Contents lists available at ScienceDirect

Toxicology and Applied Pharmacology

journal homepage: www.elsevier.com/locate/taap

Early life exposure of a biocide, CMIT/MIT causes metabolic toxicity via theO-GlcNAc transferase pathway in the nematode C. elegans

Youngho Kim, Jinhee Choi⁎

School of Environmental Engineering, University of Seoul, 163 Siripdaero, Dongdaemun-gu, Seoul 02504, Republic of Korea

A R T I C L E I N F O

Keywords:CMIT/MITEarly-life exposureMetabolic toxicityO-GlcNAc transferaseHumidifier disinfectant

A B S T R A C T

Unusual cases of fatal lung injury, later determined to be a result of exposure to chemicals used as humidifierdisinfectants, were reported among Korean children from 2006 to 2011. This resulted in considerable study ofthe pulmonary toxicity of humidifier disinfectant chemicals to establish the causal relationship between ex-posure and lung disease. However, the systemic toxicity of the former and health effects other than lung diseaseare not fully understood. Here, we investigated the effect of 5-chloro-2-methyl-4-isothiazoline-3-one and 2-methyl-4-isothiazolin-3-one (CMIT/MIT), among the humidifier disinfectants used in the accidents, on the de-velopment of metabolic toxicity in the model organism, Caenorhabditis elegans using an exposure scenariocomparison. We screened the potential of CMIT/MIT to induce metabolic toxicity using C. elegans oga-1(ok1207)and ogt-1(ok1474) mutants. We also performed a pathway analysis based on C. elegans transcription factor RNAilibrary screening to identify the underlying toxicity mechanisms. Finally, to understand the critical window ofexposure for metabolic toxicity, responses to exposure during different periods in the life cycles of the wormswere compared. We determined that CMIT/MIT could induce metabolic toxicity through O-linked N-acet-ylglucosamine transferase and early life seems to be the critical window for exposure for metabolic toxicity forthis substance. The O-linked N-acetylglucosamine transferase pathway is conserved from worms to humans; ourresults thus insinuate that early-life exposure to CMIT/MIT could cause metabolic health problems during adultlife in humans. We therefore suggest that a systemic toxicity approach should be considered to comprehensivelyunderstand the adverse health effects of humidifier disinfectant misuse.

1. Introduction

From 2006 to 2011, epidemics of fatal lung injury with unknowncause were observed in Korean children every spring, the causal agentsof which were later determined to be chemical disinfectants that wereused in household humidifiers (Lee et al., 2012; Hong et al., 2014; Kimet al., 2014a; Kim et al., 2014b). These disinfectants contain [2-(2-ethoxy) ethoxyethyl] guanidinium chloride (PGH), poly-hexamethyleneguanidine (PHMG), 5-chloro-2-methylisothiazol-3-one(CMIT)/2-methylisothiazol-3-one (MIT), and didecyldimethylamm-oniumchloride (DDAC) (Lee et al., 2012; Kim et al., 2014b). Since then,intensive toxicological and epidemiological studies have been con-ducted on pulmonary disease using the substances used in the humi-difier disinfectants. However, the systemic toxicity thereof and healtheffects other than lung disease have not yet been fully understood.Moreover, although children were the most affected population (Kimet al., 2014b; Yoon et al., 2017), many humidifier disinfectant toxicitystudies have focused on chemical-specific toxicity without accounting

for the influence of developmental stages.A known mechanism of toxicity of isothiazolone biocides, which

include CMIT/MIT, is disruption of the metabolic pathways involvingdehydrogenase enzymes leading to the inhibition of critical physiolo-gical functions, such as oxygen consumption and adenosine tripho-sphate (ATP) synthesis (Williams, 2007). Indeed, CMIT/MIT has beenreported to induce apoptosis and necrosis because of mitochondrialimpairments through perturbation of calcium homeostasis from theendoplasmic reticulum in human cells (Di Stefano et al., 2006; Frosaliet al., 2009). As cell death may be controlled through necrotic orapoptotic mechanisms relying on the depletion of energy in the mi-tochondria, CMIT/MIT is likely to affect energy metabolism and mighthave an impact on metabolic toxicity.

The nematode Caenorhabditis elegans offers a unique opportunity toinvestigate the genetic and molecular functions of human disease-re-lated genes. It is known that approximately 42% of genes involved inhuman disease are conserved in the C. elegans genome (Markaki andTavernarakis, 2010). The full availability of genomic information and

https://doi.org/10.1016/j.taap.2019.05.012Received 3 April 2019; Received in revised form 28 April 2019; Accepted 13 May 2019

⁎ Corresponding author.E-mail address: jinhchoi@uos.ac.kr (J. Choi).

Toxicology and Applied Pharmacology 376 (2019) 1–8

Available online 14 May 20190041-008X/ © 2019 Published by Elsevier Inc.

T

resources facilitates the investigation of disease development mechan-isms. Furthermore, owing to its short life cycle (3 days to reach theadult stage), C. elegans is also ideal for studying the effect of early-lifeexposure to chemicals on health outcomes in later life stage.

