Skip Navigation


ToxSci Advance Access originally published online on January 12, 2006
Toxicological Sciences 2006 90(2):400-418; doi:10.1093/toxsci/kfj101
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
90/2/400    most recent
kfj101v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (5)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Jackson, M. A.
Right arrow Articles by Dunnick, J. K.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Jackson, M. A.
Right arrow Articles by Dunnick, J. K.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2006. Published by Oxford University Press on behalf of the Society of Toxicology. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

Genetic Alterations in Cancer Knowledge System: Analysis of Gene Mutations in Mouse and Human Liver and Lung Tumors

Marcus A. Jackson*, Isabel Lea*, Asif Rashid{dagger}, Shyamal D. Peddada{ddagger} and June K. Dunnick{ddagger},1

* Integrated Laboratory Systems, Inc., Research Triangle Park, North Carolina 27709; {dagger} Alpha-Gamma Technologies Inc., Raleigh, North Carolina 27609; and {ddagger} National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina 27709

1 To whom correspondence should be addressed at National Institute of Environmental Health Sciences, MD EC-35, P.O. BOX 12233, Research Triangle Park, NC 27709. Fax: (919) 541-4255. E-mail: dunnickj{at}niehs.nih.gov.

Received September 9, 2005; accepted December 9, 2005


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUPPLEMENTARY DATA
 REFERENCES
 
Mutational incidence and spectra for genes examined in both human and mouse lung and liver tumors were analyzed using the National Institute of Environmental Health Sciences (NIEHS) Genetic Alterations in Cancer (GAC) knowledge system. GAC is a publicly available, web-based system for evaluating data obtained from peer-reviewed studies of genetic changes in tumors associated with exposure to chemical, physical, or biological agents, as well as spontaneous tumors. In mice, mutations in Kras2 and Hras-1 were the most common events reported for lung and liver tumors, respectively, whether chemically induced or spontaneous. There was a significant difference in Kras2 mutation incidence for spontaneous versus induced mouse lung tumors and in Hras-1 mutation incidence and spectrum for spontaneous versus induced mouse liver tumors. The major gene changes reported for human lung and liver tumors were in KRAS2 (lung only) and TP53. The KRAS2 mutation incidence was similar for spontaneous and asbestos-induced human lung tumors, while the TP53 mutation incidence differed significantly. Aflatoxin B1, hepatitis B virus, hepatitis C virus, and vinyl chloride all caused TP53 mutations in human liver tumors, but the mutation spectrum for each agent differed. The incidence of KRAS2 mutations in human compared to mouse lung tumors differed significantly, as did the incidence of Hras and p53 gene mutations in human compared to mouse liver tumors. Differences observed in the mutation spectra for agent-induced compared to spontaneous tumors and similarities in spectra for structurally similar agents support the concept that mutation spectra can serve as a "fingerprint" of exposure based on chemical structure.

Key Words: genetic alteration; mutation; lung tumor; liver tumor; environmental exposure; database.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUPPLEMENTARY DATA
 REFERENCES
 
Cancer is a leading cause of death in the United States (Jemal et al., 2005Go), yet it remains a very difficult and complicated disease to understand. It is a complex process that clearly involves multiple genetic changes (Balmain 2002Go; Hermeking 2003Go), and a more complete understanding of the process will help in the implementation of cancer prevention strategies. Information from the National Institutes of Health (NIH) on the human genome has greatly enhanced our understanding of cancer genes and their contribution, and future cancer-prevention efforts will undoubtedly rely on the use of this type of information. However, there are other critical factors to be considered, including epigenetic and environmental factors. The need still exists for better understanding the relationship(s) among environmental, genetic, and epigenetic factors regarding their individual and collective contribution to human cancer.

Numerous studies have reported on individual proclivity to certain types of cancers that are attributable to inherited susceptibility genes or single nucleotide polymorphisms (SNPs) (Blankenburg et al., 2005Go; Demokan et al., 2005Go; Lee et al., 2005Go). These types of genes have been associated with breast cancer (BRCA1 and BRCA2), colorectal cancer (APC), and prostate cancer (ELAC2), but it is becoming more evident that environmental factors also play an important role in cancer development (Czene et al., 2002Go; Lichtenstein et al., 2000Go). As our knowledge of the genome grows, so does the opportunity for better understanding the relationships between these environmental and genetic factors.

While many resources are available for identifying critical target genes or gene loci involved in cancer, there is a need to develop additional resources that expand our understanding of the relationships between genetic and environmental factors in tumor development (Birney et al., 2002Go). Databases such as the International Agency for Research on Cancer (IARC) TP53 Mutation Database (Olivier et al., 2002Go) provide a wealth of information on mutations in a single gene; however, to our knowledge no single database containing results from studies of multiple genes has been assembled. The NIEHS has recently developed the Genetic Alterations in Cancer (GAC) knowledge system to help meet this need. GAC is a web-based system for collecting, recombining, and summarizing gene mutation data that are extracted from studies published in the open literature (http://dir-apps.niehs.nih.gov/gac/). Results from human and rodent studies are included and are organized by species, strain, target organ, tumor type and origin, and agent. Data mining features used to query the database combine and summarize data from all studies that match the query criteria. The results are presented in profile graphs or data tables that are displayed in one of four individual chart areas to facilitate comparative analysis.

Although loss of heterozygosity (Pan et al., 2005Go; Tseng et al., 2005Go) and epigenetic events (reviewed by Jones, 2005Go) are also important factors in cancer development, the GAC system focuses primarily on gene mutations. An introduction to the GAC knowledge system is presented here along with an analysis of the mutation incidence (percentage of tumors with a mutation) and spectra (pie charts showing the percentage of each type of mutation [e.g., AT > CG, GC > TA, or deletion] based on data from all studies) for genes that have been studied in tumors associated with environmental carcinogens compared to those that occur sporadically (spontaneous). Point mutation data for multiple genes studied in lung and liver tumors from humans and mice were retrieved by the GAC program. These data were used to demonstrate chemical-, species-, and organ-specific patterns of mutations and assess possible tumor etiology. Strain differences in gene mutations have been documented for the ras genes in both chemically induced and spontaneously occurring tumors (Maronpot et al., 1995Go). However, the results presented here are from combined studies using various mouse strains and portray genetic variation in the species (Festing, 1995Go) that more closely mimics genetic variation in human populations.

