Abstract
This study was performed to examine the carcinogenic effects of benzo[a]pyrene (B[a]P) and manufactured gas plant (MGP) residues on the hamster cheek pouch (HCP). Syrian hamsters were treated topically with a suspension of 2%, 10%, or 20% B[a]P or 50% or 100% MGP-7 (a mixture of residues from 7 MGP sites) in mineral oil for eight (short-term study) and sixteen, twenty, twenty-eight, and thirty-two weeks (long-term study). The short-term study showed that B[a]P induced p53 protein accumulation, indicative of genotoxic damage, as well as increased cell proliferation, hyperplasia, and inflammation, which is usually associated with promotional activity. In contrast, the MGP-7 presented only marginal p53 accumulation and induction of BrdU incorporation. In the long-term experiments, animals treated with 2% and 10% of B[a]P continued to show p53 protein accumulation as well as hyperplasia and increased cell proliferation and inflammation. By thirty weeks, all the animals treated with B[a]P had a 100% incidence of squamous cell carcinoma (SCC). Animals treated with 50% and 100% MGP-7 showed only weak hyperplasia and a low proliferation rate and accumulation of p53 protein through thirty-two weeks. Benzo[a]pyrene was highly carcinogenic when used at adequate doses. Manufactured gas plant residue, however, was not carcinogenic in this model.
Keywords
Introduction
Oral cancer is the sixth most common cancer in the world (Parkin et al. 1988). It is predicted that in 2008, there will be more than 35,300 new cases of oral cancer in the United States, and more than 7590 oral cancer–related deaths (Ries et al. 2008). One of the most unsettling statistics is a 60%five-year survival rate in oral cancer patients (Ries et al. 2007). Over the past several decades, the incidence of many cancers has been declining and their prognoses have been improving; however, oral cancer has not significantly decreased, nor have the prognoses improved during this same period. In the past, this disease has primarily affected middle-aged and older individuals, but the affected ages are descending, owing to the increased use of tobacco products among people as young as 12 years old (CDC 2005).
Approximately 90% of oral cancers are squamous cell carcinomas (SCC) (Burch et al. 1991). These carcinomas are highly malignant and metastatic, which explains the low survival rate. It has been well documented that these carcinomas develop in stages. From a clinical point of view, the earliest recognized stage of oral carcinogenesis is the leukoplakia, a white, potentially precancerous lesion histopathologically characterized by focal hyperplasia, hyperkeratosis, and variable degrees of atypia (Regezi and Sciubba 1989). Most of these precancerous lesions are reversible if exposure is ceased and lesions are treated early; however, some of these lesions will progress to SCC and eventually metastasize to various lymph nodes in the neck region (Regezi and Sciubba 1989).
One extensively studied animal model for oral SCC is the hamster cheek pouch (HCP), which has been used for decades to analyze the development of oral SCC (Salley 1954). The development of chemically induced lesions in this model parallels the development of those in human oral mucosa. Some of these similarities include progression from a normal stratified squamous epithelium to hyperplasia and hyperkeratosis followed by development of dysplasia, which progresses to in situ and invasive carcinomas (Salley 1954). In the early 1960s, studies were conducted by Morris to optimize and standardize experimental techniques using dimethylbenz(a)anthracene (DMBA) in the HCP (Morris 1961). Although DMBA has been proven consistently effective in HCP carcinogenesis studies, it has not been identified in human exposure material, and therefore it is not an actual environmental carcinogen. For this reason we considered it important to developa protocol using a more relevant and ubiquitous environmental carcinogen such as benzo[a]pyrene (B[a]P).
