Use of computerized data listings and activity profiles of genetic and related effects in the review of 195 compounds

Computer-generated listings of data from short-term tests for genetic and related effects (activity profile listings) were prepared for 195 compounds that included for each compound, the test system (identified by a three-letter code word), qualitative results and the lowest effective dose (LED) or highest ineffective dose (HID) tested. A corresponding bar or line graph (activity profile) was also generated, in which test systems are displayed along the x-axis and the LED or HID values along the y-axis. The listings were reviewed and the data summarized by an IARC Working Group. The methodology used to generate these listings and plots is described, and results are given for one compound, benzene. The entire data base contains approximately 7000 entries from 4000 references.     

Bacterial systems for carcinogenicity testing

During the past 30 years, bacterial test systems have been extensively refined in their ability to detect not only mutagenic agents but, in many cases, carcinogenic ones as well. Since many carcinogens are known to be activated within the mammalian body, major improvements in bacterial test systems were made when representative parts of mammalian metabolism were included as part of the test protocol. Presently, systems of great simplicity and convenience are available for the efficient detection of gene mutations, lysogenic induction of prophages, and differential DNA repair. These qualities render bacterial systems potentially useful in distinguishing between carcinogens and non-carcinogens, in characterizing induced mutation spectra, and possibly in quantifying mutagenic potency that may be used to predict tumor-initiating potency. Sensitive strains of Salmonella typhimurium. Escherichia coli and Bacillus subtilis with altered DNA-repair capacities have been constructed which accurately identify many carcinogens. Comparative studies have shown that techniques using these strains can be standardized to some extent and that the majority of carcinogens are active in all adequately sensitive genetic systems. Because of this redundancy, it may be sufficient to employ only one standardized set of tester strains and methodology. However, serveral classes of known carcinogens are undetected or underestimated when assayed in standard testing procedures. Some of these chemicals can be efficiently recognized as mutagens upon varying the methodology, the genetic endpoint, or the mammalian activation system. Thus, to modify and adjust the experimental protocol to the particular type of chemical under study and to calibrate the system with appropriate carcinogenic and non-carcinogenic reference compounds is advisable. It is noteworthy that chemical carcinogens which probably act by non-genotoxic mechanisms thus far remain undetected in bacterial tests. Newly developed systems which measure specific types of genetic events, such as transpositions of DNA segments and derepression of genes, presently are being tested for their ability to detect such carcinogens. A final matter of growing concern is the increasing number of environmental chemicals that are found to be mutagenic in bacteria but for which information about carcinogenic activity in vivo is insufficient. The possible use of bacteria for quantifying mutagenic potency and extrapolating this information to tumor-initiating potency can be envisaged in three ways: (i) direct extrapolation from standard in vitro tests, (ii) indirect extrapolation making use of an in vitro/in vivo comparison of induced effects (the parallelogram method) as devised by Sobels [138] on the basis of identical dose (to DNA), and (iii) host-mediated assays to assess mutagenic potency of carcinogens in selected organs of mammals...     

An analysis by chemical class of Salmonella mutagenicity tests as predictors of animal carcinogenicity

For a number of years, investigators have recognized that humans potentially are exposed to large numbers of genotoxicants. Many efforts have attempted to validate various short-term bioassays for use as rapid, inexpensive screens for genotoxicants--especially carcinogens. In this analysis, we examine Salmonella mutagenicity as an indicator of potential carcinogenicity by comparing published (and when possible, evaluated) Salmonella results with the evaluated Gene-Tox animal carcinogen data base. The Salmonella bioassay does especially well in those cases where the level of evidence for carcinogenicity is the strongest. Analysis shows that except for specific classes of compounds, the plate-incorporation protocol and the preincubation protocol are equally efficient at detecting mutagens. This paper also demonstrates how validation values (sensitivity, specificity, etc.) vary with chemical class. Overall, this analysis demonstrates that when used and interpreted in a meaningful chemical class context, the Salmonella bioassay remains extremely useful in identifying potential animal carcinogens.     