O-GlcNAcylation, among the posttranslational modifications, iscontrolled by O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA).OGT plays a role in the addition of the O-linked GlcNAc molecule fromUDP-GlcNAc to serine and threonine residues in the target proteins, andOGA moves O-GlcNAc to the protein. These enzymes are known toregulate glucose metabolism, fatty acid metabolism, signal transduc-tions, insulin secretions, and the cell cycle (Ruan et al., 2013). Dysre-gulation of this pathway is known to be involved in the development oftype 2 diabetes (T2DM) (Hanover et al., 2010; Love and Hanover,2005). OGT and OGA pathways are known to be conserved in worms(Hanover et al., 2005; Forsythe et al., 2006). C. elegans oga-1 encodes anO-GlcNAcase. Like human, dysregulation of this gene in C. elegansshowed accumulate of O-GlcNAcylated protein (Forsythe et al., 2006).On the other hands, C. elegans ogt-1 encodes an O-GlcAc transferase,which induces depletion of protein O-GlcNAcylation (Hanover et al.,2005). Both strains showed increased carbohydrate, while decreasedlipid storage phenotypes (Hanover et al., 2005; Forsythe et al., 2006). Ithas been also reported that dysregulation of O-GlcNAc affected insulinlike signaling in C. elegans (Hanover et al., 2005; Forsythe et al., 2006).Taken together, abnormal function of the highly conserved C. elegansOGT and OGA leads to carbohydrate and lipid perturbation in C. ele-gans, similar to that seen in human T2DM, thus O-GlcNAc modificationmutants were considered as metabolic disease models (Hanover et al.,2005; Forsythe et al., 2006; Markaki and Tavernarakis, 2010). Wetherefore used C. elegans metabolic disease models, oga-1(ok1207) andogt-1(ok1474) mutants to determine whether CMIT/MIT exposurecauses metabolic toxicity.

In the present study, we investigated the effect of CMIT/MIT onmetabolic toxicity using the C. elegans metabolic disease model oga-1(ok1207) and ogt-1(ok1474) mutants. The results of screening re-vealed that CMIT/MIT toxicity was rescued in these mutants, sug-gesting the metabolic toxic potential of this chemical. To validatewhether OGT and OGA genes were involved in metabolic perturbationto CMIT/MIT exposure, ROS production and fatty acid levels were ex-amined. To further identify the mechanisms of toxicity of CMIT/MIT,pathway analysis based on C. elegans transcription factor RNAi libraryscreening was performed. To gain insights into the critical window ofexposure for the observed metabolic toxicity, in the second part of thisstudy, the responses of the worms to exposure to CMIT/MIT duringdifferent periods in their life cycles (i.e., PE, partial early-life exposure;L1–L2 stage; PL, partial late-life exposure; L3–YA; and WE, whole lifecycle exposure; L1–YA) was compared. The results showed that wormsexposed under the PE scenario had a more sensitive response than thoseexposed under the PL scenario. We then used the ogt-1(ok1474) mutantto investigate metabolic changes caused by CMIT/MIT, paying parti-cular attention to the PE exposure scenario. This study suggests thepossibility of metabolic disease developing in people who used CMIT/MIT as a humidifier disinfectant during the accident period and revealsthe necessity to monitor the long-term health of child victims of hu-midifier disinfectant accidents.

2. Materials and methods

2.1. C. elegans and chemical preparation

C. elegans were cultured in petri dishes on a nematode growthmedium (NGM) and fed OP50 strain Escherichia coli (E. coli) at 20 °Caccording to a standard protocol (Brenner, 1974). Worms were in-cubated at 20 °C and young adults (3 d old) from an age-synchronizedculture were used in all experiments. To produce the age-synchronizedcultures, eggs from mature adults were isolated using a 10% hypo-chlorite solution and rinsed with M9 buffer (4.2 mM Na2HPO4, 2.2mM

KH2PO4, 86mM NaCl, and 1mM MgSO4∙7H2O). The wild-type BristolN2 and oga-1(ok1207), ogt-1(ok1474) mutants were used in this study.All strains were provided by the Caenorhabditis Genetics Center (MN,USA).

CMIT/MIT was purchased from Sigma Aldrich (St. Louis, MO, USA).Stock solutions prepared in deuterium-depleted water were dilutedwith complete K-media (0.032M KCl, 0.051M NaCl, 1 mM CaCl2, 1 mMMgSO4, and 13mM cholesterol) and used as working solutions.

2.2. Exposure design

Three exposure conditions were examined. First, to determine theeffect of exposure during development, the effect of exposure across thewhole life cycle was investigated by exposing L1 synchronized wormsfor 60 h (WE). Next, the effect of CMIT/MIT on two different life stageswas compared. The effect on the early developmental stage was in-vestigated by exposing L1 synchronized worms for 20 h, which resultedin exposure of C. elegans in the L1 and L2 larval stages (PE). The effecton the later developmental stage was investigated by exposing wormsat 48 h after L1 synchronization for 20 h, which resulted in treatment ofC. elegans in the L3 and young adult stages (PL) (Fig. S1).

2.3. C. elegans reproduction assay

As previously described, the number of offspring of C. elegans wasinvestigated for 72 h as an indicator of reproduction (Roh et al., 2009).After exposure, age-synchronized young adult worms were transferredto 96-well plates and incubated for 72 h. The number of progeny wasdetermined as a measure of reproduction in each well using a complexobject parametric analyzer and sorter (COPAS; Union Biometrica,Holliston, MA, USA) and laser-based equipment to automate the ana-lysis of multiple endpoints for hundreds of C. elegans per minute (Hunt,2017). Twenty-four biological replications were performed for eachtreatment.

2.4. C. elegans lifespan assay

Lifespan analysis was conducted based on a modification of theprotocol described by (Solis and Petrascheck, 2011). The prepared L4stage worms were transferred into each well of a 96-well plate con-taining complete K-media, OP50, and 0.6mM Fluorodeoxyuridine(FUDR) and incubated at 20 °C. Scoring was conducted the next day andthis time point represented the first day of the lifespan analysis. Ob-servations were conducted once every five days until all worms weredead. Worms that did not respond to gentle prodding and displayed nopharyngeal pumping were scored as dead. Assays were carried out in 8-well replicates for each condition.

