Acute exposure to dihydroxyacetone promotes genotoxicity and chromosomal instability in lung, cardiac, and liver cell models

Abstract

Inhalation exposures to dihydroxyacetone (DHA) occur through spray tanning and e-cigarette aerosols. Several studies in skin models have demonstrated that millimolar doses of DHA are cytotoxic, yet the genotoxicity was unclear. We examined the genotoxicity of DHA in cell models relevant to inhalation exposures. Human bronchial epithelial cells BEAS-2B, lung carcinoma cells A549, cardiomyocyte Ac16, and hepatocellular carcinoma HepG3 were exposed to DHA, and low millimolar doses of DHA were cytotoxic. IC₉₀ DHA doses induced cell cycle arrest in all cells except the Ac16. We examined DHA’s genotoxicity using strand break markers, DNA adduct detection by Repair Assisted Damage Detection (RADD), metaphase spreads, and a forward mutation assay for mutagenesis. Similar to results for skin, DHA did not induce significant levels of strand breaks. However, RADD revealed DNA adducts were induced 24 h after DHA exposure, with BEAS-2B and Ac16 showing oxidative lesions and A549 and HepG3 showing crosslink-type lesions. Yet, only low levels of reactive oxygen species or advanced glycation end products were detected after DHA exposure. Metaphase spreads revealed significant increases in chromosomal aberrations in the BEAS-2B and HepG3 with corresponding changes in ploidy. Finally, we confirmed the mutagenesis observed using the supF reporter plasmid. DHA increased the mutation frequency, consistent with methylmethane sulfonate, a mutagen and clastogen. These data demonstrate DHA is a clastogen, inducing cell-specific genotoxicity and chromosomal instability. The specific genotoxicity measured in the BEAS-2B in this study suggests that inhalation exposures pose health risks to vapers, requiring further investigation.

Dihydroxyacetone (DHA) is a triose carbohydrate discovered to brown the skin in the 1950s. The US Food and Drug Administration (FDA) approved topical applications of up to 20% in 1977 (Braunberger et al. 2018). Soon after, DHA was demonstrated as mutagenic using an Ames test (Pham et al. 1979). Despite this early finding, limited data has confirmed its genotoxicity or mutagenic potential. The invention and popularization of spray tanning in 1999 led to renewed interest in the genotoxicity and cytotoxicity of DHA (Petersen et al. 2004; Jung et al. 2008; Perer et al. 2020; Striz et al. 2021).

Aerosolized applications raised concerns about inhalation exposures, leading the FDA to recommend protective measures (Jensen et al. 2017). More recently, DHA was found in electronic cigarettes (e-cigarettes) aerosol, produced by the free radical oxidation of glycerol, one of the two main e-liquid ingredients (Vreeke et al. 2018). Inhalation exposure to DHA is now highly likely, yet exposure effects are poorly understood.

The amount of DHA inhaled or absorbed from spray tanning is unknown. However, the European Union Safety Commission estimates a maximum of 600 µg of DHA may be inhaled in spray tanning booths (SCCS 2010). E-cigarette exposures are also undefined. However, up to 2.29 µg of DHA per 58 ml puff is generated (Vreeke et al. 2018). E-cigarettes are highly variable in their puff volumes, with up to 133.92 ml per puff found in cigarette-like devices and up to 519.6 ml for tank devices (Spindle et al. 2018; Hiler et al. 2020). Most vapers report 10 puffs per session with up to 24 smoking sessions per day (Yingst et al. 2020; Soule et al. 2023). Using these estimates, e-cigarette DHA exposures range from 38.4 µg/d to ∼1 mg/d, assuming the smaller cigarette-like devices are used (Vreeke et al. 2018; Jones et al. 2020; Yingst et al. 2020). Therefore, a vaper may inhale high micromolar to low millimolar doses per day (Lee et al. 2018; Vreeke et al. 2018). Although these estimates are based on self-reported and laboratory-measured smoking behaviors, they demonstrate that inhaled and systemic exposures to DHA occur and may pose a risk to human health.

DHA’s cytotoxicity and genotoxicity have been characterized in skin models. Initially, there was a prevailing attitude that DHA was “safe,” largely because it is a carbohydrate whose absorption was limited to the stratum corneum (Akin and Marlowe 1984). Later studies confirmed DHA penetrated the viable skin and bloodstream, entering tissues and cells (Yourick et al. 2004). In 2004, DHA, in the 25 to 50 mM range, was shown to induce apoptosis in the immortalized keratinocyte HaCaT cells through G2/M cell cycle arrest and DNA damage (Petersen et al. 2004). DNA damage measured by comet assay showed increased tail moments. DNA damage was attributed to reactive oxygen species (ROS) generated by exposure because antioxidants reduced the prevalence of strand breaks (Petersen et al. 2004). In 2008, ex vivo skin was used to examine co-exposures of DHA with UV, resulting in increased ROS (Jung et al. 2008). In 2012, the glycation properties of DHA and dihydroxyacetone phosphate (DHAP) were examined in vitro with human serum albumin (HSA) (Seneviratne et al. 2012). Both DHA and DHAP induced glycation of HSA at 2, 10, and 30 mM. DHA was more reactive in solution with HSA than DHAP, with reactions depending on pH, temperature, and time (days) (Seneviratne et al. 2012).

More recently, the cytotoxicity of DHA in primary keratinocytes and fibroblasts and human epidermal reconstructs were examined with replication stress, G2/M arrest, and the generation of advanced glycation end products (AGEs) observed (Perer et al. 2020; Striz et al. 2021). Similar to the previous results, genotoxicity was only observed at doses higher than 25 mM (Perer et al. 2020; Striz et al. 2021). Low millimolar doses of DHA-induced glycation and stress responses in these models (Perer et al. 2020; Striz et al. 2021). These studies suggest 25 to 50 mM DHA is cytotoxic in skin models but provide mixed evidence that DHA is genotoxic. Striz et al. (2021) suggested the genotoxicity observed is related to the cytotoxic mechanism.

We have extended the cytotoxicity characterization of DHA into systemic models, nontumorigenic human embryonic kidney HEK293T and the hepatocellular carcinoma C3A, also known as HepG3 (Smith et al. 2018, 2019; Hernandez et al. 2022). Cytotoxicity in these models occurs at lower doses than in the skin models, with IC₅₀ values between 5 and 10 mM compared with the 25 to 50 mM observed for the keratinocyte and fibroblast models. We determined whether the cell death mechanism differed in these cells from the skin models. HEK293T cells undergo autophagy, whereas the HepG3 showed a mixture of apoptosis and autophagy (Smith et al. 2019; Hernandez et al. 2022). In HepG3, no cleaved PARP1 or Caspase 3 was observed despite annexin-positive staining, and LCB3II and LAMP1 increased over time. Replication stress was observed in both cell lines, but the cell cycle arrest point was altered. HEK293T showed G2/M arrest consistent with the skin models, but HepG3 showed weak cell cycle blocks at both G1/S and G2/M (Smith et al. 2019; Hernandez et al. 2022).

Given the differences between these systemic models and the skin models previously characterized, we evaluated the cytotoxic and genotoxic effects of DHA across several cell lines commonly used to assess inhalation and systemic environmental exposures to understand the genotoxic mechanisms related to DHA (Danielsen et al. 2008; Park et al. 2015; Zhang et al. 2016; Lu et al. 2023). Human bronchial epithelial cells BEAS-2B, human lung carcinoma cell A549, human cardiomyocytes Ac16, and HepG3 cell line were characterized for genotoxicity.

Materials and methods

Chemicals

DHA (Cat No. PHR 1430) was purchased from Sigma Aldrich (St. Louis, MO).

Cell culture

The human bronchial epithelial cell (BEAS-2B, CRL-3588), human lung carcinoma cell (A549, CCL-185), human carcinoma liver cell line (HepG2 C3A/HepG3, CRL-10741), purchased from ATCC (Manassas, CA) and the human cardiomyocytes (Ac16) were a gift from Dr Prasanna Krishnamurthy’s lab at the University of Alabama at Birmingham (UAB). The BEAS-2B cells were cultured in RPMI 1640 supplemented with 10% FBS. The A549 and HepG3 were cultured with Dulbecco’s modified Eagle’s (DMEM; 4.5 g/l) supplemented with 10% FBS, 2% Glutamax, and 1% sodium pyruvate. The Ac16 cells were cultured with Dulbecco’s Modified Eagle medium F12 (DMEM/F12) supplemented with 10% FBS. The cells were screened periodically for mycoplasma contamination using the Lonza Mycoalert kit (Walkersville, MD, United States).

