The Role of Skin Painting in Predicting Lung Cancer

Abstract

The mouse skin cancer model provides an important system for studying mechanisms involved in the various stages of carcinogenesis and for bioassaying tobacco smoke constituents and additives for carcinogenic/cocarcinogenic and tumor-promoting properties as well as for identifying compounds that may inhibit tumor formation and malignant conversion. In addition, it is an excellent model for studying the formation of precancerous lesions as well as squamous cell carcinomas. It relates very well to other squamous cell carcinoma models and contributes to better understanding of the human epithelial cancers including lung cancer. The SENCAR mouse is an established model system demonstrated to be more sensitive than the B6C3F1 or Swiss CD-1 strains in the initiation/promotion skin-painting test method. Although the relationship between mouse skin tumors and any manifestation of the toxicity of tobacco smoke and other complex environmental mixtures in humans is unknown, the skin-painting model is the only assay that provides a practical method of obtaining a tumorigenic end point with cigarette smoke condensates and other complex mixtures. This assay provides a rapid response with relative ease of quantification of various parameters of tumorigenic response including tumor incidence, latency, multiplicity, and malignancy.

Many investigators now support the concept that carcinogenesis is a multistep and multifactorial phenomenon and animal models, especially mouse skin carcinogenesis model and mouse and rat liver carcinogenesis models, have significantly contributed to our understanding of the important events in the development of human cancer and its chemoprevention. Thus, the model mimicking this process can serve as a useful method for the study of cancer induction and chemoprevention (Slaga et al. 1995; Hursting et al. 1999; Marks and Furstenberger 2000). Treatment with 7, 12-dimethylbenz(a)anthracene (DMBA) and 12-O-tetradecanoylphorbol-13-acetate (TPA) in mice is a widely used and reliable model for skin cancer study (Slaga 1984; DiGiovanni 1992). Carcinogenesis depends on inherited and acquired susceptibility factors, on exposure to initiation factors, i.e., exogenous and endogenous carcinogens, and on promotion and progression factors.

With the experimental design of Berenblum (1941), which employed a two-step procedure (benzo[a]pyrene and croton oil) for the treatment of mouse skin, a useful animal model became available to study the multistep nature of carcinogenesis. Friedewald and Rous (1944) were among the first investigators to analyze the process of skin tumorigenesis and they defined the concept of tumor initiation (benzo[a]pyrene and 3-methylcholantrene) and promotion (coal tar) and cocarcinogenesis in general terms. Initiation-promotion studies in mouse skin were then being pursued by Shubik, Boutwell, Hecker, Slaga as well as many others (reviewed by Slaga 1984). The sequential application of croton oil (the active ingredient being TPA) and turpentine to initiated mouse skin has been reported to induce a synergistic response, suggesting a qualitative difference between these two promoters (Boutwell 1964). Similar results were observed when substituting wound healing for turpentine as a tumor-promoting stimulus.

Although the above early studies (Boutwell 1964) first suggested that the tumor promotion phase of mouse skin carcinogenesis had two operationally distinct stages, only subsequent extensive studies on mouse skin tumor promotion (Slaga et al. 1980), confirmed later by another group (Furstenberger et al. 1981), led to further development of this two-stage concept. Currently, the standard two-stage promotion protocol involves initiation followed by one to four applications of TPA (stage I) and then by multiple applications of a weak papilloma promoting agent such as mezerein or 12-O-retinylphorbol-13-acetate (RPA) (stage II) (Slaga 1983). In 1980, the term “multistage tumor promotion” was introduced (Slaga et al. 1980), leading to the concept of multistage carcinogenesis.

MULTISTAGE CARCINOGENESIS

The induction of cancer (carcinogenesis) is a multistage process that occurs over a long period of time, i.e., decades in humans. Its stages have been defined experimentally as initiation, promotion, and progression. Initiation involves mutation of cellular DNA, resulting in the activation of oncogenes (Bishop 1991) and the inactivation of tumor suppressor genes (Marshall 1991). Initiation is thought to be irreversible and consist of a single gene mutation that is caused in most cases by environmental genotoxic agents such as chemicals, radiation, and viruses. Oncogenes can also be activated by chromosomal translocations and gene amplifications. Studies in the human colon indicate that the carcinogenic process involves multiple genetic alterations in a staged fashion (Fearon and Vogelstein 1990).