2.5. C. elegans transcription factor RNAi library screening

A 384 TFs RNAi library was purchased from Source Bioscience Inc.,UK. It accounts for 41% of the predicted transcription factors in C.elegans (Reece-Hoyes et al., 2005). The RNAi library was constructedusing an E. coli system on a 96-well plate. The full list of TF RNAi ispresented in Supplementary Table S1. RNAi screening was conductedbased on the protocol previously described with modification (Squibanet al., 2012). In this study, 384 TFs RNAi was inoculated in 96-deepwell plates containing Luria–Bertani (LB) media with 100 μg/mL ofampicillin using a fixed pins replicator and incubated at 27 °C over-night. The next day, 5mM isoprophylthiogalactoside (IPTG) was addedto each well, and the plate was incubated in a shaking incubator at27 °C for 1 h. The prepared 96-deep well plate was spun down and thesupernatant was discarded. The RNAi pellet was dissolved in 1mL ofcomplete K-media with 100 μg/mL of ampicillin and 5mM IPTG. Forthe screening, 100 μL of dissolved RNAi was separated into a controlgroup and chemical treatment group in each 96-well plate using a

Y. Kim and J. Choi Toxicology and Applied Pharmacology 376 (2019) 1–8

2

multi-channel pipet. Synchronized L1 stage worms were sorted for eachwell using COPAS (Union Biometrica, Holliston, MA, USA). Wormswere grown to the young adult stage and then exposed to 100 μL ofCMIT/MIT. The exposure concentration was selected that which re-duced N2 reproduction by approximately 20% over 72 h (CMIT/MIT:2mg/L). After 72 h, reproduction was investigated using COPAS. A 5-well replication was performed in each RNAi per group.

2.6. Pathway analysis

A list of RNAi clones with greater than two-fold changes (either anincrease or decrease) identified them as potential genes involved inCMIT/MIT toxicity. Based on the list of identified genes, pathways al-tered by CMIT/MIT exposure were analyzed using KEGG (Kanehisaet al., 2012).

2.7. Measurement of glucose content

Measurement of glucose content was conducted based on the pro-tocol previously described, with modifications (Park et al., 2012).Following exposure, worms were washed three times using K-mediaand frozen at −80 °C. The worm pellet was thawed on ice and re-suspended in 300 μL of glucose assay buffer. To obtain glucose, theworm pellet was homogenized using a glass homogenizer and thencentrifuged at 13,000 rpm and 4 °C. The supernatant was used for aglucose colorimetric assay in accordance with the manufacturer'smanual using a Glucose Assay Kit (Biovision K606-100, Mountain View,CA, USA). Absorbance intensity was measured using a Glomax™ Dis-cover System GM3000 (Promega, Madison, WI, USA) at 570 nm.

2.8. Nile red staining

Age-synchronized L4 stage worms were prepared for the experi-ments. The L4 stage worms were treated to 0.6 mM of FUDR to preventegg laying during the experiments (Mitchell et al., 1979). The wormswere exposed to chemicals with E. coli OP50 for 24 h and then in-cubated with 10 μg/mL Nile red (Sigma Aldrich, St. Louis, MO, USA) at20 °C for 1 h. After staining, worms were washed three times using K-media. Twenty worms were transferred using K-media into each well of96-well plate with opaque walls and a transparent bottom. Fluorescenceintensity was measured using a Glomax™ Discover System GM3000(Promega, Madison, WI, USA) at excitation/emission wavelengths of520 and 580–640 nm. Their staining phenotypes were assessed usingfluorescence microscopy (Leica DM 2500, Wetzlar, Germany).

2.9. Measurement of ATP contents

ATP concentration in the bodies of the C. elegans was measuredusing an ATP Colorimetric/Fluorometric Assay Kit (Biovision K354,Mountain View, CA, USA). The assays were completed in accordancewith the manufacturer's manual. The worms were homogenized using400 μL of ATP assay buffer and then centrifuged at 5000 rpm for 3min.Fifty microliters of supernatant was dispensed to a 96-well plate andallowed to react with the reaction mixture buffer containing the ATPprobe, ATP converter, and Developer mix. The plates were incubated atroom temperature (24–26 °C) for 30min. The absorbance was measuredat 570 nm using a Glomax™ Discover System GM3000 (Promega,Madison, WI, USA).

2.10. Behavior analysis

Computational behavior recording, tracking, and parameter calcu-lation were conducted according to the previously described metho-dology, with some modifications (Lee et al., 2015). The behavior re-cording system selected in this study consisted of Petri dishes on anNGM medium, a camera system (IMT cam3, i-Solution Inc., USA), and

an image processing system (Virtual Dub software, http://www.virtualdub.org/). After 5min of acclimation in the arena, the beha-vior of C. elegans was recorded for 3min using a digital microscopecamera. The movements of 20 worms were examined for each condi-tion. Based on the results previously described, a time segment of 0.25 swas selected for recording (Lee et al., 2015). This was sufficiently shortto observe the location of C. elegans. After location data were gatheredusing the Image J software, parameters and tracking images were cal-culated using the MATLAB software (The Mathworks, R2018). Ac-cording to a previous study, we calculated two parameters (i.e., speed(mm/s) and stop duration (t)) (Liu et al., 2011).

2.11. Data analysis

Three biological replicates were performed for all assays. All resultsare shown as mean ± SEM (standard error of the mean). Survivalanalysis was performed using the Log-rank test. Statistical analyseswere conducted using SPSS version 13.0 (SPSS Inc., Chicago, IL, USA).Dose responses and effective concentration (EC) were calculated using alogistic model fitted with the least squares optimization method usingthe “DRC” package in R (Ritz and Streibig, 2005). p values< .05 wereconsidered significantly different from the control.