Benzo[a]pyrene is a polycyclic aromatic hydrocarbon found in abundance throughout our environment (IARC 1983), and it has recently been upgraded to a Group 1 carcinogen by the International Agency for Research on Cancer (IARC 2005). It is a product of incomplete combustion of organic materials, including those from forest fires and volcanic eruptions (Zedeck 1980), cigarette smoke produced by burning tobacco, charbroiled foods (IARC 1983), industrial pollution, automobile exhaust, and manufactured gas plant wastes (Lee et al. 1977; Stoker et al. 1975). Surprisingly, early experiments have indicated that B[a]P has weak or no carcinogenic properties in the HCP (Solt et al. 1987), despite the fact that B[a]P has been consistently shown to be a potent carcinogen for a great variety of organs and tissues, including the mouse skin model (DiGiovanni et al. 1980). As a consequence of these early studies, DMBA has been the carcinogen of choice for the HCP and has been used extensively in carcinogenesis and prevention studies (Shklar et al. 1987). We hypothesized that the failure of B[a]P to induce tumors in the HCP was owing to the use of inadequate protocols, and therefore in this study, we have optimized the experimental conditions for B[a]P carcinogenesis and have successfully established a B[a]P model of HCP carcinogenesis.
Another problem we wanted to address in this study is the effect of complex mixtures of carcinogens in the HCP model. Carcinogenesis in humans frequently occurs as a consequence of exposure to complex mixtures of carcinogens rather than a single carcinogen, and experiments with complex mixtures have been carried out in mouse skin and other relevant models, but not in the HCP or other models for the oral mucosa (Culp et al. 1998; Goldstein et al. 1994). Thus, we also explored the possibility of developing a model of oral carcinogenesis using manufactured gas plant (MGP) residues. Manufactured gas plant residues, also known as coal tar, are complex mixtures of carcinogens that contain B[a]P at concentrations of up to 6400 mg/kg (Weyand et al. 1991). It is estimated that between 3600 and 5500 MGP plants were established between 1800 and the mid-1900s, generating MGP residues and leaving behind numerous coal tar waste sites in close proximity to many heavily populated areas in the United States (U. S. Environmental Protection Agency 2004). Therefore, along with B[a]P, we also tested MGP to investigate the carcinogenicity of a complete mixture containing B[a]P and other environmental carcinogens in the HCP.
Materials and Methods
Materials
Male Syrian hamsters, four to six weeks old, were purchased from Harlan Sprague Dawley. Benzo[a]pyrene, DMBA, BrdU, and mineral oil were purchased from Sigma Chemical Company (St. Louis, MO). Number 4 camel hair paintbrushes were purchased from ThermoShandon (Pittsburgh, PA). Anti-BrdU antibody was purchased from Becton Dickinson (San Jose, CA). Anti-p53-DO7 antibody, secondary antibodies, and Avidin-Biotin Peroxidase Complex Kit were purchased from Vector Laboratories, Inc. (Burlingame, CA). 3-3′-diaminobenzidine (DAB) was purchased from BioGenex (San Ramon, CA).
MGP Residue #7
Manufactured gas plant residue #7 (MGP-7) was obtained from Electric Power Research Institute (Palo Alto, CA). This residue was obtained by mixing equal amounts (v/v) of residue from seven different MGP sites. The analytical composition of MGPs has been previously published (Hawthorne et al. 2002). The analytical composition of MGP-7 and a comparison with the composition sites was also previously published (Goldstein et al. 1998). The B[a]P content of MGP-7 is 1.8 mg/g, representing 2.3% of the total aromatic compounds. Other abundant aromatic compounds in MGP-7 are naphthalene (27.3%), 2-methylnaphthalene (13.9%), 1-methylnaphthalene (8.1%); acenaphthene (2.5%); fluorene (4.5%); anthracene (3.1%); and pyrene (6.3%), as well as many other aromatic and nonaromatic components (Goldstein et al. 1998; Hawthorne et al. 2002).