Early detection of carcinogenic substances and modifiers in rats

Over the past 20 years, we have been developing in vivo medium-term bioassay systems in rats for detecting carcinogenic and modifying effects of test compounds. The systems are based on the two-step hypothesis of carcinogenesis. In a liver model, male F344 rats are initially given a single dose of diethylnitrosamine (DEN, 200 mg/kg, i.p.) and starting 2 weeks later are treated with test compounds for 6 weeks and then killed, all rats being subjected to two-thirds partial hepatectomy at week 3. Carcinogenic potential is scored by comparing the numbers and areas per cm(2) of induced glutathione S-transferase placental form (GST-P) positive foci in the livers of groups of about 15 rats with those of corresponding control groups given DEN alone. A positive response is defined as a significant increase in the quantitative values of GST-P-positive foci, such a negative response as no change or a decrease. The results obtained have been compared with reported Salmonella/microsome and long-term carcinogenicity test findings for the same compounds. Of the liver carcinogens, 30 out of 31 (97%) mutagenic and 29 out of 33 (88%) non-mutagenic compounds gave positive results. Carcinogens other than hepatocarcinogens gave a lower proportion of positive results (9 out of 42, 21%). This bioassay also provides information concerning inhibitory potential. The practical utility and benefits of a multi-organ medium-term experimental protocol for early detection of carcinogenic agents and modifiers acting at sites other than the liver are also discussed.     

Activity profiles of developmental toxicity: design considerations and pilot implementation

The available literature was searched for quantitative test results from both in vitro and in vivo assays for developmental toxicity for five model compounds: cyclophosphamide, methotrexate, hydroxyurea, caffeine, and ethylenethiourea. These compounds were chosen on the basis of their extensive utilization in a variety of assay systems for developmental toxicity as evidenced by their representation in the ETIC database (each generally has 100-500 citations encompassing multiple test systems). Nine cellular-based assays, six assays using whole embryos in culture, as well as Segment II and abbreviated exposure tests for mammalian test species are included in the database. For each assay, the critical endpoints were identified, each of which was then provided a three-letter code, and the criteria for extraction of quantitative information were established. The extracted information was placed into a computerized reference file and subsequently plotted such that the qualitative (positive/negative) and quantitative (e.g., IC50, highest ineffective dose (HID), lowest effective dose (LED] results across all test systems could be displayed. The information contained in these profiles can be used to compare qualitative and quantitative results across multiple assay systems, to identify data gaps in the literature, to evaluate the concordance of the assays, to calculate relative potencies, and to examine structure-activity relationships.     

Evaluation of the Escherichia coli K12 inductest for detection of potential chemical carcinogens

46 chemicals of diverse classes and structures, including 30 known animal carcinogens, were evaluated for prophage-inducing ability using the Escherichia coli inductest with lysogenic strain GY5027 envA - uvrB- and indicator strain GY4015 ampR . The inductest detected 9 of 30 known carcinogens as genotoxic agents, including 3 polycyclic hydrocarbons, 2 aflatoxins, and 2 antitumor antimicrobials. Among the 21 carcinogens ineffective as prophage inducers were 3 aromatic amines (other than 2-aminoanthracene), 3 azo-aminoazo compounds, 2 methanesulfonates, and 2 nitro aromatics. In contrast, 18 and 17 of the 30 animal carcinogens were detected as genotoxic agents in the Salmonella/Ames test and E. coli WP2/ WP100 rec assay, respectively. The threshold sensitivity of the inductest was less than that of the Salmonella/Ames test for chemicals genotoxic in both tests. The ineffectiveness of the inductest as a routine test for detecting potential chemical carcinogens may be related to the nature of the DNA damage lesions formed by various genotoxic agents.     

Enhancement of bleomycin-induced micronucleus formation in V79 cells as a rapid and sensitive screen for non-covalent DNA-binding compounds.

Non-covalent drug/DNA interactions are difficult to study and because of this, the significance of such interactions from a safety standpoint and their contribution to positive genetic toxicology test findings is poorly understood. It is shown in the present study that such interactions may be detected and quantified in Chinese hamster V79 cells by an adaptation of the bleomycin amplification assay. This assay measures the ability of a test compound to enhance the DNA damaging activity of the antibiotic bleomycin using micronucleus formation as an endpoint. Results are presented examining the bleomycin amplification activity of known intercalating agents, groove-binding agents and other structurally diverse classes of compounds for which intercalative status has not been reported. The assay reveals a strong and predictable SAR for amplification activity based on number and orientation of aromatic rings. Moreover, excellent correlations are observed between DNA binding (viscometric analyses) and DNA amplification in V79 cells for a series of seven experimental compounds. The assay is shown to be useful in understanding the genotoxicity of marketed antihistamines and to help explain genetic toxicology findings observed in a series of novel pharmaceutical entities. It is proposed that assessment of bleomycin amplification activity of novel compounds in early genotoxicity prescreening may provide important information upon which to base synthesis of compounds with minimal or no genotoxic liability.     

