• Open Access

Deletion and single nucleotide substitution at G:C in the kidney of gpt delta transgenic mice after ferric nitrilotriacetate treatment


To whom correspondence should be addressed. E-mail: toyokuni@path1.med.kyoto-u.ac.jp


An iron chelate, ferric nitrilotriacetate (Fe-NTA), induces oxidative renal proximal tubular damage that subsequently leads to a high incidence of renal cell carcinoma in rodents, presenting an intriguing model of free radical-induced carcinogenesis. In the present study, we used gpt delta transgenic mice, which allow efficient detection of point mutations and deletions in vivo, to evaluate the mutation spectra, in association with the formation of 8-oxoguanine and acrolein-modified adenine during the first 3 weeks of carcinogenesis. Immunohistochemical analysis revealed the highest levels of 8-oxoguanine and acrolein-modifed adenine in the renal proximal tubules after 1 week of repeated administration. DNA immunoprecipitation and quantitative polymerase chain reaction analysis showed that the relative abundance of 8-oxoguanine and acrolein-modified adenine at the gpt reporter gene were increased at the first week in the kidney. Similarly, in both 6-thioguanine and Spi selections performed on the renal specimens after Fe-NTA administration, the mutant frequencies were increased in the Fe-NTA-treated mice at the first week. Further analyzes of 79 mutant clones and 93 positive plaques showed a high frequency of G:C pairs as preferred targets for point mutation, notably G:C to C:G transversion-type mutation followed by deletion, and of large-size (>1 kilobase) deletions with short homologous sequences in proximity to repeated sequences at the junctions. The results demonstrate that the iron-based Fenton reaction is mutagenic in vivo in the renal tubular cells and induces characteristic mutations. (Cancer Sci 2006; 97: 1159–1167)


acrolein-modified 2′-deoxyadenosine




base pairs




2′-deoxycytidine triphosphate


DNA immunoprecipitation


ethylenediaminetetraacetic acid




ferric nitrilotriacetate


mutant frequency


mitomycin C




polymerase chain reaction






ultraviolet B



Oxidative stress is associated with a variety of pathological phenomena, including infection, inflammation, ultraviolet- and γ-irradiation, overload of transition metals and certain chemical agents.(1) Many epidemiological studies have demonstrated a close association between chronically oxidative conditions and carcinogenesis. For example, chronic tuberculous pleuritis causes a high incidence of malignant lymphoma;(2) asbestosis (asbestos fibers are rich in iron),(3) is often associated with mesothelioma and lung carcinoma;(4) chronic Helicobacter pylori infection is associated with a high incidence of gastric cancer;(5,6) the incidence of colorectal cancer is increased in ulcerative colitis;(7,8) a high risk for heptocellular carcinoma is observed in patients with genetic hemochromatosis, an iron overload disease;(9,10) severe burns by ultraviolet radiation is a risk factor for skin cancer;(11,12) and γ-irradiation causes a high incidence of leukemia.(13) At least under these circumstances, and probably in other types of carcinogenesis as well, oxidative stress appears to play a major role in human carcinogenesis.

An iron chelate, ferric nitrilotriacetate (Fe-NTA), causes oxidative renal proximal tubular injury via the Fenton reaction, and this injury ultimately leads to a high incidence of renal cell carcinoma in mice(14) and rats(15) after repeated intraperitoneal administration. This is an intriguing model in the following respects: (1) more than half of the induced tumors metastasize to the lung and/or invade the peritoneal cavity, resulting in animal mortality;(16) (2) convincing evidence exists regarding the involvement of free radical reactions in the carcinogenic process, including not only an increase in covalently modified macromolecules (oxidatively modified DNA bases(17) and lipid peroxidation products)(18,19) but also preventive effects of α-tocopherol fortification against carcinogenesis;(20) (3) genetic changes in the p16INK4a tumor suppressor gene, especially homozygous deletions(21,22) and expressional alteration of several key genes, including annexin 2(23) and thioredoxin binding protein-2,(24) have been observed.

