Measurement of Endogenous versus Exogenous Formaldehyde–Induced DNA–Protein Crosslinks in Animal Tissues by Stable Isotope Labeling and Ultrasensitive Mass Spectrometry

DNA damage is a major culprit in many diseases, including cancer and aging. Toxic chemicals originating from a wide range of endogenous and exogenous sources continuously damage DNA (1). It is well known that our DNA is not pristine, and endogenous DNA damage occurs at a high frequency compared with exogenous damage, having more than 40,000 lesions in every cell in our body (2). The “endogenous exposome” was first explored by our laboratory from the concept of the “exposome” that emphasized the importance of understanding relationships between human disease and lifetime exposures to both environmental and internal chemicals (2). Recent advances have identified endogenous aldehydes as possible genotoxic agents that cause severe bone marrow failure (BMF) and leukemia in mice with mutations in both Fanconi anemia group D2 (fancd2) and aldehyde dehydrogenase 2 (Aldh2; ref. 3). Notably, Aldh2 is mutated in approximately 1 billion people, most frequently observed in Southeast Asians due to inherited genetic mutations (4). Deficiency of Aldh2 has been demonstrated to dramatically accelerate BMF in Japanese Fanconi anemia patients (5). In addition, endogenous formaldehyde has been shown to have greater genotoxicity than acetaldehyde (6–8). These findings raise new challenges for understanding the cellular environments involved in disease causation. Linear extrapolation of risk down to zero remains a common approach used by regulators, despite the fact that such models use no biology and that when biology exists, linear extrapolation will overestimate risks (9). Thus, better understanding of endogenous and exogenous DNA–protein crosslinks (DPC) data is important for advancing science-based risk assessments.

DPCs strongly disrupt normal DNA–protein interactions and interfere with DNA replication, transcription, and repair, which ultimately threatens genomic integrity and cell viability (10). DPCs can originate from exposure to environmental agents, such as ionizing radiation, UV light, formaldehyde, and transition metal ions (11). Furthermore, anticancer drugs have been developed on the basis of their ability to form DPCs, such as nitrogen mustards, platinum-containing agents (i.e., cisplatin), and 5-aza-2′-deoxycytidine. Notably, DPCs can also arise upon exposure to endogenous carcinogens, such as reactive aldehydes that are produced during various cellular processes (2, 12). Because of their bulky nature, DPCs impact practically all aspects of genomic activity and may therefore result in genomic instability or cell death. Thus, DPC formation and repair are crucial for genomic integrity in humans and consequently for understanding and preventing carcinogenesis (1, 10, 13–15). Surprisingly, however, DPC repair mechanisms have not been fully elucidated (15).

Until now, accurate and distinct measurements of both endogenous and exogenous DPCs have not been studied, which significantly limits our understanding of the biologic effects and repair of DPCs. Accurate measurement of DPCs is complicated by two critical issues. First, covalent DPCs must be completely isolated from non-covalent DNA–protein complexes, as the latter are present in a clear excess over the former throughout the genome (16). Second, DPCs need to be measured with structural specificity, given that there are numerous lesions that complicate the background of DPCs resulting from both endogenous and exogenous electrophiles. However, such stringent measurements of DPCs have not been possible using previously available DPC detection techniques, such as SDS/KCl precipitation, phenol–chloroform extraction, nitrocellulose filter binding, comet assay, alkaline elution, and CsCl density gradient centrifugation, all of which are nonselective (16, 17). Recently, mass spectrometry (MS)–based methods have emerged as powerful tools for chemical analysis of digested DPCs (11). Elegant work using MS has comprehensively characterized and ultrasensitively quantified the DPCs induced by bifunctional electrophiles, such as nitrogen mustards (18, 19), cisplatin, and diepoxybutane (20).

