The malignant capacity of skin tumours induced by expression of a mutant H-ras transgene depends on the cell type targeted

516 Research Paper

The malignant capacity of skin tumours induced by expression ofa mutant H-ras transgene depends on the cell type targetedKen Brown, Douglas Strathdee*, Sheila Bryson, Wendy Lambie and Allan Balmain†

Background: Pinpointing the cells from which tumours arise is a major challengein tumour biology. Previous work has shown that the targeted expression of amutant ras gene within the interfollicular cell compartment of mouse skin inducesthe formation of benign papillomas, but these do not spontaneously progress tomalignancy. We have investigated the carcinogenic effects of expressing thesame oncogene in a different population of epidermal cells.

Results: Expression of mutant ras from a truncated keratin 5 gene promoter,which directs expression to the follicular and interfollicular cells of newborn miceand the hair follicle cells of adults, stimulated the development of acanthotic areasin newborn mice. Within one week of birth, the acanthotic skin developed areas ofcarcinoma in situ and adult mice developed papillomas and keratoacanthomas,the latter having a high frequency of spontaneous malignant transformation tosquamous and occasionally spindle carcinomas. The benign tumours that arosehad several hallmarks of tumours at a high risk of malignant progression, includingsuprabasal cell proliferation and heterogeneous expression of keratin 13. Incontrast to tumours induced by expressing mutant ras under the control of thekeratin 10 or keratin 1 gene promoters, the formation of these lesions was notdependent on wounding or a tumour promoter.

Conclusions: Benign tumours that are at a risk of malignant conversion areprimarily derived from cells located within the hair follicle, and the nature of the cellin which tumour initiation occurs is a major determinant of malignant potential.

BackgroundUnderstanding the nature and location of the ‘target cell’for carcinogenesis — the cell in a tissue from which atumour is most likely to arise — is a central problem oftumour biology which has yet to be fully resolved. It isgenerally assumed that the major target must be one of arelatively small population of stem cells which reside,usually within a protected niche, in mammalian tissues[1]. Recent studies on the biology of immortalisation haveshown, however, that more differentiated cells, fromrenewing epithelia such as skin, are capable of acquiringan infinite life span, at least in cell culture [2]. This raisesthe possibility that many cells that have left the stem cellcompartment and are further along the lineage to the com-pletely differentiated state may, through acquisition ofsufficient genetic alterations, become capable of givingrise to malignant tumours. It is also possible that theremay be a continuum of target cells in tissues in which thestage of differentiation determines the malignant capacity,as is seen in cells of the haematopoietic lineage [3]. Nev-ertheless, most human tumours arise in self-renewingepithelial tissues, and our knowledge of the relationshipsbetween stem cells, target cells, and the malignant capac-ity of tumours is extremely sparse.

We have previously attempted to address some of thesequestions using two-stage chemical carcinogenesis inmouse skin — in which papillomas are triggered by thetopical application of a tumour initiator followed by atumour promoter [4]. Several investigations have shownthat tumour initiation is irreversible and that the initiatedcell can persist within the epidermis for months to yearsafter the initial carcinogenic insult [5]. This agrees withthe generally accepted notion that tumour initiation mustalter a cell belonging to a long-lived population within astem cell compartment. Two separate populations of ker-atinocytes are implicated in the renewal of the epidermis.A number of studies have located putative stem cells inthe basal layer of the interfollicular epidermis [6–8],though others have suggested the bulge region of the hairfollicle as the location of the stem cell compartment [9,10].Significantly, much of the available data implicate the hairfollicle as playing a central role in skin tumorigenesis [11].

The malignant conversion of papillomas to carcinomas isa relatively rare event in two-stage carcinogenesis, occur-ring at a frequency of around 5–10%. Furthermore,subsets of papillomas have either a reduced or anincreased [12–14] risk of progression, and these can be

Address: CRC Beatson Laboratories, Departmentof Medical Oncology, Alexander Stone Building,University of Glasgow, Garscube Estate,Switchback Road, Bearsden, Glasgow, G61 1BD,UK.

Present addresses: *Roslin Institute, Roslin,Midlothian, EH25 9PS, UK. †Onyx Pharmaceuticals,3031 Research Drive, Richmond, California 94806,USA.

Correspondence: Allan BalmainE-mail: [email protected]

Received: 20 October 1997Revised: 16 March 1998Accepted: 17 March 1998

Published: 9 April 1998

Current Biology 1998, 8:516–524http://biomednet.com/elecref/0960982200800516

© Current Biology Ltd ISSN 0960-9822

distinguished by changes in the expression of phenotypicmarkers such as TGF-β [15,16], α6β4 integrin [17] andkeratin 13 [18]. The relationship between these low-riskand high-risk papillomas is presently unclear. Some dif-ferences in progression frequencies might arise as a con-sequence of different initiating events within cells of thesame target population [15]. It seems extremely unlikelythat this is the explanation for the different progressionfrequencies of papillomas initiated with dimethylbenzan-thracene (DMBA) as the vast majority of tumours thatarise after treatment with this carcinogen have the sameinitiating mutation at codon 61 of the Harvey ras (H-ras)gene [19].

