Formaldehyde (Methanal, CAS No.: 50-00-0) is abundantly produced in the chemical industry, and its annual global production is about 20 million tons (IARC, 2012). It is used in industrial and consumer products, and is also generated as a byproduct of biomass and petroleum burning, cigarette smoke and automobile exhaust. Cigarette smoke contains 12–106 μg of formaldehyde per cigarette (IARC, 2012). In medical and biological fields, it has been widely used as a fixative or preservative for pathological specimens. It also occurs as a natural product in most organisms and in the environment. There are several metabolic pathways in humans leading to formaldehyde synthesis, in particular, amino acid and methanol metabolisms, lipid peroxidation, and P450-dependent demethylation (Swenberg et al., 2013).

The major sources of formaldehyde exposure can be roughly classified into three categories: fumes generated from industry, those released in residential housing and those in various occupational indoor settings (Ma and Harris, 1988). Non-occupational exposure in the household is mainly due to the fumes released from formaldehyde-based resin for binding of plywood, particle board, paint, and textile fibers. Formaldehyde fumes have caused health problems across the country. People inhaling formaldehyde fumes indoors often show symptoms such as headache, dizziness, wheezing, runny nose, and nausea. In severe cases, they have sore throat, burning eyes, and skin irritation. These symptoms are called “sick house syndrome” or “sick building syndrome” (Norbäck, 2009). The levels of formaldehyde in the air inside houses are typically 0.02–0.06 mg/m³, but these levels have been declining since the late 1980s as a result of standards for building materials (IARC, 2012). The World Health Organization now recommends a formaldehyde level of 0.1 mg/m³ in indoor air (Kaden et al., 2010).

Formaldehyde also has activities that cause chronic adverse effects, such as genotoxicity, carcinogenicity, and teratogenicity. The health effects have been extensively studied for a long time and reviewed in many articles (Ma and Harris, 1988; Conaway et al., 1996; Swenberg et al., 2013). In this review, we focused on molecular basis of DNA damage and mutations induced by formaldehyde exposure. In this sense, our original data on the mutation spectrum of formaldehyde in human cells, have been clearly showed further details concerning the mechanism of formaldehyde carcinogenesis.

In animal experiments using mice and rats, chronic inhalation of formaldehyde at high concentrations caused cancers, particularly, squamous cell carcinomas of the nasal cavities (IARC, 2006). In other studies, the incidences of lymphoma and leukemia were also increased. In animals administered formaldehyde in drinking water, increased incidences of forestomach papillomas, leukemias, and gastrointestinal tract tumors were reported, but their frequencies varied. These studies have been summarized in IARC Monograph (IARC, 2006).

In the human body, many endogenous and exogenous compounds release formaldehyde through catabolic reactions. The primary metabolic system for formaldehyde involves an initial spontaneous reaction with glutathione to form S-hydroxymethylglutathione, followed by reaction facilitated by alcohol dehydrogenase-3 to convert the intermediate to S-formylglutathione. This intermediate is then further metabolized by S-formylglutathione hydrolase to yield nontoxic formate, which is excreted in the urine (Hedberg et al., 2002). There have been numerous cohort and case-control studies on formaldehyde exposure and an increase in nasopharynx cancer or leukemia in humans (IARC, 2012). The IARC (2006) judged that there was sufficient evidence for the carcinogenicity of formaldehyde, based primarily on its association with nasopharyngeal cancer, and that there was strong but not sufficient evidence for the leukemogenic effects of formaldehyde. Owing to advancements via recent epidemiological studies (Coggon et al., 2003; Hauptmann et al., 2004, 2009; Pinkerton et al., 2004; Marsh and Youk, 2005; Marsh et al., 2007a,b; Beane Freeman et al., 2009) and meta-analyses (Bosetti et al., 2008; Zhang et al., 2009; Bachand et al., 2010), the IARC (2012) concluded that occupational exposure to formaldehyde causes nasopharyngeal cancer and leukemia, and classified formaldehyde as a human carcinogen Group 1.

The genotoxicity of formaldehyde has been examined by various methods. In most reports using Salmonella typhimurium strains TA98, TA100, TA1535, TA1537, and TA1538 (Ames test), formaldehyde induced both base change and frame-shift mutations without metabolic activation, but its mutagenicity is not as strong as that of typical environmental mutagens such as 4-nitroquinoline-1-oxide and N-methyl-2-nitro-N'-nitrosoguanidine (Ma and Harris, 1988; Conaway et al., 1996). Formaldehyde also shows mutagenicity in assays using Escherichia coli WP2 (Takahashi et al., 1985; O'Donovan and Mee, 1993).