Fe-NTA itself is Ames test-negative,(14) but is positive in other cell culture systems detecting mutations.(25,26) Thus far, its mutation spectrum has not been comprehensively studied. Since the Ames test is a system involving prokaryotes, an assay system with the ability to detect mutations under in vivo conditions in which eukaryotic DNA repair mechanisms, metabolic pathways and other physiological systems are operative would offer significant advantages with respect to reliability. Based on this premise, several transgenic mouse mutagenesis assay systems have been developed, including Muta mice,(27) Big Blue mice(28) and HITEC mice.(29) These systems employ a recoverable transgenic lambda phage vector containing a reporter gene from bacteria. However, these systems all have the limitation that large deletions cannot be efficiently detected. We have developed a novel mutagenesis test system named gpt delta transgenic mice, which are transgenic for the lambda EG 10 gene containing the gpt gene of Escherichia coli.(30) An important feature of this system is that both point mutations and large deletions can be tested concurrently in the targeted organs of the mice; point mutations are detected by 6-thioguanine (6-TG) selection and deletions larger than 1 kb can be identified by Spi (sensitive to P2 interference) selection. Thus far, various mutagens, including γ-ray irradiation, UVB, mitomycin C and PhIP, have been studied by using this in vivo system.(31)

In the present study, we used gpt transgenic mice to investigate the early genetic changes in Fe-NTA-induced renal carcinogenesis. Furthermore, we studied the relative abundance of two different types of DNA base modifications in several limited genomic loci with a novel technique called DNA immunoprecipitation (DnaIP), which selectively collects enzyme-digested DNA fragments containing the target oxidative DNA base modification with specific monoclonal antibody. The present study for the first time revealed characteristics of the mutation spectrum in the kidney following repeated episodes of the Fenton reaction.

Materials and Methods

Animals and chemicals. Gpt delta C57BL/6 J transgenic mice were provided by Dr Takehiko Nohmi (Division of Genetics and Mutagenesis, National Institute of Health Sciences, Tokyo, Japan) and maintained in Kyoto University under specific-pathogen free and light-, temperature- and humidity-controlled conditions. The animal experiment committee of the Graduate School of Medicine, Kyoto University, approved the present experiments. Fe(NO3)39H2O was obtained from Wako (Osaka, Japan). Nitrilotriacetic acid, disodium salt, was purchased from Nacalai Tesque (Kyoto, Japan). Fe-NTA was prepared immediately before use as described previously.(18) A total of 12 4-week-old male mice were used; nine mice were subjected to repetitive Fe-NTA administration and three mice were used as untreated controls. Mice were injected intraperitoneally with 3 mg iron/kg of Fe-NTA daily for three days, and the dose was increased to 5 mg iron/kg of Fe-NTA from the fourth day according to the established carcinogenesis protocol.(16) The injections were performed five times a week at approximately 10.00 hours. The animals were killed 48 h after the final administration. Both kidneys and the central lobe of the liver were immediately excised. Half of one kidney and a portion of the liver were used for histological and immunohistochemical analysis, and the rest of the kidney was frozen in liquid nitrogen and stored at −80°C for mutational analyzes.

Monoclonal antibodies.  Monoclonal antibody N45.1 recognizing 8-hydroxy-2′-deoxyguanosine (8-OHdG)(32) and monoclonal antibody mAb21 recognizing acrolein-2′-deoxyadenosine adduct (acrolein-dA)(33) were used.

Histological and immunohistochemical analyzes.  Kidney specimens were fixed with phosphate-buffered 10% formalin and embedded in paraffin, cut at 3-µm thickness and stained with hematoxylin and eosin staining. For immunohistochemical analyzes, the avidin-biotin complex method with peroxidase was used as described previously.(32,33)