In order to address these potentially related questions oftarget cell location and malignant conversion, we haveexamined the consequences of expressing mutant H-rasin different cellular compartments of the epidermis. Pre-viously, transgenic mice were generated in which expres-sion of a mutant H-ras gene was targeted to theinterfollicular compartment of the epidermis by using acytokeratin 10 promoter [20]. This promoter directsexpression mainly to suprabasal cells of the epidermis,but also to a small population of basal cells [21]. The phe-notype of the animals expressing the transgene (K10ras)was, primarily, increased differentiation of the epidermis,but they also developed benign papillomas at sites of pro-motional stimuli such as scratching, wounding or treat-ment with 12-O-tetradecanoylphorbol-13-acetate (TPA).This result indicated that interfollicular cells can give riseat least to benign tumours. Similar observations werereported by Greenhalgh et al. [22] for animals (HK1.rasmice) in which ras expression was driven by a modifiedpromoter from a human keratin 1 gene, which has aslightly more widespread expression in basal cells but isstill restricted to the interfollicular epidermis. A commonfeature, however, in all of these studies was the absenceof spontaneous malignant conversion in the benigntumours which arose.

We describe here the results of expressing a mutantH-ras transgene from a truncated keratin 5 promoter

element. In adult mice, this promoter induced transgeneexpression almost exclusively in the hair follicle, whereone of the putative epidermal stem cell compartments islocated. A total of four transgenic mouse lines wereobtained, each of which developed spontaneous papil-loma-like or keratoacanthoma-like skin tumours. Thelatter type frequently underwent spontaneous conversionto squamous carcinomas and occasionally to undifferenti-ated spindle cell carcinomas. Targeted expression ofmutant ras in the hair follicle region of mouse skin cantherefore reproduce the complete spectrum of events inmultistage skin carcinogenesis.

ResultsHair follicle expression of lacZ driven by a truncatedkeratin 5 promoterA 1.3 kb fragment (K5) of promoter sequence from thebovine cytokeratin III gene [23] (equivalent to humankeratin 5) was used to target transgene expression to thebasal layer of the epidermis. Cytokeratin 5 is normallyfound in the basal layers of stratified epithelia and in theouter root sheath cells of the hair follicle [24]. In order toanalyse the expression pattern driven by the K5 promotersequence, two lines of transgenic mice were produced(K5lacZ mice) which contained a reporter construct con-sisting of the gene encoding β-galactosidase under thecontrol of the K5 sequence. In newborn mouse skin, theextent of transgene expression, as detected by β-galactosi-dase staining, was highly variable. The occasional areas ofexpression that were detected were highly mosaic andstaining was restricted to cells within the outer root sheathof the hair follicle and to patches of interfollicular basalcells (Figure 1a). Analysis of adult skin samples demon-strated that fewer hair follicles showed positive β-galac-tosidase staining, and staining in the interfollicularepidermis was almost undetectable (Figure 1b). Therestricted expression pattern conferred by the shortenedpromoter, which differs from the pattern described previ-ously [25], might have been advantageous for our studies,as in previous experiments we were unable to obtainviable transgenic animals expressing a potent oncogenesuch as H-ras from a full length (5.3 kb) promoter [20].

Research Paper The malignant capacity of H-ras-induced skin tumours Brown et al. 517

Figure 1

LacZ expression in the skin of K5lacZ mice.Sections of dorsal skin from (a) neonate and(b) adult K5lacZ animals, stained forβ-galactosidase activity and counterstainedwith eosin. LacZ expression can be seen inthe outer root sheath cells of the hair follicleand in the interfollicular basal epidermis inneonates; in the adult epidermis, however, theexpression is restricted to the hair follicle.

Skin abnormalities in neonatal K5ras transgenic miceThe 1.3 kb K5 promoter fragment was subsequently usedto drive the expression of a cDNA fragment encoding con-stitutively activated H-Ras. From 30 founder transgenic(K5ras) animals that contained this transgene, 15 had anabnormal skin phenotype at birth. This typically pre-sented as weal-like regions of raised and thickened skin(Figure 2a), which developed within several days intogrossly thickened and hypopigmented papillomatousmasses (Figure 2b). Histological examination of the skinsshowed hyperplastic acanthotic downgrowths(Figure 3c,d). Normal hair follicles and associatedmelanocytes were absent from the acanthotic areas,although remnants of follicle structures were occasionallyfound (Figure 3c). Non-acanthotic skin from thesenewborn animals was usually hyperplastic, with appar-ently normal hair follicles. In several sections, thickeningof the skin was due to regions of papillomatous hyperpla-sia (Figure 3b) or sessile papillomas, which similarlylacked normal hair follicles, but which contained remnantsof follicle structures. These findings are consistent withabnormal proliferation and differentiation of the hair folli-cle being involved in the development of both the acan-thosis and papillomatosis.