Formaldehyde induced chromosome aberrations (CA) and sister chromatid exchanges (SCE) in cultured Chinese hamster (Natarajan et al., 1983; Basler et al., 1985; Galloway et al., 1985; Merk and Speit, 1998, 1999) and human lymphocyte cells (Kreiger and Garry, 1983). In rats treated with formaldehyde through inhalation (0.5–12 ppm), no increase in the frequencies of CA and SCE was observed in lymphocytes (Kligerman et al., 1984). In rats inhaling 15 ppm formaldehyde, CA was induced in pulmonary lavage cells but not in bone marrow cells (Dallas et al., 1992). In mice treated with formaldehyde by oral administration, no CA or micronucleus (MN) formation was detected in bone marrow and spleen cells (Gocke et al., 1981; Natarajan et al., 1983). In rats administered a single dose of 200 mg/kg formaldehyde by gavage, gastrointestinal organs exhibited local irritation and their cells formed MN (Migliore et al., 1989). In people occupationally exposed to formaldehyde, increases in the frequencies of MN and SCE in the peripheral lymphocytes were reported (Costa et al., 2008), but no statistically significant increases were also reported elsewhere (Ye et al., 2005; Pala et al., 2008).

Formaldehyde induces mutations in human cultured lymphoblastoid TK6 cells and Chinese hamster ovary (CHO) cells, which were detected by resistance to toxic nucleoside analogs, trifluorothymidine (Goldmacher and Thilly, 1983), 8-azaguanine and 6-thioguanine (Liber et al., 1989; Graves et al., 1996; Speit and Merk, 2002).

Formaldehyde reacts immediately with primary and secondary amines, thiols, hydroxyls, and amides to form methylol derivatives. It acts as an electrophile and can react with macromolecules such as DNA, RNA, and protein to form adducts and cross-links. Guanine, adenine and cytosine have amino group in their molecules. When deoxyadenosine, deoxyguanosine, deoxy-cytidine, and thymidine were reacted with formaldehyde, the major products determined by high-performance liquid chromatography and electrospray mass spectrometry (LC-MS/MS) were N⁶-hydroxymethyldeoxyadenosine, N⁴-hydroxymethyl-deoxycytidine and N²-hydroxymethyldeoxyguanosine (Beland et al., 1984; Cheng et al., 2003). In human and Chinese hamster cells treated with formaldehyde, N⁶-hydroxymethyldeoxy-adenosine and N²-hydroxymethyldeoxyguanosine were detected in their genome (Beland et al., 1984; Zhong and Que Hee, 2004). N⁶-hydroxymethyldeoxyadenosine was detected in leukocyte DNA from 29 of 32 heavy smokers (Wang et al., 2009). The reaction of formaldehyde with deoxyadenosine and deoxyguanosine or DNA also produces three major methylene crosslinks: di-(N⁶-deoxyadenosyl)methane, (N⁶-deoxyadenosyl-N²-deoxyguanosyl)methane and di-(N²-deoxyguanosyl)methane (Huang et al., 1992; Huang and Hopkins, 1993; Cheng et al., 2003). In a study of the in vitro reactions between formaldehyde and all amino acids, deoxyguanosine crosslinks with cysteine were abundant and stable, while those with lysine were readily formed but rapidly disintegrated (Lu et al., 2010). There are many reports showing that formaldehyde-protein crosslinks were detected in cells treated with formaldehyde (IARC, 2012), but their molecular structures have not been fully clarified. The DNA-protein crosslinks with protein sizes less than 12–14 kDa can be removed by the nucleotide excision repair pathway, whereas oversized DNA-protein crosslinks are processed exclusively by RecBCD-dependent homologous recombination in bacteria (Nakano et al., 2007). In mammalian cells, the upper size limit of crosslinked proteins amendable to the nucleotide excision repair is relatively small (8–10 kDa), therefore, homologous recombination would be the major pathway in tolerance of chromosomal DNA-protein crosslinks (Nakano et al., 2009).