The genetic toxicology of putative nongenotoxic carcinogens

This report examines a group of putative nongenotoxic carcinogens that have been cited in the published literature. Using short-term test data from the U.S. Environmental Protection Agency/International Agency for Research on Cancer genetic activity profile (EPA/IARC GAP) database we have classified these agents on the basis of their mutagenicity emphasizing three genetic endpoints: gene mutation, chromosomal aberration and aneuploidy. On the basis of results of short-term tests for these effects, we have defined criteria for evidence of mutagenicity (and nonmutagenicity) and have applied these criteria in classifying the group of putative nongenotoxic carcinogens. The results from this evaluation based on the EPA/IARC GAP database are presented along with a summary of the short-term test data for each chemical and the relevant carcinogenicity results from the NTP, Gene-Tox and IARC databases. The data clearly demonstrate that many of the putative nongenotoxic carcinogens that have been adequately tested in short-term bioassays induce gene or chromosomal mutations or aneuploidy.     

An evaluation of the L5178Y TK+/− mouse lymophoma forward mutation assay using 42 chemicals

The L5178YTK^+/? mouse lymphoma assay (MLA) has been utilized in several laboratories as a short-term test for chemical-induced forward mutation in cultured mammalian cells. In order to evaluate several technical modifications to the MLA, 42 chemicals representing 9 chemical classes were tested and the results were compared with those published elsewhere as well as with findings in a genetic toxicology test battery currently used in this laboratory. A positive response for the induction of TK^+/? mutants was obtained for 26 chemicals. With the exception of p-aminophenol, all of these compounds were recognized mutagens or carcinogens and were represented of direct-acting and activation-dependent genotoxins. 16 compounds did not induce IK^?/? mutanants and among these were 5 compounds that were considered to be mutagens or carcinogens. A comparison of the results of this study with those published elsewhere revealed a strong agreement among findings for this test irrespective of minor technical variations. It was concluded that th MLA is a useful system for identifying chemical mutagens in mammalian cells and can serve as a valuabel component in a genetic toxicology test battery.     

The Escherichia coli K-12 SOS chromotest agar spot test for simple, rapid detection of genotoxic agents

The Escherichia coli K-12 SOS chromotest is a colorimetric (beta-galactosidase induction) system for detecting genotoxic chemicals as agents which induce filamentation in response to DNA damage. The chromotest was modified from a liquid suspension assay to a simple, convenient agar spot test, which was performed in the manner of a related colorimetric prophage induction assay (BIA). Chromotest agar dishes yielded optimal results after 16-18 h incubation, presumably because of the agar growth characteristics of tester strain PQ37. Of 44 tested chemicals, nitro aromatics, cytotoxic/antitumor agents, polycyclic hydrocarbons and aflatoxins showed good activity. Alkylating agents such as MNNG and MMS were active only at high concentrations. Compounds active in both the chromotest and BIA were active at 10-100-fold lower concentrations in the chromotest. The chromotest appeared to be less effective than the Salmonella Ames mutagenicity test in the detection of diverse classes of chemical carcinogens. The chromotest may be a useful alternative to the BIA in the study of particular classes of genotoxic compounds.     

Assessing the use of known mutagens to calibrate the Salmonella typhimurium mutagenicity assay: I. Without exogenous activation