Within several days, the acanthotic lesions usually became3–10 mm thick and highly disorganised in structure, withthe surface becoming macroscopically papillomatous(Figure 2b). Histological sections showed the rapid devel-opment of dysplasia, with areas typical of carcinoma in situ

and invasive carcinoma (Figure 3e,f) becoming progres-sively apparent between 7 and 10 days after birth, atwhich age the animals died or had to be sacrificed. Thisdiffers from the results found with HK1.ras mice [22] forwhich the initial hyperplasia converted to massive hyperk-eratosis within several weeks.

Spontaneous tumour progression in adult transgenic miceFive K5ras founder animals that appeared normal at birthgradually developed phenotypic abnormalities in adult-hood, including papillomatous hyperplasia, vibrissae folliclehyperplasia (Figure 2c), and overt tumour formation. Sevenfounders gave rise to lines in which a proportion of trans-gene-positive offspring were born with acanthotic lesions ordeveloped spontaneous skin tumours at anywhere frombetween one week and one year of age. Four transgeniclines chosen for further study, KT932, KT754, KT760, andKT769, showed variable transgene copy number(1–15 copies per cell) but very similar phenotypes. Only afraction of the mice that contained the transgene devel-oped skin tumours, and from an initial study which exam-ined 464 animals over 65 weeks, 141 (30%) developed skintumours. The tumour number in individual mice variedfrom a solitary tumour to cases of animals with more thanten tumours. Though occasional skin lesions were histolog-ically characterised as papillomas, the tumours more com-monly exhibited prominent acanthotic characteristics andclosely resembled keratoacanthomas (Figure 4a). Occasion-ally, keratoacanthomas were observed in association withabnormal or metaplastic hair follicles. Additionally, cystic

518 Current Biology, Vol 8 No 9

Figure 2

Phenotype of K5ras mice. (a) Normal (left)and transgenic (right) littermates 2 days afterbirth. Note the abnormal skin wrinkling andthickening in the transgenic pup. (b) Examplesof 6 day old pups with extensive (top) andlocalised (centre) areas of abnormalepidermis, compared to a normal littermate(bottom). The gross thickening and lack ofhair in the affected regions is clearly apparent.(c) The muzzle area of an adult transgenicanimal in which a small papilloma-type lesionis associated with each vibrissa follicle.(d) Three keratoacanthomas in various stagesof exfoliation in the dorsal skin of an adulttransgenic animal.

epithelial lesions lined by squamous epithelium, classifiedas type 3 keratoacanthomas [26] (Figure 2d), wereobtained. None of the spontaneous tumours that arose inthe mice showed any signs of regression. Examination ofhistological sections from keratoacanthomas often showedareas of local invasion (Figure 4b), and about half of thelesions progressed to squamous (Figure 4c) or spindle cell(Figure 4d) carcinomas within a few months. In contrast,true pendunculated papillomas which arose at lower fre-quency in the K5ras mice showed no evidence of malig-nant progression. The notion that many of the spontaneoustumours had a high probability of malignant progressionwas supported by labelling tumour sections with bromod-eoxyuridine (BrdU) or staining them for keratin 13. Anti-BrdU antibody staining of BrdU-labelled acanthotic lesionsand keratoacanthomas showed that they had a high BrdU-labelling index and frequently showed labelled cells in thesuprabasal layers of the lesions (data not shown). Thesefeatures are characteristic of the high-risk tumours thatarise during two-stage chemical carcinogenesis in mouseskin [15]. Similarly, results from the staining of frozen

tumour sections with an antibody directed against keratin13 showed that many of the acanthotic lesions and keratoa-canthomas contained regions of heterogeneous expressionof keratin 13, similar to that observed in high-risk papillo-mas or carcinomas (data not shown) [18,27].

Transgene expression in tumours and adjacent hairfolliclesThe presence of mutant human Ras protein in the affectedtissues of K5ras transgenic mice was confirmed by westernblotting. Protein extracts from tumours and normal skinwere immunoprecipitated with an anti-H-Ras antibody andvisualised using a pan-Ras antibody (Figure 5a). MutantH-Ras protein was seen in tumours, but was undetectable innormal tissue. In order to characterise more fully the expres-sion pattern of the transgene in the tumours, in situ hybridis-ation, using a probe specific for the SV40 intron andpolyadenylation sequences of the transgene construct, wascarried out on sections from normal skin and tumours fromtransgenic mice. No hybridisation signal was detected in fol-licles or interfollicular epidermis in unaffected skin


Figure 3

Histopathology of neoplasms in newbornmice. (a) Normal skin and (b–f) lesions,stained with haematoxylin and eosin, fromK5ras mice. (a) Normal dorsal skin from anewborn mouse. (b) Papillomatoushyperplasia from a 6 day old transgenicanimal. Note the absence of normal hairfollicles, though some hyperplastic follicularstructures can be seen on the right hand sideof the section (arrow). (c) Region ofacanthotic dorsal skin from a 6 day old animal,with the vestigial remnant of a hair follicle bulb(arrow). (d) Area from the margin of a lesionshowing the transition from hyperplasticfollicles containing intact hair shafts (left), tosquamous epithelial structures which containonly occasional remnants of normal follicles(centre and right). (e) Section of an acanthoticlesion from a 7 day old mouse which exhibitsa more disorganised and dysplasticepithelium, but which has not yet progressedto invasive carcinoma. (f) Acanthotic lesiontaken from a 6 day old animal with an area ofinvasive squamous carcinoma (bottomcentre).