We attempted to determine the likelihood of formaldehyde-induced intra-strand crosslink formation between four kinds of nucleotide in DNA using the unique method developed by Kawanishi (1995) and Matsuda et al. (1998). Single-strand DNA (175 base pairs) and two 20-mer complementary oligonucleotides: one a 5′ [³²P]-labeled sequence and the other an unlabeled sequence following the [³²P]-labeled sequence, were annealed as shown in Figure 1A. If the oligonucleotides are bound between the 3′-end base of the [³²P]-labeled oligonucleotide and the adjacent 5′-end base of the unlabeled oligonucleotide by a formaldehyde crosslink, a 40-mer single-strand oligonucleotide band should appear on the denatured polyacrylamide gel. As shown in Figures 1B,C, formaldehyde formed crosslinks between deoxyguanosines (5′-GG) of the ends of the oligonucleotides. Fewer crosslinks were formed between deoxyguanosine and deoxyadenosine (5′-GA), but no crosslinks were formed between deoxycytidines (5′-CC). Figures 1B,C also shows formaldehyde-produced inter-strand crosslinks. The repair mechanism of the formaldehyde-induced DNA inter-strand crosslinks has not been reported so far, but the inter-strand crosslinks might be removed by the Fanconi anemia (FA) pathway and homologous recombination, which is shown in the studies of mitomycin C-induced inter-strand crosslinks (Kottemann and Smogorzewska, 2013; Unno et al., 2014).

Human and Chinese hamster mutant cells deficient in XPF or ERCC1 are hypersensitive to formaldehyde (Kumari et al., 2012), and E. coli uvrA mutant is more sensitive to formaldehyde than wild-type E. coli (Takahashi et al., 1985). These lines of evidence indicate that intra-strand crosslinks constitute a type of damage that could be fatal to cell survival and are removed by the nucleotide excision repair pathway; whereas mono-adducts, hydroxymethyldeoxyadenosines and hydroxymethyldeoxyguanosines, would be removed by the base excision repair pathway. Types of DNA damage induced by formaldehyde and their repair pathways are summarized in Figure 2.

In E. coli, mutations induced by formaldehyde in the xanthine guanine phosphoribosyl transferase (gpt) gene were characterized (Crosby et al., 1988). Exposure of E. coli to 4 mM formaldehyde induced insertions (41%), deletions (18%), and point mutations (41%). DNA sequencing revealed that most of the point mutations were transversions at the G:C base pair in the gene. In contrast, exposure of E. coli to 40 mM formaldehyde produced point mutations (92%), 62% of which were transitions at the single A:T base-pair in the gene, indicating that there were concentration-dependent variations in mutation spectra. When naked pSV2gpt plasmid was exposed to 3.3 or 10 mM formaldehyde and transfected into E. coli, most mutations were frameshifts, which also suggests the presence of various mechanisms in formaldehyde-induced mutation.

In human lymphoblastoid TK6 cells treated with 150 μM formaldehyde, 30 hprt mutant colonies were characterized by Southern blot analysis (Liber et al., 1989). Fourteen (47%) of these mutants had large deletions, while the other 16 mutants (53%) showed normal restriction patterns. Of these, 12 mutants produced hprt mRNA and 3 mutants produced no detectable mRNA. Sequence analyses of the mRNA showed that 6 mutants had base substitutions occurring at the A:T base pair. Four mutants had A:T to C:G transversions at a specific site that appears to be a mutation hot spot. The remaining 2 mutants lost exon 8 of the hprt gene.

CHO cells treated with 1 mM formaldehyde increased the hprt mutation frequency to 4.7-fold higher than the spontaneous mutation frequency, and mutations of 6 mutant clones were identified. All mutations were single base transversions with 3 A:T to T:A, 2 A:T to C:G, and 1 G:C to T:A (Graves et al., 1996).

Mutations in the p53 gene from rat primary nasal squamous cell carcinomas induced by formaldehyde inhalation (15 ppm up to 2 years) were analyzed (Recio et al., 1992). Five tumors had point mutations, namely, 3 G:C to T:A transversions, 1 G:C to A:T transition, and 1 G:C to C:G transversion.