The analysis shows that in mice Kras2 had the highest incidence of mutation in both spontaneous and induced lung tumors, whereas in liver Hras-1 had the highest incidence. The spectrum of mutations in genes from mouse lung and liver tumors were exposure specific as were the mutation spectra for human liver tumors associated with exposure to aflatoxin B1 (AFB1), hepatitis B or C virus (HBV or HCV), or vinyl chloride (VCl). Each produced a unique mutation spectrum. Comparative analysis of the mutation spectra based on the molecular structure of the test chemical showed that the overall pattern of mutations for structurally related chemicals were remarkably alike. This information can be used to estimate the mutagenic and carcinogenic potential of structurally similar agents which have not been well studied.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 SUPPLEMENTARY DATA
 REFERENCES
 
Data selection and entry.
Comprehensive literature search strategies were used to identify studies of gene alterations in tumors associated with exposure to specific environmental agents. The search strategies were applied to the peer-reviewed literature for selecting journal articles that met three critical criteria: (1) a description of the tumor(s) and indication of which ones were associated with exposure to a specific agent (e.g., occupational exposure or general population from epidemiologic or case series studies) and which were spontaneous; (2) a molecular analysis of the tumor sample for genetic alterations; and (3) the identification of the affected gene(s) and description of the gene change(s).

Studies were not included when a sufficient description of the study design, including tumor topography and morphology, methods used to analyze gene changes, and appropriate control information were not given; original data were not reported (e.g., reviews); the specific base mutation was not described (e.g., A > T); or the data were from in vitro studies.

Lung and liver tumors.
Data from studies of gene mutations in lung tumors from individuals evaluated for asbestos exposure and in liver tumors of individuals exposed to AFB1, HBV, and/or HCV were extracted and entered into the system. Gene mutation data from studies of chemically induced and spontaneous (control) lung and liver tumors in mice were also included. Since data are continually added to the database and new data made available to the user each quarter, the number of studies displayed in the GAC database, particularly for spontaneous tumors, will be greater than the numbers included in the analysis presented here.

The data were organized into two Agent Groups, "Induced" and "Spontaneous." Tumors from unexposed human subjects (e.g., result from questionnaire, personal monitoring device, control subjects) or those with no reported exposure, along with tumors from control animals, were assigned to the "Spontaneous" group. Tumors that were associated with exposure of individuals to a specified agent (e.g., AFB1 from living in a high exposure area; biomarker of exposure such as protein adducts) and tumors induced in experimental animals by treatment with a specific test agent were assigned to the "Induced" group.

The primary data fields used for data entry and summaries include study information: species and strain; tumor information: topography, morphology, tumor origin (e.g., primary or recurrent), number of tumors, and alteration category (gene or chromosome); subject group information: exposure category (e.g., acute or chronic), route, dose, frequency, exposure time, sampling time, geographic region, and ethnicity; subject tumor information: individual subject identification, genetic disorder (if applicable), gender, age, and confounding factor (e.g., scar tissue); and alteration data: gene symbol, exons evaluated, alteration type (e.g., point mutation), analytical technique, affected codon, type of base change (e.g., C > T), alteration description (e.g., GC > AT), wild-type codon sequence and amino acid, and altered codon sequence and predicted amino acid. Note fields for clarifying discrepancies between the published data and that shown in GAC (e.g., amino acid predicted in a study was wrong) or to provide additional study information that could affect data comparisons are also included.

Statistical analysis.
Data for human and mouse lung and liver tumors were retrieved from the database using the data mining features and were organized by species, tumor site, and agent group (spontaneous or induced). Data from detailed data lists showing the results for each tumor sample from all studies retrieved were generated by the GAC program, captured, and prepared for broad-based statistical analysis. This included data from 46 studies of induced and spontaneous mouse lung tumors and 38 studies of mouse liver tumors. Data from 53 studies of human lung and 106 studies of liver tumors associated with exposure to specific agents or reported to be sporadic were also prepared for statistical analysis.

The mutation incidence was compared for pairs of data groups (e.g., human lung and mouse lung, mouse liver induced and spontaneous, or mouse Kras2 and Hras1) using a standard two-sample Z-statistic for proportions. However, since sample proportions were derived from multiple studies, and variables between studies (e.g., strain, dose, and treatment time) were not specifically factored in, extra variation between studies was taken into account by using a nonparametric estimator for extra binomial variability (McCullagh and Nelder, 1997Go). The p values for the tests were derived using a nonparametric bootstrap methodology (Efron and Ribshirani, 1993Go).

Mutation spectra for concordant genes in the different data groups were also statistically analyzed. Mutations for which a description was not given (NG) in the study or that represented <1% of the total mutations were not included in these analyses. Mutation spectra of pairs of data groups were compared by computing the Euclidean distance between the two multinomial vectors. As with the analysis of mutation incidence, the p values for the test were derived using nonparametric bootstrap methodology (Efron and Ribshirani, 1993Go). The distribution of the mutation spectrum was taken to be multinomial. Experiments were used as a covariate. Other covariates such as strain or race and gender were not modeled due to data limitations. The results from the statistical analyses are summarized in Table 1. Differences as well as similarities between species, tumor sites, and/or agent groups can be used for planning future study needs or developing surrogate models/systems for determining the carcinogenic potential of other agents that have similar chemical characteristics.