(Figure 5b). A detectable signal was observed, however, intumours and in follicles closely adjacent to tumours(Figure 5c). This finding was consistent in acanthoticlesions, papillomas, keratoacanthomas and carcinomas. Theexpression pattern was not uniform in all basal cellsthroughout the tumour, but was mosaic, with some areas ofthe tumour expressing very high transgene levels. Thehyperplastic epidermis at the margins of the tumour was fre-quently negative for detectable H-ras transgene expression.

Tumour development is not dependent on wounding or apromotional stimulusOne prevalent feature of previously described transgeniclines expressing ras constructs in the epidermis under thecontrol of the keratin 10 [20] or keratin 1 [22] promoter wasthat papillomas tended to develop at sites of wounding orscratching and, in particular, in response to treatment witha tumour promoter. In contrast, spontaneous skin tumoursarising in the K5ras mice were neither more prevalent in,nor restricted to, areas of grooming, scratching or wound-ing, such as at the edges of ear-tag or tail-biopsy sites. Agroup of 94 transgene-positive mice, some with pre-exist-ing spontaneous tumours, were treated twice weekly withTPA for 14 weeks to examine the effect of tumour promo-tion. The results are summarised in Table 1. Though thehyperplastic response to TPA treatment was normal, thegreat majority of these animals appeared refractory to anypromotional stimulus, and TPA treatment did not result inan increase in tumour incidence over that found inuntreated control animals. In contrast, five mice, two withpre-existing papillomas, developed extensive papillomato-sis after only two applications of TPA, similar to the reac-tion reported for HK1.ras mice [28]. Unlike the reaction in

the HK1.ras mice, however, the papillomatosis in K5rasmice did not regress and persisted for four weeks aftertreatment, at which stage the mice were sacrificed.

DiscussionConsequences of mutant ras expression in differentepidermal target cellsThe mouse H-ras gene can initiate skin tumours whenmutated by specific chemical carcinogens [13,19] or intro-duced into epidermal cells by retroviral application in vivo[29] or in vitro [30]. The critical target cell which suffers themutation and gives rise to the tumours has not been identi-fied. Several groups have now attempted to reproduce thefeatures of chemical tumour initiation by expressing amutated H-ras gene in the epidermis of transgenic miceusing different keratin promoters. Initial studies, whichused either the keratin 10 or the keratin 1 promoter to targetras expression to the interfollicular epidermis, demonstratedthat this cell population can give rise to papillomas at sitesexposed to wounding or tumour promoters [20,22]; malig-nant progression of these tumours occurs very infrequently,however, and only after a long latency period [28].

A substantially different phenotype is seen in animalsexpressing the same mutant ras gene in hair folliclesunder the control of a truncated keratin 5 promoter. Inadult mice, in which K5-directed expression is seen onlyin the hair follicle, with virtually undetectable levels inthe interfollicular epidermis, there was no evidence ofhyperkeratosis, as reported for K10ras transgenic mice.Spontaneous keratoacanthomas frequently developed,consistent with a follicular origin. Importantly, many ofthese tumours progressed to squamous carcinomas and/or

Figure 4

Histopathology of spontaneous neoplasms inadult mice. Sections of skin tumours fromadult K5ras mice were stained withhaematoxylin and eosin. (a) A spontaneoustumour from an adult animal, illustrating thetypical keratoacanthoma-like nature of thelesions. (b) A higher magnification view of aregion from the base of the keratoacanthomashown in (a). Note the epithelial cell nests inthe dermal stroma and the lack of a cleartumour–stroma interface, typical of acarcinoma. (c) Section of a squamouscarcinoma. (d) Area of undifferentiatedspindle carcinoma.


spindle cell carcinomas, thereby reconstituting all thestages of tumour induction in mouse skin.

It could be argued that the dramatic contrast between theconsequences of expressing the same Ras protein using

the keratin 10 promoter and the K5 sequence is due toquantitative differences in the levels of expression drivenby the two promoters in the same target cell population.Quantitative alterations in the level of signalling throughthe Ras pathway can influence the decision between pro-liferation and differentiation in mouse PC12 cells [31], ordetermine cell fate during development [32]. We thinkthat this is extremely unlikely to be the explanation of theobserved results, however. The levels of ras transgeneexpression in unaffected skin are extremely low in bothK10ras and K5ras mice, as assessed both by in situ hybridi-sation and by western blotting, suggesting that major dif-ferences in the Ras protein levels cannot account for thephenotypes seen. In contrast, the use of both in situhybridisation and reporter gene constructs have clearlydemonstrated the cell specificity in the expression pat-terns driven by the K10 promoter and the K5 sequence.Although expression levels of the human ras transgenewere too low to be detected in morphologically normalskins of the K5ras mice, in situ hybridisation clearlyshowed specific transgene expression in hair follicles adja-cent to areas of dysplasia, but not in hyperproliferativeinterfollicular epidermis. These data, together with thepatterns of hair follicle expression seen in K5lacZ mice andin transgenic mice expressing the viral E1A protein underthe control of the keratin 5 promoter [33], argue stronglythat the contrasting phenotypes seen in the different ras-expressing transgenic animals are due to differential tar-geting of epidermal cell populations by the respectivekeratin promoters, rather than to quantitative differencesin the levels of transgene expression in these mice.