There has been an increasing need in genetic toxicology to progress from strictly qualitative tests to more quantitative tests. This, in turn, has increased the need to develop better quality assurance and comparative bioassay methods. In this paper, two laboratories tested 10 Salmonella mutagens in order to determine the usefulness of selected chemicals as potential reference materials to calibrate the Salmonella assay. If variance within a bioassay is sufficiently low and the rankings of the compounds are of acceptable consistency, the chemicals later could be evaluated for use as standard control compounds, as audit materials, and as standard reference materials for comparative bioassay efforts. The results demonstrated that the chosen chemicals (with the possible exception of dimethylcarbamylchloride) provide such consistent results in the Salmonella mutagenicity bioassay that they can be used for semi-quantitative calibration and as possible bioassay controls, special audit chemicals, and potentially as reference standards in comparative bioassay efforts. Reference standards, whether used as audit materials or in comparative bioassays, must be used concurrently with the test substances of interest; used without bias; used in a standardized, highly controlled bioassay; and be tested across an appropriate dose range. The study also shows that when these compounds are used as reference standards much care must be given to the number and spacing of doses if highly reproducible slope values are to be generated. We recommend use of a pilot test to establish a dose range for definitive tests and the placement of doses for the definitive tests within the first half of the linear dose-response curve. For appropriate comparisons, one should replicate the tests using the defined dose range and analyze the results in a non-biased statistical manner.     

How many high production chemicals are rodent carcinogens? Why should we care? What do we need to do about it?

High production volume (HPV) chemicals are produced in or imported to the US in amounts greater than 1 million pounds per chemical per year. The EPA has identified thousands of HPVs. Due to their abundance, such chemicals bring a substantial risk for human exposure, and as a result some level of adverse consequences to health are to be expected. In order to examine the potential for cancer risk associated with HPVs, this paper examines HPVs that have been tested in the National Toxicology Program's rodent cancer bioassay. The chemicals tested in the bioassay represent a small sample of the universe of environmental chemicals to which people are exposed. Unexpectedly, 60% of the 128 HPVs evaluated in the bioassay proved to be rodent carcinogens. This value compares to a predicted prevalence of only 16.5% carcinogens in a previous study. The previous study concluded that HPVs were less likely to be toxic by a variety of other test criteria as well. However, the approach involved identifying putative carcinogens using quantitative chemical structure-activity relationships (QSAR) in contrast to the direct tabulation of bioassay test results performed here. Detailed examination of bioassay results reveals that test outcomes depend heavily on route of administration as well as on dose level, sex, strain, and species used. Since most of these factors have little or no apparent relationship to chemical structure, results of this study suggest that QSAR, as well as virtually all other alternative methods, may not generally provide accurate predictions of carcinogenic potential. As we wait for efficient and effective methods for predicting carcinogens to be developed, people continue to be exposed to environmental carcinogens. Progress on regulating environmental carcinogens as well as on developing more effective test methods has been minimal since "war on cancer" began approximately 30 years ago. The present study questions whether the current strategy for dealing with environmental carcinogens will ever be successful. Close examination of rodent cancer test results seems to suggest that almost all chemicals may have carcinogenic potential in some genotypes under some exposure circumstances. If this hypothesis is correct, it would explain the general lack of progress in developing methods to differentiate carcinogens from noncarcinogens. A completely new strategy for dealing with cancer caused by exposures to environmental chemicals seems to be needed.     

An evaluation of tests using DNA repair-deficient bacteria for predicting genotoxicity and carcinogenicity. A report of the U.S. EPA's Gene-TOX Program

The detection of DNA-damaging agents by repair-deficient bacterial assays is based on the differential inhibition of growth of repair-proficient and repair-deficient bacterial pairs. The various methodologies used are described and recommendations are made for their improved use. In a survey of the literature through April 1979, 91 of 276 papers evaluated contained usable data, resulting in an analysis of 611 compounds that had been assayed in 1 or more of 55 pairs of repair-proficient and repair-deficient strains. The results indicate that (1) a liquid suspension assay is more sensitive than a spot (diffusion) test. In a review of the Escherichia coli polA assay, 45 compounds that gave "No Test" in the spot test were clearly positive or negative in the liquid suspension assay. (2) Of the 21 compounds analyzed by the E. coli polA assay and by other E. coli repair-deficient strains (e.g., rec, uvr, hcr, and exr derivatives of WP2 and AB1157), 10 were in complete agreement in all strains except uvrA strains. This indicates that strains other than polA+/polA- are useful for detecting DNA-damaging agents. However, in selecting strains for use in these assays, care should be taken to consider repair pathway specificity for particular compounds. (3) There was a 78% correspondence between results obtained with E. coli polA and Bacillus subtilis (H17/M45, 17A/45T) rec assay and between E. coli polA and Proteus mirabilis. (4) In a comparison of test results with carcinogenicity data, 44 of 71 (62%) carcinogenic compounds assayed by the polA system were positive, 10 (14%) were negative, and 17 (24%) gave No Test or doubtful results. 7 carcinogens were assayed by other E. coli strains and all were positive. 56 carcinogens were assayed in B. subtilis: 24 (43%) were positive, 9 (16%) were negative, and 23 (41%) gave No Test or doubtful results. Of the 7 carcinogens assayed in P. mirabilis, 6 (86%) were positive and 1 (14%) was negative. (5) The results were analyzed with respect to chemical classes. E. coli polA detected the highest percentage of hydroxylamines and alkyl epoxides. The B. subtilis rec assay detected the highest percentage of nitrosamines and sulfur and nitrogen oxides. It is concluded that some of these test systems are effective tools for the detection of DNA-damaging and potentially carcinogenic compounds, especially if the assay is done in liquid suspension and if more than 1 pair of tester strains is used. Advantages and disadvantages of the assay are discussed and suggestions are made for improvements in the system.     