Cytotoxicity

Cytotoxicity was determined using a growth inhibition assay as previously described (Hernandez et al. 2022). BEAS-2B, A549, and Ac16 cells were plated at 5,000, 2,500, and 2,500 cells per well in a 12-well, respectively, and left overnight (ON) in a 5% CO₂ incubator at 37 °C to attach. The next day, cells were exposed to increasing concentrations of DHA dissolved in media, up to 20 mM for the BEAS-2B cells and up to 10 mM DHA for the A549 and Ac16 cells. The cells were dosed in triplicates and grown for 5 to 7 d. Cells were collected after exposure using 0.25% Trypsin-EDTA (Gibco) and resuspended in 1× phosphate-buffered saline (PBS, VWR Life Sciences, Radnor, PA, United States). Cells were counted using the Bio-Rad TC-20 automated cell counter (Bio-Rad, Hercules, CA). The results are displayed as the percentage mean of survival and standard error of the mean (SEM) of 3 biological replicates. GraphPad Prism was used to plot the values, and a nonlinear regression was used to calculate the IC₅₀ and IC₉₀ values.

Immunoblotting

Immunoblotting was performed as previously described (Hernandez et al. 2022). BEAS-2B, A549, and Ac16 cells were plated at 0.5×10⁶, 0.3×10⁶, and 0.5×10⁶ in 10-cm dishes, respectively, and left in a 5% CO₂ incubator at 37 °C to attach for 2 d. After attachment, cells were mock-treated with media or dosed with corresponding IC₉₀ doses of DHA for 24, 48, or 72 h, or dosed with camptothecin (CPT) at 1 µM for 24 h. After the exposure period, the cell medium was collected in a 15 ml conical tube, and the cells were detached using cell scrapers and resuspended in the collected cell medium. Cells were pelleted down at 2,000 rpm for 5 min. Supernatants were aspirated, and the cell pellets were stored at −80 °C ON.

The next day, cells were lysed, and protein quantification was performed using a Bradford assay (Bio-Rad). Twenty micrograms of protein were loaded in 4% to 20% Mini-Protein TGX pre-cast gels (Bio-Rad) with Precision Plus Protein Dual color standards (1610364 Bio-Rad) as molecular weight marker and run at 120 V for 1 h. The gels were transferred using a nitrocellulose membrane and blocked for 1 h using 5% skim milk in 1× Tris-buffered saline (TBS, VWR Life Sciences, Radnor, PA) using 0.1% Tween 20 (TBST). The membranes were then incubated ON with corresponding primary antibodies (Table 1). The next day, membranes were washed three times with TBST for 5 min each and incubated with corresponding HRP-conjugated secondary antibodies (Cell Signaling, Danvers, MA) for 1 h at RT (∼21 °C). After incubation, membranes were washed three times again with TBST, and chemiluminescence (ECL, Advansta, San Jose, CA, United States) was used to image the membranes using the Bio-Rad ChemiDoc imagining system (Bio-Rad). Detected bands of 3 biological replicates were quantified using the Bio-Rad Image Lab software and calculated relative to the loading control and then the untreated cells. Values were plotted using GraphPad Prism and displayed as the mean and SEM values. Statistical analysis was performed on values using the 1-way analysis of variance with Dunnett’s post hoc test.

Table 1.

Antibodies, dilutions, and sources used in the study.

Antibodies	Source
1:5,000
α-Tubulin (T9026)	Millipore Sigma, St. Louis, MO
1:1,000
ATG5 (12994)	Cell Signaling, Danvers, MA
Caspase-3 (GTX23585)	Genetex, Irvine, CA
Cyclin D1 (ab16603)	Abcam, Cambridge, United Kingdom
GAPDH (ab181602)	Abcam
LAMP1 (D2D11) (9091)	Cell Signaling
LC3B (PA1-46286)	Life Technologies, Carlsbad, CA
mTOR (2972)	Cell Signaling
p-mTOR (Ser2448) (5536)	Cell Signaling
p16 (92803)	Cell Signaling
p21 (Sc-397)	Santa Cruz
PARP-1 (556494)	BD Biosciences, Franklin Lakes, NJ
Tdp1 (ab4166)	Abcam
TKFC (HPA039486)	Sigma Aldrich, St. Louis, MO
TOM20 (42406)	Cell Signaling
TPI (ab135533)	Abcam
1:500
Cyclin B1 (12231)	Cell Signaling
TOPI (ab85038)	Abcam
TOP2 α (11327)	BD Biosciences

The same procedure was used for shorter time periods in BEAS-2B and HepG3 cells, which were plated at 0.75×10⁶ and dosed with respective IC₉₀ values for 1, 4, and 24 h. After exposure, cells were detached using cell scrapers and resuspended in 1× PBS. After collection, the protocol was followed as detailed above.

Cell cycle analysis

To evaluate cell cycle arrest, BEAS-2B, A549, and Ac16 cells were plated at 1×10⁶, 0.75×10⁶, and 0.5×10^6, respectively, and incubated ON at 37 °C in a 5% CO₂ incubator for attachment. Cells were then mock-treated as a control or exposed to their corresponding IC₉₀ values of DHA for 24, 48, or 72 h. The cell cycle arrest was confirmed for selected cell lines by dosing with 100 nM CPT for 24 h. After exposure, cells were collected and prepped for cell cycle analysis through flow cytometry, as previously described (Hernandez et al. 2022). Briefly, the medium was collected from plates, and cells were washed using 1× PBS once. The cells were detached using 0.25% Trypsin-EDTA and resuspended in the collected medium. The samples were centrifuged, the pellet was washed with PBS and centrifuged again. The cells were counted, and 10⁶ were aliquoted into a new tube and centrifuged. The pellet was resuspended in ice-cold PBS, and the cells were fixed using 70% ethanol and left ON at 4 °C. The next day, samples were centrifuged, and the pellet was washed once with 1× PBS and centrifuged again. The samples were resuspended in RNase A, incubated for 10 min at 37 °C, and then stained with propidium iodide (PI) for 15 min at RT. The samples were run for cell cycle analysis using the BD FACS Symphony (BD Biosciences Franklin Lakes, NJ) and analyzed using the Flowjo software. The graphs displayed are a representative run, and the values calculated are the average percentage of each cycle ± SEM of the 3 biological replicates.

Reactive oxygen species measurements

Total ROS was measured in BEAS-2B, A549, Ac16, and HepG3 cells. Each cell line was plated at 5,000, 3,000, 1,000, and 5,000 per well in a clear bottom, black 96-well plate, respectively, and left to grow ON in a 5% CO₂ incubator at 37 °C. After attachment, cells were dosed with corresponding IC₉₀ concentrations of DHA for 1, 4, 24, or 48 h. As a positive control, cells were dosed with tert-butyl hydrogen peroxide (TBHP) at 250 µM concentration for 1 h. After exposure, 10 µM of CM-H₂DCFDA was added to each well and incubated for 30 min at 37 °C (Thermo Fisher Scientific, Waltham, MA). Once the incubation period was complete, fluorescence was measured at 493/522 nm using the BioTek Synergy HR microplate reader (Agilent Technologies, Santa Clara, CA). After reading the total ROS measurement, cells were stained using Hoechst staining solution at 1:1,000 and incubated for 15 min at RT (Thermo Fisher). Fluorescence was read again at 366/460 nm to normalize fluorescence to the cell count in the well. Values were graphed relative to control using GraphPad Prism and displayed as mean and SEM values of 3 biological replicates. Significance was calculated using the 1-way analysis of variance with Dunnett’s post hoc test.

Immunofluorescence

BEAS-2B, A549, and Ac16 cells were examined for histone H2AX phosphorylation on Ser 139 (γH2AX) and BEAS-2B, A549, Ac16, and HepG3 cells were examined for tumor-suppressor p53 binding protein 1 (53BP1). Cells were plated in 8-well chambers (Thermo Fisher) at a density of 3,000 cells per well for the BEAS-2B, 1,000 cells per well for the A549 and Ac16, and 5,000 cells per well for the HepG3. The chambers were allowed to adhere ON in a 5% CO₂ incubator at 37 °C. After attachment, cells were dosed with IC₉₀ concentrations of DHA for 24, 48, or 72 h. After exposure, cells were fixed with 3.7% formaldehyde (Thermo Fisher) in PBS for 10 min at RT and washed three times with 1× PBS. Then, cells were permeabilized using Biotum permeabilization buffer for 10 min at RT (Fremont, CA) and washed three times with 1× PBS. Chambers were blocked for 30 min at RT using 2% BSA in PBS. The cells were then incubated with the corresponding primary antibody, phosphoSer139-Histone γH2AX (1:500 9718 Cell Signaling) and 53BP1 (1:750 NB100-304 Novus Biologicals) for 1 h at RT. After incubation, cells were washed three times with 1× PBS to remove any excess antibodies; then, they were incubated for 1 h with antirabbit Alexa Fluor 546 (1:400 Thermo Fisher) for 1 h at RT. Nuclear staining was performed using Hoechst solution (1:800, Thermo Fisher) for 10 min before the end of the incubation period. Then, cells were washed three times with 1× PBS and imaged. Imaging was conducted with the all-in-one fluorescence Keyence (BZ-X800) (Keyence, Osaka, Japan) using the 10× objective (NA 0.45). A minimum of 100 cells were imaged for each condition over 3 biological replicates. The Nikon Elements software was used to define the region of interest (ROI) for the nucleus, and the mean fluorescent intensity was calculated with this ROI for each condition. GraphPad Prism was used to display the values ± SEM over the replicates. Significance was calculated using 1-way analysis variance and Dunnett’s post hoc test.