The role of tumour promotersAn important feature that distinguishes the K5ras trans-genic animals from others previously reported is theabsence of a strong effect of treatment with a tumour pro-moter on tumour yield. Both the K10ras [20] and theHK1.ras [22] mice developed papillomas at sites of wound-ing or irritation, such as the tail, footpad or adjacent to eartags. The HK1.ras mice were extremely sensitive to treat-ment with TPA, forming very large papillomatous lesions,which regressed if treatment was interrupted. In a study of

Table 1

The effect of TPA treatment upon tumour formation in K5ras mice.

Before TPA treatment After TPA treatmentNumber of mice Number of tumours Number of tumours Mice with new tumours

TPA-treated group With tumours 39 63 63 0*Without tumours 55 7 7†

Untreated group With tumours 31 50 51 1Without tumours 43 4 4

*Three mice in this group with papillomas developed acute papillomatosis in response to TPA treatment. †Two mice in this group developed acutepapillomatosis in response to TPA treatment.

Figure 5

H-ras transgene expression and localisation in tumours of K5ras mice.(a) Western blot of immunoprecipitated H-Ras protein from fourspontaneous lesions arising in K5ras mice: lane 1, human mutantH-Ras from cell line 1d1a; lane 2, acanthotic lesion; lane 3,keratoacanthoma; lane 4, squamous carcinoma; lane 5, papilloma.The upper band corresponds to the mutant human H-Ras protein,which contains an amino acid substitution at Val12, and the lowerband is the endogenous mouse H-Ras protein, which has a fastermobility. Only the lower band is detected in extracts from normalmouse cells (data not shown). (b,c) Bright-field images showing insitu hybridisation of transgene message to 35S-labelled SV40antisense riboprobe. (b) In a section from unaffected dorsal skin of atransgenic animal, H-ras transgene mRNA is not detectable. (c) Asection from the edge of a keratoacanthoma. Note that H-ras mRNAis expressed in a follicle adjacent to the lesion (centre), and in someareas of the tumour (right), but is absent from hyperplastic skin at thetumour margin (top left).

Current Biology

(b)

(a) 1

MutantNormal

2 3 4 5

(c)

more than 90 K5ras animals, however, only 5 showed astrong papilloma response to TPA similar to that reportedfor several lines of HK1.ras mice. The remainder of theanimals, although their skins showed the usual hyperplas-tic response as a consequence of chronic TPA treatment,did not develop more tumours than untreated transgenicmice. These results suggest that the target cell populationthat gives rise to carcinomas is not strongly stimulated bytumour promoters, a result compatible with observationsthat the incidence of chemically induced carcinomas is notincreased [34] by more extensive promoter treatment.

The hair follicle in carcinogenesisAn association between hair follicle expression of ras andskin tumour development has also been noted in a singleline of transgenic mice containing a mutant H-ras alleleunder the control of a ζ-globin promoter. These animalsdo not develop spontaneous tumours, but papillomas, car-cinomas and fibrosarcomas arise in response to woundingor TPA treatment [35]. H-ras expression has been shownto occur only in response to such promotional stimuli inthese mice, and in situ hybridisation has shown that thesite of induced transgene expression is in the upper part ofthe hair follicle [35]. The unique features of this mouseline make comparisons with the present results, and withchemically induced tumour development in normal mice,very difficult.

A great deal of earlier evidence implicates the hair follicleas a source of target cells for mouse skin tumour develop-ment. Treatment, with tumour initiators and promoters, ofmice that had been subjected to epidermal abrasion,resulting in removal of most of the interfollicular epider-mis but leaving the follicles intact, still gave rise to sub-stantial numbers of papillomas and carcinomas [36],showing that some target cells must exist within the folli-cles. More recently, Lavker and colleagues showed thatthe bulge region of the hair follicle is the most likely loca-tion of the epidermal stem cells [9], and that stimulationof hair follicle growth before chemical initiation canincrease the number of papillomas which develop [11].Additional evidence that target cells exist within hair folli-cles came from analysis of tumour development in trans-genic animals expressing ornithine decarboxylase underthe control of keratin 6 or keratin 5 promoters [37]. Malig-nant progression frequency was not assessed in any ofthese studies, however, and no conclusions could bereached as to the differential locations of cells capable ofgiving rise to either benign papillomas or carcinomas. Evi-dence also exists for the presence within the epidermalbasal cell compartment of both a putative stem cell popu-lation [6,7], and of target cells for tumour development[38]. It has now been shown by specific targeting of theinitiating ras gene to the interfollicular cells [20,22] or tothe follicular compartment (this report) that target cellscan exist in both locations, but that those within the hair

follicle region are much more likely to form premalignanttumours which progress to carcinomas.