Promotion follows initiation and involves the process of gene activation such that the latent phenotype of the initiated cell becomes expressed through cellular selection and clonal expansion. This can occur through a variety of mechanisms including toxicity, terminal differentiation, or mitoinhibition of the non-initiated cells and mitogenesis of the initiated cells (Slaga et al. 1995). Promotion occurs over a long period of time and is reversible in its early stages. The breadth of the available data as well as the multistage nature of tumor promotion suggests that this process, which is now thought to occur in most tissues where cancer can be induced or where it occurs spontaneously, may involve the interaction of a number of endogenous factors as well as environmental factors such as chemicals, radiation, viruses, and diet and nutrition, thus, unifying all current areas of cancer research (Slaga et al. 1995). In human cancer, smoking, environmental factors such as asbestos, hydrocarbons, radiation, hormones, alcoholic beverages, as well as diet and nutrition, are now thought to have more of a promotional influence in the multistage carcinogenesis process (Walaszek et al. 2004).

The last step leading to cancer is called progression. Progression involves genetic alterations that result in the conversion of benign tumors into malignant neoplasms capable of invading adjacent tissues and metastasizing to distant sites (Slaga et al. 1995). The additional genetic alterations thought to be required for neoplastic progression often occur faster that would be expected from the statistics of accidental genotoxic insults due to so called genetic instability. The concept of genetic instability implies that while environmental genotoxic agents generally cause cancer initiation, the additional mutations required for neoplastic progression may be attributed to endogenous reaction and factors such as detoxification and removal of damaged cells by programmed cell death. Genetic instability may happen due to the errors in DNA replication, spontaneous hydrolytic alterations of DNA such as depurination and deamination in combination with an impaired ability of premalignant cells to repair DNA damage, or due to oxidative DNA damage (Li et al. 1997). Oxidative DNA damage occurs through an overactivation or disregulation of metabolic reactions, which in turn give rise to reactive oxygen species (ROS) such as hydroxyl and superoxide anion radicals, singlet oxygen, hydrogen peroxides, and nitrogen oxide as well as to free radicals and peroxides. Growing evidence implicates both oxygen and organic free radical intermediates in the biomolecular interactions, which contribute to each of the multistep stages of carcinogenesis (Hanausek et al. 2003). Many exogenous chemicals implicated in initiation, promotion, and progression can be activated to radical intermediates, which can serve as electrophiles or participate in reactive oxygen generating redox cycling processes. Similarly, the increased and/or continuous production of reactive oxygen species by endogenous sources can create and increase oxidative state in cells and organs contributing to the promotion and progression of cancer (Li et al. 1997).

Cancer is a disease in which there is an uncontrolled proliferation of cells that express varying degrees of fidelity to their precursor cell of origin (Marks and Furstenberger 2000). Most human tumors have a long history of pathological development during which they pass through several preneoplastic and premalignant stages before they become malignant. The clinical stages of tumor development have been correlated with specific genetic alterations such as activation of proto-oncogenes and deletion of tumor suppressor genes. By providing confirmation of molecular defects, human cancer genetics strongly, though indirectly, supports the concept of multistage carcinogenesis. The implication here is that malignant neoplasia is the result of multiple genetic defects successively accumulating over a period of time (Marks and Furstenberger 2000). Multiple progressive stages have been defined for colon cancer, i.e., aberrant crypt foci, polyps, adenomas, and carcinomas (Fearon and Vogelstein 1990). However, only animal models can provide direct information on the underlying mechanisms and enable a final proof of the multistage concept. As indicated above, the animal models of multistage carcinogenesis that are presently fairly well defined are skin cancer in mice and liver cancer in rats (Slaga et al. 1995). Both models allow a clear-cut distinction and mechanistic insight of individual stages of carcinogenesis.