Formaldehyde is classified as an animal and human carcinogen. Humans are exposed to formaldehyde originating from environmental sources due to its broad range of applications and formation from various chemical processes, as well as internal sources due to its role as an essential intermediate of various cellular processes (21). As a strong electrophile, formaldehyde is a well-known crosslinking agent that is cytotoxic, resulting in increased cell proliferation, mutagenesis, and nasal carcinomas in rats exposed by inhalation (22). Formaldehyde has been widely used to elucidate the cellular pathways of DPC repair (8, 10, 23, 24). Furthermore, endogenous formaldehyde is an important source of DNA damage (6, 8). It thus poses unique challenges for understanding the risks associated with exposure. One major question that must be addressed is whether inhaled formaldehyde can increase the risk for leukemia. A critical dilemma has been to determine whether exogenous formaldehyde reaches bone marrow. To this end, we focused on the tissue distribution of endogenous and exogenous formaldehyde–induced DPC formation in rats and nonhuman primates (NHP) at exposure concentrations that had been studied in earlier DPC experiments, as well as in carcinogenicity studies in rats (25–27). For the first time, the incorporation of stable isotope–labeled formaldehyde ([¹³CD₂]-formaldehyde) in animal exposures permitted readily distinguishable measurements of endogenous and exogenous DPCs based on their mass differences through the use of ultrasensitive and selective LC/MS. These measurements of endogenous and exogenous DPCs provide critical new data that advance science-based risk assessment, as well as providing improved methods for understanding pathways for DPC repair, such as the Fanconi anemia pathways, that may lead to better disease treatment.

2′-Deoxyguanosine, reduced l-glutathione (GSH), sodium phosphate monobasic (BioXtra), sodium phosphate dibasic (BioXtra), magnesium chloride solution (MgCl₂; 1 mol/L), calcium chloride solution (CaCl₂; 1 mol/L), sucrose, ammonium acetate, acetic acid, formic acid, formaldehyde–2,4-dinitrophenylhydrazine (DNPH) standards, sodium cyanoborohydride (NaCNBH₃), pronase, leucine aminopeptidase M, carboxypeptidase Y, prolidase, alkaline phosphatase, and phosphodiesterase were all purchased from Sigma. DNAzol and Turbo DNase were obtained from Life Technologies. Methanol, acetonitrile, high-performance liquid chromatography (HPLC) grade water, and formaldehyde solution (37%, w/w) were all purchased from Thermo Fisher Scientific. Formaldehyde solution was used to synthesize the dG-Me-Cys standard. [¹⁵N₅, ¹³C₁₀]-deoxyguanosine, [¹⁵N₅]-deoxyguanosine, [¹³CD₂]-paraformaldehyde (≥98% purity), and [¹³CD₂]-formaldehyde solution (20% w/w in D₂O) were ordered from Cambridge Isotope Laboratories, Inc. [¹⁵N₅]-deoxyguanosine and [¹³CD₂]-formaldehyde solution were used to synthesize the [¹⁵N₅]-dG-[¹³CD₂]-Me-Cys (internal standard). DNPH cartridges were ordered from Waters. Proteinase K (20 mg/mL) was obtained from 5 PRIME, Inc. Amicon Ultra Centrifugal Filters (0.5 mL, 3K) were purchased from EMD Millipore. Nanosep Centrifugal Devices (MWCO 3K) were purchased from Pall Life Sciences.

A synthesized standard, dG-Me-Cys, was prepared by digestion of dG-Me-GSH according to our previous study (28), using the experimental details described in the Supplementary Information. As demonstrated in a previous study, dG-Me-Cys showed a very close UV absorbance spectrum to that of dG (28). Thus, quantification of dG-Me-Cys was based on a dG standard calibration curve created using a HPLC–UV method with a detection wavelength at 254 nm. For the preparation of internal standard, [¹⁵N₅]-deoxyguanosine and [¹³CD₂]-formaldehyde were used to synthesize [¹⁵N₅]-dG-[¹³CD₂]-Me-GSH. Specifically, GSH (100 mmol/L) was incubated with [¹³CD₂]-formaldehyde (60 mmol/L) in 1.0 mL sodium phosphate buffer (100 mmol/L, pH 7.2) at 37°C for 3 hours. [¹⁵N₅]-dG was then added at a final concentration of 5 mmol/L, and the crosslinking reaction was performed at 37°C for 14 hours. The same methods described in our previous study were applied for the purification and digestion of [¹⁵N₅]-dG-[¹³CD₂]-Me-Glutathione (28). The isolated [¹⁵N₅]-dG-[¹³CD₂]-Me-GSH was digested in 0.4 mL sodium phosphate buffer (40 mmol/L, pH 6.0) by carboxypeptidase Y (50 μg/mL) and leucine aminopeptidase M (250 μg/mL) in the presence of MgCl₂ (10 mmol/L) and CaCl₂ (10 mmol/L) at room temperature for 15 hours. The internal standard, [¹⁵N₅]-dG-[¹³CD₂]-Me-Cys, was purified and quantified by the same method as described in the Supplementary Information.