DNA immunoprecipitation and quantitative PCR analysis.  To evaluate the relative abundance of Fe-NTA-induced oxidative DNA base modifications (8-OHdG and acrolein-dA) at desired genomic loci, we developed a technique designated as DnaIP (DNA immunoprecipitation).(34) More details will be published elsewhere.(35) Briefly, genomic DNA was extracted from each kidney of gpt delta transgenic mice with the NaI method (Wako) using argon gas-saturated buffer to avoid further oxidation during the extraction procedures.(36) Twenty µg of genomic DNA was digested with HaeIII (TakaraBio, Shiga, Japan), and incubated with each antibody (10 µg of N45.1 or 2 µg of mAb21) in 10 mM phosphate-buffered saline, pH 7.4, containing 0.1% bovine serum albumin, for 3 h at 4°C in a 900-µL volume. The mixture was then incubated with 100 µL of Dynabeads M-280 sheep antimouse IgG (Dynal, Oslo, Norway) for another 3 h, washed sequentially with four different buffers (buffer 1: 0.1% sodium deoxycholate, 1% Triton X-100, 1 mM EDTA, 50 mM HEPES-KOH, 140 mM NaCl, pH 7.5; buffer 2: 0.1% sodium deoxycholate, 1% Triton X-100, 1 mM EDTA, 50 mM HEPES-KOH, 500 mM NaCl, pH 7.5; buffer 3: 0.1% sodium deoxycholate, 0.5% Nonidet P-40, 1 mM EDTA, 250 mM LiCl, 10 mM Tris-HCl, pH 8.0; and buffer 4: 1 × TE). Beads-bound DNA was recovered by incubating the beads with 80 µL of elution buffer (10 mM EDTA, 1% SDS, 50 mM Tris-HCl, pH 8.0) at 65°C for 10 min, and was amplified twice by PCR after ligation to an adaptor (sense, 5′-OH-GGAATTCGGCGGCCGCGGATCC-3′; antisense, 5′-GGATCCGCGGCCGCCG-3′; sense oligonucleotides were used as primers for amplification), treated with exonuclease I (TakaraBio) and purified with phenol-chloroform extraction. The purified products were subjected to Real-Time PCR (7300 Real Time PCR System, Applied Biosystems, Tokyo) using Platinum SYBR Green qPCR SuperMix UDG (Invitrogen, Tokyo). The primer pairs used were as follows: gpt, forward-5′-GCCTTCTGAACAATGGAAAGG-3′, reverse-5′-CGTGATCGTAGCTGGAAATAC-3′ (125 bp); β-actin, forward-5′-TCCAACAAACCAAGAGAAATCC-3′, reverse-5′-CGACCTCTGAAACAATTCTGGT-3′ (108 bp); C15-49–5 (chromosome 15, extrageneic region), forward-5′-TGGTACCTGAGTAAGGCAAGGT-3′, reverse-5′-CCCACTTGTGATTGCTTTCTTC-3′ (107 bp); C16-47–2 (chromosome 16, extrageneic region): forward-5′-CACACACACATGCACACTGTACT-3′, reverse-5′-GCATTTCTCCTCACATTCAGACT-3′ (114 bp); C16-47–5 (chromosome 16, extrageneic region): forward-5′-CCAATTGGAGCTAACAGAAACC-3′, reverse-5-AGCTGGTCAACTGCCTACTCTC-3′ (125 bp). These three extrageneic areas were selected based on our observations that chromosome 15 is peripherally located and chromosome 16 is centrally located in the murine renal tubular cells at interphase.(35)

In vitro phage packaging.  Genomic DNAs were extracted with the phenol-chloroform extraction protocol.(37) Trangenic lambda EG10 DNA was rescued from the host genomic DNA using Transpack Packaging Extract (Stratagene, La Jolla, CA) according to the manufacturer's instructions.(30)

Mutation analysis.  The 6-TG mutation assay protocol has been described elsewhere.(38–40) Briefly, rescued phage was infected into YG 6020 E. coli expressing Cre enzyme, converted into a plasmid carrying the Cm-resistance gene and gpt gene, and poured on plates containing chloramphenicol (Cm) with or without 6-TG. The positive clones carrying the mutant gpt gene were obtained from 6-TG selection plates by incubating at 37°C for 96 h. Selected clones were confirmed again by plating on 6-TG selection plates. The whole gpt sequence was amplified from positive clones and identified by sequencing with an ABI PRIZM 377 sequencer. The primers used for amplifying and sequencing were as follows: forward-5′-GCGCAACCTATTTTCCCCTCGA-3′ and reverse-5′-TGGAAACTATTGTAACCCGCCTG-3′. The same primer pair was used for direct sequencing.(41) E. coli XL1-Blue MRA and XL1-Blue MRA (P2) were infected with the packaged phage. E. coli XL1-Blue MRA was poured onto NZY plates and XL1-Blue MRA (P2) was poured onto I-trypticase agar plates. Plaques that grew on the XL1-Blue MRA (P2) plates were selected and further confirmed with E. coli XL1-Blue MRA, E. coli WL95 (P2) and XL1-Blue MRA (P2). Positive plaques were recovered and used for determining the deletion position of the red/gam gene. Clones or plaques were counted for determining mutant frequencies (MFs). MFs were calculated by using established methods as described previously.(30,42,43)