ConclusionsExpression of mutant H-ras in the hair follicle of the skinproduces epithelial tumours with a high malignant capac-ity. This contrasts markedly with the effects seen in previ-ous studies of targeted H-ras expression in theinterfollicular epidermis, in which malignant conversion israrely observed. Tumour promoters appear to be moreimportant for the continued outgrowth of cells from theinterfollicular population that express mutant ras genes,but these are less likely to become malignant. We con-clude that, within the skin, there is a continuum of cellpopulations capable of being initiated by mutation and ofgiving rise to tumours, and that the nature and location ofthe initiated cell determines the malignant capacity of theresultant tumour.

Materials and methodsGeneration of the K5ras gene construct A 1.3 kb Hind III–NruI fragment of the bovine keratin 5 promoter (K5)was isolated from plasmid pKIII-5′ (provided by J. Jorcano) and placedinto the plasmid pIC20R. Subsequently, a 0.85 kb Bgl II–BamHI frag-ment from pRSVβglobin [39], containing the SV40 small t intron andpolyadenylation signal (SV40poly(A)), was inserted into the Bgl II site.Finally, a human H-ras cDNA fragment, containing a Val12 mutation,was isolated as an EcoRI fragment from plasmid pEXV-KT (a gift from C.Marshall) and cloned into the XhoI site of the K5–SV40poly(A) cassette.

Generation and identification of transgenic miceThe 3.1 kb K5ras construct was excised with BamHI and ClaI, purified,and microinjected into the pronuclei of B6CBF2 hybrid fertilisedembryos, using established techniques [40]. Genomic DNA wasobtained from mice by tail-tip biopsy [41] and transgenic animals wereidentified by PCR analysis. A 258 bp fragment of the H-ras transgenewas amplified using primers 5′-GATGGGGAGACGTGCCTGTTG(bases 139–159) and 3′-GTCCTGAGCCTGCCGAGATTC (bases396–376). A 386 bp fragment of the thyroid stimulating hormone βchain was co-amplified as an internal control for the efficiency of theamplification reaction [42]. The amplification conditions used were 35cycles of 94°C for 30 s, 55°C for 30 s and 72°C for 30 s. Transgenecopy number was determined by Southern blotting of KpnI-digestedDNA and hybridisation with the H-ras cDNA probe.

HistologyTissues for histological examination were fixed in 4% phosphate-buffered formalin overnight, dehydrated and embedded in paraffin bystandard methods. Sections (6 µm) were stained with haematoxylinand eosin. Photomicrography was performed on an Olympus BX2microscope using Ektachrome film (Kodak).

Immunoprecipitation and western blottingStandard procedures were followed. Briefly, tumour samples were pul-verised after snap-freezing in liquid nitrogen and dissolved in lysisbuffer containing Triton X-100 (Sigma). Samples were preclearedusing anti-rat IgG Protein-A–Sepharose, followed by immunoprecipita-tion using the anti-H-Ras monoclonal antibody YA6-172 [43] and anti-rat IgG Protein-A–Sepharose. Resin pellets were washed in RIPAbuffer, and the samples resuspended in Laemmli buffer, prior to beingrun on 12.5% SDS–polyacrylamide gels and electroblotted onto nitro-cellulose (Schleicher & Schuell). Immunoblotting was carried out usingmouse monoclonal pan-Ras antibody Ras11 (Oncogene Science), andbinding visualised using horseradish-peroxidase-conjugated secondary


antibody with ECL (Amersham) chemiluminescence detection andautoradiography.

In situ hybridisationTumour samples were fixed overnight in fresh 4% paraformaldehyde at4°C, dehydrated and embedded in paraffin wax. Sections (6 µm) wereused for in situ hybridisation with 35S-labelled SV40 antisense orsense riboprobes [44]. Autoradiographic exposures were for 10 days.Photomicrography was performed on an Olympus BX2 microscopeusing Delta film (Ilford).

AcknowledgementsWe thank Chris Marshall for providing the H-ras cDNA construct, JoséJorcano for the keratin 5 promoter plasmid and Dennis Roop for the anti-K13 antibody. Our thanks to Stephen Bell and Maria Henry for assistancewith animal care, to David Tallach for artwork and to Elizabeth Duffie andRosemary Akhurst for assistance with in situ hybridisation. This work wassupported by the Cancer Research Campaign of Great Britain. All animalexperiments were performed in accordance with the UK Animals (ScientificProcedures) Act 1986.

References1. Cairns J: Mutation selection and the natural history of cancer.

Nature 1975, 255:197-200.2. Barrandon Y, Morgan JR, Mulligan RC, Green H: Restoration of

growth potential in paraclones of human keratinocytes by a viraloncogene. Proc Natl Acad Sci USA 1989, 86:4102-4106.