View this table:
[in this window]
[in a new window]
  		TABLE 1 Statistical Analysis of Mutation Incidence and Mutation Spectra

 

Figure 5
View larger version (59K):
[in this window]
[in a new window]
  		FIG. 5. Mutation spectra for spontaneous and agent-induced liver tumors for (A) the human TP53 gene, and (B) the mouse Hras-1 gene. (A) Human liver TP53 references—Spontaneous (underlined papers have data for spontaneous and induced tumors): An et al., 2001Go; Boix-Ferrero et al., 1999Go; Bourdon et al., 1995Go; Challen et al., 1992Go; Deng et al., 1997Go; Hayashi et al., 1993Go; Hollstein et al., 1993Go; Hsieh and Atkinson, 1995Go; Jin et al., 2002Go; Kar et al., 1993Go; Karachristos et al., 1999Go; Kazachkov et al., 1996Go; Kennedy et al., 1994Go; Kress et al., 1992Go; Kubicka et al., 1995Go; Lai et al., 1993Go; Laurent-Puig et al., 2001Go; Murakami et al., 1991Go; Nishida et al., 1993Go; Oda et al., 1992bGo; Park et al., 1996Go; Pogribny and James, 2002Go; Sahoo et al., 1993Go; Shi et al., 1995Go; Shieh et al., 1993Go; Shimizu et al., 1999Go; Soini et al., 1996Go; Tanaka et al., 1993Go; Tannapfel et al., 2001Go; Vautier et al., 1999Go; Vesey et al., 1994Go; Volkmann et al., 2001Go; Wong et al., 2000Go; Zhu et al., 2002Go. Human liver TP53 references—Induced: Bressac et al., 1991Go; Buetow et al., 1992Go; Chao et al., 1999Go; De Benedetti et al., 1995Go; Diamantis et al., 1994Go; Fujimoto et al., 1994Go; Goldblum et al., 1993Go; Greenblatt et al., 1997Go; Hollstein et al., 1994Go; Hoque et al., 1999Go; Hosono et al., 1993Go; Hsia et al., 2000Go; Hsu et al., 1991Go; Katiyar et al., 2000Go; Lee et al., 2002Go; Li et al., 1993Go; Lunn et al., 1997Go; Ng et al., 1994aGo,bGo; Nose et al., 1993Go; Oda et al., 1992aGo; Ozturk, 1991Go; Rashid et al., 1999Go; Scorsone et al., 1992Go; Sheu et al., 1992Go; Stern et al., 2001Go; Unsal et al., 1994Go; Weihrauch et al., 2000Go, 2002Go; Yang et al., 1997Go. (B) Mouse Liver Hras-1 references—Spontaneous (underlined papers have data for spontaneous and induced tumors): Buchmann et al., 1991Go; Devereux et al., 1993bGo; Dragani et al., 1991Go; Enomoto et al., 1993Go; Fox et al., 1990Go; Frey et al., 2000Go; Herzog et al., 1993Go; Hong et al., 1998Go; Iida et al., 2000Go; Johansson et al., 1997Go; Lord et al., 1992Go; Malarkey et al., 1995Go; Manam et al., 1995Go; Manjanatha et al., 1996Go; Mori et al., 1995Go; Richardson et al., 1992aGo; Rumsby et al., 1991Go; Shinder et al., 1993Go; Stanley et al., 1992Go; Von Tungeln et al., 1999Go; Watson et al., 1995Go. Mouse Liver Hras-1 references—Induced: Aydinlik et al., 2001Go; Bauer-Hofmann et al., 1990Go, 1992Go; Chen et al., 1993Go; Gotoh et al., 2003Go; Huang et al., 2003Go; Kalkuhl et al., 1996Go; 1998Go; Lee and Drinkwater, 1995Go; Manam et al., 1992bGo, 1995Go; Mitchell and Warshawsky, 1999Go; Richardson et al., 1992bGo; Schroeder et al., 1997Go; Stowers et al., 1988Go; Wang et al., 1993Go; Wiseman et al., 1986Go; Xia et al., 1998Go.

 

Figure 6
View larger version (33K):
[in this window]
[in a new window]
  		FIG. 6. TP53 gene mutation spectra for human spontaneous lung and liver tumors. Human TP53 Spontaneous tumors—Lung: Hayashi et al., 1996Go; Kishimoto et al., 1992Go; Mao et al., 1994Go; Miller et al., 1992Go; Mitsudomi et al., 1993Go; Mor et al., 1997Go; Reichel et al., 1994Go; Sameshima et al., 1992Go; Sanchez-Cespedes et al., 1999Go; Segers et al., 1995Go; Takahashi et al., 1991Go; Taniguchi et al., 1996Go; Tomizawa et al., 1999Go; Top et al., 1995Go. Human TP53 Spontaneous tumors—Liver: An et al., 2001Go; Boix-Ferrero et al., 1999Go; Bourdon et al., 1995Go; Challen et al., 1992Go; Deng et al., 1997Go; Hayashi et al., 1993Go; Hollstein et al., 1993Go; Hsieh and Atkinson, 1995Go; Jin et al., 2002Go; Kar et al., 1993Go; Karachristos et al., 1999Go; Kazachkov et al., 1996Go; Kennedy et al., 1994Go; Kress et al., 1992Go; Kubicka et al., 1995Go; Lai et al., 1993Go; Laurent-Puig et al., 2001Go; Murakami et al., 1991Go; Nishida et al., 1993Go; Oda et al., 1992bGo; Park et al., 1996Go; Pogribny and James, 2002Go; Sahoo et al., 1993Go; Shi et al., 1995Go; Shieh et al., 1993Go; Shimizu et al., 1999Go; Soini et al., 1996Go; Tanaka et al., 1993Go; Tannapfel et al., 2001Go; Vautier et al., 1999Go; Vesey et al., 1994Go; Volkmann et al., 2001Go; Wong et al., 2000Go; Zhu et al., 2002Go.

 

   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 SUPPLEMENTARY DATA
 REFERENCES
 
Genetic Changes in Lung Tumors
An overview of the incidence of point mutations in different genes studied in lung tumors (spontaneous plus induced) from mice (46 studies total) and humans (53 studies total) is shown in the gene mutation profiles and corresponding data tables in Figures 1A and 1B. It is evident from the profiles that the incidence of mutation in concordant genes is different for these two species, especially for Kras and p53. It is also apparent from the tumor numbers shown in the summary table that some genes (e.g., Apc, Cdkn2a, CDK4, and RB1) have not been studied as thoroughly as others and the data may represent only one or two studies. For example, the Apc gene was evaluated in lung tumors in only one mouse study (Oreffo et al., 1998Go) and two human studies (Cooper et al., 1996Go; Horii et al., 1992Go). Although no mutations were reported for either species, additional studies are needed before it can be determined whether this gene could be mutated and play a role in lung tumor development.