Hybridization assay and PCR analysis for Spi mutant analysis.  A protocol for Southern blot analysis for Spi (sensitive to P2 interference) mutants has been established.(43) Seventeen oligomers located within ∼14 kb flanking sequence of the red/gam gene were used as probes for identifying the deletion junctions. These oligomers were named 18874R, 19258R, 20341R, 21328R, 22556R, 22869R, 23921R, 24858R, 25389F, 26704F, 27096F, 28165F, 29290F, 30104F, 31070F, 31879F and 32890F according to their position as described.(43) The oligomers were spotted onto HybondTM-N+ membrane (Amersham) and cross-linked with UV. PCR products, which were amplified by primer 18874R and 32890F using positive individual plaques as templates, were labeled with (α-32P) dCTP using the Megaprime DNA labeling System (Amersham). The membranes were incubated with labeled PCR products at 50°C overnight, washed three times and exposed to BioMax film (Kodak, New York, NY). Deleted regions were located within those oligomers whose signals could not be observed on the film. The nearest primers were selected for PCR amplification and the PCR products were subjected to sequencing to determine the exact deletion junction.

Statistical analysis.  Statistical analyzes were performed with an unpaired t-test, which was modified for unequal variances when necessary.


Renal histology after repeated Fe-NTA administration.  As shown in Fig. 1a, no significant histological changes were observed in the kidney of the untreated control group. In contrast, pyknotic nuclei of proximal tubular cells revealing degeneration were scattered in the kidney of mice after 1 week of Fe-NTA treatment (Fig. 1b). Degenerative tubular cells were no longer observed there after 2 or 3 weeks of repeated Fe-NTA treatment, but atypical regenerative cells with a large nucleus containing prominent nucleoli were gradually increased (Fig. 1c,d). In either case, histological evaluation of the liver showed no apparent alterations (data not shown).

Figure 1.

Immunohistochemical analysis of 8-hydroxy-2′-deoxyguanosine (8-OHdG) and acrolein-modified 2′-deoxyadenosine after repeated administration of ferric nitrilotriacetate (Fe-NTA). (a–d) Hematoxylin and eosin (HE) staining. Regenerative proximal tubular cells were prominent at the first week, together with some necrotic cells (b, inline image). At the second and third week, necrotic cells were no longer observed but increasing numbers of karyomegalic cells (c and d, inline image) appeared. (e–h) Immunohistochemistry of 8-OHdG. Nuclear immunopositivity was observed after Fe-NTA administration, with the highest level after repeated administration for 1 week (f). (i–l) Immunohistochemistry of acrolein-dA. Nuclear immunopositivity was observed after Fe-NTA administration with that of repeated administration of 1 week the highest level (j). Refer to the Materials and Methods section for details (bar in l, 50 µm).

Oxidative DNA damage induced by repeated Fe-NTA administration.  Two major oxidative DNA base modifications, 8-OHdG and acrolein-dA, were evaluated with immunohistochemistry and DnaIP. Intense diffuse nuclear immunostaining of 8-OHdG and acrolein-dA was prominent in the renal proximal tubules after repeated Fe-NTA administration for 1 week, and gradually decreased thereafter (Fig. 1e–l). To assess whether these oxidative modifications were increased in the gpt gene locus, quantitative PCR analysis after DnaIP was performed. The gpt reporter gene locus after 1 week of repeated Fe-NTA administration showed higher amounts of 8-OHdG and acrolein-dA than that in the untreated control group. Similar patterns were also observed in the other loci examined, but the gpt locus was the most sensitive at the first week (Fig. 2), consistently with the immunohistochemical data (Fig. 1e–l).

Figure 2.

