, attempts to block HaCaT differentiation in an analogous LSE model by ectopic Id expression were unsuccessful because we were unable to generate lines overexpressing Id-1, Id-2, or Id-3 (data not shown),indicating that Id expression can promote keratinocyte death under certain circumstances. Indeed, Norton et al.(4) reported that stable transfection of rat embryo fibroblast cells by Id vectors resulted in apoptosis.

In vivo, proliferation is a feature of transit amplifying cells in the basal layer of the epidermis, and cell cycle withdrawal and accumulation of differential markers are concomitant with migration to the spinous layer. We observed significant Id expression in the basal layer, consistent with proliferation, although staining persisted in a subset of keratinocytes in the spinous layer. This may represent either continued proliferation in the suprabasal compartment or postmitotic Id expression, a phenomenon reported in spermatogenesis (19). Complete Id down-regulation was seen only prior to terminal differentiation and death, suggesting that Id proteins exert their influence at two discrete stages: in the promotion of proliferation and in the later regulation of terminal differentiation. Moreover, the transition of Id-1 from the cytoplasm of basal cells to the nucleus of spinous keratinocytes suggests a discrete mechanism of action. Unlike Id-1, Id-2 and Id-3 are known to be phosphorylated by cyclin E and cyclin A-cdk2 complexes in a cycle-dependent fashion, and sequestration of Id-1 in the cytoplasm may indicate an additional regulatory mechanism (4). Indeed,contrasting subcellular Id localization was observed in intestinal crypt epithelia, chondrocytes, and in spermatogenesis (15, 17, 19). Id proteins lack a nuclear localization sequence, and dimerization with E-proteins was shown to be a mechanism of nuclear translocation (20). However, nuclear colocalization of E2A with Id-2 and Id-3 but not Id-1 in the basal layer was observed in our experiments. The Ids dimerize avidly with all E-proteins; hence, it is possible that an active mechanism sequesters Id-1 from E-proteins in the cytoplasm, preventing inappropriate transit to the nucleus(21). This discrete pattern of Id localization may play a pivotal role in the homeostasis of the epidermis and other tissues.

Another novel observation was the columnar organization delineated by Id-2 and Id-3 expression. This suggests a complex architecture of epidermal organization and may be related to the epidermal proliferative unit as postulated by Potten (22).

Transformed SCC9 and HaCaT keratinocytes undergo only very limited differentiation in culture, and this correlated with continued Id expression under conditions of high calcium and high cell densities, which may result from autocrine stimulation of the EGF receptor (14). We reasoned that such Id dysregulation may be a feature of SCC in situ. SCC is thought to derive from a common compartment, the basal layer of the epidermis; however,different lesions were characterized by distinct levels of Id expression that correlated directly with disease course. In particular,poorly differentiated SCC exhibited the greatest Id immunoreactivity;such tumors have the greatest propensity to metastasize and hence, have the poorest prognosis. Lin et al. (16) reported a correlation between Id-1 expression and the aggressive phenotype of breast cancer cells, providing further evidence of an Id dosage effect in disease progression. However, we cannot exclude the possibility that Id expression reflects the stage of developmental arrest, and further analysis of the mechanisms of Id regulation in SCC is required. Id expression was absent from the foci of keratinized cells in well-differentiated SCC, and the association between decreased Id expression with the capacity for spontaneous differentiation provides further evidence that Id expression acts as a restriction point for keratinocyte maturation.

Although there is some redundancy in Id expression, their distinct expression profiles are indicative of discrete roles in epidermal maintenance. It is conceivable that the balance of Id proteins acts as a critical determinant in the creation of a permissive environment for keratinocyte maturation, perhaps via interaction with cell cycle components or regulation of an epidermis-specific class B HLH factor. Furthermore, we have evidence for at least two discrete modes of action, via regulation of proliferation and a later activity in the inhibition of terminal differentiation. The study of these mechanisms may provide much insight into the regulation of normal epidermal development and malignant transformation.

We thank Debbie Sanders for the preparation of LSEs and members of the laboratory, particularly Dr. Van Hoang and Andrew McLellan, for useful discussions. We are grateful to Lesley Morris of the Department of Histopathology, Addenbrooke’s Hospital, for help with IHC. We also extend our thanks to Drs. David Pawson, Gill Westgate, Paul Slusarewicz, and Scott Nicol at Unilever Research for critical reading of the manuscript and continued support.