MURINE MODELS OF CHEMICAL SKIN CARCINOGENESIS

Mouse skin tumors can be induced by sequential application of subthreshold dose of carcinogen (initiation stage) followed by repetitive treatment with a noncarcinogenic promoter (promotion stage). The initiation stage requires only a single application of either direct or an indirect carcinogen and it is in essence irreversible step; the promotion stage, however, is initially reversible but later becomes irreversible. A single large dose of a carcinogen, such as DMBA, is capable of inducing skin tumors in mice, in which papillomas occur after a relatively short latency period (10 to 20 weeks). Carcinomas develop after a longer period (20 to 60 weeks). If the dose is lowered, it becomes necessary to administer DMBA repeatedly to induce tumors. If it is progressively reduced, a subthreshold dose of DMBA is reached that will not give rise to tumors over the lifespan of the mouse. If a tumor promoter, such TPA, is then applied repeatedly to the back of the mice previously initiated with a single subthreshold dose of DMBA, multiple papillomas appear after a short latency period, followed by squamous cell carcinomas after a much longer period. If the promoter is repeatedly applied but there is no initiation by DMBA, generally no tumors are seen or only a few, but they never exhibit a dose-response relationship (Slaga 1984; DiGiovanni 1992; Slaga et al. 1995, 1996). If initiation of tumors in mice is accomplished with a subthreshold dose of carcinogen, such as DMBA, an excellent dose response is seen using TPA as the promoter. Likewise, a good dose response is seen with benzo[a]pyrene, or DMBA, as tumor initiators, when the promoter dose is held constant (Slaga et al. 1982).

In summary, the real hallmark of the two-stage carcinogenesis in mouse skin relates to the irreversibility of tumor initiation. A delay of up to 1 year between the application of the initiator and the beginning of the promoter treatment provides a tumor response similar to that observed when the promoter is given only one week following initiation. Unlike the initiation stage, the promotion stage is reversible and requires a certain frequency of application to induce tumors. The progression stage is defined as those events occurring after the initial appearance of a skin tumor and represents the transition from benign (papilloma) to malignant (squamous cell carcinoma) lesion and finally to metastatic tumors. The results from the murine skin models of carcinogenesis appear to be of relevance for a more in-depth understanding of the human epithelial cancers including colorectal (Marks and Furstenberger 2000) as well as lung cancer. Similarities of the mechanisms involved in chemically induced experimental skin and lung carcinogenesis are shown in Tab. 1.

Because of the stratum corneum, there is a difference in absorption between skin and pulmonary epithelia. In skin studies, solvents such as acetone are used to get most lipid-soluble carcinogens into the skin. Most lipid-soluble carcinogens are easily absorbed into pulmonary cells. There have not been any studies on the mucocilliary element related to different carcinogens as a class. The polycyclic aromatic hydrocarcarbon (PAH) carcinogens give similar DNA-binding kinetics in skin cells and pulmonary cells. In terms of removing the stratum corneum, carcinogens of various types get into basal cells easier but this causes the skin to proliferate and replace the removed stratum corneum, which increases the carcinogenic effect of many carcinogens.

LUNG CARCINOGENESIS AND SMOKING

Today, lung cancer is the most common fatal malignancy of both men and women in the United States. Cigarette smoking remains the largest risk factor in the causation of lung cancer. Eighty-five percent to 90% of all lung cancers occur in smokers. Therefore, an especially high-risk population can be identified, for example a smoking history of 30 pack years or greater. The risk of developing lung cancer is 15-fold greater in smokers than in non-smokers. This relative risk of developing lung cancer decreases from the 15-fold level to about 3-fold in former smokers who have quit for 15 years or more (Baron and Rohan 1996). Despite this decreased risk, we do know that more than 50% of all new lung cancers occur in former smokers (Blott and Fraumeni 1996).

The morphologic changes in the bronchial epithelium caused by cigarette smoking were first noted four decades ago (Auerbach et al. 1957, 1961). Since then, considerable attention has been given to cytological evaluation of early changes predating the appearance of invasive cancer (Knudson 1960; Auerbach et al. 1978). Histological changes, which have been described in such individuals, include loss of cilia, basal cell hyperplasia, the presence of atypical nuclei, and squamous cell hyperplasia (Nasiell 1963; Auerbach et al. 1979). Because of the strong association of these bronchial epithelial changes with carcinogen exposure, and because of the widespread presence of these changes in the bronchial tree of lung cancer patients, squamous cell metaplasia and dysplasia has been proposed as a histological and cytological marker of lung carcinogenesis.