Real-time polymerase chain reaction (RT-PCR) analysis after DNA immunoprecipitation for quantitation of oxidatively modified DNA bases at specific genomic loci. Renal genomic DNA was digested with HaeIII and subjected to immunoprecipitation (IP) with specific monoclonal antibodies against 8-hydroxy-2′-deoxyguanosine (8-OHdG) and acrolein-dA. The recovered DNA fragments were amplified after ligation to an adapter and were used as substrates for RT-PCR analyses of gpt, β-actin and three extrageneic regions at chromosome 15 or 16. Data are shown as relative abundance of PCR products amplified from recovered genomic DNA by IP per those amplified from the original genomic DNA in the same amounts. (a) 8-OHdG. (b) Acrolein-dA. Refer to the Materials and Methods section for details (N = 3, means ± SEM; *P < 0.05, **P < 0.01 versus untreated control kidney at the same genomic region; #P < 0.05, ##P < 0.01 versus gpt locus data of the same treatment group).

Fe-NTA-induced mutant frequencies in gpt and red/gam genes.  We then investigated the reporter genes, gpt and red/gam, to analyze Fe-NTA-induced mutations using the 6-TG and Spi selection systems. In both 6-TG and Spi selections, the mutation frequencies were significantly increased (2.44-fold increase in 6-TG selection and 1.72-fold increase in Spi selection) after 1 week of repeated Fe-NTA administration (Fig. 3), which was consistent with the results of immunohistochemistry (Fig. 1e–l) and DnaIP (Fig. 2).

Figure 3.

Mutant frequency (MF) of 6-TG and Spi seletion. 6-TG selection was used for the detection of base substitutions in the gpt gene; Spi selection was used for the detection of large-size deletions. Refer to the Materials and Methods section for details. (N = 3, means ± SEM; *P < 0.05, **P < 0.01 versus untreated control kidney).

Fe-NTA-induced gpt gene mutations.  To further characterize the exact gpt mutations caused by Fe-NTA, 79 mutant clones, in which 69 clones were from Fe-NTA-treated mice and 10 from untreated control mice, were analyzed (Table 1 and Fig. 4). Among the mutations induced by Fe-NTA, 75.4% (52/69) were single base substitutions, of which more than half (32/52 = 61.5%) occurred at G:C base pairs, whereas GC content of gpt gene was 46.6%. Among the Fe-NTA-induced substitutions, 40.4% (21/52) were transitions, including G:C to A:T (13/21) and A:T to G:C (8/21), whereas the rest of substitutions (31/52 = 59.6%) were transversions, including G:C to T:A (5/31), G:C to C:G (14/31), A:T to T:A (4/31) and A:T to C:G (8/31) (Table 1). In addition, 17.4% (12/69) of mutant clones were identified as carrying single- or mutiple-base deletions. Among them, 9/12 were single-base deletions, which occurred preferentially at repeated sequences (Table 1 and Fig. 4). Four insertional mutations and one tandem base substitution were also observed. In contrast, analyses of a total of 10 clones from the untreated control kidney showed that 8/10 were single-base substitutions with a single-base deletion and an insertion. In either case, complex mutations were not observed. Therefore, the results indicated that Fe-NTA-induced gpt gene mutation preferentially consisted of single-base substitutions occurring at G:C base pairs, in which transversions were more frequent than transitions (Table 1).

Table 1. Spectrum of Fe-NTA-induced mutations in the kidney of gpt delta transgenic mice
Mutation typeNucleotide positionSequence changeAmino acid changeFe-NTAControl
Transition    30.4% 50.0%
G:C–A:T 27G–ATrp–STOP1   
 39G–AGln–Gln    1
 64C–TArg–STOP1   1
110G–AArg–His5   1
113G–AArg–His    1
116G–AArg–Arg  1 
128G–AVal–Met  11 
356G–AArg–His  1  
447C–TIle–Ile  1 
A:T–G:C  2T–CMet–Thr  1  
 25T–CTrp–Ser1 2  
188A–GTyr–Cys  1  
275A–GAsp–Gly    1
410A–GGln–Arg  1  
415T–CTrp–Arg  1 
Transversion    44.9% 30.0%
G:C–T:A  3G–TSer–Ile  1  
110G–TArg–His  1  
324C–AHis–Gln    1
G:C–C:G109C–GArg–Gly  1 
125C–GPro–Arg  1 
238G–CAsp–His  1 
413C–GPro–Arg2 1  
427G–CVal–Leu1 1 
430G–CVal–Leu2   1
A:T–T:A 52A–TLys–STOP1   
 66A–TArg–Arg  1 
179T–AIle–Asn  11 
A:T–C:G 94A–CIle–Leu  1  
133T–GPhe–Val  1  
134T–GGly–STOP  11 
146A–CGlu–Ala  1  
286A–CThr–Pro1 1 
375T–GTyr–STOP    1
Deletions    17.4% 10.0%
1 base pair  8–12AAAAA–AAAA 1 31 1
 88–90AAAGG–AAGG   1 
423–425GGGCG–GGCG 1   
430TCGTA–TCTA   1 
437CGTCC–CGCC   1 
>2 base pairs156–162ATTCGTCATGT–ATCG   1  
170–171TACCG–TAG   1  
252–253TTCATC–TTTC   1  
Insertions     5.7% 10.0%
 74–75CCTT–CCAATT   1 
122–123GTAC–GTTAC 1   
310–311ATCC–ATTCC   11 
440–441CCGC–CCCGC     1
Other     1.4%  0.0%
Figure 4.