The procedure described above for immunofluorescent staining was also used to detect AGEs (1:500 ab23722 Abcam) and methylglyoxal (MG) (1:200 MABN1838 Millipore Sigma) in the BEAS-2B, A549, Ac16, and HepG3 cells at 24 and 48 h. Cells were mock-treated, treated with the IC₉₀ of DHA, or treated with MG for 24 h at 2.5 or 25 µM, as indicated. The initial procedure remained the same, and the corresponding primary antibody was incubated for 1 h. After incubation, cells were washed three times with 1× PBS and incubated for 1 h with a secondary antibody. Anti-rabbit Alexa Fluor 546 (1:400 Thermo Fisher) and anti-mouse Alexa Fluor 488 (1:400 Thermo Fisher) were used. Then, cells were washed with 1× PBS three times and imaged. Cells were imaged using the Keyence with a 20× objective (NA 0.75). For analysis, at least 4 images were taken, with at least 100 cells for each time point for each biological replicate. The total intensity was measured using a binary threshold to define the total cell area within the image field using the Nikon Elements software, and the values are reported as sum fluorescence intensity ± SEM for 3 biological repeats. Significance was calculated to its matched control using a Student’s t-test.

Repair Assisted Damage Detection

DNA damage was evaluated using Repair Assisted Damage Detection (RADD), which detects DNA lesions using a specific cocktail of DNA repair enzymes. These enzymes remove the DNA lesion, and then the gap or strand break is tagged using a digoxigenin-labeled dUTP inserted with Klenow exo-, which lacks proofreading (Holton et al. 2018; Lee et al. 2019). We used specific lesion cocktails to evaluate oxidative lesions, crosslinks, alkylation, or uracils within the genomic DNA. For oxidative lesions only (oxRADD), Fapy-DNA glycosylase (FPG), Endonuclease IV (Endo IV), and Endonuclease VIII (Endo VII) are used. For crosslinks (T4PDG), only T4 pyrimidine dimer glycosylase (T4PDG) and Endo IV are used. For uracil detection (UDG), uracil DNA glycosylase (UDG) and Endo IV are added. 3-Alkyladenine DNA glycosylase (AAG) and Endo IV are added for alkylation damage. We assessed DNA lesions within the BEAS-2B, A549, Ac16, and HepG3 cells (Holton et al. 2018). The cells were plated in 8-well chambers at a density of 5,000 cells for the BEAS-2B, 3,000 cells for the A549 and Ac16, and 10,000 cells for the HepG3 and left to attach ON in a 5% CO₂ incubator at 37 °C. The next day, cells were mock-treated or dosed with corresponding IC₉₀ concentrations of DHA for 24 h. After the exposure, cells were fixed using 3.7% formaldehyde for 10 min and washed three times with PBS following fixation. Biotum permeabilization buffer with 0.05% Triton-X was used to permeabilize the cells for 10 min at 37 °C. The cells were then washed three times with 1× PBS.

Chambers were then incubated with lesion removal cocktails described above and resuspended in 1× ThermPol buffer+BSA (New England BioLabs, Ipswich, MA, United States) for 1 h at 37 °C in a hybridization FISH oven (Lee et al. 2019). Once the incubation period has finished, the gap-filling mixture of Klenow exo- and digoxigenin-labeled dUTP is directly added to the chambers and placed again within the oven at 37 °C for an additional 1 h. The chambers are washed three times with 1× PBS and blocked with 2% BSA in PBS for 30 min. After blocking, chambers are incubated with primary antibody anti-digoxigenin (1:250 ab420 Abcam) or anti-mouse IgG1 isotype control (1:625 5415 Cell Signaling) for 1 h at RT. IgG1 is used as a negative control in these cells. Once the primary antibody has finished incubation, the cells are washed three times with 1× PBS and incubated with secondary antibody anti-mouse AlexaFluor 546 (1:400 Thermo Fisher) for 1 h at RT. The cells are then incubated with Hoechst solution (1:800, Thermo Fisher) for 15 min at RT and washed with 1× PBS three times.

Cells are then imaged using the Keyence microscope with the 20× objective (NA 0.75). At least 30 cells of each condition for each biological replicate are taken over 4 different fields. For analysis, the Nikon Elements software created an ROI around the nucleus, and the total intensity for the RADD channel within the nucleus was recorded for at least 100 cells. The sum of the fluorescent intensity for all the nucleus was graphed in GraphPad Prism, and significance was calculated using a Student’s t-test for each DNA adduct category versus its matched control.

Rapid approach to DNA adduct recovery

Rapid approach to DNA adduct recovery (RADAR) assay protocol was adapted from Kiianitsa and Maizels (2013) and Sonavane et al. (2018) and optimized for BEAS-2B, A549, Ac16, and HepG3 cells. Cells were cultured in 10-cm dishes with 1×10⁶, 0.75×10⁶, 0.75×10⁶, and 1×10⁶ cells and incubated ON at 37 °C for BEAS-2B, A549, Ac16, and HepG3, respectively, in a 5% CO₂ incubator to allow cells to adhere. Cells were then exposed to their corresponding IC₉₀ values of DHA for 1, 4, or 24 h or mock treated. After exposure, the cell culture media was aspirated from the plates, and cells were immediately lysed with 2 ml of DNAzol (Life Technologies) with 1% Sarkosyl (VWR). DNA–protein covalent complexes (DPCs) were recovered by taking 0.8 ml of cell lysate mixed with 0.4 ml of 200-proof ethanol (Fisher) and inverted. Samples were left to incubate at −80 °C ON. DPCs were centrifuged at 4 °C for 15 min the following day at 14,000 × g. The supernatant was aspirated, and pellets were washed twice with 1 ml of 75% ethanol and followed with a 10-min centrifugation at 14,000 × g set to 4 °C. The supernatant was aspirated, and the pellets were dissolved in fresh 8 mM NaOH (Sigma).

DNA content was measured using the AccuBlue Broad Range dsDNA Quantitation kit (Biotum) with 9 standards per the manufacturer’s instructions. Five hundred nanograms were used per condition for DPC immunodetection. Equal volumes (0.2 ml) of DPCs in 1× TBS were loaded into the slots of the BioDot SF Microfiltration Apparatus containing a 0.45-µM nitrocellulose membrane (Bio-Rad) cut to fit the apparatus. Samples were applied using vacuum to the membrane. For detection of total DPCs using SYPRO Ruby protein stain, the membrane was then incubated with agitation in a 7% acetic acid and 10% methanol mixture for 15 min, then washed in MilliQ water four times for 5 min each. Then, the membrane was gently rocked in SYPRO ruby blot stain reagent (Life Technologies) for 15 min to detect total protein loaded onto the membrane, then washed in MilliQ water four times for 1 min each and air dried ON. The dry membrane was imaged using the Bio-Rad ChemiDoc imagining system under the Sypro Ruby filter setting. Once imaging was completed, the membrane was washed 3 times using TBST for 5 min each and then incubated with Sybr Gold for loading control (Life Technologies) diluted in TBST (1:10,000) for 1 h. After incubation, the membranes were washed with TBST and imaged under fluorescence with the Sybr Gold excitation and emission settings using the Bio-Rad ChemiDoc imagining system.

The protocol was conducted as described above for specific DPC protein detection until after the samples were loaded onto the membrane. The membrane was then blocked using 5% skim milk in TBST for 1 h at RT followed by ON incubation at 4 °C with anti-TOPI (ab28432, abcam) at 1:500 or anti-TOP2α (611327, BD Biosciences, Franklin Lakes, NJ) at 1:500. The following day, membranes were washed three times in TBST for 5 min each. HRP-conjugated secondary antibodies (Cell Signaling) were added and incubated at RT for 1 h. Enhanced chemiluminescence (ECL, Advansta) detected specific DPCs on each membrane. Membranes were washed three times in TBST before applying the Sybr Gold and imaged in the Bio-Rad ChemiDoc imaging system.

Image Lab software quantified results for both total protein and specific antibodies. Values were normalized to loading control to determine the percentage of protein crosslinks and then calculated relative to control. The final values of 3 biological replicates were plotted in GraphPad Prism and displayed as mean ± SEM. Significance was calculated using the 1-way analysis of variance with Dunnett’s post hoc test.

Chromosomal aberrations

The chromosomal aberrations protocol was adapted from Matijasevic et al. (2008). Mitotic cells were cultured in 10-cm dishes with 1×10⁶ for A549 or 0.75×10⁶ for BEAS-2B and HepG3. Dishes were incubated ON at 37 °C in a 5% CO₂ incubator to allow cells to adhere. Cells were then mock treated or exposed to their corresponding IC₉₀ values of DHA for 48 h for the BEAS-2B and A549 or 96 h for HepG3 cells to evaluate aberrations after at least 2 cell cycles. Colcemid (Gibco) was spiked into the dishes with a final concentration of 0.02 µg/ml and incubated at 37 °C in a 5% CO₂ incubator for 90 min. The cells were collected by trypsinization and incubated with a hypotonic solution (0.075 KCl) in a 37 °C water bath for 20 min. Cells were fixed in a 3:1 methanol and acetic acid mix and dropped onto clean, dry slides. Slides were quickly passed through a Bunsen burner flame before a 45-min incubation on a slide warmer at ∼65 °C. Slides were stained with a 5% Giemsa stain (Gibco) diluted in 1× Gurr buffer (Gibco) for 10 min, washed in Gurr buffer, dried, and mounted.