Figure 1
View larger version (38K):
[in this window]
[in a new window]
  		FIG. 1. Gene mutation profiles and summary tables for (A) mouse and (B) human lung tumors.

 
On the other hand, although results shown for BRAF in human lung tumors include data from just four studies (Brose et al., 2002Go; Davies et al., 2002Go; Dote et al., 2004Go; Naoki et al., 2002Go), the total number of tumors evaluated (465 tumors) is relatively high compared to the number shown for other genes such as TP53 (569 tumors), which represents 30 studies. It is reasonable to expect that the low incidence of mutation (1.51%) shown for BRAF is not likely to change significantly as new data become available. Other genes such as Kras (43 mouse and 23 human studies) and p53 (seven mouse and 30 human studies) have been extensively evaluated for mutations in lung tumors. The data show that approximately 40% of the human lung tumors included here had TP53 mutations, while <10% (21/246 tumors) of mouse lung tumors had mutations in this gene. Conversely, while over 70% of all mouse lung tumors had Kras2 mutations, <20% of the human tumors had mutations. The results from statistical analysis of the data (Table 1) show that this difference in mutation incidence was significant (p < 0.0001).

The KRAS2 mutation spectrum presented in Figure 2A for all of the human lung tumors was also significantly different from that for mouse lung tumors (p < 0.0001). Virtually all of the mutations (90%) in the human lung samples were GC base changes (transitions and transversions; ~1:3) observed predominantly at codon 12, while in the mouse 53% of the mutations were GC base changes (transitions and transversions; ~2:1) at codon 12, and 46% were AT base changes (transitions and transversions; ~2:1) at codon 61. Approximately 10% of the mouse tumors were spontaneous, while 90% of the human tumors were considered spontaneous and 10% were associated with exposure to asbestos. Of the 102 tumors from five studies (Hayashi et al., 1996Go: Husgafvel-Pursiainen et al., 1993Go; Kitamura et al., 1998Go; Nelson et al., 1999Go; Ni et al., 2000Go) that were reported to be related to asbestos exposure, only 20% had KRAS2 mutations; 52% of the mutations were GC base changes at codon 12 (transitions and transversions; ~1:1.75), and 33% were not described (NG). The frequency of base changes for spontaneous lung tumors also differed significantly: humans—0% AT > GC; 61% GC > TA (codon 12); mice—40% AT > GC (codon 61); 11% GC > TA (codon 12).


Figure 2
View larger version (46K):
[in this window]
[in a new window]
  		FIG. 2. Kras mutation spectra for all lung tumors from (A) mice and humans, and (B) EC-induced, NNK-induced, and spontaneous mouse lung tumors (see references in Table 2). (A) Lung Kras2 spectra—mouse (Bai et al., 1998Go, 2003Go; Candrian et al., 1991Go; Chen et al., 1993Go, 1994Go; Devereux et al., 1991Go; 1993a,b; Doi et al., 1994Go; Gray et al., 2001Go; Hayashi et al., 2001Go; Herzog et al., 1993Go; Horio et al., 1996Go; Karasaki et al., 1997Go; Kawano et al., 1995Go, 1996Go; Lin et al., 1998Go; Manam et al., 1992aGo; Manenti et al., 1995Go; Mass et al., 1993Go, 1996Go; Massey et al., 1995Go; Matzinger et al., 1994Go, 1995Go; Nesnow et al., 1994Go; Nuzum et al., 1990Go; Ohmori et al., 1992Go; Prahalad et al., 1997Go; Ramakrishna et al., 2000Go, 2002Go; Re et al., 1992Go; Ronai et al., 1993Go; Sills et al., 1995Go, 1999Go; Stowers et al., 1987Go; Ton et al., 2004Go; Wang and Witschi, 1995Go; Wang et al., 1993Go; Warshawsky et al., 1996Go; You et al., 1992aGo,bGo, 1993Go, 1994Go); and lung Kras2 spectra—human (Brose et al., 2002Go; Cooper et al., 1997Go; Hayashi et al., 1996Go; Husgafvel-Pursiainen et al., 1993Go; Keohavong et al., 2001Go; Kitamura et al., 1998Go; Maeshima et al., 1997Go, 2002Go; Mao et al., 1994Go; Naito et al., 1992Go; Nelson et al., 1999Go; Ni et al., 2000Go; Redondo et al., 1997Go; Reichel et al., 1994Go; Rodenhuis et al., 1987Go, 1988Go; Siegfried et al., 1997Go; Silini et al., 1994Go; Slebos et al., 1990Go; Somers et al., 1998Go; Sugio et al., 1998Go; Suzuki et al., 1990Go; Yanez et al., 1987Go).

 

View this table:
[in this window]
[in a new window]
  		TABLE 2 Summary Data for Kras2 Mutations in Spontaneous, EC-, and NNK-Induced Mouse Lung Tumors

 
Further analysis of the mouse Kras2 data showed that, although the mutation incidence in spontaneous compared to all induced lung tumors was not significantly different (p = 0.1233), the mutation spectra were (p = 0.0008). Data from studies by the NIEHS of lung tumors induced in mice by exposure to urethan (ethyl carbamate: EC) or 4-(N-nitrosomethylamino)-1-(3-pyridyl)-1-butanone (NNK) are summarized in Table 2 along with the results from 16 studies of spontaneous lung tumors. The number of studies, number of tumors evaluated, and number of tumors reported to have Kras2 mutations are listed. The mutation incidence for each group is comparable, ranging from 60 to 81%; however, the mutation spectra shown in Figure 2B for these three agents are uniquely different.