Animal exposure was described in the Supplementary Information. DPCs were isolated from animal tissue samples using DNAzol from Life Technologies following the manufacturer's instructions with some modifications. First, animal tissues (30–50 mg) were homogenized in 1 mL of sucrose buffer (20 mmol/L sodium phosphate; 250 mmol/L sucrose, pH 7.2). The nuclei were isolated from homogenate solution by centrifugation at 1,500 × g and 4°C for 10 minutes. The nuclear pellets were washed with 1 mL of 20 mmol/L sodium phosphate buffer (pH 7.2). The washed nuclear pellets were dissolved in 50 μL of water, followed by the addition of 800 μL of DNAzol and 20 μL of proteinase K (20 mg/mL). After protein digestion at room temperature for 15 hours, DPCs were precipitated by adding 600 μL of 100% ethanol. The precipitated DPCs were washed with 800 μL of 75% ethanol 4 times. The isolated DPCs were dissolved in 450 μL digestion buffer (20 mmol/L ammonium acetate, pH 6.0) containing MgCl₂ (5 mmol/L), CaCl₂ (5 mmol/L), DNase I (40 U/mL), alkaline phosphatase (4 U/mL), phosphodiesterase I (0.001 U/mL), and pronase (0.5 mg/mL). DNA digestion was performed at room temperature for 20 hours, resulting in the formation of nucleoside–peptide crosslinks. The enzymes and undigested DNA were removed using a Nanosep Centrifugal Device (MWCO, 3K) at 10,000 × g and 4°C for 40 minutes. The resultant filtrate was combined with 30 μL of an enzyme mixture containing carboxypeptidase Y (0.3 mg/mL), aminopeptidase M (0.6 mg/mL), and prolidase (0.3 mg/mL) to further digest peptides to amino acids. Following a 20-hour incubation at room temperature, the reaction was terminated with the addition of 4 μL of 30% acetic acid. After adding 8 fmol of internal standard (4 μL, 2 nmol/L), the final reaction mixtures were subjected to centrifugation at 10,000 × g and 4°C for 40 minutes in a Nanosep Centrifugal Device to remove enzymes prior to HPLC purification. To eliminate matrix interferences, digestion enzymes were washed before use as described in the Supplementary Information.

The target analyte, dG-Me-Cys, was purified from the reaction mixture on an Agilent 1200 Series UV HPLC System by using two C₁₈ reverse-phase columns connected in series (Waters Atlantis T3, 3 μm, 150 mm × 4.6 mm). Following digestion, there were two products detected by HPLC–UV, both of which had similar retention times to that of DPCs. The use of two HPLC columns connected in series was found to greatly increase the resolution of HPLC separation between the dG-Me-Cys and the other digestion products. This increased the detection sensitivity of the DPC using nano-LC/ESI/MS-MS. The detection wavelength and column temperature were set at 254 nm and 15°C, respectively. The mobile phases consisted of 0.05% acetic acid in water (A) and pure acetonitrile (B). The flow rate was 0.45 mL/minute, and elution gradient conditions were set as follows: 0 minute, 1% B; 3 minutes, 1% B; 42 minutes, 3% B; 86 minutes, 3.5% B; 86.5 minutes, 80% B; 95 minutes, 80% B; 95.5 minutes, 1% B; 110 minutes, 1% B. dG-Me-Cys was eluted at a retention time of 81.5 minutes. The fractions containing target compounds were combined and concentrated to approximately 10 to 20 μL using a vacuum concentrator. The amount of digested dG in each sample was quantitated by the UV peak area (λ = 254 nm) at the corresponding retention time using a calibration curve ranging from 4 to 80 nmole dG on column.