More than 75 images were obtained over 3 biological replicates for each condition and cell line using the Keyence microscope with the 60× oil objective (NA 1.40). Each image obtained included at least one metaphase spread. Counts of the aberrations were obtained in each of the following categories: chromosome breaks, chromosome gaps, chromosome rings, chromatid exchange, and centromeric disruptions (Registre and Proudlock 2016). Four independent readers scored the metaphase spreads for aberrations. Values used were the lowest final counts per image and category.

The chromosomal aberrations for each cell were averaged from the lowest scores and graphed with GraphPad Prism. The graph shows the average number of chromosomal aberrations per cell ± SD for each category and the total number of aberrations. A 2-way Student’s t-test was used to determine the significance of each aberration category. The number of chromosomes was also counted in each image with ImageJ software to evaluate for polyploidy (>2n) or aneuploidy (<2n). The ploidy graph displays the frequency distribution for all images of the ordinal scoring (1 = aneuploid, 2 = normal, and 3 = polyploid) for each image.

supF mutation analyses

A forward mutation assay was performed as described by Fukushima et al. (2020). Briefly, the pSB189KL-BmBl plasmid was transfected into BEAS-2B or human embryonic kidney cells (HEK293T) using FuGene 6 according to the manufacturer’s instructions (Fukushima et al. 2022). After 6 h of transfection, cells were treated with 12.5 mM DHA for the BEAS-2B or 10 mM DHA (IC₉₀) for the HEK293T or 100 µM methyl methanesulfonate (MMS) for both cell lines for 48 h. MMS is used as a positive control. The plasmids were extracted as previously described by Stary and Sarasin (1992). The extracted plasmids were introduced into E. coli RF01 strain using electroporation, and cells were plated into titer and selection plates. Colonies were allowed to grow for 18 to 20 h at 37 °C, and the supF mutant frequencies were calculated as number of colonies in section plate × dilution factor/number of colonies in titer plate×dilution factor.

The numbers of colonies were far lower for the BEAS-2B due to low transfection efficiency compared with the HEK293T. No condition could be optimized to equalize the transfection efficiency between the 2 cell lines, so the HEK293T data is presented in the text, and the BEAS-2B results are presented in the Supplementary Materials. GraphPad prism was to display the mutant frequencies calculated over 3 biological replicates. Significance was calculated using the 1-way analysis of variance and Dunnett’s post hoc test.

Results

DHA induces dose-dependent cytotoxicity across the cell models

We evaluated the cytotoxicity of DHA using a growth inhibition assay. The BEAS-2B, A549, and Ac16 cells were exposed to increasing concentrations of DHA for 5 to 7 d, and viability was evaluated by continued growth (Fig. 1). Survival decreased at low millimolar concentrations of DHA across all cell lines. The BEAS-2B has an IC₅₀ of 7.5 ± 0.33 mM and IC₉₀ of 12.5 ± 0.92 mM (Fig. 1A). The A549 showed an IC₅₀ value of 6.5 ± 0.95 mM and an IC₉₀ value of 10 ± 1.6 mM of DHA (Fig. 1B). The Ac16 were the most sensitive to DHA, with an IC₅₀ of 2.0 ± 0.33 mM and an IC₉₀ of 4.4 ± 0.56 mM (Fig. 1C). HepG3 were previously characterized and was found to have an IC₅₀ of 3.9 ± 2.0 mM and IC₉₀ of 12.5 ± 3.6 mM (Fig. 1D). The variance calculated in the HepG3 cells may be due to the senescence population in these cells. A table summarizes the values calculated for each cell line (Fig. 1E).

Fig. 1.

DHA induces cytotoxicity across the cell line panel. Cells were dosed with increasing concentrations of DHA, and the survival percentage was calculated relative to control, mock-treated cells. Survival curves are shown for BEAS-2B (A), A549 (B), Ac16 (C), and HepG3 (D). HepG3 was previously characterized, and the graph was adapted from (Hernandez et al. 2022). The calculated IC₅₀ and IC₉₀ values for each cell line are shown in the table (E).

Previous work has shown that DHA induces cell type-dependent cell death mechanisms. A375P, dosed with 5 mM DHA, showed induced senescence and apoptotic cell death, whereas HEK293T, dosed with 5 mM DHA, showed autophagy with increased cleavage of LC3B (Smith et al. 2018, 2019). HepG3 cells had a hybrid cell death with autophagy shown through LC3B cleavage and apoptotic cells detected by annexin V but without PARP1 cleavage after DHA exposure (Hernandez et al. 2022). Each cell line was dosed with their corresponding IC₉₀ concentrations of DHA for 24, 48, or 72 h. This concentration was selected to evaluate the cytotoxic and genotoxic effects of DHA consistent with previous skin model studies. The time points selected for the genotoxicity assays do not show signs of cytotoxicity, and cell death initiation occurs at 72 h or 96 h, allowing us to examine both genotoxicity and the processes contributing to cell death. We characterized the cell death mechanisms for each cell line. PARP-1 was used to evaluate apoptotic cell death, whereas cleavage of LC3B was used as an autophagy indicator.

BEAS2-B cells were dosed with 12.5 mM for 24, 48, or 72 h, and 1 µM CPT was used as a positive apoptotic control. A gradual increase was observed in cleaved PARP-1 starting at 24 h, resulting in almost a 2-fold increase at 48 h compared with control cells (Fig. 2A). An increase in total PARP-1 was observed in DHA-exposed BEAS-2B cells at 24 and 48 h (Fig. S1). Corresponding cleavage of caspase 3 was not observed, although there was a slight increase in total caspase 3 in the DHA-exposed cells, whereas the cells dosed with CPT showed cleaved caspase-3 (Fig. 2A). These data suggest a noncaspase-dependent apoptosis pathway occurs in the DHA-exposed BEAS-2B cells.

Fig. 2.

DHA exposure promoted cell-type-dependent cell death mechanisms. A) Total and cleaved PARP-1 protein probed in BEAS-2B cells exposed to 12.5 mM DHA at 24, 48, 72 h or 1 µM of camptothecin (CPT) for 24 h. CPT is a positive control for apoptosis. B) LC3BII expression levels were measured in A549 cells exposed to DHA at 24, 48, and 72 h. C) PARP-1 and LC3BII were measured in Ac16 cells exposed to DHA at 24, 48, and 72 h. Graphs are displayed as the mean±SEM over 3 biological replicates. Significance is displayed as follows: *P < 0.05 and ****P < 0.0001.

DHA exposure of 10 mM to A549 cells showed a significant increase in LC3B cleavage, with LC3BII increasing at 24 h and continuing to increase at 48 and 72 h, resulting in autophagic cell death (Fig. 2B). The LC3B ratio was also calculated in these cells and showed an increase at 24 h and a significant increase at 48 and 72 h after DHA exposure (Fig. S2A). Additional autophagy markers were also examined, and an increase in the lysosomal marker LAMP-1 was also observed 24 h after DHA exposure and continued until 72 h (Fig. S2B). Autophagy-linked mTOR signaling was also measured with p-mTOR levels slightly increasing after DHA exposure, whereas total mTOR protein levels only slightly increased 72 h after exposure (Fig. S2C). No cleaved PARP-1 protein was observed in these cells, confirming that autophagy is the cell death mechanism in the A549 cells (Fig. S3).

Ac16 cells dosed with 4.4 mM DHA showed increased autophagic and apoptotic markers (Fig. 2C). Cleaved PARP-1 increased at 24 and 48 h in DHA-exposed cells but decreased at 72 h (Fig. 2C). Total caspase-3 remained constant, and no cleavage was observed in cells dosed with DHA (Fig. S4). LC3BII increased significantly at 24 h and nonsignificantly at 48 and 72 h, though the ratio of LC3BII to LC3BI was not as significantly altered as the A549 cells (Fig. 2C and Fig. S5A). Other autophagy markers, autophagy-related gene 5 (ATG5), and mTOR signaling were also measured. ATG5 gradually decreased at 24 h and significantly at 72 h (Fig. S5B). p-mTOR protein levels increased almost 3-fold at 72 h, consistent with the increase in total mTOR levels at 72 h (Fig. S5C). Ac16 cells were found to have a hybrid cell death mechanism similar to the one characterized in the HepG3 cells after DHA exposure with the mTOR-linked autophagy (Hernandez et al. 2022).