3. Greaves MF: Differentiation-linked leukemogenesis inlymphocytes. Science 1986, 234:697-704.

4. Hecker E, Fusenig NE, Kunz W, Marks F, Thielmann HW:Carcinogenesis. A Comprehensive Survey. Volume 7;Cocarcinogenesis and Biological Effects of Tumor Promoters. NewYork: Raven Press; 1982.

5. Van Duuren BL, Sivak A, Katz C, Seidman I, Melchionine S: Theeffect of aging and interval between primary and secondarytreatment in two stage carcinogenesis in mouse skin. Cancer Res1975, 35:502-505.

6. Potten CS: The epidermal proliferative unit: the possible role ofthe central basal cell. Cell Tissue Kinet 1974, 7:77-88.

7. Mackenzie IC, Fusenig NE: Regeneration of organized epithelialstructure. J Invest Dermatol 1983, 81(Suppl):189S-194S.

8. Jones PH, Harper S, Watt FM: Stem cell patterning and fate inhuman epidermis. Cell 1995, 80:83-93.

9. Sun TT, Cotsarelis G, Lavker RM: Hair follicular stem cells: thebulge-activation hypothesis. J Invest Dermatol 1991,96(Suppl):77S-78S.

10. Rochat A, Kobayashi K, Barrandon Y: Location of stem cells ofhuman hair follicles by clonal analysis. Cell 1994, 76:1063-1073.

11. Miller SJ, Wei Z, Wilson C, Dzubow L, Sun T, Lavker RM: Mouse skinis particularly susceptible to tumor initiation during early anagenof the hair cycle: possible involvement of hair follicle stem cells. JInvest Dermatol 1993, 101:591-594.

12. Scribner JD, Scribner NK, McKnight B, Mottet NK: Evidence for anew model of tumor progression from carcinogenesis and tumorpromotion studies with 7-bromomethylbenz(a)anthracene. CancerRes 1983, 43:2034-2041.

13. Brown K, Buchmann A, Balmain A: Carcinogen-induced mutations inthe mouse c-Ha-ras gene provide evidence of multiple pathwaysfor tumour progression. Proc Natl Acad Sci USA 1990, 87:538-542.

14. Hennings H, Shores R, Mitchell P, Spangler EF, Yuspa SH: Inductionof papillomas with a high probability of conversion to malignancy.Carcinogenesis 1985, 6:1607-1610.

15. Glick AB, Kulkarni AB, Tennenbaum T, Hennings H, Flanders KC,O’Reilly M, et al.: Loss of TGF-

b expression in mouse epidermisand epidermal tumors is associated with hyperproliferation and ahigh risk for malignant conversion. Proc Natl Acad Sci USA 1993,90:6076-6080.

16. Cui W, Kemp CJ, Duffie E, Balmain A, Akhurst RJ: Lack oftransforming growth factor-b1 expression in benign skin tumorsof p53null mice is prognostic for a high risk of malignantconversion. Cancer Res 1994, 54:5831-5836.

17. Tennenbaum T, Yuspa SH, Grover A, Castronovo V, Sobel ME,Yamada Y, et al.: Extracellular matrix receptors and mouse skincarcinogenesis: altered expression linked to appearance of earlymarkers of tumor progression. Cancer Res 1992, 52:2966-2976.

18. Nischt R, Roop DR, Mehrel T, Yuspa SH, Rentrop M, Winter H, et al.:Aberrant expression during two-stage mouse skin carcinogenesisof a type I 47-kDa keratin, K13, normally associated with terminaldifferentiation of internal stratified epithelia. Mol Carcinogen 1988,1:96-108.

19. Quintanilla M, Brown K, Ramsden M, Balmain A: Carcinogen-specificmutation and amplification of Ha-ras during mouse skincarcinogenesis. Nature 1986, 322:78-80.

20. Bailleul B, Surani MA, White S, Barton SC, Brown K, Blessing M, etal.: Skin hyperkeratosis and papilloma formation in transgenicmice expressing a ras oncogene from a suprabasal keratinpromoter. Cell 1990, 62:697-708.

21. Schweizer J, Kinjo M, Furstenberger G, Winter H: Sequentialexpression of mRNA-encoded keratin sets in neonatal mouseepidermis: basal cells with properties of terminally differentiatingcells. Cell 1984, 37:159-170.

22. Greenhalgh DA, Rothnagel JA, Quintanilla MI, Orengo CC, Gagne TA,Bundman DS, et al.: Induction of epidermal hyperplasia,hyperkeratosis, and papillomas in transgenic mice by a targetedv-Ha-ras oncogene. Mol Carcinogen 1993, 7:99-110.

23. Casatorres J, Navarro JM, Blessing M, Jorcano JL: Analysis of thecontrol of expression and tissue specificity of the keratin 5 gene,characteristics of basal keratinocytes. J Biol Chem 1994,269:20489-20496.

24. Byrne C, Fuchs E: Probing keratinocyte and differentiationspecificity of the human K5 promoter in vitro and in transgenicmice. Mol Cell Biol 1993, 13:3176-3190.