The position of point mutations in the gpt gene. Refer to the Materials and Methods section for details (x, deletion; ^, insertion; underline, more than one base pair deletion within the same case; the number in parenthesis indicates the multiplicity of the same mutation. Square (in one letter), mutation-prone area with the same sequence.

Fe-NTA-induced Spi mutations.  To characterize the Spi mutations induced by Fe-NTA, 93 positive plaques obtained from either the kidneys of Fe-NTA-treated mice or untreated control mice were screened by Southern blot analysis followed by sequencing that resulted in the confirmation of 21 large-size deletions (Fig. 5a). Large-size deletions were at first roughly positioned on ∼14 kb of sequence spanning the red/gam gene by the use of 17 different oligomers as probes. We detected signals for all the 17 oligomer probes in the blot with hybridization to the wild-type lambda EG 10 (Fig. 5b i). Signals for certain oligomers were absent with Spi mutant plaques containing large-size deletions, as shown in Fig. 5(b ii–iv). Most of the large deletions induced by Fe-NTA were more than 1 kb in size (Class I mutation,(31) (Fig. 5a). Furthermore, the majority of them (70.6%) had short homologous sequences of 1–6 bp at the junctions (Class I-A), and in many cases showed three bp or longer running sequences at the junction or its vicinity. Five cases of large-size deletions were accompanied by simultaneous single-base deletion in the red/gam gene (Fig. 5a).

Figure 5.

Size and position of large-size deletions after Spi selection with each junctional sequence. (a) Summary of the strategy and the observed deletions. P1–P4, untreated control; P5–P21, ferric nitrilotriacetate (Fe-NTA)-induced deletions. Blank areas between two double-dash lines indicate large-size deletions; short longitudinal lines indicate accompanying 1-base deletion; □, short homologous sequence; underline, short run more than 3 bp. Classification of the deletion type was done as described.(31) (b) Representative results of Southern blotting analysis for screening the deleted positions. Arabic numerals (1–17) indicate the probes used as described in the Materials and Methods section and P1, P12 and P20 correspond to the a section.


In the present experiments we have for the first time studied the mutation spectrum of the Fenton reaction-based renal tubular damage in a model of oxidative stress-induced carcinogenesis mediated by Fe-NTA. We intentionally avoided the acute periods for evaluation because of the abundance of necrosis and apoptosis,(20,44) and thus used the subacute phase when the majority of the tubular cells become resistant to oxidative stress with rare cell death present (Fig. 1b–d), though mutation spectrum might be slightly different between the acute and subacute phases. The accumulation of two different kinds of oxidative DNA base modifications, 8-OHdG and acrolein-dA, was most evident with immunohistochemistry after the first week of repeated administration of Fe-NTA and gradually decreased thereafter (Fig. 1e–l). This is probably due to the activation in the cellular metabolic pathways for those either suppressing the Fenton reaction or promoting DNA repair mechanisms. It is also possible that cellular selective processes worked to remove heavily damaged cells.