International Commission for Protection against Environmental Mutagens and Carcinogens. ICPEMC working paper No. 5. Genotoxicity tests as predictors of carcinogens: an analysis

Differences between the results of numerical validation studies comparing in vitro and in vivo genotoxicity tests with the rodent cancer bioassay are leading to the perception that short-term tests predict carcinogenicity only with uncertainty. Consideration of factors such as the pharmacokinetic distribution of chemicals, the systems available for metabolic activation and detoxification, the ability of the active metabolite to move from the site of production to the target DNA, and the potential for expression of the induced lesions, strongly suggests that the disparate sensitivity of the different test systems is a major reason why numerical validation is not more successful. Furthermore, genotoxicity tests should be expected to detect only a subset of carcinogens, namely genotoxic carcinogens, rather than those carcinogens that appear to act by non-genetic mechanisms. Instead of relying primarily on short-term in vitro genotoxicity tests to predict carcinogenic activity, these tests should be used in a manner that emphasizes the accurate determination of mutagenicity or clastogenicity. It must then be determined whether the mutagenic activity is further expressed as carcinogenicity in the appropriate studies using test animals. The prospects for quantitative extrapolation of in vitro or in vivo genotoxicity test results to carcinogenicity requires a much more precise understanding of the critical molecular events in both processes.     

An overview of bioactivation of chemical carcinogens

Most environmental carcinogens require metabolic activation to reactive intermediates and are mutagenic in appropriate test systems. During the last decade, the cDNAs of numerous xenobiotic-metabolizing enzymes have been cloned. The individually expressed enzymes were used to study their substrate specificities and their inhibition by other compounds. Various enzymes were expressed directly in target cells of in vitro mutagenicity tests. This is illustrated in the present study for rat and human sulphotransferases (SULTs) expressed in Salmonella typhimurium TA1538. Numerous compounds were mutagenic in the new test system. Some of these promutagens were activated by several different SULT forms, whereas many other promutagens were activated with high selectivity by a specific enzyme form, but not by genetically closely related forms from the same species (e.g. allelic variants) or orthologous enzymes from other species. Similar findings have been made using recombinant test systems for specific forms of other classes of enzymes (e.g. cytochromes P450). This high selectivity in activation (and inactivation) may explain some organotropisms as well as species and inter-individual differences in the action of carcinogens. Many carcinogen-metabolizing enzymes are induced or inhibited by other xenobiotics. Such interactions can be exploited for chemo-prevention, which however may be carcinogen- and tissue-dependent.     

Induction of genetic tandem duplications in Salmonella by polycyclic aromatic hydrocarbon and amine carcinogens

6 polycyclic aromatic hydrocarbon or similar amine carcinogens were tested as inducers of genetic tandem duplications in a rough strain of Salmonella typhimurium. When metabolically activated by rat-liver microsomes, all 6 were active in inducing genetic tandem duplications, yielding from over 3 times to almost 14 times as many tandem duplicants per viable bacterium as did concurrent uninduced control cultures. These results extend the number and chemical diversity of carcinogens shown to induce genetic duplications in bacterial tester systems. We suggest that polycyclic hydrocarbon carcinogens may act in carcinogenesis by inducing genetic duplications or other genetic rearrangements. Duplication induction may be a useful genetic endpoint for screening potential carcinogens.     