Nano-LC/ESI/MS-MS analysis of dG-Me-Cys was performed on a TSQ Quantum Ultra Triple-Stage Quadrupole Mass Spectrometer (Thermo Scientific) in positive-mode electrospray ionization. Selected reaction monitoring (SRM) mode was used to detect and quantify dG-Me-Cys. Sample introduction and separation was accomplished on a nanoACQUITY Ultra Performance Liquid Chromatography system from Waters Corporation. Following injection, the compounds were retained on two Symmetry C₁₈ trap columns (5 μm, 20 × 0.18 mm) connected in series, followed by sample separation on a HSS T3 analytical column (1.8 μm, 100 × 0.1 mm) from Waters Corporation at room temperature. Mobile phases consisted of water with 0.05% acetic acid (A) and acetonitrile (B). Analytes were first retained on two trap columns with a flow rate of 1.8 μL/minute of 1% mobile phase B for 3 minutes, followed by transfer to the analytical column. The flow rate was 0.4 μL/minute, and elution gradient conditions were set as follows: 0 minute, 1% B; 3 minutes, 1% B; 16 minutes, 50% B; 20 minutes, 1% B; 35 minutes, 1% B. Mass spectrometer conditions were set as the following: source voltage, 2,200 V; temperature of ion transfer tube, 280°C; skimmer offset, 0; scan speed, 75 ms; scan width, 0.7 m/z; Q1 and Q3 peak width, 0.7 m/z; collision energy, 31 eV; collision gas (argon), 1.5 arbitrary units. A sample volume of 5 μL was injected for analysis. Endogenous DPC (dG-Me-Cys) and exogenous DPC (dG-[¹³CD₂]-Me-Cys) were quantified by using the transition of m/z 401.1 to m/z 164.1 and m/z 404.1 to m/z 167.1, respectively. The transition of m/z 409.1 to m/z 172.1 was monitored for the internal standard ([¹⁵N₅]-dG-[¹³CD₂]-Me-Cys). Linear calibration curves were obtained using the ratio of integrated peak area of the analytic standard over that of the internal standard.

Standard curves were established by plotting the peak area ratios of solutions containing a fixed amount of internal standard ([¹⁵N₅]-dG-[¹³CD₂]-Me-Cys, 8.0 fmol) and increasing amounts of dG-Me-Cys from 0.15 to 6 fmol. The limit of quantification was determined as the amount of standard that produced a signal-to-noise ratio >10. To evaluate method accuracy and precision, analytic standard was spiked into the digested mixture of rat liver samples at three different levels (0.8, 2.0, and 5.0 fmol), followed by HPLC purification and concentration using a vacuum concentrator.

The digestion process of DPCs isolated from animal tissues has the potential to generate some chemically reactive species, including nucleotides containing a hydroxymethyl group, hydroxymethyl-dG, dG, and peptides containing cysteine. These products could potentially crosslink to form artifactual dG-Me-Cys during sample preparation. Control studies were performed to determine the potential for artifactual formation of dG-Me-Cys. The extent of possible artifacts was evaluated in two ways. The first technique involved spiking isotope–labeled [¹³C₁₀, ¹⁵N₅]-dG into the DPC solution to determine the potential for extrinsic formation of [¹³C₁₀, ¹⁵N₅]-dG-Me-Cys. Specifically, [¹³C₁₀, ¹⁵N₅]-deoxyguanosine (4 mmol/L) was treated with 600 μL of 0.5 mol/L NaCNBH₃ for 48 hours at 37°C. The treated [¹³C₁₀, ¹⁵N₅]-dG was spiked into an equal amount of dG produced from DNA digestion. This spiked sample was subjected to DNA and protein digestion, followed by HPLC purification and LC/ESI/MS-MS analysis as described above. The potential artifact, [¹³C₁₀, ¹⁵N₅]-dG-Me-Cys, was monitored using the corresponding transition of m/z 416.1 to m/z 174.1. The second technique to determine artifact formation employed NaCNBH₃ to reduce active hydroxymethyl groups in DNA and protein during the isolation of DPCs. Specifically, all solutions used in the isolation of DPCs, including homogenate buffer, wash solution, and DNazol, contained 20 mmol/L NaCNBH₃. For control samples, NaCNBH₃ was excluded in all solutions used for DPC isolation.