The majority of the mutations (73–98%) were distributed between AT > GC or GC > AT transitions or AT > TA transversions in each group, but the distribution of the mutation types was uniquely different. The percentage of AT > GC, GC > AT, and AT > TA mutations was 24, 29, and 20%, respectively, for spontaneous tumors; 31, 3, and 64%, respectively, for EC-induced tumors; and 1, 96, and <1%, respectively, for NNK-induced tumors. The difference in the frequency of transitions and transversions for these two structurally different agents indicates that the mutation pattern is related to chemical structure. This observation is further supported by the similarity in the pattern of alterations for EC and one of its metabolites, vinyl carbamate (VC); the only structural difference in these two chemicals is a double bond on the terminal carbon of VC where EC has a single bond. Data from nine studies of EC-induced lung tumors in mice showed that the Kras2 mutation incidence (66%) was comparable to that for VC-induced tumors (82%) summarized from three studies (see Figs. 3A and 3B) as was the percentage of AT > GC and GC > AT transitions and AT > TA transversions (31, 3, and 64%, respectively), compared to that for VC-induced tumors (42, 13, and 33%, respectively). Due to the limited amount of data available for VC a statistical comparison was not possible.


Figure 3
View larger version (46K):
[in this window]
[in a new window]
  		FIG. 3. Kras2 mutation spectra from mouse lung tumors.

 
Mouse lung tumor Kras2 mutation spectra by chemical structure (Fig. 3).
The Kras2 mutation spectra for the other mouse lung carcinogens are given in Figures 3C–3L and are grouped by chemical structure. Spectra for three nonaromatic amines, diethylnitrosamine (DEN; Fig. 3C), ethylnitrosourea (ENU; Fig. 3D), and N-nitrosodimethylamine (DMN; Fig. 3E) showed that ~99% of the mutations induced by each agent were AT > GC or GC > AT transitions. The mutation pattern induced the two alkylating agents DEN and ENU are very similar (71% and 64% AT > GC and 18% and 27% GC > AT, respectively) but quite different from the spectrum induced by the nonalkylating agent DMN (5% AT > GC and 94% GC > AT) and from the spectra for the unsaturated nonaromatic amines EC and VC (discussed above). The activity of DMN is in fact comparable to that of NNK (Fig. 3F; 1% AT > GC and 96% GC > AT), a heterocyclic aromatic nitrosamine. Both of these chemicals have a methyl group attached to the N-nitroso moiety.

Kras2 mutations induced by the nonnitrosated heterocyclic aromatic amines, 2-amino-3-methylimidazo[4,5-f]quinoline (IQ) and 7H-dibenzo[c,g]carbazole (DBC) are also similar (80% and 92% AT > TA transversions, respectively), but are uniquely different from NNK (Fig. 3F) as well as the aromatic azo-dyes 4-aminoazobenzene and benzidine, which induced 95% and 85% GC > TA transversions, respectively (data not shown). Two of three polycyclic aromatic hydrocarbons (PAH) with a cyclopentane ring, cyclopenta[cd]pyrene (Fig. 3G) and benz[j]aceanthrylene (Fig. 3H), have no bay region in their structure and induced both GC > TA and CG transversions (40% and 50% compared to 35% and 65%, respectively), while benzo[b]fluoranthene, which has a bay region that can form an epoxide, induced 92% GC > TA transversions (Fig. 3I). This activity was comparable to that of the PAH benzo[a]pyrene (B[a]P) (Fig. 3J) and 5-methylchrysene (Fig. 3K) that induced 81% and 73% GC > TA transversions, respectively. The mutation spectrum for 1-nitropyrene (1-NP; Fig. 3L), the only nitro-PAH evaluated for Kras2 mutations, was similar to that seen for the nitrosamines DEN and ENU. It induced 83% AT > GC and 17% GC > AT transitions. Like 6-nitrochrysene (6-NC), 1-NP is primarily metabolized via nitroreductases to the aromatic amine 1-aminopyrene, which forms an N-(deoxyguanosin-8-yl)-1-aminopyrene adduct (Bai et al., 1998Go). Two other DNA adducts may also be formed from N-hydroxy-1-aminopyrene; they are 6-(deoxyguanosin-N2-yl)-1-aminopyrene and 8-(deoxyguanosin-N2-yl)-1-aminopyrene (Herreno-Saenz et al., 1995Go). The reduction of 1-NP to an aromatic amine is most likely the reason for the similarity in the mutation spectra for 1-NP, DEN, and ENU.

Genetic Changes in Liver Tumors
An overview of the incidence of point mutations in different genes from liver tumors (spontaneous plus induced) of mice (38 studies total) and humans (106 studies total) is shown in the mutation profiles and corresponding data tables in Figures 4A and 4B. As with the profiles for the lung tumors, species differences and similarities are readily apparent, especially for the incidence of Hras, Kras, and p53 mutations.


Figure 4
View larger version (35K):
[in this window]
[in a new window]
  		FIG. 4. Gene mutation profiles and summary tables for (A) mouse and (B) human liver tumors.

 
The incidence of Apc, Kras, and Nras gene mutations was comparable in both mouse and human liver tumors. The incidence of ß-catenin mutations appeared to be higher in mouse (31%) than in human (17%) tumors; however the mouse data represent six studies (Anna et al., 2000Go; Aydinlik et al., 2001Go; Gotoh et al., 2003Go; Hayashi et al., 2003Go; Huang et al., 2003Go; Ogawa et al., 1999Go) compared to 17 human studies. Due to the limited amount of data from the mouse studies, statistical analysis was not possible. The median mutation incidence for the mouse compared to human data was 27% and 19%, respectively, indicating that a statistically significant difference is unlikely.

The incidence of Hras and p53 gene mutations appeared to be significantly different in the liver tumor samples from humans compared to mice for this data set. Hras-1 mutations were induced in 31% (812/2605 tumors) of the mouse tumors evaluated in 40 studies but not in any of the 111 human tumor samples from seven studies. Conversely, the incidence of human liver tumors with TP53 mutations was 26% (559/2153 tumors) according to results from 64 studies compared to 0% in 250 mouse liver tumors (of which 235 were agent induced) from five studies.