Consideration of both genotoxic and nongenotoxic mechanisms in predicting carcinogenic potential

Bacterial and cell culture genotoxicity assays have proven to be valuable in the identification of DNA reactive carcinogens because mutational events that alter the activity or expression of growth control genes are a key step in carcinogenesis. The addition of metabolizing enzymes to these assays have expanded the ability to identify agents that require metabolic activation. However, chemical carcinogenesis is a complex process dependent on toxicokinetics and involving at least steps of initiation, promotion and progression. Identification of those carcinogens that are activated in a manner unique to the whole animal, such as 2,6-dinitrotoluene, require in vivo genotoxicity assays. There are many different classes of non-DNA reactive carcinogens ranging from the potent promoter 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) that acts through a specific receptor, to compounds that alter growth control, such as phenobarbital. Many compounds, such as saccharin, appear to exhibit initiating, promotional and/or carcinogenic activity as events secondary to induced cytotoxicity and cell proliferation seen only at the chronic lifetime maximum tolerated doses mandated in rodent bioassays. Simple plus/minus vs. carcinogen/noncarcinogen comparisons used to validate the predictivity of bacterial and cell culture genotoxicity assays have revealed that a more comprehensive analysis will be required to account for the carcinogenicity of so many diverse chemical agents. Predictive assays and risk assessments for the numerous types of nongenotoxic carcinogens will require understanding of their mechanism of action, reasons for target organ and species specificity, and the quantitative dose-response relationships between endpoints such as induced cell proliferation and carcinogenic potential.     

Potential identification of chemical carcinogens in a viral transformation system

Chemical carcinogens from several diverse chemical classes i.e.; aromatic amines, polycyclic hydrocarbons, nitrosamines, hormonal derivatives, metals and direct alkylating agents cause a 6.2–60.5-fold increase in the frequency of murine sarcoma virus (MSV)-induced transformation in a normal rat kidney (NRK) cell system. Exogenous metabolic activation with a rat liver S-9 homogenate is required for expression of this activity by procarcinogens. Non-carcinogenic analogs of these compounds fail to cause significant increases in the transformation frequency either with or without prior metabolic activation. Iododeoxyuridine, a mutagen also does not cause enhancement of transformation. This system may serve as the basis for a rapid and quantifiable means of identifying chemical carcinogens while introducing a new model for the understanding of the interactions between oncorna-viruses and chemical carcinogens.     

Tradescantia-micronucleus bioassay for detection of carcinogens

The Tradescantia-Micronucleus (Trad-MCN) bioassay is one of the tests used in the International Program on Plant bioassays (IPPB) under the auspices of the United Nations Environment Programme (UNEP). Using chromosome damage as the indicator of the carcinogenic properties of environmental agents, the Trad-MCN bioassay is a quick and efficient tool for screening carcinogens in gaseous, liquid and solid forms. Test results can be obtained within 24-48 hr after the exposure either on site or in the laboratory. The international standard protocol of this test was published in 1994 and a list of test results of carcinogens and clastogens compiled from publications in the last 23 years will be presented. Under the IPPB/UNEP, more than 40 institutes including public health, medical and cancer research in the major countries of the world are carrying on the monitoring task on genotoxicity of polluted air, water and soil. At the same time, the Trad-MCN can be used in a global scale to detect carcinogens as a preventive measure of cancer.     

The effect of N-nitroso carcinogens and triazenes on protein synthesis in cell-free systems

The effect of a series of carcinogenic nitrosamines, alkylnitrosoureas and alkaryltriazenes on enzymatic reactions involved in protein synthesis was studied in cell-free systems from rat liver. The addition of most compounds stimulated the formation of aminoacyl-tRNA complex in test systems from rat liver whereas analogous preparations from Escherichia coli did not show this effect. The polymerization of phenylalanine and the binding of aminoacyl-tRNA to ribosome were only slightly and apparently non-specifically inhibited in the presence of the test compounds. On the other hand, the binding of nRNA to ribosome was markedly stimulated after the addition of most carcinogens tested. It appears that the carcinogens intervene specifically with one of the early steps in peptide initiation. Since the binding of mRNA to ribosome is known to be an important rate-limiting step in protein synthesis, the N-nitroso carcinogens and triazenes may thus control the expression of genetic message at the translation level.