Gpt delta transgenic mice are an established model for analyzing mutations in vivo, and have been used to analyze several possible mutagens.(31) Here we have used a technique designated as DnaIP to evaluate the relative abundance of the two DNA base modifications at the gpt loci. Approximately 80 copies of the transgenes are included per haploid genome in the gpt delta transgenic mice.(30) Among genomic loci examined, including β-actin and three extrageneic regions, the gpt loci showed the highest level of 8-OHdG and acrolein-dA after one week of repeated Fe-NTA administration (Fig. 2). This consistency with the immunohistochemical data demonstrates that the transgenic gpt loci are indeed vulnerable and suitable for mutational analyzes. We believe that the high copy number of the gpt gene contained in these mice is at least partially responsible for this reliable sensitivity. In contrast to the findings at one week, certain extrageneic loci showed significantly higher levels of DNA base modifications than the gpt gene locus at other time points, suggesting that further studies would be necessary to elucidate the principles governing the distribution of oxidative DNA base modifications over the whole genome.(35,45)

Mutation frequencies both for the 6-TG selection and Spi selection also were the highest at the first week of repeated Fe-NTA administration (Fig. 3). This confirms the usefulness of 8-OHdG and acrolein-dA, which were detected both by immunohistochemistry and DnaIP, as reliable markers of mutation. In 22.7% of the Spi plaques after Fe-NTA treatment, large-size deletions (>1 kb) were observed and most of them were class I-A mutants (Fig. 5). This preference for large-size deletions with short homologous sequences at the junctions might be a prominent feature of the results obtained in this renal carcinogenesis model in that the patterns of Spi mutations are similar to that of the untreated colon.(31) With γ-rays, shorter deletions than 1 kb are prominent; with UVB and MMC, large-size deletions with or without short homologous sequences at the junctions are more frequently observed (>40%); whereas with PhIP and APNH, large-size deletions were rare.(31)

In the Fe-NTA-induced renal cell carcinoma of rats, homozygous deletion of the p16INK4A tumor suppressor gene was frequently observed,(21) and the allelic loss of this locus was observed at a high frequency one to three weeks after the repeated administration of Fe-NTA in rats.(22) We believe that the iron-mediated Fenton reaction is mainly responsible for this characteristic deletion. Short deletions were also increased after Fe-NTA administration (Table 1). Probably, the free radical reaction associated with iron is distinct from the reactions associated with other agents studied so far in the gpt lambda transgenic mice in that this is a universal reaction, though exaggerated through iron overload, involving the generation of hydroxyl radical and lipid peroxidation products. This kind of reaction is always taking place in the body under conditions of normal metabolism associated with oxygen consumption and, though it results in only minor consequences under physiological conditions, can be a driving force of carcinogenesis.

The mutation spectrum detected in the gpt gene was also quite distinctive. G:C pairs were the preferred bases for mutation, and especially G:C to C:G transversion-type mutation was characteristic (Fig. 4 and Table 1). This type of mutation was observed in PhIP and MMC as a minor type, but has not been reported as a major type of mutation (Table 2). We observed a low incidence of G:C to T:A transversion-type mutation that results from 8-OHdG formation.(46,47) This may be explained by the fact that this mutagenic process is strongly inhibited by a DNA repair enzyme, Mutyh.(48) Here we may propose a mechanism in which certain oxidative modification to guanine/cytosine may cause abnormal pairing with the same corresponding base. Recently, it was reported that formamidopyrimidine (FaPy)-guanine, another oxidative DNA base modification,(49,50) would not be responsible for this type of mutation.(51) We suspect that 5,6-dihydroxyuracil and 5-hydroxycytosine which are increased in this model(17) or other aldehyde-modified bases than acrolein-dA are among the possible candidates.