Statistical analyses were done on the TP53 mutation incidence and spectrum for human spontaneous compared to induced liver tumors and on the Hras-1 mutation incidence and spectrum for mouse spontaneous compared to induced liver tumors to determine if the frequency and types of mutations seen in each group for a given species were similar (see Table 1). Tumors considered to be induced in humans included those from individuals reported to live in areas of high AFB1 exposure and/or to have HBV or HCV. Results from 19 tumors in three studies of workers occupationally exposed to VCl are also in this group (Hollstein et al., 1994Go; Weihrauch et al., 2000Go, 2002Go). The difference between the TP53 mutation incidence in human spontaneous liver tumors (118/612 = 19%; 34 studies) compared to induced tumors (441/1541 = 29%; 57 studies) was marginal (p = 0.0485). Perhaps the reason for this lies in the heterogeneous nature of the human spontaneous tumor group; a variety of environmental exposures are probable in these patients, but are not reported in the literature.

The overall TP53 mutation spectra for the human liver tumor data are shown in Figure 5A. The TP53 mutation spectrum for spontaneous tumors was significantly different from the spectrum for the induced tumors (p < 0.0001). A major factor that contributed to this difference is the extensive amount of data reported in numerous studies that specifically analyzed codon 249ser mutations in hepatocellular carcinomas (HCC). This is the most frequent mutation seen in HCC from populations exposed to relatively high levels of AFB1 in the diet. The GC > TA missense mutation in codon 249 causes an AGGarg -> AGTser sequence change. Over 50% of the mutations shown in the TP53 mutation spectra in Figure 5A for induced tumors are GC > TA transversions that have been reported in studies from moderate to high AFB1 exposure areas.

The Hras-1 mutation incidence calculated from data in 21 studies of spontaneous mouse liver tumors (290/670 = 43%) was also significantly different from that calculated for chemically induced tumors (522/1935 = 27%) reported in 32 studies (p = 0.021). The mutation spectra for several classes of the chemicals represented in the induced group and the results from the analysis of these spectra are described below.

Data presented in Figure 5B for Hras-1 mutations in spontaneous compared to chemically induced mouse liver tumors showed a significant difference (p < 0.0001; Table 1) between the two mutation spectra. Approximately 60% of the induced tumor data are for three agents, DEN, VC, and methylene chloride (MeCl) from 18 studies. The mutation spectrum for each of these agents was also unique. The percentage of AT > GC and GC > TA transitions and AT > TA transversions for DEN-induced tumors from 14 studies (see Fig. 7D for citations) was 38, 36, and 24%, respectively; for VC-induced tumors (Stanley et al., 1992Go; Watson et al., 1995Go; Wiseman et al., 1986Go) it was 28, 12, and 60%, respectively; and for MeCl-induced tumors (Hegi et al., 1993Go) it was 42, 42, and 16%, respectively.


Figure 7
View larger version (42K):
[in this window]
[in a new window]
  		FIG. 7. Hras-1 mutation spectra from mouse liver tumors.

 
Finally, statistical analyses were done on selected gene mutation spectra for lung compared to liver tumors from the same species. The TP53 mutation spectra for human spontaneous lung compared to liver tumors are shown in Figure 6. The spectra were significantly different (p = 0.0094), as was the incidence of TP53 mutations in lung (42%) compared to liver (19%; p = 0.0076). The incidence of mutations in Kras2 also differed significantly in mouse lung compared to liver tumors for both spontaneous (p = 0.0065) and induced (p < 0.0001) tumors, as did their mutation spectra (p < 0.0001).

Mouse liver tumor Hras-1 mutation spectra by chemical structure (Fig. 7).
The mutation spectra for Hras-1 showed that the nonaromatic amines induced primarily AT > TA or GC transversions and transitions in contrast to the aromatic amines that predominantly induced GC > TA transversions (60 to 100%). The mutation spectra for EC and its metabolite VC, both unsaturated nonaromatic amines, are almost identical (~60% AT > TA transversions and 28% AT > GC transitions; Figs. 7A and 7B), yet they are noticeably different from the spectra for the nonaromatic nitrosamines, DMN and DEN (Figs. 7C and 7D). These nitrosamines had a higher incidence of AT > GC transitions (55 and 38%, respectively) and a lower incidence of AT > TA transversions (14 and 24%, respectively). The incidence of GC > TA transversions was 32 and 36%, giving an overall transversion incidence of 46 to 60%.

The two amines with nonfused benzene rings, benzidine and 4-aminobiphenyl, had a lower incidence of GC > TA transversions (85% and 60%, respectively; Figs.7E and 7F) compared to the two aromatic amines, 2-acetylaminofluorene (2-AAF) and N-hydroxy-2-acetylaminofluorene (OH-2-AAF), (93% and 100%, respectively; Figs. 7G and 7H).

Two PAHs, B[a]P and 7,12-dimethylbenz[a]anthracene (DMBA), primarily induced AT > TA transversions (80% and 92%; Figs. 7I and 7J), and the incidence was higher than that seen with the nonaromatic amines. Conversely, the mutation spectra for the sole nitro-PAH, 6-NC, showed only 5% AT > TA transversions and 95% GC > TA transversions (Fig. 7K). This spectrum matched that of the two aromatic amines, 2-AAF and OH-2-AAF (93–100% GC > TA transversions). This similarity is likely due to 6-NC being activated by nitro reduction to intermediate aromatic amines (i.e., N-hydroxy-6-aminochrysene or trans-1,2-dihydroxy-1,2-dihydro-6-aminochrysene) that react with DNA much like 2-AAF and OH-2-AAF (Delclos et al., 1989Go).