Experimental Design
Short-term Study
Twenty-eight male Syrian hamsters were housed four per cage at room temperature with wood chip bedding. They were provided food and water ad libitum. It has been experimentally estimated that each brushing contains 200 μL of designated compound to be applied topically into the right cheek pouch three times per week using a number 4 camel hair paintbrush. The hamsters were divided into seven groups of four hamsters each (Table 1). Group 1, the negative control group, was treated with mineral oil. Group 2, the positive control group, was treated with 0.5% DMBA solution (20 mM). Group 3 was treated with a 50% solution of MGP-7 mixed with mineral oil, Group 4 was treated with a 100% MGP-7, Group 5 was treated with a 2% B[a]P suspension (80 mM), Group 6 was treated with a 10% B[a]P suspension (400 mM), and Group 7 was treated with a 20% B[a]P suspension (800 mM). Mineral oil was used as the vehicle for B[a]P and DMBA. All groups were treated for eight weeks. Twenty-four hours after the last treatment they were injected intraperitoneally (i.p.) with BrdU (50 mg/kg body weight [b.w.]) one hour prior to euthanasia. The selection of doses of the carcinogens was based on published literature (including our own papers for the DMBA protocols) (Gimenez-Conti et al. 1990; Morris 1961). In the case of B[a]P, we selected the doses based on the previous work of Solt et al. (1987) that gave negative results, as well as the relative potency of DMBA vs. B[a]P in mouse skin, as shown previously in our laboratories (DiGiovanni et al. 1980). Doses of MGP were established by Drs. Rodriguez and Goldstein based on their previous experience with this compound (Rodriguez et al. 1997).
Long-term Study
One-hundred-twenty male Syrian hamsters were housed four per cage at room temperature with wood chip bedding. They were provided food and water ad libitum. Using a number 4 camel hair paintbrush, 200 μL of designated compound was applied topically into the right cheek pouch three times per week. Hamsters were divided into six groups of twenty animals each (Table 2). Group 1, the negative control group, was treated with mineral oil and Group 2, the positive control group, was treated with 0.5% DMBA. Group 3 was treated a 50% solution of MGP-7 mixed with mineral oil. Group 4 was treated with 100% MGP-7, Group 5 was treated with 2% B[a]P suspension, and Group 6 was treated with 10% B[a]P suspension. In the long-term experiment, we omitted treatment with 20% concentration of B[a]P after the short-term experiment showed intense inflammation and microulcerations in the HCP. Mineral oil was used as the vehicle for B[a]P. All animals were injected ip with BrdU (50 mg/kg b.w.) one hour prior to euthanasia. Four animals from each group were euthanized at the following time points: twelve weeks, sixteen weeks, twenty weeks, twenty-eight weeks, and thirty-two weeks.
Animal Welfare
Animal experiments were performed according to the Policy on Human Care and Use of Laboratory Animals guidelines for animal welfare (Grants.NIH.gov/grants/olaw/References/phspol.htm), which included approval by the Institutional IACUC.
Histopathology and Immunohistochemical Staining
Animals were euthanized and the right cheek pouch from each hamster was harvested, fixed in 10% neutral buffered formalin, processed, and embedded in paraffin. Sections 4 μm thick were cut and put on slides for hematoxylin and eosin (H & E) staining and immunohistochemistry. It was estimated that each slide contained approximately 13 cm of buccal epithelium. Hematoxylin and eosin–stained sections were employed for histopathological analysis according to the criteria previously described (Gimenez-Conti et al. 1990).
Cell proliferation was determined by BrdU incorporation following immunohistochemistry with an anti-BrdU antibody as previously described (Benavides et al. 2002). The slides were deparaffinized through several changes of solvent and rehydrated through gradients of ethanol and water. The tissue was then put into a solution of 3.0% H2O2 for ten minutes to block endogenous peroxidase. They were boiled in citrate buffer for ten minutes for antigen retrieval, incubated in 10.0%horse serum for twenty minutes, and then incubated with anti-BrdU at a dilution of 1:150 for one hour at room temperature. Biotinylated anti-mouse IgG was applied for thirty minutes. The reaction was developed with an Avidin-Biotin Peroxidase Complex kit and visualized with DAB. The BrdU labeling index (LI) was scored as the percentage of BrdU-positive cells in 600 cells along the basal layer, choosing five or six different random fields in 13 cm of epithelial pouch at 20× magnification (Benavides et al. 2002).
The accumulation of p53 protein was used as an indicator of genotoxic damage as previously reported by ours and other laboratories (Chang et al. 1996; Gimenez-Conti et al. 1996). After deparaffinization and 3% H2O2 treatment, the tissue was boiled in citrate buffer for ten minutes. The slides were incubated in 10% goat serum for thirty minutes followed by anti-p53-DO7 monoclonal antibody at a 1:100 dilution overnight at 4°C. Biotinylated anti-mouse IgG was applied for thirty minutes. The reaction was developed with an Avidin-Biotin Peroxidase Complex kit and visualized with DAB. The percentage of p53 positive cells was counted per 1500 cells of random fields in 13 cm of epithelial pouch at 20× magnification.