A number of studies have begun to examine the expression of tumor suppressor genes that are frequently dysregulated in lung cancer in the setting of preneoplastic lesions. The p53 tumor suppressor gene, at chromosome p17, is one of the most frequent mutated genes in many cancers. It has been found to be abnormal in about one third of mild to moderate bronchial dysplasias and carcinomas in situ (Bennett et al. 1993). Deletions of portions of chromosome 3p have been found in a large percentage of lung cancer specimens (Chung et al. 1995). Numerous studies have shown that 3p loss is an early and frequent event during carcinogenesis (Hung et al. 1995). Inactivation of the p53 gene (17p13) appears to have a role in lung cancer progression similar to its role in colorectal and other epithelial cancers (Chung et al. 1995). One area in which human lung tumors differ from mouse skin tumors from DMBA/TPA protocols is the early and frequent mutation of p53; this is a late event in the DMBA/TPA model. However, p53 mutations were recently shown (Rebel et al. 2001) to be early events in skin carcinogenesis induced by chronic ultraviolet B (UVB) irradiation in SKH-1 mice.

Of the dominant proto-oncogenes that have been characterized in lung carcinogenesis, K-ras is the most extensively studied. K-ras mutations have been documented in atypical alveolar hyperplasia (a possible preneoplastic lesion for adenocarcinoma) (Westra et al. 1996), and in 30% to 50% of lung adenocarcinomas (Mills et al. 1993). Mutations of the K-ras oncogene can play a role in the development of not only lung adenocarcinomas but also of a subset (about 8%) of squamous cell carcinomas (Vachtenheim et al. 1995). K-ras mutations have also been detected in the sputum of patients up to four months prior to the diagnosis of lung cancers positive for K-ras (Mao et al. 1994). In contrast, the c-jun oncogene has been found to be expressed in 88% of bronchial atypical lesions, in 40% of alveolar atypical lesions, but only in 31% of invasive lung cancers, implying an early but reversible expression.

As to the murine models of lung cancer, the preponderance of lung adenocarcinomas induced in A/J mice by a wide variety of chemical carcinogens exhibit mutations in the K-ras oncogene (You et al. 1989). However, a recent study (Wang et al. 2004) failed to observe K-ras mutations in squamous cell carcinomas of the lung from either NIH Swiss or A/J mice treated topically with nitrosoalkylureas.

The inability to quantitatively assign the increased risk associated with smoking to a specific chemical constituent or class of chemical constituents has frustrated smoking and health researchers for almost 50 years. Part of this frustration stems from the inability to induce squamous cell pulmonary carcinoma in laboratory animals exposed via inhalation to tobacco smoke in early studies conducted from 1936 through the late 1980s (Campbell 1936; Henry and Kouri 1984, 1986). Recently, the model developed by Witschi (1998, 2000) has been recommended to study the effects of smoking. In this model, tumors (alveolar/bronchiolar adenomas and carcinomas) are induced in A/J mice by exposure to a mixture of 89% cigarette side-stream smoke and 11% mainstream smoke. The animals are exposed for 5 months, and then allowed a 4-month recovery period. A small but reproducible and significant increase in lung tumor multiplicity in this model was observed, from one tumor per mouse to about 2.5. Studies have shown that the increase in tumor multiplicity observed in this model is due to a component of the gas phase of tobacco smoke (Witschi 2005). A disadvantage of this model is that it is difficult to differentiate individual contributions of various tobacco smoke constituents to its overall tumorigenicity. Strengths and weaknesses of the A/J mice inhalation model are discussed elsewhere (Witchi 2005).

On the other hand, there have been some notable successes in extrapolating biological activity from cigarette smoke composition. For example, the mutagenicity of cigarette smoke condensate (CSC) as measured by some strains of Salmonella in the Ames assay was found to results from the pyrolysis of tobacco proteins (Clapp et al. 1999). In addition, sister-chromatid exchange (SCE) clastogenic activity has been reported to be influenced by aldehydes and ketones in the vapor phase of cigarette smoke (Bombick et al. 1997). Although there are these and other examples of specific classes of smoke components that can be correlated with adverse responses in the in vitro tests, the goal of mechanistically explaining the carcinogenic risk associated with the complex cigarette smoke aerosol has remained elusive. Although many reports have addressed cigarette mainstream (MS) components categorized as tumorigenic agents, cocarcinogens, or tumor promoters, the nature and concentrations of MS components listed as tumorigens have received much less attention (Rodgman 1992; Smith et al. 1997, 2000, 2001). To put the biological properties of the complex mainstream mixture cigarette in perspective, not only should the presence and level of components defined as carcinogens, mutagens, and promoters be considered, but also the presence of anticarcinogens, inhibitors, and antimutagens should be given equivalent consideration. The mouse skin painting with MS CSC or its constituents remains the bioassay of choice in the mechanistic studies of the biological effects of tobacco smoking. It appears to be a very predictive in vivo assay to assess MS CSC and its constituents for carcinogenic, cocarcinogenic, and tumor-promoting activities as well as anticarcinogenic and antipromoting activities.