To study the chemical identity of formaldehyde-induced DPCs, we investigated the in vitro crosslinking reaction between nucleosides and amino acids in the presence of formaldehyde (29). Our results demonstrated that dG and cysteine crosslinked in the presence of formaldehyde to form dG-Me-Cys. We hypothesized that this crosslink would serve as a specific DPC biomarker of formaldehyde exposure based on its relatively high stability and abundance. To test this hypothesis, we examined formaldehyde-induced DPCs, specifically dG-Me-Cys, in both rats and NHPs that were exposed to [¹³CD₂]-formaldehyde using nose-only and whole-body inhalation chambers, respectively (Fig. 1A and Supplementary Figs. S2 and S3). The use of [¹³CD₂]-formaldehyde permitted the differential measurement of endogenous and exogenous dG-Me-Cys based on their mass difference (+3 m/z) by MS. A challenging step in structurally specific analysis of DPCs is the digestion or hydrolysis of bulky DPCs into small molecules suitable for MS-MS analysis. Many DPCs are unstable, and chemical hydrolysis is generally too harsh to preserve the DPC linkages for their chemical identities. To this end, we optimized a mixture of enzymes to digest large DPCs into small nucleoside–amino acid crosslinks (dG-Me-Cys) under mild conditions (pH 6.0 and room temperature), thereby enabling preservation of the DPC chemical identity before MS analysis (Fig. 1B and C). After complete digestion, formaldehyde-induced DPCs were isolated by offline HPLC fraction collection along with the quantification of digested dG using UV absorbance at 256 nm. Using authentic standards, the isolated endogenous and exogenous dG-Me-Cys were differentially quantified by MS (Fig. 1D).

Given the inherent advantage of improved sensitivity, we adopted a nanospray UPLC/ESI/MS-MS methodology to measure dG-Me-Cys. The validated method exhibits ultrasensitivity, as well as excellent accuracy and precision. Both the standard (dG-Me-Cys) and the internal standard were detected at the same retention time (Fig. 2). The limit of quantification was 37.5 amol on the column (S/N = 10). Good linearity was observed in the concentration range from 0.15 to 6.0 fmol with an R² value of 0.9993 (Supplementary Fig. S1). The accuracy of this method was greater than 93.4%, with precision being less than 11.2% RSD using three spiking concentrations: 0.8, 2.0, and 5 fmol (Supplementary Table S1).

The validated method was then utilized to identify endogenous and exogenous formaldehyde–specific DPCs in rat nasal tissue (respiratory epithelium) as shown in Fig. 3. Only the endogenous peak (dG-Me-Cys) corresponding to the specific transition of m/z 401.1 to m/z 164.1 was detected in control rat nasal tissue at the same retention time as the internal standard at 14.20 minutes (Fig. 3A). Both endogenous and exogenous crosslinks were detected in [¹³CD₂]-formaldehyde–exposed rat nasal tissue at the same chromatographic retention time (Fig. 3B). The exogenous DPC fragment ions retained the stable isotope label, demonstrating the predicted isotope mass shift (+3 m/z) and were clearly distinguished from fragment ions of endogenous origin.

The potential for artifactual dG-Me-Cys formation during sample processing was investigated in two ways. First, stable isotope–labeled [¹³C₁₀, ¹⁵N₅]-dG was spiked into the reaction mixture to investigate whether free dG from the DNA digestion could form the crosslink. The potential formation of artifact, [¹³C₁₀, ¹⁵N₅]-dG-Me-Cys, was monitored using the corresponding transition of m/z 416.1 to m/z 174.1. The results showed that endogenous dG-Me-Cys and internal standard were observed as shown in the top and middle panels of Fig. 4A, respectively. On the contrary, no peak corresponding to the artifactual formation of [¹³C₁₀, ¹⁵N₅]-dG-Me-Cys was found in the same sample (Fig. 4A, bottom). This evidence demonstrated that the digestion of dG does not lead to artifactual formation of dG-Me-Cys during sample preparation. The second technique employed NaCNBH₃ to reduce active hydroxymethyl groups that could possibly form artifactual dG-Me-Cys (Fig. 4B). The results demonstrated that the reduction of hydroxymethyl groups did not change the amount of dG-Me-Cys in rat liver samples. These data imply that the digested dG and other products do not generate artifacts that interfere with the detection of DPCs.