Since no previous reports had described that formaldehyde induces tandem base substitutions that would arise from intra-strand crosslinks, we adopted the shuttle vector plasmid pMY189 (Matsuda et al., 1995; Yagi, 2013) that can be treated with a high concentration of formaldehyde. The plasmid pMY189 can replicate in both human and E. coli cells, and has an ampicillin-resistant marker gene and a mutation marker gene supF that can be selected as colored colonies on bacterial agar plates. The result was published by Kawanishi (1995), and we describe its outline below. The formaldehyde-treated pMY189 propagated in human fibroblast cells caused reduction of the number of ampicillin-resistant bacterial colonies and enhanced the mutation frequency of the supF gene (Figure 3). The mutation frequency of the plasmid was increased 4- to 50-fold following treatment of the plasmid with 70–270 mM formaldehyde. Nucleotide sequences of the supF mutation marker gene of 94 plasmids were determined. Mutated plasmids induced by 200–270 mM formaldehyde were classified and are shown in the third column of Table 1. About half (49%) of the plasmids had tandem base substitutions. Among the mutant plasmids with single base substitutions, 89% were substitutions of G:C base pairs (Table 1). The most common type of single base substitution mutation was G:C to T:A transversions (52%). Types of tandem base substitution are shown in Figure 4A. Thirty-nine percent were 5′-GG to 5′-TT (5′-CC to 5′-AA), and 15% were 5′-GA to 5′-AT (5′-TC to 5′-AT), and 5′-GA to 5′-TT (5′-TC to 5′-TT) substitutions. About half (47%) of them occurred at 5′-GG (5′-CC) sequences and 44% occurred at 5′-GA (5′-TC) sequences. The distribution of the formaldehyde-induced base substitutions in the supF gene is shown in Figure 4B. Underlined letters indicate that these base changes were found in the same mutated plasmid. The “d” or “del” means a deletion. The tandem base substitutions were not randomly distributed in the supF gene, but appeared preferentially at certain sites. About half of the tandem base substitutions occurred at 4 5′-GG (5′-CC) sequences (at base pairs 102–103, 103–104, 109–110, and 159–160) and 2 5′-GA (5′-TC) sequences (107–108 and 171–172).

Mutagenesis of various aldehydes such as acetaldehyde, 2-chloroacetaldehyde, crotonaldehyde, acrolein, glyoxal, and methylglyoxal has been examined using supF shuttle vector plasmids (Matsuda et al., 1995, 1998; Murata-Kamiya et al., 1997, 2000; Kawanishi et al., 1998a,b). These aldehydes react with DNA and form exocyclic guanines, ethenoguanine, and propanoguanine, as major adducts (Minko et al., 2009; Voulgaridou et al., 2011). These DNA lesions generate mutations mainly at G:C base pairs, but the spectra of base substitutions differ among aldehydes with different chemical structures. Table 1 summarizes mutation spectra of various aldehydes, which are induced in the shuttle vector plasmids propagated in mammalian cells. All aldehydes induced base substitutions manly at G:C base pairs. It should be noted that acetaldehyde induced tandem base substitutions at 5′-GG sequences like formaldehyde, suggesting that acetaldehyde forms intra-strand crosslinks between adjacent guanines, probably through propanoguanine formation (Matsuda et al., 1998). Glyoxal is an endogenous mutagen produced abundantly in the process of DNA oxidation by reactive oxygen free radicals that have been thought to contribute to the pathogenesis of many age-related human diseases including cancer. Features of glyoxal-induced mutations are similar to those of spontaneous mutations (Table 1), which may suggest that glyoxal is a cause of spontaneous mutations.

Formaldehyde is one of the most abundantly exposed environmental mutagens for humans, and shown to be carcinogenic in experimental animals. Formaldehyde causes MN, SCE, and CA in mammalian cells and induces gene mutations in mammalian and bacterial cells. Formaldehyde acts as an electrophile to react mostly with guanine and adenine in DNA, and forms various kinds of DNA lesions. The DNA lesions are classified into DNA adducts, DNA intra-strand crosslinks, DNA inter-strand crosslinks and DNA-protein crosslinks. Organisms have mechanisms to remove these DNA lesions, which are specific to each type of DNA lesions. If these DNA lesions are not repaired, mutations are induced during DNA replication. Formaldehyde-induced mutations in human cells were analyzed using shuttle vector plasmids (Kawanishi, 1995). Interestingly, about half of the formaldehyde-induced mutations were tandem base substitutions, and most of them occurred at 5′-GG (or 5′-CC) and 5′-GA (or 5′-TC) sequences. These mutation features agree with the predominance of formaldehyde-induced 5′-GG and 5′-GA crosslinks, indicating that the intra-strand crosslinks lead to tandem base substitutions. The prevalence of N²-hydroxymethyl adduct to guanine would lead to the G to T transversion that is the major single base substitution mutation. In early studies using bacteria, cultured cells and animals, exposure concentrations of formaldehyde would not have been sufficiently high to manifest the features of the mutation spectrum because formaldehyde mutagenicity is relatively weak and its acute cytotoxic effect is strong compared with those of other environmental mutagens. Our findings as well as the accumulation of molecular data provide clues to promote our understanding of the mechanisms behind the genotoxicity and carcinogenicity of formaldehyde.