3. Results and discussion

3.1. Identification of metabolic toxicity of CMIT/MIT using C. elegans O-GlcNAc modification defective mutants, oga-1(ok1207) and ogt-1(ok1474)

To gain an insight regarding the link between the occurrence ofmetabolic disease and exposure of CMIT/MIT, CMIT/MIT toxicity wasinvestigated using C. elegans oga-1(ok1207) and ogt-1(ok1474) mu-tants. As shown in Fig. 1, compared to N2 worms, an approxi-mately> 1.3-fold increase in reproduction potential was observed inthe oga-1(ok1207) and ogt-1(ok1474) mutants to CMIT/MIT exposure(p < .01). This result suggests the potential of CMIT/MIT to inducetoxicity through O-GlcNAc modification. Concentration–response ana-lysis showed that the calculated effective concentration (EC) 50 valueswere 4.46mg/L (95%; CI: 2.17–6.76), 5.11mg/L (95%; CI: 4.82–5.39)and 3.81mg/L (95%; CI: 0.93–6.69) on N2, oga-1(ok1207), and ogt-1(ok1474), respectively. We thus postulated that the O-GlcNAc mod-ification might be a mechanism underlying the metabolic toxicity of theCMIT/MIT exposure.

In Fig. 1, comparing the EC values calculated from the con-centration–response assay revealed that the response of oga-1(ok1207)to CMIT/MIT exposure was less sensitive than that of N2, whereas thatof ogt-1(ok1474) was more sensitive than that of N2 at 8mg/L below.The O-GlcNAc modification is reversibly controlled by the enzymatic

Fig. 1. Effect of CMIT/MIT on reproduction in C. elegans. Wild-type N2, oga-1(ok1207), and ogt-1(ok1474) were exposed to 1, 2, 4, and 8mg/L of CMIT/MIT for 72 h and the number of offspring were then measured. The derived ECvalues were obtained using the “DRC” package in R.

Y. Kim and J. Choi Toxicology and Applied Pharmacology 376 (2019) 1–8

3

activity of OGT and OGA (Ferrer et al., 2014). Therefore, our resultssuggest that the O-GlcNAc modification is involved in CMIT/MITtoxicity.

3.2. Oxidative stress and fatty acid accumulation through OGT via CMIT/MIT exposure in C. elegans

As oxidative stress and fatty acid accumulation are biomarkers inpatients with metabolic disease (Wakil and Abu-Elheiga, 2009;Mittendorfer, 2011; Ruan et al., 2013; Bonomini et al., 2015), to de-termine whether the involvement of O-GlcNAc modification in CMIT/MIT induces metabolic stress, oxidative stress, and fatty acid accumu-lation were investigated in CMIT/MIT-exposed C. elegans. Results areshown in Fig. 2A. The relative fluorescent intensities of both N2 andoga-1(ok1207) worms increased in a concentration-dependent mannercompared to those of the control group whereas CMIT/MIT-inducedROS formation was not observed in the ogt-1(ok1474) mutant. Pre-viously, Kim et al.(Kim et al., 2017) reported that inhibition of OGTdecreased ROS generation in neural stem cells and diminished oxidativestress in the embryos. Darley-Usmar et al. (2012) also reported thatincreased O-GlcNacylation of mitochondrial protein because of OGTleads to impaired mitochondrial function and generates ROS. Our resultindicates that O-GlcNAcylation via OGT is associated with an increasein ROS generation in C. elegans via CMIT/MIT exposure.

Next, we investigated whether CMIT/MIT exposure alters fatty acidaccumulation via O-GlcNAc modification. As shown in Fig. 2B, C, thefluorescent intensity of wild-type worms exposed to 2 and 4mg/L ofCMIT/MIT was significantly higher than that of the control group.However, such a trend was not observed in the oga-1(ok1207) and ogt-1(ok1474) mutants. Only oga-1(ok1207) exposed to 1mg/L of CMIT/MIT exhibited fatty acid accumulation. At the highest exposure

concentration (8mg/L), all strains showed a significant decrease influorescent intensity, which may reflect significant physiological dete-rioration because of high toxicity. These findings demonstrate that OGTinduced fatty acid accumulation in C. elegans following CMIT/MIT ex-posure. Despite the lack of dedicated adipocytes in C. elegans, increasedfat accumulation is a hallmark of the stress response and implicated inhuman obesity (Ashrafi, 2007). Furthermore, the increased O-GlcNA-cylation through hexosamine flux and overexpression of OGT are re-lated to metabolic diseases such as obesity and T2DM (McClain, 2002;Coomer and Essop, 2014). Taken together, our results suggest thatCMIT/MIT exposure could induce metabolic disease via O-GlcNAcyla-tion.

3.3. FOXO signaling pathway via CMIT/MIT exposure shown by C. eleganstranscription factor RNAi library screening

Evidence shows that O-GlcNAc modification is associated with theinsulin signaling pathway and insulin resistance and that this process ishighly conserved in C. elegans (Love and Hanover, 2005; Hanover et al.,2010). In C. elegans, a highly conserved insulin-like signaling pathwaymodulates stress response, immune response macronutrient storage,longevity, and Dauer formation (Love and Hanover, 2005). The role ofOGT in this pathway in C. elegans was identified using the OGTknockout mutant. OGT knockout inhibits the formation of Dauer larvae,which are induced by the insulin-like receptor gene daf-2; this indicatesthat OGT regulates the daf-2 receptor mediated signal pathway. Thesefindings demonstrate that OGT regulates stress response, macronutrientstorage, and Dauer formation in C. elegans and induces a phenotypesimilar to insulin resistance.