Results
Short-term Study
This study was performed to evaluate the chosen doses of the test articles using short-term markers as well as detecting doses with high toxicity before committing to a long-term experiment. Demonstrated in figure 1 Histopathologic analysis was performed on the HCP of all the animals of each experimental group (four hamsters per group). The HCP samples of Group 1 (negative control group treated with vehicle) were histologically normal. As expected, Group 2, the DMBA-treated hamsters, showed epithelial hyperplasia with increased hyperkeratosis after eight weeks of treatment. The epithelia also showed rete ridges, which is a characteristic reaction to irritation. The underlying connective tissue (lamina propria) was thicker than normal and was characterized by edema, congestion, and mild inflammation (increased cellularity, and abundance of polymorphonuclear cells [PMNs]). At the same time point, Groups 3 and 4 (the experimental groups treated with 50% MGP-7 and 100% MGP-7, respectively) presented only a slight increase in epithelial thickness and mild inflammation. The histopathological changes observed in the HCP treated with B[a]P at the eight-week time point (Group 5, 2% B[a]P; Group 6, 10% B[a]P; and Group 7, 20% B[a]P) were similar to the changes described for the eight-week DMBA-treated animals. The HCP samples from animals treated with higher doses of B[a]P (Groups 6 and 7) were characterized by enhanced inflammatory changes in the lamina propria. In particular, Group 7 also showed microulcerations in addition to more severe inflammation.
Proliferative changes in the epithelium were evaluated by BrdU incorporation. The percentage of basal cells positive for BrdU (BrdU LI) was 5.0% ± 1.74% for the mineral oil–treated animals. Dimethylbenz(a)anthracene induced a strong proliferative response (LI of 25.0% ± 2.70%), but the cheek pouches treated with 50% and 100% MGP-7 residue mixes were not statistically different compared to the normal epithelia LI of 6.4% ± 2.23% to 7.0% ± 2.11%. Benzo[a]-pyrene induced a proliferative response similar to DMBA, with LIs ranging from 20.5% ± 2.23% to 24.8% ± 3.45% (Table 3).
To determine the degree of genotoxic damage produced by the different treatments, we evaluated p53 nuclear accumulation by immunohistochemistry (Gimenez-Conti et al. 1996). Our data showed 100% of the animals in the mineral oil group were negative for p53 staining. In contrast, the DMBA-treated animals were positive for nuclear p53 accumulation in epithelial cells. In the groups treated with MGP-7, p53 protein accumulation was evident only in a small percentage of cells (7%). The B[a]P-treated hamsters, like the DMBA groups, also produced accumulation of p53 protein in the epithelium of the treated cheek pouches (Figure 2). Results of BrdU LI and p53 protein accumulation are represented in Table 3.
The results of the short-term experiments predicted that the chosen doses of B[a]P would be carcinogenic because they elicit genotoxic damage (as measured by p53 protein accumulation) and several markers of promotional activity such as hyperplasia, increased proliferation, and inflammation comparable to the carcinogenic dose of DMBA. Since the lower doses of B[a]P (2% and 10%) were positive for short-term markers and the highest dose (20%) showed signs of excessive toxicity, we omitted the 20% dose in the long-term experiments.