Given the differences between the cell lines and their different tissue or tumor origin, we examined proteins critical for incorporating DHA in metabolic pathways. Triose kinase FMN/cyclase (TKFC) converts DHA to DHAP, which can then enter glycolysis or other metabolic pathways (Moreno et al. 2014; Marco-Rius et al. 2017). Triosephosphate isomerase (TPI) converts DHAP to glyceraldehyde-3-phosphate to enter the glycolytic pathway. Another key enzyme is glyceraldehyde-3-phosphate dehydrogenase (GAPDH), which converts glyceraldehyde-3-phosphate to 1,3 biphosphoglycerate downstream within the glycolysis pathway. TKFC, TPI, and GAPDH levels were measured using immunoblotting (Fig. S6). TKFC protein levels vary by cell line, with HepG3 showing the highest protein levels, followed by Ac16, A549, and then BEAS-2B (Fig. S6A). TPI and GAPDH protein levels remained constant across the cell lines (Fig. S6B).

No clear correlation exists between these protein levels and the cytotoxicity observed across the cell line panel. Overall, DHA is cytotoxic across the different cell lines. Sensitivity to DHA and the cell death mechanism were cell-type dependent, consistent with other nonskin models.

Replication stress is observed in cells dosed with DHA

Given the observed cytotoxicity, we evaluated cell cycle changes after DHA exposure. Cells were dosed with IC₉₀ DHA concentrations, and cell cycle phases were measured using flow cytometry at 24, 48, and 72 h (Fig. 3). BEAS-2B cells showed a decrease in G1 24 h after dosing, then a significant increase in S phase at 48 h (23.7 ± 1.9%; P = 0.01) and 72 h (25.4 ± 3.5%; P = 0.003) compared with control (16.3 ± 1.9%) (Fig. 3A and Table S1). As previously reported for BEAS-2B cells, a population of polyploidy cells after G2/M could be observed in the control (Conery and Harlow 2010). This population increased 24 h after DHA exposure (Fig. S7).

Fig. 3.

Cell-dependent cell cycle arrest was found due to DHA exposure. Cell cycle phases were observed in cells dosed with corresponding IC₉₀ values at 24, 48, or 72 h using propidium iodide (PI) staining and flow cytometry. A) BEAS-2B cells were dosed with 12.5 mM DHA, and the cell population increased in the S-phase starting at 48 h. B) A549 cells were dosed with 10 mM DHA, and an increase in the G1 phase was observed starting at 24 h. C) Ac16 cells were dosed with 4.4 mM DHA, and no cell cycle arrest was observed. Representative cell cycle analysis graphs are displayed for each time point with the mean±SEM over 3 biological replicates in the bar graphs.

A549 cells showed a weak G1 arrest with a slight increase in cells in this population starting at 24 h (50.6 ± 10.1%) and continuing until 72 h (57.6 ± 6.4%) compared with control (48.8 ± 3.2%) (Fig. 3B and Table S1). Although Ac16 cells showed no change in the cell cycle phase over the 72-h exposure period and no cell cycle arrest (Fig. 3C and Table S1). Previously characterized HepG3 cells had a weak cell cycle arrest in both the G1 and G2/M phases, with a significant increase at 72 h in the percentage of cells arrested in these phases (Fig. S8).

We examined cell cycle checkpoint markers, cyclin B1, cyclin D1, and p21 at 24, 48, and 72 h of DHA exposure to confirm cell cycle changes. In BEAS-2B, an increase in cyclin B1 started at 24 h, along with a significant decrease in p21 (Fig. 4A). Cyclin B1 and p21 levels returned to normal levels after 72 h of DHA exposure. Cyclin D1 was also measured, and no change was found at any time points (Fig. S9A). Similar to the BEAS-2B, the A549 cells showed a significant decrease in p21 24 h after DHA exposure, but cyclin B1 and cyclin D1 showed no change (Fig. 4B and Fig. S9B). Ac16 cells, which had no cell cycle arrest, showed decreased cyclin D1 and cyclin B1 over the exposure period (Fig. 4C). No change in p21 occurred over the exposure period (Fig. S9C). HepG3 cells showed changes in cell cycle markers consistent with the weak cell cycle arrest at G1 and G2, as characterized previously (Fig. S8C). The weak cell cycle arrest in the A549 and HepG3 and the S phase increase in BEAS-2B suggest DHA is inducing replication stress in most of the cell panel, consistent with previous skin models and HEK293T cell characterizations (Petersen et al. 2004; Smith et al. 2019; Perer et al. 2020; Striz et al. 2021). The notable exception is the Ac16, given that these cells show more sensitivity to DHA.

Fig. 4.

Cell cycle checkpoint markers are altered by DHA exposure. A) BEAS-2B cells were probed with cell cycle markers, cyclin B1, and p21 using immunoblotting after exposure to 12.5 mM DHA for 24, 48, and 72 h. B) Cell cycle marker, p21 was probed using immunoblotting in A549 cells dosed with 10 mM DHA for 24, 48, and 72 h. C) Cell cycle markers, cyclin B1, and cyclin D1 were probed using immunoblotting. Graphs are displayed as the mean ± SEM over 3 biological replicates. Significance displayed: *P < 0.05; **P < 0.01.

DHA promotes cell type-dependent genotoxic effects

Characterization of DHA in skin models and HEK293T showed strand breaks detected by γH2AX only occurred at cytotoxic doses after 72 h of exposure, consistent with cell death (Petersen et al. 2004; Smith et al. 2018, 2019; Perer et al. 2020; Striz et al. 2021). We first examined strand break signaling by γH2AX, which signals single- and double-strand breaks, and tumor-suppressor p53 binding protein 1 (53BP1) recruitment after DHA exposure in this cell panel (Panier and Boulton 2014; Kopp et al. 2019).

BEAS-2B, A549, and Ac16 cells were dosed with their corresponding values of IC₉₀ concentrations for 24, 48, and 72 h and immunofluorescent staining was performed. Only very low levels of γH2AX were observed across the cell panel (Fig. S10). Similarly, HepG3 cells only showed low levels of γH2AX staining starting at 24 h with significantly increased levels at 72 and 96 h, again consistent with cell death (Hernandez et al. 2022).

However, 53BP1 recruitment and protein levels were significantly increased at all time points in BEAS-2B cells exposed to DHA (Fig. 5A). A549 cells only showed a significantly increased 53BP1 at 72 h (Fig. 5B). Ac16 cells showed a modest but significant increase in 53BP1 48 h after DHA exposure and decreased levels back to the level of control at 72 h (Fig. 5C). We also measured 53BP1 levels in HepG3 cells, which was not previously done. There was no change in 53BP1 at any point in time for the HepG3 (Fig. 5D). Except for the BEAS-2B, no sustained increase in strand break generation was observed within the cell line panel.

Fig. 5.

53BP1 recruitment and nuclear intensity after DHA exposure in the cell line panel. Cells were stained with 53BP1, and the nuclear intensity of the 53BP1 immunofluorescence was quantified in cells dosed with their corresponding IC₉₀ concentrations of DHA for 24, 48, and 72 h. BEAS-2B (A), A549 (B), Ac16 (C), and HepG3 cells (D). The scale bar is 50 µM. The graphs are displayed as the mean fluorescence intensity of the cells and SEM. Significance displayed as follows: **P<0.01; ***P<0.001; ****P<0.0001.

We then expanded our analysis to consider DHA’s induction of DNA adducts, which were previously inferred by Petersen et al. (2004) to be caused by the generation of ROS. We first confirmed that DHA exposure induced ROS within our cell line panel using the fluorescent reporter CM-H₂DCFDA to measure intracellular ROS after DHA exposure. Measurements were obtained at 1, 4, 24, 48, and 72 h after DHA exposure and tert-butyl hydrogen peroxide (TBHP) was used as a positive control (Fig. 6). BEAS-2B cells had fluctuating levels of total ROS from 24 h until 72 h (Fig. 6A). Total ROS levels in A549 cells gradually increased, starting at 1 h and significantly increasing at 24 h, then decreasing back to control levels at 48 and 72 h (Fig. 6B). Both Ac16 and HepG3 cells showed no increase in ROS at all time points (Fig. 6C and D). Only minimal ROS generation was observed across the cell line panel compared with TBHP.

Fig. 6.

Cell line-dependent changes in total ROS after DHA exposure. Total reactive oxygen species (ROS) measured by CM-H₂DCFDA assay in cells exposed to DHA for 1, 4, 24, 48, and 72 h. Tert-butyl hydrogen peroxide (TBHP) is dosed at 250 µM for 1 h and used as a positive control for ROS generation. A) BEAS-2B, B) A549, C) Ac16, and D) HepG3 cells. Graphs are displayed relative to control of the mean intensity ± SEM over 3 biological replicates. Significance is shown as follows: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Later DHA papers by Perer et al. (2020) and Striz et al. (2021) suggested AGEs are formed in DHA-exposed skin cells, contributing to cytotoxic effects. Thus, immunofluorescence was used to measure methylglyoxal (MG), a precursor for AGEs, and protein-conjugated AGEs in our cell panel (Fig. 7). The MG antibody selected was raised specifically against MG, and we confirmed using MG dosing as a positive control. The AGE antibody was raised against glycolaldehyde-conjugated BSA and detects AGE-protein conjugates.

Fig. 7.