We found that CMIT/MIT toxicity is highly related to O-GlcNAcmodification, particularly via OGT. To gain an insight into the

Fig. 2. Effect of CMIT/MIT on metabolic stress in C. elegans. Under the PE scenario, C. elegans were exposed to 0.5 and 2mg/L concentrations. (A) The ROS level in itsentire body was measured. (B) The fatty acid level was investigated using Nile-red staining. (C) Staining phenotypes were observed via fluorescence microscopy. Thestatistical difference among groups was analyzed using one-way ANOVA followed by a Tukey post-hoc test (control= 1, n=3, mean ± SEM, p value: p * < 0.05, p** < 0.01, and p *** < 0.001).

Y. Kim and J. Choi Toxicology and Applied Pharmacology 376 (2019) 1–8

4

mechanism of observed metabolic toxicity induced by CMIT/MIT, weconducted C. elegans RNAi screening using a transcription factor RNAibacterial library. Exposure to CMIT/MIT led to a 20% decrease in re-production (response of empty vector, EV). Based on this, we identifiedRNAi clones that showed statistically significant differences (clonesshowing either rescued or exacerbated toxicity compared to that of EV).The selected list is shown in Table 1 (the full list is in SupplementaryTable 1).

Pathway analysis was then conducted to identify the specific toxicmechanisms associated with CMIT/MIT exposure using the KEGG basedon the RNAi clones belonging to rescued or exacerbated toxicity listedin Table 1. As a result, the FOXO signaling pathway was determined tobe associated with CMIT/MIT exposure, which seems to be related torecovery from CMIT/MIT toxicity. Interestingly, the involvement ofFOXO signaling, also called the insulin-like signaling pathway in thenematode, supports our results that CMIT/MIT induces metabolictoxicity via OGT.

3.4. Effect of early-life exposure of CMIT/MIT on metabolic toxicity viaOGT

The effects of early-life (i.e., in utero or childhood) exposure tochemicals on health during adulthood, known as the developmentalorigins of adult health and disease (DOHaD) (Silveira et al., 2007), hasbeen receiving much attention. It has been reported that early-lifechemical exposure is associated with development of metabolic diseaselater in life (Braun and Gray, 2017; De Long and Holloway, 2017). Forinstance, exposure to early bisphenol A in children is associated withincreased adiposity in children (Vafeiadi et al., 2016). Early-life ex-posure to di(2-ethylhexyl) phthalate is linked to excess adiposity inrodents (Wassenaar and Legler, 2017).

Once we confirmed CMIT/MIT caused O-GlcNAcylation-mediatedmetabolic toxicity to C. elegans, in the second part of this study, wesought to determine windows of susceptibility for CMIT/MIT-inducedmetabolic toxicity in C. elegans. We exposed C. elegans to CMIT/MITduring PE and during PL; their response was compared using re-production and lifespan (Fig. 3). In wild-type C. elegans, a decreasedreproduction and lifespan was observed with a severe degree at a highexposure concentration; however, worms under the PE scenario weremore susceptible to CMIT/MIT exposure than under PL in both end-points (Fig. 3). As far as the ogt-1(ok1474) mutant, no difference wasobserved between the PE and PL scenarios for reproduction (Fig. 3A)whereas for lifespan the PE scenario was more susceptible to CMT/MITthan the PL (Fig. 3B).

Some studies have showed that OGT genes on X chromosomes havebeen highly conserved and are involved in organism development.Webster et al. (Webster et al., 2009) reported that OGT over-expression

induced impaired zebrafish embryo development. In particular, Liuet al. (Liu et al., 2012) reported that OGT regulated brain developmentduring the early life stages of mice. Mariappa et al. (Mariappa et al.,2015) reported that OGT catalytic activity is needed for early-larvaldevelopment in Drosophila. Interestingly, these studies suggest thatOGT affects particularly the early-life stage of development, whichleads to attenuation of development and adulthood. Taking the litera-ture and our results together, exposure to CMIT/MIT during the early-life stage may result in a critical effect during later life through O-GlcNAc transferase.

As reproduction and lifespan analysis showed that PE is a moresusceptible window of exposure for CMIT/MIT, we then investigatedthe consequence of the PE scenario on the metabolic toxicity exerted byCMIT/MIT. OGT regulates both glucose and fat metabolism through O-GlcNAc modification of transcription fact (Ruan et al., 2013); our re-sults showed OGT-1 is involved in CMIT/MIT toxicity during early-lifeexposure. Therefore, we hypothesized that OGT affects nutrient meta-bolism during early-life exposure to CMIT/MIT. For determination, wemeasured the glucose and fatty acid level in CMIT/MIT-exposed N2 andogt-1(ok1474) under the PE scenario. Under the PE scenario, the whole-body glucose level decreased via CMIT/MIT exposure in N2 but not inogt-1(ok1474) (Fig. 4A) whereas no difference was observed in fattyacid contents in both N2 and ogt-1(ok1474) (Fig. 4B, C). AlthoughCMIT/MIT exposure showed an alteration of adult C. elegans fatty acidaccumulation through OGT, it does not appear during early-life ex-posure. Because O-GlcNAc modification has been well demonstrated toregulate glucose uptake and gluconeogenesis (Ruan et al., 2013) andwas shown under the PE scenario, our results suggest that OGT altersglucose metabolism during early-life exposure to CMIT/MIT in C. ele-gans.

3.5. Effect of early-life exposure of CMIT/MIT on the locomotive behaviorof C. elegans via OGT

Our window of exposure results collectively suggest that early-lifeexposure of CMIT/MIT led to a decrease in the glucose level, and inturn, affected the energy metabolism via OGT. Finally, to identify thephysiological consequences of early-life exposure to CMIT/MIT, weinvestigated the effect of CMIT/MIT on the C. elegans ATP level andlocomotion under the PE scenario. As shown in Fig. 5A, the ATP level ofN2 decreased by approximately 40% compared to control, while theogt-1(ok1474) did not significantly change.