Long-term Study
Histopathological analysis was also performed in long-term–treated animals. The mineral oil–treated animals did not present any histologic alteration in the epithelium or stroma at any time point of the experiment. In accordance with institutional and federal regulations, all animals treated with 0.5% DMBA had to be sacrificed at sixteen weeks of treatment owing to the presence of multiple tumors. These tumors were classified histologically as SCC. Approximately 80% were well- to moderately differentiated exophytic SCC, as previously described (Gimenez-Conti et al. 1990). A smaller percentage of the tumors was endophytic and moderately to poorly differentiated. In addition to the multiple neoplastic lesions, the hamsters in this group also presented inflammatory changes similar to those described for the short-term experiments and multiple focal hyperplasias with various degree of dysplasia (Gimenez-Conti et al. 1990). The animals treated with MGP did not exhibit dysplastic or neoplastic lesions, and only diffuse epithelial hyperplasia was observed throughout the thirty-two–week experiment (Figure 3). In contrast, at twenty weeks, 100% of the 2% B[a]P-treated pouches had inflammatory changes and multiple foci of dysplasia in the epithelium. By thirty-two weeks, 100% of the animals had SCC in addition to the inflammatory changes described for twenty weeks. The histopathological analysis of the SCC showed exophytic and endophytic tumors with a variable degree of differentiation, from well differentiated to poorly differentiated. Unlike the tumors from the DMBA-treated hamsters, in this group, approximately 90% of the SCC were endophytic (well to moderately differentiated). The animals treated with 10.0% B[a]P showed similar results to the 2% B[a]P-treated animals, however, the higher dose produced a marked increase in hyperplasia, hyperkeratosis, and a mixed inflammatory infiltration characterized by the presence of PMNs and mononuclear cells (Figure 4). The histopathologic differences are illustrated in Table 4.
Discussion
The present study was carried out to examine the carcinogenic effects of B[a]P and an MGP, a complex mixture containing B[a]P, on the HCP. Unlike previous studies using B[a]P, these experiments demonstrated that B[a]P is highly carcinogenic in the HCP, however, MGP lacked carcinogenic activity in this model despite its effects on previously studied animal models (Culp et al. 1998; Rodriguez et al. 1997; Vesselinovitch 1990; Weyand et al. 1995).
Previous studies had suggested that B[a]P has weak or no carcinogenic activity in the HCP. Solt et al. (1987) used doses of 20 mM (500 mg) B[a]P twice per week, which resulted in minimal microscopic dysplastic lesions and only 1 SCC after forty weeks of treatment. As a result of these studies, DMBA has been considered as the standard carcinogen in the cheek pouch, whereas B[a]P was considered not to be active in this model. Based on our experience in mouse skin (DiGiovanni et al. 1980), we have speculated that the dose of B[a]P used by Solt et al. was too low to produce a carcinogenic response in the HCP. As predicted, when we used higher doses of B[a]P (80 and 400 mM), the HCP developed 100% incidence of premalignant lesions at twenty weeks of treatment and a 100% incidence of SCCs after thirty-two weeks. More interestingly, this study also showed that the lesions induced by B[a]P in the HCP are more similar to those in the human oral mucosa (Ackerman 1948; Smith et al. 1990) than lesions produced by treatment with DMBA. Most of the precancerous lesions in the B[a]P-treated hamsters were non-nodular (non-exophytic), with a smooth and even transition to adjacent normal epithelium. These lesions are remarkably similar to human precancerous lesions such as focal hyperkeratosis, clinically referred as leukoplakias, and carcinomas in situ (Regezi and Sciubba 1989). These lesions are characterized by focal hyperplasia with hyperkeratosis, irregular stratification, drop-shaped epithelial ridges, and varying degrees of dysplasia. These similarities extended into the fully invasive carcinoma, which, in most cases, are endophytic (Regezi and Sciubba 1989). In contrast, DMBA-produced lesions are primarily exophytic in nature (Gimenez-Conti et al. 1996; Morris 1961; Salley 1954; Shklar 1968).
We have previously shown that DMBA induced tumors are characterized by a signature mutation of the H-ras gene (Gimenez-Conti et al. 1992). In this regard, DMBA-induced exophytic tumors are similar to mouse skin tumors induced by DMBA and particularly to those induced by DMBA in combination with TPA (2-stage carcinogenesis protocols) (Balmain et al. 1984; Conti et al. 1989). In contrast, the majority of human oral carcinomas do not present alterations in the H-ras gene (Chang et al. 1991; Das et al. 2000; Rohatgi et al. 2005; Rumsby et al. 1990; Xu et al. 1998) and, as discussed previously, are essentially endophytic (Regezi and Sciubba 1989). The only exception is oral carcinomas associated with the use of betel, particularly in India. Those tumors are exuberant and exophytic and present the characteristic mutations of the H-ras gene (Das et al. 2000; Saranath et al. 1991). It would be interesting to investigate what genetic alterations occur during initiation or at very early stages of tumor development in B[a]P-treated pouches, and whether differences in the mutational spectrum of this tumor could explain the different biological behavior of the B[a]P-induced tumors.