After DHA exposure, the cell line panel does not significantly produce methylglyoxal (MG) or advanced glycation end products (AGEs). MG and AGEs were quantified by immunofluorescence using specific antibodies after dosing with the IC₉₀ of DHA for 24 or 48 h. MG at 2.5 or 25 µM was used as a positive control for MG generation. A–D) show the MG quantification for the BEAS-2B (A), A549 (B), Ac16 (C), and HepG3 (D). E–H) show the AGE levels for the BEAS-2B (E), A549 (F), Ac16 (G), and HepG3 (H). Total intensity was measured using a binary threshold and fluorescence intensity for a minimum of 100 cells is quantified and averaged over 3 biological replicates. The graph displays the mean fluorescence intensity for the cells ± SEM. Significance displayed as follows: *P < 0.05.

BEAS-2B cells were dosed with 12.5 mM DHA for 24 and 48 h, and no MG or AGEs were detected at any time (Fig. 7A and E, respectively). DHA-exposed A549 showed no increase in MG and only a slight increase in AGEs at 24 h and 48 h (Fig. 7B and F, respectively). Ac16 also showed no change in MG and a slight increase in AGEs at 24 and 48 h (Fig. 7C and G, respectively). HepG3 cells exposed to DHA showed decreased MG levels at 24 and 48 h and slightly increased AGEs (Fig. 7D and H, respectively). Despite observation in the skin, the systemic models showed only a minimal generation of MG or AGEs.

Since we could not confirm the proposed genotoxic mechanisms for DHA from the skin in these systemic models, we used a novel DNA adduct detection method, RADD, to identify lesion classes induced after DHA exposure in our cell panel. RADD detects DNA adducts using bacterial DNA glycosylases to excise the DNA adducts from the DNA. The gapped DNA or strand breaks that result from the DHA-induced DNA damage are then filled using Klenow exo-, which inserts a tagged dUTP for fluorescent detection of the DNA damage (Holton et al. 2018; Lee et al. 2019). We used specific cocktails of the DNA glycosylases to detect oxidative lesions (FPG + Endo VIII + Endo VI, OXO), pyrimidine crosslinks (T4PDG + Endo IV, T4PDG), uracils (UDG + Endo IV, UDG), and alkylation (AAG + Endo IV, AAG) damage.

BEAS-2B cells dosed with 12.5 mM DHA were found to have a significant increase in oxidative and crosslink-type lesions (Fig. 8A). A549 cells were found to have a significant increase in crosslink-type lesions and a significant decrease in uracil and alkylation lesions (Fig. 8B). Ac16 cells showed increased oxidative lesions and significantly decreased crosslink and alkylation lesions (Fig. 8C). HepG3 cells have a significant increase in crosslink lesions with a significant decrease in oxidative and alkylation lesions (Fig. 8D). Since RADD detects DNA lesions using bacterial glycosylases, independent of cellular DNA repair mechanisms, we can also compare the relative levels of DNA damage across the cell models. The BEAS-2B showed the lowest levels of DNA damage within the cell panel, the A549 and HepG3 showed similar DNA damage levels, and the Ac16 showed the highest basal DNA damage levels and induced lesions in the cell panel. The RADD assay confirmed DHA induces DNA adducts within 24 h of exposure, though the lesion type is cell-type dependent.

Fig. 8.

DHA exposure for 24 h induces oxidative and crosslink-type lesions within the cell panel. The Repair Assisted Damage Detection (RADD) assay measured specific classes of DNA adducts after DHA exposure for 24 h across cell lines. The OXO reaction measured oxidative lesions, T4PDG measured crosslink-type lesions, UDG measured uracil lesions, and AAG was used to measure alkylation lesions. Adduct levels are measured by immunofluorescence of the modified nucleotide inserted into the damage site. The total nuclear intensity for at least 100 cells is averaged over 3 biological replicates. The images are representative with a scale bar of 50 µM. The graphs display the fluorescence intensity for the cells ± SEM. A) BEAS-2B cells, B) A549 cells, C) Ac16 cells, and D) HepG3 cells. Significance displayed as follows: *P < 0.05; **P < 0.01; ****P < 0.0001.

Given the DNA adduct results, we performed one final genotoxicity evaluation for DPCs, which would not be detected in the previous assays. The generation of MG or AGEs may induce DPCs, or the metabolism of DHA could generate metabolites that promote DPC formation (Perry and Ghosal 2022). To measure DPCs, we performed the RADAR assay. RADAR isolated DPCs using DNAzol, and DPCs were measured using a slot blot (Kiianitsa and Maizels 2013). We examined total protein–DNA crosslinks and specific types of DPCs, topoisomerase I (TOPI) and topoisomerase II-α (TOP2α). TOPI and TOP2α catalyze the relaxation of supercoiled DNA involved in the replication fork arrest and double-strand break formation through covalent binding to the DNA. These complexes can become trapped on the DNA, forming DPCs (Kiianitsa and Maizels 2013).

BEAS-2B, A549, Ac16, and HepG3 cells were dosed with their corresponding IC₉₀ concentrations for 1, 4, and 24 h to detect TOPI, TOP2α, or total DPCs (Fig. 9). BEAS-2B cells have no change in TOPI-related crosslinks after DHA exposure and a significant decrease in TOP2α at 1 and 4 h with continued decreased formation at 24 h (Fig. 9A). Total DPCs were also measured at these time points using Sypro Ruby to stain the proteins bound to the membrane, and the total levels of DPCs were also significantly decreased at 1 and 24 h after 12.5 mM DHA (Fig. S11A). We confirmed these changes in DPCs were not related to protein changes in TOPI, TOP2α, or TDP1, which resolves topoisomerase DPCs by immunoblotting (Sun et al. 2020). BEAS-2B dosed with 12.5 mM DHA showed no significant changes in TOPI, TOP2α, or TDP1 at 1, 4, or 24 h (Fig. S12). A549 cells showed no significant changes in TOPI or TOP2α DPCs, consistent with the total DPCs measured (Fig. 9B and Fig. S11B). Similar to the A549, Ac16 cells showed no change in TOPI, TOP2α, or total DPCs (Fig. 9C and Fig. S11C). HepG3 cells showed no change in TOPI DPCs and a slight increase in TOP2α DPCs at 1 h compared with control (Fig. 9D). HepG3 total DPCs decreased formation at 1 and 24 h with a significant decrease at 4 h compared with control (Fig. S11D). Again, with this observed decrease in DPCs, we examined the TOPI, TOP2α, and TDP1 protein levels in HepG3 after DHA exposure. No significant changes in these proteins were observed across the time points (Fig. S13).

Fig. 9.

TopI and Top2α related DNA–protein crosslinks in cells dosed with DHA. The rapid approach for DNA adduct recovery (RADAR) assay measured specific DNA–protein crosslinks from topoisomerase I (TOPI) and topoisomerase II-α (TOP2α). The cells were dosed with corresponding IC₉₀ values of DHA for 1, 4, and 24 h in A) BEAS-2B cells, B) A549 cells, C) Ac16 cells, and D) HepG3 cells. A representative image of each type of crosslinks is shown and graphs are displayed relative to control using the mean intensity ± SEM of 3 biological replicates. Significance displayed as follows: *P < 0.05; **P < 0.01.

These assays demonstrate that DHA induces genotoxicity through specific DNA adducts within the first 24 h of exposure. The type of adducts is dependent on cell type. No other significant DNA damage event, ROS, or AGEs was measured across the cell line panel.

DHA promotes chromosomal instability across the cell lines

The increase in polyploidy in the cell cycle analysis for BEAS-2B suggested chromosomal changes may also occur after DHA exposures. Therefore, we used metaphase spreads to examine chromosomal aberrations in the cells experiencing cell cycle arrest, BEAS-2B, A549, and HepG3 (Fig. 10). BEAS-2B and A549 have doubling times close to 24 h, so we dosed these cells with their IC₉₀ DHA dose for 48 h to examine aberrations after 2 doublings. The HepG3 has a doubling time of 40 h, so we dosed these cells with their IC₉₀ DHA dose for 96 h. Chromosomal spreads were prepared as described in the Materials and Methods section, and metaphase spreads were imaged. Each spread was categorized by type of aberration, chromosomal breaks, gaps, rings, chromatid exchanges, or centromere disruptions, as outlined in Registre and Proudlock (2016). The aberrations were then compared within treatment groups by 4 independent scorers. The lowest score per aberration type was selected from the grouped scores as the final score for each metaphase spread, and the total number of aberrations per cell was calculated. At least 70 spreads were scored per treatment group for each cell line.

Fig. 10.

Chromosomal aberrations increased across cell lines exposed to DHA. Metaphase spreads are used to measure chromosomal aberrations after DHA dosing. Aberrations were scored by the type of event: A: Chromatid Break, B: Chromatid Gap, C: Chromosome Ring, D: Chromatid Exchange, E: Centromere Disruption. Representative images for A) BEAS-2B, C) A549, and E) HepG3 are shown with aberrations indicated with the letter code noted above. The mean aberration counts for B) BEAS-2B, D) A549, and F) HepG3 are shown. The graph displays the total and categories of aberrations in DHA-dosed cells compared with control, mock-treated cells. Significance displayed as follows: *P < 0.05; **P < 0.01; ****P < 0.0001.