We finally investigated locomotive behavior, as a physiologicalendpoint associated with energy consumption. In wild-type N2, thespeed of movement of the worms was significantly lower (p < .01),whereas the stop duration was higher than that of the control under thePE scenario. However, such a behavior alteration via CMIT/MIT

Table 1Selected list of RNAi candidates that caused exacerbated or rescued reproductive toxicity to CMIT/MIT exposure compared to EV-fed C. elegans (cut-off value > 2-fold). The full list is shown in Table S1. The statistical difference between the control group and treated group was analyzed using the two-tailed t-test (p value: p* < 0.05, p ** < 0.01, and p *** < 0.001). Two independent experiments were performed per condition with five worms per RNAi.

Sequence Gene Human ortholog Relative unit (Empty vector= 1) p value

List of RNAi leading to rescued toxicityY5H2B.h nhr-137 (Nuclear Hormone Receptor family) 2.93 0.22R06F6.6 ceh-62 (C. elegans Homeobox) 2.68 0.01F10C1.5 dmd-5 (Doublesex/MAB-3 Domain family) DMRT2 2.49 0.04H27C11.1 nhr-97 (Nuclear Hormone Receptor family) 2.47 0.02ZK6.1 nhr-278 (Nuclear Hormone Receptor family) 2.42 0.01F44C8.5 nhr-128 (Nuclear Hormone Receptor family) 2.18 0.03F59E11.10 nhr-195 (Nuclear Hormone Receptor family) MLPH 2.06 0.03

List of RNAi leading to exacerbated toxicityF54F7.1 taf-7.1 (TBP-associated transcription factor) family) TAF7 0.44 0.01F09C6.8 nhr-262 (Nuclear Hormone Receptor family) 0.45 0.01T14G12.4 fkh-2 (ForKHead transcription factor family) FOXG1 0.46 0.01ZC64.4 lim-4 (LIM domain family) LHX6 0.46 0.01

Y. Kim and J. Choi Toxicology and Applied Pharmacology 376 (2019) 1–8

5

exposure was annulled in ogt-1(ok1474) (Fig. 5B and C).Locomotion is closely related to energy consumption. Therefore, the

decreased ATP according to the impaired energy metabolism wouldlead to locomotive disability. In our results, glucose level as well as ATPlevel decreased via CMIT/MIT under the PE scenario and OGT seems tobe involved. Taken together, our results suggest that the OGT-decreasedglucose level subsequently decreased the ATP level, which finally mayhave had consequences on locomotive behavior. The overall result alsosuggests that early-life exposure of CMIT/MIT affects phenotypes re-lated to energy metabolism.

Although some previous studies have reported that CMIT/MIT ex-posure cause skin sensitization through immune response and apoptosis(Ettorre et al., 2003; Di Stefano et al., 2006; Isaksson et al., 2008; Williet al., 2011), very limited information is available regarding the dis-ease-causing possibility of CMIT/MIT, much less its underlying me-chanisms. Our study showed that CMIT/MIT exposure causes metabolictoxicity through the O-linked N-acetylglucosamine transferase. SinceOGT is a fundamental post-translational modification enzyme in cellsignaling pathway in metazoans, imbalanced O-GlcNAc modification is

well studied in development of many human disease, such as, cancer,diabetes, neurological disease and cardiovascular disease (Zhao et al.,2018). For example, elevated O-GlcNAcylation was reported to leadcolonic inflammation by modulating NF-κB signaling (Yang et al.,2015). OGT was reported to facilitate tumorigenesis by interacting withkey transcription factors that regulate the cell cycle, such as MYC, P53and SP1 transcription factors (Slawson and Hart, 2011). In addition,increased O-GlcNAc level was reported to act as an intermediate pro-cess in the insulin-AKT signaling, subsequently leading to regulation oftranscription factor, FOXO1 translocation and insulin resistance (Zhaoet al., 2018). This relationship between O-GlcNAc modification andFOXO signaling is also revealed in our study (Table 1). Although herewe investigated whether CMIT/MIT exposure is associated with meta-bolic toxicity, it is necessary to further investigate whether CMIT/MITtoxicity might lead to occurrence other diseases, because O-GlcNAcy-lation modification is involved in many signaling and disease devel-opment process, as described above. Moreover, as CMIT/MIT was asubstance used during the Korean humidifier disinfectant accident,findings from this toxicology study are worth translating to

Fig. 3. Windows of susceptibility to CMIT/MIT during C. elegans development. C. elegans were exposed to 0.5 and 2mg/L concentrations under the PE and PLscenarios. (A) The worm's reproduction and (B) lifespan were investigated in wild type N2 and ogt-1(ok1474). The statistical difference in reproduction among thegroups was analyzed using one-way ANOVA followed by a Tukey post hoc test (control= 1, n=3, mean ± SEM, p value: p * < 0.05, p ** < 0.01, and p*** < 0.001).

Fig. 4. Effect of early-life exposure to CMIT/MIT on metabolism in C. elegans. Under the PE scenario, C. elegans were exposed to 0.5 and 2mg/L concentrations. (A)The glucose contents and (B) fatty acid level in the whole body were measured in wild type N2 and ogt-1(ok1474). (C) Staining phenotypes were observed viafluorescence microscopy. The statistical difference among the groups was analyzed using one-way ANOVA followed by a Tukey post-hoc test (control= 1, n= 3,mean ± SEM, p value: p * < 0.05, p ** < 0.01, and p *** < 0.001).