Using this sensitive method, endogenous and exogenous formaldehyde–induced DPCs were distinctly quantified in selected tissues [nasal epithelium, peripheral blood mononuclear cells (PBMC), bone marrow, and liver] of animals exposed to either filtered air only or [¹³CD₂]-formaldehyde. Consistent results from rat and NHP studies clearly demonstrate that exogenous DPCs were only found in the nasal epithelium of animals exposed to varying amounts of [¹³CD₂]-formaldehyde. Specifically, exogenous DPCs were detected at 1.36 ± 0.20 crosslinks/10⁸ dG in NHP nasal tissues exposed to [¹³CD₂]-formaldehyde, with a targeted aerosol concentration of 6.0 ppm for 2 consecutive days (6 hours/day; Table 1). A much higher number of exogenous DPCs were measured at 18.18 ± 7.23 crosslinks/10⁸ dG in the rat nasal tissues from animals exposed to [¹³CD₂]-formaldehyde at a targeted concentration of 15 ppm for 4 consecutive days (6 hours/day; Table 2). In contrast to the exogenous DPCs, endogenous DPCs were present in all examined tissues in both air control and exposed animals. Surprisingly, endogenous DPCs were present at higher amounts, compared with exogenous DPCs in exposed NHP nasal tissue (Table 1). More strikingly, the number of endogenous DPCs (15.46 ± 1.98 crosslinks/10⁸ dG) in air control NHP livers was close to the highest amounts of exogenous DPCs (18.18 ± 7.23 crosslinks/10⁸ dG) observed in the rat nasal tissues exposed to 15 ppm [¹³CD₂]-formaldehyde for 4 consecutive days, a highly carcinogenic exposure concentration in the rat cancer bioassays (30–32).

To monitor the induction and elimination of DPCs, we applied our method to investigate the accumulation and persistence of formaldehyde-induced DPCs in rat nasal tissues. As shown in Tables 2 and 3, both high and low exposure concentrations of inhaled formaldehyde gave rise to the accumulation of exogenous DPCs in rat nasal tissues. At a high exposure concentration of 15 ppm [¹³CD₂]-formaldehyde, similar amounts of exogenous DPCs were detected in 1-day and 2-day exposed rat nasal tissues (5.52 ± 0.80 and 4.69 ± 1.76 crosslinks/10⁸ dG), whereas 4-day exposed rat nasal tissues showed a dramatic increase in the formation of exogenous DPCs (18.18 ± 7.23 crosslinks/10⁸ dG; Table 2). In contrast to the large changes in exogenous DPC formation, less difference in the amounts of endogenous DPCs were present in the same samples, suggesting that the endogenous DPCs were at steady-state concentrations. Most importantly, exogenous DPCs showed little change in the animals exposed for 28 days with either a 24- or 168-hour (7 days) postexposure recovery period (Table 3), indicating high stability and long-term persistence of DPCs with the dG-Me-Cys linkage. The discovery of accumulation and persistence of formaldehyde-DPCs highlights the need for a better understanding of DPC stability and repair.

Formaldehyde was first shown to be an animal carcinogen in 1980, causing squamous cell carcinomas in the nasal passages of exposed rats at concentrations at or above 6 ppm (30, 31). Recent epidemiologic studies have suggested that exposure to formaldehyde vapors may lead to the development of hematopoietic cancers, such as leukemia; however, these findings remain under debate (28, 33–37). Although leukemia has been a major finding in epidemiology studies of inhaled formaldehyde, no mechanism for leukemia has been established. The International Agency for Research on Cancer working group also was not in full agreement on the evaluation of formaldehyde causing leukemia in humans (22, 38). A recent article by Coggon and colleagues did not find any leukemias in one of the largest cohorts of formaldehyde workers, even though some of those workers were exposed to inhaled formaldehyde at concentrations much higher than 2 ppm (37). In addition, the induction of leukemia by inhaled formaldehyde exposures has not been supported in rat carcinogenicity studies (39). Previous studies in rats and NHP also have not found any evidence that inhaled [¹³CD₂]-formaldehyde reached the bone marrow as exogenous DNA adducts (28, 34).