Liver Hras-1 mutations versus lung Kras2 mutations in mouse tumors.
Five chemicals had mutation spectra for both Hras-1 in liver tumors and Kras2 in lung tumors. For three of these chemicals, B[a]P, DMN, and DEN, the spectrum for each gene is visibly different (Figs. 3C, 3E, 3J, 7C, 7D, and 7I). As shown in Table 3, B[a]P induced 100% GC > TA or AT mutations in Kras2 and 90% AT > TA or GC mutations in Hras-1; DMN induced 94% GC > AT mutations in Kras2 and 69% AT > GC or TA and 32% GC > TA mutations in Hras-1; and DEN induced 71% AT > GC and 18% GC > AT mutations in Kras2 compared to 62% AT > GC or TA and 36% GC > TA mutations in Hras-1.


View this table:
[in this window]
[in a new window]
  		TABLE 3 Summary of ras Gene Mutation Profiles for B[a]P-, DMN-, DEN-, EC- and VC-Induced Mouse Lung and Liver Tumors

 
The Hras-1 mutation spectra for the remaining two agents, EC and its metabolite VC, were essentially the same for mouse liver tumors (~60% AT > TA and 28% AT > GC mutations), but the Kras2 mutation spectra for lung tumors differed. EC induced 64% AT > TA and 31% AT > GC mutations, and VC 33% AT > TA and 42% AT > GC mutations. It is interesting to note that the Kras2 and Hras-1 mutation spectra of EC in lung and liver, respectively, are the same as that for Hras-1 in VC-induced liver tumors, suggesting that EC metabolism in the lung may produce mutagenic metabolites other than VC. Studies have shown organ susceptibilities to tumor formation which may be based on differences in metabolic capabilities (Xue and Warshawsky, 2005Go) and formation of DNA adducts (Phillips, 2005) due to strain and organ site differences (Festing, 1995Go; Festing et al., 1998Go).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
SUPPLEMENTARY DATA
 REFERENCES
 
Results from recent studies demonstrate that environmental factors are a major contributor to human cancers and may even play a greater role than hereditary factors in the development of some types of cancers (Czene et al., 2002Go; Lichtenstein et al., 2000Go). It is well established that gene–environment interactions underlie almost all human diseases (Khoury et al., 2005Go). As efforts in genomic research continue toward establishing a better understanding of the role of individual genes, gene–gene interactions, and how alterations in gene sequences contribute to the disease, new approaches are also needed for expanding our understanding of controllable environmental factors that alter such critical genes so that prevention and treatment of disease can be improved. As part of this effort, we have designed a system that merges data from peer-reviewed studies of mutations in multiple genes from different types of tumors that are associated with exposure to specific agents. This system allows a retrospective analysis of various factors, including environmental exposure that can contribute to cancer development. Data can be queried by species, strain, topography, morphology, agent of interest, tumor origin, heritable factors for known disorders, or any combination of these fields.

The results from the analysis of lung and liver tumor data showed that the mutation incidence and spectra for concordant sets of genes from human and mouse tumors differed significantly between agents, species, and/or topographies. The predominant mutations in mouse lung tumors were in the Kras2 gene, followed by mutations in Hras-1 and Trp53. In human lung tumors the predominant mutations occurred in the TP53 gene, followed by mutations in KRAS2 and other genes. Less than 1% (3/308 tumors) of the human lung tumors had HRAS-1 mutations. The mutational spectrum for the p53 gene was also different in mouse and human lung tumors. TP53 appears to play the most significant role in lung tumor development in humans, while Kras2, and to a lesser degree Hras-1, seem to be more important in mouse lung tumor development. This is based on the statistically significant difference seen in both the mutational incidence and spectra for these two genes that were evaluated in both species. It is of course possible that these differences reflect the disparity in the types of agent exposures in human populations compared to the experimentally controlled exposures of mice. As more data become available in the database, statistical tests that take account for many of these variables can be applied.

The role of KRAS2 in human lung and liver appears to be similar based on the evidence that there is no statistically significant difference in the mutation incidence for spontaneous tumors in these two tissues. However, the exact mutations that occur in the KRAS2 gene in lung compared to liver tumors do differ significantly, indicating that the mutations may be mediated by different factors in the different tissue types.

The analysis of Hras and p53 gene mutation incidence in human compared to mouse liver tumors also showed a difference between the two species. The predominant mutations in mouse liver were found in the Hras-1 gene, followed by Catnb and Kras2. In human liver tumors, mutations were found in TP53, KRAS2, and CTNNB1. The mutation incidence was similar in both species for ß-catenin and the K- and Nras genes. The Hras-1 mutation spectra for mouse liver tumors showed a significant difference between spontaneous and induced tumors. Likewise, analysis of TP53 mutations in human liver tumors showed a significant difference in the mutation spectrum for sporadic compared to induced tumors.

Our analysis of mutation patterns in chemically induced mouse lung and liver tumors corroborates the large body of data from work dating back to 1958 (Benzer and Freese, 1958Go), where 5-bromouracil–specific mutations were shown to be induced in phage T4, and subsequent studies that reported similar observations in human cell populations (Albertini and Hayes, 1997Go; Cariello et al., 1990Go). These findings provide strong evidence that chemical exposures often induce mutations not normally found in spontaneous tumors that can be important events contributing to tumor development. Using the GAC knowledge system, we compared the mutation spectra from a wide variety of chemicals based on chemical structures. In doing so, we were able to show that specific common chemical characteristics among agents often produce remarkably similar mutation profiles. Further development of the system should allow estimates to be made regarding the carcinogenic potential of structurally similar chemicals based on mutation spectra, but this remains to be tested.

A parallel mechanism of carcinogenesis introduced by Slaughter et al. (1953)Go, some 50 years ago, suggested that when a cell which acquires a genetic alteration, such as a mutation, the change can confer a growth advantage on that cell, leading to clonal expansion and tumor development (Prevo et al., 1999Go; Stern et al., 2002Go). Elucidating which types of mutations and genes can confer a growth advantage to a cell is also an important future step toward understanding key events in the initiation of carcinogenesis.