Y. Kim and J. Choi Toxicology and Applied Pharmacology 376 (2019) 1–8

6

epidemiological study. Epidemiological studies of the Korean humidi-fier disinfectant accidents thus far mainly focused on patients with lungdisease. Related toxicology studies are also needed for lung toxicityusing animal models and there is very limited information on systemictoxicity.

Our study also showed the early life stage is a more susceptiblewindow of exposure for exerting metabolic toxicity of CMIT/MIT. Fromthe DoHAD point of view, early-life exposure to chemicals is a veryimportant environmental health issue. Infants and children were thehighly affected groups during the Korean humidifier disinfectant acci-dents. Our results also suggest special care is needed for these groupsvia monitoring their health during their development and later in life.

4. Conclusion

Our results provide evidence that early-life exposure to CMIT/MITcould cause metabolic health problem in later life stage. Based on ourresults, we suggest that the study of humidifier disinfectant toxicityshould be considered via the systemic toxicity and early-life exposureconcepts. In addition to being an active substance in the Korean hu-midifier disinfectant accident, CMIT/MIT is among the most commonbiocides used in products such as personal cleaners, cosmetics, andpreservatives. Therefore, its systemic toxicity and disease-causing po-tential should be investigated thoroughly.

Acknowledgements

This work was supported by the Mid-Career Researcher Program(2017R1A2B3002242) through the National Research Foundation ofKorea funded by the Ministry of Science and ICT.

Conflict of interestThe authors declare that there are no conflicts of interest.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.taap.2019.05.012.

References

Ashrafi, K., 2007. Obesity and the regulation of fat metabolism. In: WormBook, pp. 1–20.Bonomini, F., Rodella, L.F., Rezzani, R., 2015. Metabolic syndrome, aging and involve-

ment of oxidative stress. Aging Dis. 6, 109–120.Braun, J.M., Gray, K., 2017. Challenges to studying the health effects of early life en-

vironmental chemical exposures on children's health. PLoS Biol. 15, e2002800.Brenner, S., 1974. The genetics of Caenorhabditis elegans. Genetics 77, 71–94.

Coomer, M., Essop, M.F., 2014. Differential hexosamine biosynthetic pathway gene ex-pression with type 2 diabetes. Mol. Genet. Metab. Rep. 1, 158–169.

Darley-Usmar, V.M., Ball, L.E., Chatham, J.C., 2012. Protein O-linked beta-N-acet-ylglucosamine: a novel effector of cardiomyocyte metabolism and function. J. Mol.Cell. Cardiol. 52, 538–549.

De Long, N.E., Holloway, A.C., 2017. Early-life chemical exposures and risk of metabolicsyndrome. Diabet. Metabol. Syndr. Obesity 10, 101–109.

Di Stefano, A., Frosali, S., Leonini, A., Ettorre, A., Priora, R., Di Simplicio, F.C., DiSimplicio, P., 2006. GSH depletion, protein S-glutathionylation and mitochondrialtransmembrane potential hyperpolarization are early events in initiation of cell deathinduced by a mixture of isothiazolinones in HL60 cells. Biochim. Biophys. Acta 1763,214–225.

Ettorre, A., Andreassi, M., Anselmi, C., Neri, P., Andreassi, L., Di Stefano, A., 2003.Involvement of oxidative stress in apoptosis induced by a mixture of isothiazolinonesin normal human keratinocytes. J. Invest. Dermatol. 121, 328–336.

Ferrer, C.M., Lynch, T.P., Sodi, V.L., Falcone, J.N., Schwab, L.P., Peacock, D.L., Vocadlo,D.J., Seagroves, T.N., Reginato, M.J., 2014. O-GlcNAcylation regulates cancer me-tabolism and survival stress signaling via regulation of the HIF-1 pathway. Mol. Cell54, 820–831.

Forsythe, M.E., Love, D.C., Lazarus, B.D., Kim, E.J., Prinz, W.A., Ashwell, G., Krause,M.W., Hanover, J.A., 2006. Caenorhabditis elegans ortholog of a diabetes suscept-ibility locus: oga-1 (O-GlcNAcase) knockout impacts O-GlcNAc cycling, metabolism,and dauer. Proc. Natl. Acad. Sci. U. S. A. 103, 11952–11957.

Frosali, S., Leonini, A., Ettorre, A., Di Maio, G., Nuti, S., Tavarini, S., Di Simplicio, P., DiStefano, A., 2009. Role of intracellular calcium and S-glutathionylation in cell deathinduced by a mixture of isothiazolinones in HL60 cells. Biochim. Biophys. Acta 1793,572–583.

Hanover, J.A., Forsythe, M.E., Hennessey, P.T., Brodigan, T.M., Love, D.C., Ashwell, G.,Krause, M., 2005. A Caenorhabditis elegans model of insulin resistance: alteredmacronutrient storage and dauer formation in an OGT-1 knockout. Proc. Natl. Acad.Sci. U. S. A. 102, 11266–11271.

Hanover, J.A., Krause, M.W., Love, D.C., 2010. The hexosamine signaling pathway: O-GlcNAc cycling in feast or famine. Biochim. Biophys. Acta 1800, 80–95.

Hong, S.B., Kim, H.J., Huh, J.W., Do, K.H., Jang, S.J., Song, J.S., Choi, S.J., Heo, Y., Kim,Y.B., Lim, C.M., Chae, E.J., Lee, H., Jung, M., Lee, K., Lee, M.S., Koh, Y., KoreanUnknown Severe Respiratory Failure, C, Korean Study Group of Respiratory, F, 2014.A cluster of lung injury associated with home humidifier use: clinical, radiologicaland pathological description of a new syndrome. Thorax 69, 694–702.