Endogenous formaldehyde is produced as an essential metabolic intermediate by enzymatic and nonenzymatic pathways and as a detoxification product of xenobiotics during cellular metabolism (40). It is well understood that endogenous formaldehyde originates from numerous sources including one-carbon pool metabolism, amino acid metabolism, methanol metabolism, lipid peroxidation, cytochrome P450–catalyzed demethylation, and histone demethylation reactions (34, 41, 42). Endogenous formaldehyde has been assumed to be present in all aqueous body fluids because of its water solubility. Its half-life in humans is estimated to be 1 to 1.5 minutes (43). In particular, the concentration of endogenous formaldehyde was detected at approximately 0.1 mmol/L in the blood of rats, monkeys, and humans (22, 44–46). In addition, endogenous formaldehyde was found to be 2 to 4 times higher in the liver and nasal mucosa than in the blood of a rat (47). Endogenous formaldehyde is also formed in cellular nuclei, secondary to demethylation of histone III (42). These formaldehyde molecules are released in close proximity to DNA, providing an important source for bone marrow DPC and formaldehyde monoadducts (42). Thus, with endogenous formaldehyde–induced DNA adducts and DPC present in all tissues examined, but a complete lack of exogenous DNA adducts and DPC in tissues distant to the portal of entry (6, 28), the key issue that must be addressed by risk assessors and epidemiologists is whether or how exogenous formaldehyde exposure could increase cancer risks at distal sites.

Because of their significant biologic consequences, formaldehyde-induced DPCs have long been recognized as a highly mutagenic form of formaldehyde DNA damage. The amounts of formaldehyde-induced DPCs are considered to represent a good molecular biomarker of formaldehyde exposure (6, 48). However, the lack of robust DPC measurements has limited our understanding of the genotoxic activity of formaldehyde. Previous studies using chloroform/isoamyl alcohol/phenol extraction–based DPC isolation following inhaled formaldehyde exposures in rats and NHPs found increased amounts of DPC formation in the nasal tissue, but not in tissues distant from the portal of entry (25, 27). In contrast, increased numbers of DPCs were detected in circulating lymphocytes of workers occupationally exposed to formaldehyde and in remote tissues, such as bone marrow, liver, kidney, and testes of mice exposed to inhaled formaldehyde using SDS/KCl precipitation–based DPC isolation (49, 50). This scientific dispute may be due to the major disadvantage of previous DPC assays that were unable to provide chemical identity for specific and accurate measurements of DPCs. Furthermore, it is has been impossible to distinguish endogenous from exogenous formaldehyde–induced DPCs prior to the use of stable isotopes.

In particular, the National Research Council (NRC) committee considered our earlier work using sensitive and distinct measurements between exogenous and endogenous formaldehyde–induced DNA adducts as being highly informative and should be incorporated in the U.S. Environmental Protection Agency (EPA) draft Integrated Risk Information System (IRIS) assessment (34, 48). Moreover, the NRC committee suggested that DPCs play a more important role in formaldehyde genotoxicity and carcinogenicity compared with DNA adducts (48). This study provides accurate measurements that discriminate between endogenous and exogenous DPCs through the use of stable isotope formaldehyde exposures and ultrasensitive MS. Our data demonstrate that exogenous DPCs were only found in nasal samples of exposed animals, but not in sites remote to the portal of entry. Consistently, our previous studies reported that inhaled formaldehyde–induced DNA adducts and protein adducts were only detected in rat and NHP nasal tissues, but not in other tissues, such as lung, liver, bone marrow, and PBMC (28, 34, 51). These findings suggest that there is no additional dose to the bone marrow beyond that from endogenous formaldehyde. The lack of exogenous DPCs in remote tissues does not support a plausible relationship between inhaled formaldehyde and increased risks of leukemia. On the other hand, with high amounts of endogenous formaldehyde always present in all cells, one would expect the induction of disease states by this highly reactive aldehyde. Our previous work identified specific DNA adducts (N²-Me-dG) and DNA–DNA crosslinks (dG-Me-dG) induced by endogenous formaldehyde (34). This study provides the first evidence to the existence of endogenous formaldehyde–induced DPCs in all examined tissues. Furthermore, the recent article by our laboratory and Pontel and colleagues clearly demonstrated that endogenous formaldehyde is a hematopoietic genotoxin and a metabolic carcinogen (6). In addition, as recently reported, as many as 1 of 3 acute myeloid leukemia patients have deficiencies in aldehyde dehydrogenases that play important roles in preventing DNA damage by detoxifying endogenous aldehydes (52). Thus, recent medical research has provided strong data that better explain the formaldehyde epidemiology findings.