25. Ramirez A, Bravo A, Jorcano JL, Vidal M: Sequences 5¢ of the bovinekeratin 5 gene direct tissue- and cell-type-specific expression ofa lacZ gene in the adult and during development. Differentiation1994, 58:53-64.

26. Ghadially FH: The role of the hair follicle in the origin andevolution of some cutaneous neoplasms of man andexperimental animals. Cancer 1961, 14:801-816.

27. Cano A, Gamallo C, Kemp CJ, Benito N, Palacios J, Quintanilla M, et al.:The expression pattern of the cell adhesion molecules E-cadherin,P-cadherin and a6b4 integrin is altered in premalignant skintumours of p53-deficient mice. Int J Cancer 1996, 65:254-262.

28. Greenhalgh DA, Wang X, Dominey AM, Rothnagel JA, Roop DR:Transgenic mouse models for skin carcinogenesis. In SkinCancer: Mechanisms and Human Relevance. Edited by Mukhtar H.Boca Raton: CRC Press, Inc; 1995:317-330.

29. Brown K, Quintanilla M, Ramsden M, Kerr IB, Young S, Balmain A: v-ras genes from Harvey and BALB murine sarcoma viruses canact as initiators of two-stage mouse skin carcinogenesis. Cell1986, 46:447-456.

30. Roop DR, Lowy DR, Tambourin PE, Strickland J, Harper JR, BalaschakM, et al.: An activated Harvey ras oncogene produces benigntumours on mouse epidermal tissue. Nature 1986, 323:822-824.

31. Traverse S, Gomez N, Paterson H, Marshall C, Cohen P: Sustainedactivation of the mitogen-activated protein (MAP) kinase cascademay be required for differentiation of PC12 cells. Comparison ofthe effects of nerve growth factor and epidermal growth factor.Biochem J 1992, 288:351-355.

32. Som T, Klaparthi VSR: Developmental decisions in Aspergillusnidulans are modulated by ras activity. Mol Cell Biol 1994,14:5333-5348.

33. Missero C, Serra C, Stenn K, Dotto GP: Skin-specific expression ofa truncated E1a oncoprotein binding to P105-Rb leads toabnormal hair follicle maturation without increased epidermalproliferation. J Cell Biol 1993, 121:1109-1120.

34. Battalora MSJ, Conti CJ, Aldaz CM, Slaga TJ, Johnston DA,DiGiovanni J: Regression and progression characteristics ofpapillomas induced by chryserobin in SENCAR mice.Carcinogenesis 1996, 17:955-960.

35. Hansen LA, Tennant RW: Follicular origin of epidermal papillomasin v-Ha-ras transgenic TG.AC mouse skin. Proc Natl Acad Sci USA1994, 91:7822-7826.

36. Argyris TS, Slaga TJ: Promotion of carcinomas by repeatedabrasion in initiated skin of mice. Cancer Res 1981, 41:5193-5195.

37. O’Brien TG, Megosh LC, Gilliard G, Soler AP: Ornithinedecarboxylase overexpression is a sufficient condition for tumorpromotion in mouse skin. Cancer Res 1997, 57:2630-2637.

38. Morris RJ, Fischer SM, Slaga TJ: Evidence that a slowly cyclingsubpopulation of adult murine epidermal cells retains carcinogen.Cancer Res 1986, 46:3061-3066.


39. Mulligan RC, Howard BH, Berg P: Synthesis of rabbit b-globin incultured monkey kidney cells following infection with a SV40 b-globin recombinant genome. Nature 1979, 277:108-114.

40. Hogan B, Beddington R, Constantini F, Lacy E: Manipulating theMouse Embryo, 2nd edn. New York: Cold Spring Harbour LaboratoryPress; 1994.

41. Laird PW, Zijderveld A, Linders K, Rudnicki MA, Jaenisch R, Berns A:Simplified mammalian DNA isolation procedure. Nucleic AcidsRes 1991, 19:4293.

42. Walter CA, Nasr-Schirf D, Luna VJ: Identification of transgenic micecarrying the CAT gene with PCR amplification. Cancer Res 1989,7:1065-1070.

43. Furth ME, Davis LJ, Fleurdelys B, Scolnick EM: Monoclonalantibodies to the P21 products of the transforming gene ofHarvey murine sarcoma virus and of the cellular ras gene family.J Virol 1982, 43:294-304.

44. Cui W, Fowlis DJ, Cousins FM, Duffie E, Bryson S, Balmain A, et al.:Concerted action of the TGF-b1 and its type II receptor in controlof epidermal homoeostasis in transgenic mice. Genes Dev 1995,9:945-955.


Because Current Biology operates a ‘Continuous PublicationSystem’ for Research Papers, this paper has been publishedon the internet before being printed. The paper can beaccessed from http://biomednet.com/cbiology/cub — forfurther information, see the explanation on the contents page.

Documents

The malignant capacity of skin tumours induced by expression of a mutant H-ras transgene depends on the cell type targeted