BEAS-2B exposed to 12.5 mM DHA for 48 h showed significantly increased chromosomal aberrations in every category (Fig. 10A). A significant increase in the mean total number of events per cell (0.86 ± 1.3 vs 7.6 ± 3.4) was also observed (Fig. 10B and Table S2). A549 dosed with 10 mM DHA for 48 h showed only a slight elevation in chromosomal rings and a significant decrease in centromere disruption (1.6 ± 2.0 vs 0.84 ± 1.1), resulting in no significant change in the mean total number of events per cell (Fig. 10C and D). HepG3 dosed with 14 mM DHA for 96 h showed elevated aberrations, with chromosomal rings and centromere disruption significantly increased in DHA-dosed cells (Fig. 10E). We also observed a significant decrease in other types of aberrations, including chromosomal breaks (1.1 ± 2.5 vs 0.30 ± 0.60), gaps (0.21 ± 0.45 vs 0.06 ± 0.23) and exchanges (0.31 ± 0.55 vs 0.06 ± 0.29) (Fig. 10F). Overall, an increase in the total mean number of events per cell (2.7 ± 3.0 vs 8.4 ± 4.5) was observed (Table S2).

Ploidy in these cells was also measured based on reported chromosome numbers for each cell line, BEAS-2B (n = 46), A549 (n = 66), and HepG3 (n = 55) (Fig. 11). BEAS-2B cells showed a decrease in 2n chromosome number and an increase in polyploidy cells (>2n) after DHA exposure (Fig. 11A). A549 cells showed a decrease in polyploidy cells (>2n) and an increase in 2n chromosome number after DHA exposure (Fig. 11B). HepG3 cells dosed with 14 mM DHA for 96 h showed an increase in aneuploidy cells (<2n), a decrease in 2n chromosome number, and a decrease in polyploid number (>2n) (Fig. 11C). We also noted the presence of micronuclei in some of the spreads. Although this assay is not specific for quantifying micronuclei, a small percentage of spreads showed micronuclei, consistent with our observed chromosomal changes.

Fig. 11.

The chromosomal count is also altered by DHA exposure within the cell panel. Chromosomes were counted for each cell line in the metaphase spreads and categorized as aneuploidy (<2n), normal (2n), or polyploidy (>2n). Ordinal scoring (1 = aneuploid, 2 = normal, and 3 = polyploid) was used to quantify the number of events per metaphase. The graph displays the frequency distribution over all the spreads for the DHA-exposed and control, mock-treated cells. A) BEAS-2B cells, B) A549 cells, and C) HepG3 cells.

Mutagenesis induced by DHA in HEK293T cells

Previous work in bacteria established DHA was mutagenic with and without metabolic activation (Pham et al. 1979). With the observed DNA adducts and chromosomal aberrations, we performed a forward mutation assay. The E. coli supF reporter system is widely used to measure forward mutation in bacterial and mammalian cells (Fukushima et al. 2020). The assay relies on the nucleotide mutations in the supF gene, which allows bacterial growth on agar plates under antibiotic selection. Recently, a more robust indicator strain of E. coli RF01 was developed to select for supF gene mutations on nutrient-rich agar plates using nalidixic acid and streptomycin, improving sensitivity and selectivity for mutations (Fukushima et al. 2020).

We measured the mutagenic potential of DHA in the nontumorigenic BEAS-2B cells using the supF reporter plasmid and the RF01 indicator strain. We dosed the BEAS-2B with 12.5 mM DHA or 100 µM MMS as a positive control for 48 h for 2 biological replicates, as shown in Fig. S14. An increase in mutations was observed in both BEAS-2B biological replicates compared with the control. However, the frequency of mutations was low for both DHA and MMS, leading to the inability to calculate the mutation frequency in these cells (Fig. S14A and B).

We wanted to confirm the mutagenic frequency, so we also transfected the supF reporter plasmid into the nontumorigenic HEK293T, which we previously characterized for DHA exposure (Smith et al. 2019). HEK293T cells were transfected with supF plasmid and treated with 10 mM DHA or 100 µM MMS (Fig. 12). In the HEK293T, we observed a larger number of colonies, consistent with the expectations for MMS exposures (White et al. 2019). We then compared the transfection efficiency of the supF-containing plasmid in the BEAS-2B and HEK293T cells, and we noted significantly diminished transfection efficiency for the BEAS-2B (∼30%) compared with the HEK293T (∼80%) using a GFP containing plasmid, which diminished the numbers of colonies for mutation screening. We could not increase the transfection efficiency for the BEAS-2B, so we proceeded with the mutagenic analysis using the HEK293T cells.

Fig. 12.

Increased supF mutation in HEK293T cells dosed with DHA. HEK293T cells were transfected with the supF reporter plasmid and then exposed to 10 mM DHA or 100 µM of methyl methanosulfonate (MMS) for 48 h. Plasmids were extracted, transformed into the indicator strain, and plated in parallel on the titer (A) and selection plates (B). C) Mutation frequency in supF is displayed in the graph and averaged over the 3 biological replicates. Significance displayed as follows: ****P < 0.0001.

Post-DHA treatment, the plasmid was extracted from the HEK293T cells and transformed into the RF01 cells. Transformed RF01 were spread onto LB agar plates containing streptomycin and nalidixic acid (selection plate) and allowed to grow for 18 to 20 h at 37 °C. The next day, colonies were counted on both titer and selection plates. The mutation frequency was calculated as described in the Materials and Methods section. Mock treatment of HEK293T cells produced a background mutation frequency of 6.1×10⁻⁶, well below previously reported results (Rocque et al. 2023). DHA and MMS treatment generated mutations in the supF gene, resulting in colony formation on the section plate (Fig. 12A and B). The mutation frequency after DHA and MMS treatment was significantly increased, with values calculated at 7.8×10⁻⁴ and 4.1×10⁻⁴, respectively (Fig. 12C). The data from both these cell lines demonstrates that DHA can have a mutagenic effect in mammalian cells for the first time.

Discussion

Since the discovery of DHA as a sunless tanning product and its approval to be used for sunless tanning products (STPs) by the FDA, there has been some debate about the genotoxicity of DHA (Petersen et al. 2004; Braunberger et al. 2018; Perer et al. 2020; Striz et al. 2021). Pham et al. (1979) demonstrated DHA in 2 STP lotions was mutagenic in Salmonella typhimurium strain TA100 without metabolic inactivation. This work also demonstrated DHA-induced DNA damage in a Bacillus subtilis rec-assay (Pham et al. 1979). Despite these early findings, there was limited follow-up data in animals or humans confirming the genotoxicity of DHA or understanding its mutagenic potential.

Most studies assumed the Maillard reaction, which produces DHA’s browning effect, limited exposure to the stratum corneum (Goldman et al. 1960; Levy 1992; SCCS 2010; Braunberger et al. 2018). Given the mutagenicity of ultraviolet light, STPs offered a safer alternative. The development of spray tanning in the late 1990s revived research into DHA, with several studies from 2004 until today demonstrating DHA is cytotoxic in skin models (Petersen et al. 2004; Jung et al. 2008; Perer et al. 2020; Striz et al. 2021). Petersen et al. (2004) inferred ROS-mediated DNA damage using a modified comet assay and the addition of antioxidant agents, which reduced the measured strand breaks. However, Striz et al. (2021) reported that alkaline comet assays showed no significant DNA damage below highly cytotoxic doses. Across DHA-exposed skin model papers, there is inconsistent evidence that DHA exposures generate significant levels of ROS or AGEs to produce genotoxic effects. There is also inconsistent evidence of DHA-inducing strand breaks through γH2AX or direct DNA damage (Gunko et al. 1975; Pham et al. 1979).

The identification of DHA within e-cigarette aerosol further raises concerns about inhalation exposures. Therefore, we have examined DHA exposures in systemic models HEK293T and HepG3 and determined DHA was cytotoxic at low millimolar (5 to 15 mM) doses in these cells (Smith et al. 2019; Hernandez et al. 2022). We also determined DHA promoted metabolic reprogramming and mitochondrial injury (Smith et al. 2019; Hernandez et al. 2022). However, these studies showed little evidence of genotoxicity beyond cycle arrest at G2/M for the HEK293T and weak cell cycle arrest at G1/S and G2/M for HepG3 (Smith et al. 2019; Hernandez et al. 2022).

This work aimed to bridge the gap in the literature to determine genotoxicity in systemic cell models relevant to potential inhalation exposures, 2 lung models (BEAS-2B and A549), a cardiomyocyte model (Ac16), and the metabolically active liver model HepG3. These models have been previously used to examine the cytotoxicity and genotoxicity of airborne environmental agents (Danielsen et al. 2008; Park et al. 2015; Zhang et al. 2016; Lu et al. 2023). Although these models have limitations due to their immortalized state (BEAS-2B and Ac16) and oncogenic transformation (A549 and HepG3), they offer a starting point for examining genotoxicity and comparing with the results of the existing skin models, which have also used immortalized cell models.