The NIEHS focus on environmental causes of cancer helps promote prevention strategies by identifying modifiable risk factors. Information in GAC, along with data from other studies such as gene expression and SNP analysis, provides a resource for investigation of these types of environmental and genetic factors that are involved in tumor development. The comprehensive data available in the system can be recombined to evaluate gene responses in specific types of tumors from different species exposed to a common agent, in tumors from different target sites in the same species exposed to a common agent, or in tumors from the same species exposed to different agents. This system is dynamic, with new data being added regularly; future additions to the database will include genetic alterations in preneoplastic lesions and alteration such as loss of heterozygosity in chromosomes.

The GAC knowledge system is available on the NIEHS website. Nominations for additional gene analysis studies may be made to the NIEHS National Toxicology Program nomination office (http://ntp-server.niehs.nih.gov/).


   SUPPLEMENTARY DATA
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUPPLEMENTARY DATA
REFERENCES
 
Supplementary data are available online at http://toxsci.oxfordjournals.org/.


   ACKNOWLEDGMENTS
 
We thank Dr. R. Sills, Dr. M. Waters, and Dr. R. Maronpot for their review of the manuscript. This research was supported by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences Contract No. N43-ES-15477.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUPPLEMENTARY DATA
 REFERENCES
 
Albertini, R. J., and Hayes, R. B. (1997). Somatic cell mutations in cancer epidemiology. IARC Sci. Publ. 142, 159–184.[Medline]

An, F. Q., Matsuda, M., Fujii, H., Tang, R. F., Amemiya, H., Dai, Y. M., and Matsumoto, Y. (2001). Tumor heterogeneity in small hepatocellular carcinoma: Analysis of tumor cell proliferation, expression and mutation of p53 AND beta-catenin. Int. J. Cancer 93, 468–474.[CrossRef][ISI][Medline]

Anna, C. H., Sills, R. C., Foley, J. F., Stockton, P. S., Ton, T. V., and Devereux, T. R. (2000). Beta-catenin mutations and protein accumulation in all hepatoblastomas examined from B6C3F1 mice treated with anthraquinone or oxazepam. Cancer Res. 60, 2864–2868.[Abstract/Free Full Text]

Aydinlik, H., Nguyen, T. D., Moennikes, O., Buchmann, A., and Schwarz, M. (2001). Selective pressure during tumor promotion by phenobarbital leads to clonal outgrowth of beta-catenin-mutated mouse liver tumors. Oncogene 20, 7812–7816.[CrossRef][ISI][Medline]

Bai, F., Nakanishi, Y., Takayama, K., Pei, X. H., Inoue, K., Harada, T., Izumi, M., and Hara, N. (2003). Codon 64 of K-ras gene mutation pattern in hepatocellular carcinomas induced by bleomycin and 1-nitropyrene in A/J mice. Teratog. Carcinog. Mutagen (Suppl. 1), 161–170.

Bai, F., Nakanishi, Y., Takayama, K., Pei, X. H., Tokiwa, H., and Hara, N. (1998). Ki-ras mutation and cell proliferation of lung lesions induced by 1-nitropyrene in A/J mice. Mol. Carcinog. 22, 258–264.[CrossRef][ISI][Medline]

Balmain, A. (2002). Cancer as a complex genetic trait: Tumor susceptibility in humans and mouse models. Cell 108, 145–152.[CrossRef][ISI][Medline]

Bauer-Hofmann, R., Buchmann, A., Wright, A. S., and Schwarz, M. (1990). Mutations in the Ha-ras proto-oncogene in spontaneous and chemically induced liver tumours of the CF1 mouse. Carcinogenesis 11, 1875–1877.[Abstract/Free Full Text]

Bauer-Hofmann, R., Klimek, F., Buchmann, A., Muller, O., Bannasch, P., and Schwarz, M. (1992). Role of mutations at codon 61 of the c-Ha-ras gene during diethylnitrosamine-induced hepatocarcinogenesis in C3H/He mice. Mol. Carcinog. 6, 60–67.[ISI][Medline]

Benzer, S., and Freese, E. (1958). Induction of specific mutations with 5-bromouracil. Proc. Natl. Acad. Sci. U.S.A. 44, 112–119.[Free Full Text]

Birney, E., Clamp, M., and Hubbard, T. (2002). Databases and tools for browsing genomes. Annu. Rev. Genomics Hum. Genet. 3, 293–310.[CrossRef][ISI][Medline]

Blankenburg, S., Konig, I. R., Moessner, R., Laspe, P., Thoms, K. M., Krueger, U., Khan, S. G., Westphal, G., Berking, C., Volkenandt, M., et al. (2005). Assessment of 3 xeroderma pigmentosum group C gene polymorphisms and risk of cutaneous melanoma: A case-control study. Carcinogenesis 26, 1085–1090.[Abstract/Free Full Text]

Boix-Ferrero, J., Pellin, A., Blesa, R., Adrados, M., and Llombart-Bosch, A. (1999). Absence of p53 gene mutations in hepatocarcinomas from a Mediterranean area of Spain. A study of 129 archival tumour samples. Virchows Arch. 434, 497–501.[CrossRef][ISI][Medline]

Bourdon, J. C., D'Errico, A., Paterlini, P., Grigioni, W., May, E., and Debuire, B. (1995). p53 Protein accumulation in European hepatocellular carcinoma is not always dependent on p53 gene mutation. Gastroenterology 108, 1176–1182.[CrossRef][ISI][Medline]

Bressac, B., Kew, M., Wands, J., and Ozturk, M. (1991). Selective G to T mutations of p53 gene in hepatocellular carcinoma from southern Africa. Nature 350, 429–431.[CrossRef][Medline]

Brose, M. S., Volpe, P., Feldman, M., Kumar, M., Rishi, I., Gerrero, R., Einhorn, E., Herlyn, M., Minna, J., Nicholson, A., et al. (2002). BRAF and RAS mutations in human lung cancer and melanoma. Cancer Res. 62, 6997–7000.[Abstract/Free Full Text]

Buchmann, A., Bauer-Hofmann, R., Mahr, J., Drinkwater, N. R., Luz, A., and Schwarz, M. (1991). Mutational activation of the c-Ha-ras gene in liver tumors of different rodent strains: Corr