Table 2. Comparison of mutations induced by various mutagens in gpt delta transgenic mice
TargetKidneyBone marrow(38)Colon(39)
Control (MF = 1.0)Fe-NTA-1 W (MF = 2.4)Fe-NTA-2 W (MF = 1.4)Fe-NTA-3 W (MF = 1.1)Control (MF = 1.0)PhIP (MF = 10.9)Control (MF = 1.0)ENU (MF = 3.2)
G:C–A:T40%28.6% (1.72) 8.7% (0.30)16.7% (0.46)43.1%14.1%26.9%28.3%
A:T–G:C10% 7.1% (1.70)21.7% (3.04) 5.6% (0.62)11.1% 0.0% 3.8%20.0%
G:C–T:A10%10.7% (2.57) 8.7% (1.21)   0% (0.00)26.4%52.5%11.5%15.0%
G:C–C:G10%32.1% (7.70) 4.3% (0.60)22.2% (2.44) 0.0%13.1% 7.7% 0.0%
A:T–T:A 0% 3.6% (∞) 4.3% (∞)11.1% (∞) 5.6% 1.0% 3.8%28.3%
A:T–C:G10% 7.1% (1.70)17.4% (2.44)11.1% (0.62) 4.2% 0.0% 0.0% 5.0%
Deletion10% 7.1% (1.70)26.1% (3.65)22.2% (2.44) 4.2%15.1%38.5% 3.3%
Insertion10% 3.6% (0.86) 4.3% (0.60)11.1% (1.22) 5.6% 1.0% 7.7% 0.0%
Others 0% 0.0% (NA) 4.3% (∞) 0.0% (NA) 0.0% 3.0% 0.0% 0.0%
TargetBone marrow(41)Liver(53)Liver(54)Epidermis(55)Liver(56)
Control (MF = 1.0)MCC (MF = 2.9)Control (MF = 1.0)APNH (MF = 10.3)Control (MF = 1.0)γ-ray (MF = 3.2)Control (MF = 1.0)UVB (MF = 7.7)Control (MF = 1.0)MelQx (MF = 8.6)
  • present study. The number in parenthesis is the relative mutation frequency in comparison to the untreated control. MF, mutation frequency; NA, not applied; W, week(s); Fe-NTA, ferric nitrilotriacetate; PhIP, 2-amino-1-methyl-6-phenylimidazo [4,5-b]pyridine; ENU, ethylnitrosourea; MMC, mitomycin C; APNH, aminophenylnorharman; UVB, ultraviolet B; MelQx, 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline.

A:T–G:C 3.4% 6.7%10% 1%15% 0% 0% 3% 8% 0%
G:C–T:A31.0%26.7%14%51%12%25% 9% 0%10%54%
G:C–C:G10.3% 6.7% 2% 1% 4% 0% 0% 1% 4% 5%
A:T–T:A 6.9% 3.3% 8% 0% 4% 0% 9% 4% 8% 3%
A:T–C:G10.3% 3.3% 4% 0%23%10%10% 0% 2% 0%
Deletion13.8% 6.7%18%16%12%35%18% 0%12%16%
Insertion 0.0% 0.0% 2% 0% 4%10% 0% 0% 2% 0%
Others 0.0%33.3% 2% 7% 0% 0% 0% 5%11% 6%

When we reviewed the spectrum of point mutation observed in the p53, tsc2, p15, p16 and tbp-2 tumor suppressor genes of Fe-NTA-induced rat renal carcinoma, we observed no G:C to C:G transversions, but G:C to T:A (p53, tbp-2),(16,24) T:A to C:G (tsc2), G:C to A:T (p15 and p16, tbp-2),(21) A:T to T:A (tbp-2)(24) and one nucleotide insertion/deletion at repeat sequences (p16, tbp-2)(21,24) were observed despite the limited available data. There are at least several possibilities to explain this: (i) we have not yet identified the target genes with G:C to C:G mutations; (ii) there are some species-differences between mice and rats; (iii) this mutation spectrum detected in this gpt transgenic system is largely reflected in non-geneic genome areas; and (iv) the last possibilities are that G:C to C:G mutations are preferentially repaired by mismatch repair enzyme(s) in non-transgene areas or abundant mutations of this kind lead to lethal effects, affecting fundamental transcriptional activity in the expressed genes. Regarding species differences, another study using the gpt delta transgenic rat(52) would answer the question. The data obtained with DnaIP is of note in that 8-OHdG and acrolein-dA were increased in some non-geneic regions after three weeks of repeated administration of Fe-NTA, warranting further studies.

In conclusion, we used the gpt delta transgenic mice to evaluate the mutation spectrum of the Fenton reaction-based oxidative renal tubular injury, and found that the major mutations consist of large-size deletions with short homologous sequences at the junctions and transversion-type point mutations at G:C base pairs. The mutant frequency was the highest at the first week of repeated Fe-NTA administration, as shown by the immunohistochemistry of 8-OHdG and acrolein-dA as well as the presence of these two modified bases at the gpt loci, indicating that this early stage is one of the critical periods in this Fenton reaction-induced carcinogenesis.


This work was supported in part by a Grant-in-Aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan, a Grant-in-Aid for Cancer Research from the Ministry of Health, Labour and Welfare of Japan, and a grant of Long-range Research Initiative (LRI) by Japan Chemical Industry Association (JCIA).