MOUSE SKIN PAINTING IN CIGARETTE SMOKE CONDENSATES STUDIES

The mouse skin-painting model has been used extensively over the years to investigate the tumorigenic activity of cigarette smoke condensates (Wynder and Hoffman 1964) as well as other complex mixtures such as particulate emissions (Nesnow, Triplett, and Slaga 1982). The SENCAR mouse is an established model system demonstrated to be more sensitive than the B6C3F1 or Swiss CD-1 strains in the initiation/promotion skin-painting test method (Slaga 1986; National Toxicology Program 1996). Although the relationship between mouse skin tumors and any manifestation of the toxicity of complex mixtures in humans is unknown, the skin-painting model is the only assay that provides a practical method of obtaining a tumorigenic end point with cigarette smoke condensates. This assay provides a rapid response with relative ease of quantification of various parameters of tumorigenic response including tumor incidence, latency, multiplicity, and malignancy.

The model has also been reported to be applicable to assigning relative lung tumor potencies for some complex mixture carcinogens (Albert et al. 1983; Lewtas et al. 1983; Nesnow and Lewtas 1991). Specifically, four complex environmental mixtures that have been associated with human lung cancer have been bioassayed in mouse skin: coke oven emissions, diesel exhaust, roofing tar emissions, and cigarette smoke condensate (Nesnow, Triplett, and Slaga 1982). These emissions have been shown to be associated with the induction of respiratory cancer in exposed populations. The tumorigenic effects of these complex environmental mixtures were determined in SENCAR mice using the tumor initiation/promotion and complete carcinogenesis protocols. Based on both mouse skin papilloma multiplicity data and the tumor incidence data, cigarette smoke condensate was found to be the least tumorigenic (Nesnow and Lewtas 1991). When the relative human cancer risk of these respiratory carcinogens was calculated (coke oven:roofing tar:diesel:cigarette smoke relative risks equal to 1.0:0.39:0.075:0.0024), cigarette smoke was also found to be the least carcinogenic. The correlation between the mouse skin tumor incidence/tumor multiplicity data and the cancer risk estimates was quite good, with a correlation constant of 0.95 and the slope value of 0.89 (see Figure 2 in Nesnow and Lewtas 1991). Thus, mouse skin data may be used to predict the risk of lung cancer from exposure to potential human respiratory carcinogens.

The great bulk of early experimental studies on lung carcinogenesis by cigarette smoking has been directed toward the polycyclic aromatic hydrocarbons (PAH) in the so called “tar” fraction or condensate and most of them have used a standardized, repeated painting of mouse skin for tumor production (reviewed by Rubin 2002). These studies led to the conclusion that tobacco smoke condensate is primarily a tumor-promoting agent with weak carcinogenic activity (van Duuren et al. 1971). However, the promoters in cigarette smoke are not as strong as TPA. Also, the initiation-promotion procedure does not approximate the long-term exposure of cigarette smokers, which exposes the smoker repeatedly to all the constituents at the same time. It may appear that cocarcinogenesis, which came to be defined as repeated, simultaneous application of a carcinogen and agents that enhance its activity, more closely simulates the conditions of exposure of the smoker and that procedure demonstrates a number of co-carcinogens in cigarette smoke (van Duuren and Goldschmidt 1976; Slaga et al. 1979). Although it is unknown how cocarcinogens produce their effect, it is certainly not by direct action on DNA. The mode of action may resemble the proliferogenic, selective stage of promotion (Rubin 2002). The cocarcinogens that have been extensively studied such as phorbol esters, decane, and pyrene do not covalently bind to DNA but bring about epigenetic changes leading to an enhancement of tumor response.

The mouse skin tumor model developed for use in short-term carcinogenicity testing utilizing the SENCAR mouse as the standard strain for both initiating and promoting effects (Slaga 1986; National Toxicology Program 1996) may have some limitations. The use of a strain that is more sensitive than other strains may increase the chance of detecting strain-specific effects without relevance to other strains and species, including humans (Enzmann et al. 1998). Although both initiation and promotion stages are monitored, the stage of progression is not and the interference of irritant activities may not be excluded. Finally, the topical route of exposure is not the major route of potential human exposure. Nevertheless, if the dose selection is appropriate, it can be a very meaningful short-term bioassay (Enzmann et al. 1998).