Similar to the HEK293T, low millimolar concentrations of DHA decreased cell growth and viability across the cell lines (Fig. 1). Interestingly, we also observed more variation in cell death mechanisms. Skin models routinely demonstrate G2/M cell cycle arrest and apoptotic cell death with cleaved PARP-1 and caspase 3 (Petersen et al. 2004; Smith et al. 2018; Perer et al. 2020; Striz et al. 2021). Only the BEAS-2B showed some evidence of apoptotic cell death, though only a minimal amount of cleaved PARP-1 was observed, and no cleaved caspase 3 was measured (Fig. 2A). A549 underwent autophagy similar to the previously characterized HEK293T, though notably, the A549 showed weak G1/S arrest, and the HEK293T showed G2/M arrest similar to the skin models (Figs 2B and 3B) (Smith et al. 2019). Finally, the Ac16 and HepG3 showed a hybrid cell death mechanism of apoptosis and autophagy, though signals for these pathways were weak (Fig. 2C and Fig. S5) (Hernandez et al. 2022).

Ac16 is the first cell line we have observed not to undergo cell cycle arrest after cytotoxic DHA exposure despite being more sensitive to DHA (Figs 1C and 3C). Unlike the previous studies in skin models and HEK293T, the cell cycle arrest is weak for BEAS-2B, A549, and HepG3 and varies at the arrest point, S, G1/S, and G1/S and G2/M, respectively (Fig. 3). These weak arrest points likely indicate replication stress is occurring across these 3 cell lines. More work is needed to determine the mechanism underlying the replication stress for each cell line, but it appears the source of stress varies in a cell-type-dependent manner.

To specifically confirm the genotoxicity of DHA, we examined strand break markers, measured reactive species, and measured specific DNA adduct types after DHA exposure. Similar to the work in skin models and HEK293T, we do not observe strand break signaling by γH2AX, which signals both single- and double-strand breaks (Fig. S10) (Petersen et al. 2004; Smith et al. 2018, 2019; Kopp et al. 2019; Perer et al. 2020; Striz et al. 2021; Hernandez et al. 2022). However, we also measured the recruitment of 53BP1, which can form foci at double breaks in cells (Panier and Boulton 2014). We observed a significant elevation of 53BP1 protein levels within the nucleus in the BEAS-2B cells 24, 48, and 72 h after DHA exposure (Fig. 5A). The 53BP1 levels were more mixed in the other cell lines, with Ac16 only showing elevation at 48 h, A549 only elevated at 72 h, consistent with the onset of cell death, and no change in the HepG3 (Fig. 5B–D). These results confirm strand break formation is weak after DHA exposure. The more significant elevation in 53BP1 in the BEAS-2B may correspond to the S-phase increases observed in the cell cycle analysis, confirming replication stress for this cell line (Fig. 3A).

Unlike previous results in the skin models, we did not observe significant elevation in ROS or AGEs within the first 24 or 48 h of DHA exposure (Figs 6 and 7) (Petersen et al. 2004; Perer et al. 2020; Striz et al. 2021). However, we did measure significantly increased oxidative DNA adducts for the BEAS-2B and Ac16 using RADD and significantly elevated crosslink-type adducts for the A549 and HepG3 (Fig. 8). RADD uses DNA cocktails to measure DNA lesion types, so we are limited to narrowing the identity of the lesions to the known specificity of the enzymes used. For the OXO cocktail, these enzymes recognize oxidized purines, urea, 5, 6-dihydroxythymine, thymine glycol, 5-hydroxy-5-methylhydantoin, 6-hydroxy-5,6-dihydrothymine, and abasic sites. The T4PDG cocktail is most commonly associated with ultraviolet crosslinks; in the context of DHA exposure, it may recognize DNA crosslinks generated by metabolic products or elevated abasic sites produced by metabolic products. Identifying the specific DNA lesions generated by DHA or its metabolites and recognized by the OXO and T4PDG cocktails requires more work and advanced methodologies. However, this is the first time specific classes of DNA adducts have been identified after DHA exposure, confirming genotoxicity.

We also examined if DPCs were generated after DHA exposure. In yeast, the metabolism of DHA has been suggested to produce MG and formaldehyde, though, like the mammalian systems, the generation of these products is highly dependent on the metabolic state of the yeast (Molin and Blomberg 2006; Zhang et al. 2013). Yeast has 2 specific enzymes for converting DHA to DHAP and other detoxification enzymes (Molin and Blomberg 2006). In humans, only TKFC has been identified to have this function, so human cells are less robust in their conversion of DHA to DHAP (Diao et al. 2007; Rodrigues et al. 2019; Wortmann et al. 2020). If generated, MG and aldehydes cause base substitutions, frameshift mutations, DNA breaks, chromosomal aberrations, and DPCs (Moretton and Loizou 2020). Consistent with our AGEs and MG results, we did not measure significant increases in total DPCs or TOPI-specific crosslinks (Figs S9 and S11). TOP2α-specific DPCs were decreased in some cell lines, which may indicate changes in the chromatin or topology of the chromatin.

Chromatin or chromosomal changes have not been previously examined after DHA exposures. However, Striz et al. (2021) observed down-regulation of aurora kinase A, BRCA1, cyclin B1, cyclin B2, CDC25C, CDK1, CKS2, and TOP2α 24 h after 5 mM DHA (IC₂₀) in human primary keratinocytes. These gene changes support chromosomal aberrations potentially induced by DHA exposure. The BEAS-2B showed a low population of altered ploidy cells in the control, mock-treated cells, which has been previously reported (Conery and Harlow 2010). Still, we observed an increase in this population after DHA exposure (Fig. 3A and Fig. S7), suggesting chromosomal instability.

BEAS-B cells showed a significant increase 48 h after IC₉₀ DHA exposure in all types of chromosomal aberrations (Fig. 10A and B). A significant increase in polyploidy cells is also confirmed in these cells (Fig. 11A). The 2 oncogenic cell lines had existing chromosomal aberrations and ploidy changes, and DHA exposure changed the composition of these events. DHA-exposed A549 cells showed fewer aberration events overall (Fig. 10C and D). There was also a chromosome number shift between aneuploidy and the normal chromosome number for A549, which has unknown consequences for the cell line (Fig. 11B). The HepG3 also showed altered aberration profiles after 96 h of DHA exposure, with a significant increase in total aberrations and events such as centromere disruptions (Fig. 10E and F). There were also increased aneuploidy cells in the HepG3 (Fig. 11C). We did not examine the Ac16 cells because they showed no evidence of cell cycle disruption (Fig. 3C).

Finally, we confirmed that DHA is mutagenic in the BEAS-2B and HEK293T. When we initially tested the BEAS-2B, we noted mutagenic colony formation, but the colony number was very low for DHA and MMS. Therefore, we also examined the HEK293T, which demonstrated colony formation with MMS to the predicted levels. We then examined the transfection efficiency between BEAS-2B and HEK293T and determined the transfection efficiency for BEAS-2B was too low for the supF reporter to calculate the mutation frequency (Fig. S14). Therefore, we proceeded with the HEK293T, where we previously demonstrated DHA’s cytotoxicity and measured replication stress, a lack of ROS, and limited γH2AX foci consistent with the cell lines examined here. Using the supF reporter with the HEK293T, we showed an increased mutation frequency 48 h after DHA exposure, consistent with exposures to known genotoxin MMS (Fig. 12). DHA is mutagenic in the HEK293T and has the potential to be mutagenic in other models. Further work will determine the DHA’s cell type-specific effects for mutagenesis.

Across the cell line panel, we observed that DHA induces genotoxicity in a cell-specific manner, likely dictated by the generation of currently unknown metabolites or reactive species. A limitation of this work is that we could not identify the reactive species being generated, but we have eliminated the commonly attributed mechanisms for these models. Future work can more specifically identify the driving reactive species. We also observed chromosomal aberrations for the first time, which again show features of cell type specificity. The drivers of chromosomal instability will also need to be confirmed in future work.

Author contributions

Conceptualization: AH, JH-D, SH, MS, NRG; Data Curation: AH, JH-D, SH, HL, MS, NRG; Formal Analysis: AH, JH-D, SH, HL, MS, NRG; Methodology: AH, JH-D, SH, MS, TS, HK, NRG; Writing—original draft: AH, JH-D, SH, NRG; Writing—review & editing: AH, JH-D, SH, HL, MS, TS, HK, NRG; Funding: NRG; Acquisition: NRG; Supervision: NRG; Validation: NRG.

Supplementary material

Supplementary material is available at Toxicological Sciences online.

Funding

This study was supported by NIH/NIEHS (R01 ES032450). HL is supported by R01 ES032450-04S1 and AH is supported by CCTS (TL1TR003106). The authors would also like to acknowledge the Flow Cytometry and Single Cell core facility (FCSC) and the support by the Center for AIDS Research (grant AI027767) and the O’Neal Comprehensive Cancer Center (grant CA013148) for the flow cytometry assistance.

Conflicts of interest

None declared.

Data availability

All relevant data are within the manuscript and its Supplementary Material files.