Often chemical carcinogenesis on human tissues occurs by the insult of complex mixtures rather than a single carcinogen that is often the focus in experimental models. Since MGP residues are widely spread throughout nature, and the oral cavity is one of the first points of contact, we have considered it important to determine the carcinogenic effects of multiple doses of MGP on the HCP. We used here a well-characterized MGP mixture produced by mixing the residues of seven different sites (Goldstein et al. 1998) and that has been used in other carcinogenic models (Benavides et al. 2003; Culp et al. 1998; Goldstein et al. 1998; Rodriguez et al. 1997; Weyand et al. 1995). Based on the previous data of our collaborators, we expected MGP-7 to be carcinogenic in the HCP. However, our data showed that after thirty-two weeks of exposure, MGP produced only minimal hyperplasia, proliferation, and p53 protein accumulation in the buccal epithelium and did not induce dysplastic or neoplastic lesions.
Our results are in contrast with previous studies by Robinson et al. (1987) demonstrating the development of lung adenomas and forestomach SCCs produced by gavaging coal tar in A/J mice; Weyand et al. (1995) reporting that ingested MGP produced lung tumors in mice; and Rodriguez et al. (1997) showing MGP introduced by intraperitoneal injection was tumorigenic in mouse liver. However, it is not clear why previous MGP experiments resulted in tumors in the lung, liver, and forestomach, but only minimal hyperplasia in the HCP. Goldstein et al. (1994) have suggested that MGP potency varies between strains of mice, route of exposure, and dose rates, and therefore a direct extrapolation of results between different experimental systems is not warranted. Other explanations based on histological and physiological differences between different target tissues are also difficult to substantiate, particularly because MGP-7 has been carcinogenic also in the rodent forestomach, which bears a reasonably histologic similarity with the cheek pouch. Further complexity in understanding the lack of carcinogenicity of MGP-7 in the HCP is the highly complex chemical composition of this mixture, including multiple compounds with potential carcinogenic activity, others that may act as inhibitors, as well as others that can affect the metabolic pathway in the HCP.
Since other mixtures have not been assayed in the HCP, it is too early to advance any hypotheses. However, it will be important to test other complex mixtures in this system to advance our knowledge of the carcinogenic effect of the simultaneous action of more than one carcinogen in the oral mucosa. From the mechanistic point of view, it is important to mention that our experiments showed that the lack of carcinogenicity of MGP-7 correlates with very weak genotoxic activity of the compound as evaluated by p53 induction, as well as the absence of markers suggesting tumor promotional activity. Whether this lack of both genotoxic and promotional activity of MGP-7 in the HCP is the result of an intrinsic resistance of the HCP or of a lack of efficient metabolic activation of carcinogens present in MPG7 remains to be determined.
In conclusion, we have shown that at appropriate doses, B[a]P is a strong carcinogen in the HCP, but MGP is not carcinogenic in this model, even at doses that have been carcinogenic in other tissue sites. We have also shown that B[a]P-induced lesions, compared to DMBA-induced lesions, are more similar to those that develop in the human oral mucosa. The lack of carcinogenicity of MGP-7 warrants further investigations of other complex mixtures to determine whether the lack of activity of MGP-7 represents an intrinsic resistance of the HCP or the oral mucosa or just a unique characteristic of the particular complex mixture used in this study.
Footnotes
Figures and Tables
Acknowledgments
We thank R. Deen for manuscript preparation and C. Brown for figure preparation. This work was funded by the Electric Power Research Institute contract WO2963-04. This publication is made possible by grant P30ES007784 from the National Institute of Environmental Health Sciences.
Conflict of interest: The authors have not declared any conflict of interest.