A review of previous CSC studies using the mouse skin initiation/promotion assay indicates that a number of factors are necessary to produce a reproducible, quantitative assay. These include (a) a standardized CSC collection method (i.e., freshly collected and painted CSC); (b) a quantitative dose-response; (c) short-term results; (d) a test species and strain that is readily available and is sensitive to the materials under evaluation; and (e) the ability to periodically verify responsiveness with a reference cigarette. The protocol presented in a recent report (Meckley et al. 2004) appears to meet these criteria and provides a highly reproducible, quantitative study design to assess the tumor promotion activity of CSC, as shown in the following examples.

A comparative study (Mecklay et al. 2004) of tobacco burning and heating cigarettes found that repeated application of smoke condensate from tobacco-burning reference cigarettes to chemically initiated SENCAR mouse skin promoted the development of tumors in a statistically significant and dose-dependent manner, whereas condensate from prototype cigarettes that primarily heat tobacco promoted statistically fewer tumors. Based on the recognized correlation between sustained, potentiated epidermal hyperplasia and tumor promotion, more in-depth tests were conducted (Curtin et al. 2004, 2006) to examine the utility of selected short-term analyses for discriminating between condensates exhibiting significantly different promotion activities. In vitro analyses assessing the potential for inducing cytotoxicity (ATP bioluminescence) or free radical production (cytochrome c reduction, salicylate trapping) demonstrated significant reductions when comparing condensate collected from prototype cigarettes to reference condensate. Short-term in vivo analyses conducted within the context of a mouse skin initiation/promotion protocol (i.e., comparative measures of epidermal thickness, proliferative index, ornithine decarboxylase expression, myeloperoxidase activity, leukocyte invasion, mutation of Ha-ras, and formation of modified DNA bases) provided similar results. Reference condensate induced statistically significant and dose-dependent increases (relative to vehicle control) for nearly all indices examined, whereas prototype condensate possessed a significantly reduced potential for inducing changes that are regarded as consistent with sustained epidermal hyperplasia and/or inflammation. Collectively, these data support the contention that selected short-term analyses associated with sustained hyperplasia and/or inflammation are capable of discriminating between smoke condensates with dissimilar tumor-promotion potentials.

Thus, based on excellent dose-response relationships observed in different studies, the mouse skin-painting assay might be suitable for assessing the relative biological activities of CSC from divergent tobacco-burning cigarettes and tobacco-heating cigarettes (Meckley et. al. 2004; Curtin et al. 2004, 2006) as well as CSC from cigarettes containing different additives (Coggins et al. 1982; Gaworski et al. 1999). The mouse skin painting may also serve to study individual tobacco smoke constituents including carcinogens, cocarcinogens, mutagens, and tumor promoters (Rodgman 1992). It may also be helpful in identifying natural and synthetic chemopreventive agents able to suppress lung carcinogenesis caused by tobacco smoking (Walaszek et al. 2004; Wang et al. 2004).

SUMMARY

Mouse skin painting with main stream cigarette smoke condensate or its constituents remains the bioassay of choice in the mechanistic studies of the biological effects of tobacco smoking. In fact, it detects more lung carcinogens, cocarcinogens, and tumor promoters than any other model. The overall biological effects of the complex mainstream tobacco smoke mixture depend not only on the presence and level of components identified as carcinogens, cocarcinogens, mutagens, and tumor promoters, but also on the presence of components identified as anticarcinogens, inhibitors, and antimutagens. The mouse skin cancer model provides an important system for studying mechanisms involved in the various stages of carcinogenesis and for bioassaying particulate phase constituents of tobacco smoke and cigarette additives for carcinogenic/cocarcinogenic and tumor-promoting properties as well as anticarcinogenic and anti–tumor-promoting properties. A mouse skin model to evaluate gaseous phase components of tobacco smoke needs yet to be developed. The mouse skin-painting model is an excellent model for studying the formation of precancerous lesions (papillomas) as well as squamous cell carcinomas. It relates very well to other squamous cell carcinoma models, and contributes to better understanding of the human epithelial cancers including lung cancer.