Formaldehyde-induced DPCs had been considered to be unstable in cell and animal studies. A rapid decline of the formaldehyde-induced DPCs was reported in cell culture, and there was no accumulation of DPCs observed in rat nasal tissues after repeated exposure (16, 53, 54). However, this study demonstrated that exogenous DPCs accumulated with exposure time and showed little change after one week postexposure. Previous inconsistent results may be due to the use of nonstructurally specific measurements, which are highly susceptible to the interference of non-covalent DNA protein complexes. Although DPC repair mechanisms have not been well understood, proteolysis of the protein components of the DPC to small peptides has been reported as a protease-based DNA repair pathway specific for DPCs (10, 11, 15, 55). This study measures single amino acid–nucleoside crosslinks, specifically dG-Me-Cys. The small change in the amount of dG-Me-Cys measured one week postexposure does not rule out the possibility that the protein components of DPCs as a whole do not undergo proteolytic degradation to small peptides. In a recent study, synthesized DPCs, including dG-Me-cysteine, dG-Me-GSH, and dG-Me-peptide crosslinks, were found to undergo rapid hydrolysis in vitro (28). The long-term persistence of DPCs observed in vivo is likely due to the steric hindrance and localized hydrophobic conditions of DNA and the associated protein that greatly increase DPC stability.

The success of distinctly measuring endogenous and exogenous formaldehyde DPCs has demonstrated the importance of structurally specific DPC data. The lack of exogenous DPCs in remote tissues away from the portal of entry suggests that the effects of inhaled formaldehyde may have been overemphasized, while endogenous formaldehyde has been underappreciated as a source of exposure leading to the induction of leukemia. The finding of exogenous DPC accumulation during inhalation exposure and its minimal repair following one week postexposure provides the first evidence of long-term persistence of DPCs. The discovery of endogenous DPCs in all tissues examined raises an important perspective on the potential health risks posed by endogenous formaldehyde. Considering the high background of endogenous DPCs, it would be impossible to measure the induction, distribution, and elimination of exogenous DPCs without exogenous DPC measurements, particularly at low levels of exposure.

Future applications of this method include, but are not limited to, (i) investigation of DPC repair mechanism(s) in vivo using formaldehyde-induced DPCs; (ii) understanding whether and how the Fanconi anemia genes are involved in aldehyde-induced DPC repair; (iii) application of such data for science-based risk assessment of aldehydes in a manner that contributes to the EPA's IRIS program assessment of formaldehyde and aldehydes in general; (iv) utilizing structurally specific methodologies to determine the makeup of DPCs induced by various endogenous and exogenous agents, such as malondialdehyde, 4-hydroxynonenal, acrolein, and aldehydic lesions present on abasic sites. Likewise, DPCs associated with exogenous anticancer drugs can be understood better, contributing to further advancements of chemotherapy. In summary, this study provides a new approach to accurately determine the roles of endogenous and environmental DPCs to form mutagenic DNA damage as factors that contribute to disease and provide more accurate data that can improve cancer risk assessments.

No potential conflicts of interest were disclosed.

Conception and design: Y. Lai, R. Yu, B.C. Moeller, J.A. Swenberg

Development of methodology: Y. Lai, H.J. Hartwell, B.C. Moeller

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Lai, H.J. Hartwell, B.C. Moeller, W.M. Bodnar, J.A. Swenberg

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Lai, R. Yu, H.J. Hartwell, B.C. Moeller, J.A. Swenberg

Writing, review, and/or revision of the manuscript: Y. Lai, R. Yu, H.J. Hartwell, B.C. Moeller, W.M. Bodnar, J.A. Swenberg

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Lai, H.J. Hartwell, B.C. Moeller, J.A. Swenberg

Study supervision: B.C. Moeller, J.A. Swenberg

The authors thank Dean Kracko, Raul Romero, and Dr. Melanie Doyle-Eisele at Lovelace Respiratory Research Institute for providing technical assistance and guidance in setting up and conducting the animal exposures. The costs of animal exposures at Lovelace Respiratory Research Institute were generously provided by Formacare and the Research Foundation for Health and Environmental Effects. The authors also thank Leonard B. Collins for his assistance with HPLC purification and nano-UPLC-MS-MS.

This work was supported by the National Institutes of Environmental Health Sciences (NIEHS) Superfund Basic Research Program (P42 ES005948), NIEHS Center for Environmental Health and Susceptibility (P30 ES010126), the Texas Commission for Environmental Quality (582-12-21861), and Formacare.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.