Hunt, P.R., 2017. The C. elegans model in toxicity testing. J. Appl. Toxicol. 37, 50–59.Isaksson, M., Bruze, M., Gruvberger, B., 2008. Cross-reactivity between methyl-

chloroisothiazolinone/methylisothiazolinone, methylisothiazolinone, and other iso-thiazolinones in workers at a plant producing binders for paints and glues. ContactDermatitis 58, 60–U68.

Kanehisa, M., Goto, S., Sato, Y., Furumichi, M., Tanabe, M., 2012. KEGG for integrationand interpretation of large-scale molecular data sets. Nucleic Acids Res. 40,D109–D114.

Kim, H.J., Lee, M.S., Hong, S.B., Huh, J.W., Do, K.H., Jang, S.J., Lim, C.M., Chae, E.J.,Lee, H., Jung, M., Park, Y.J., Park, J.H., Kwon, G.Y., Gwack, J., Youn, S.K., Kwon,J.W., Yang, B.G., Jun, B.Y., Kim, Y., Cheong, H.K., Chun, B.C., Kim, H., Lee, K., Koh,Y., 2014a. A cluster of lung injury cases associated with home humidifier use: anepidemiological investigation. Thorax 69, 703–708.

Kim, K.W., Ahn, K., Yang, H.J., Lee, S., Park, J.D., Kim, W.K., Kim, J.T., Kim, H.H., Rha,Y.H., Park, Y.M., Sohn, M.H., Oh, J.W., Lee, H.R., Lim, D.H., Choung, J.T., Han, M.Y.,Lee, E., Kim, H.Y., Seo, J.H., Kim, B.J., Cho, Y.A., Do, K.H., Kim, S.A., Jang, S.J., Lee,M.S., Kim, H.J., Kwon, G.Y., Park, J.H., Gwack, J., Youn, S.K., Kwon, J.W., Jun, B.Y.,Pyun, B.Y., Hong, S.J., 2014b. Humidifier disinfectant-associated Children's inter-stitial lung disease. Am. J. Resp. Crit. Care 189, 48–56.

Kim, G., Cao, L., Reece, E.A., Zhao, Z., 2017. Impact of protein O-GlcNAcylation on neuraltube malformation in diabetic embryopathy. Sci. Rep. 7, 11107.

Fig. 5. Effect of early-life exposure to CMIT/MIT on the locomotion of C. elegans. In PE scenario, C. elegans were exposed to 0.5 and 2mg/L concentration. (A) TheATP level in the whole body was measured and (B) locomotion behavior was investigated in (A) wild type N2 and ogt-1(ok1474). Speed and stop duration wereobserved as a locomotion behavior endpoint. (C) The worm's moving path were schematized. The statistical difference among groups was analyzed using one-wayANOVA followed by Tukey post-hoc test (control= 1, n= 3, mean ± SEM, p value: p * < 0.05, p ** < 0.01, and p *** < 0.001).

Y. Kim and J. Choi Toxicology and Applied Pharmacology 376 (2019) 1–8

7

Lee, J.H., Kim, Y.H., Kwon, J.H., 2012. Fatal misuse of humidifier disinfectants in Korea:importance of screening risk assessment and implications for management of che-micals in consumer products. Environ. Sci. Technol. 46, 2498–2500.

Lee, S.B., Choe, Y., Chon, T.S., Kang, H.Y., 2015. Analysis of zebrafish (Danio rerio) be-havior in response to bacterial infection using a self-organizing map. BMC Vet.Res. 11.

Liu, Y., Lee, S.H., Chon, T.S., 2011. Analysis of behavioral changes of zebrafish (Daniorerio) in response to formaldehyde using self-organizing map and a hidden Markovmodel. Ecol. Model. 222, 2191–2201.

Liu, Y., Li, X., Yu, Y., Shi, J., Liang, Z., Run, X., Li, Y., Dai, C.L., Grundke-Iqbal, I., Iqbal,K., Liu, F., Gong, C.X., 2012. Developmental regulation of protein O-GlcNAcylation,O-GlcNAc transferase, and O-GlcNAcase in mammalian brain. PLoS One 7, e43724.

Love, D.C., Hanover, J.A., 2005. The hexosamine signaling pathway: deciphering the "O-GlcNAc code". Sci. STKE 2005 (re13).

Mariappa, D., Zheng, X., Schimpl, M., Raimi, O., Ferenbach, A.T., Muller, H.A., vanAalten, D.M., 2015. Dual functionality of O-GlcNAc transferase is required forDrosophila development. Open Biol. 5, 150234.

Markaki, M., Tavernarakis, N., 2010. Modeling human diseases in Caenorhabditis ele-gans. Biotechnol. J. 5, 1261–1276.

McClain, D.A., 2002. Hexosamines as mediators of nutrient sensing and regulation indiabetes. J. Diabetes Complicat. 16, 72–80.

Mitchell, D.H., Stiles, J.W., Santelli, J., Sanadi, D.R., 1979. Synchronous growth andaging of Caenorhabditis elegans in the presence of fluorodeoxyuridine. J. Gerontol.34, 28–36.

Mittendorfer, B., 2011. Origins of metabolic complications in obesity: adipose tissue andfree fatty acid trafficking. Curr. Opin. Clin. Nutr. Metabol. Care 14, 535–541.

Park, D., Jones, K.L., Lee, H., Snutch, T.P., Taubert, S., Riddle, D.L., 2012. Repression of apotassium channel by nuclear hormone receptor and TGF-beta signaling modulatesinsulin signaling in Caenorhabditis elegans. PLoS Genet. 8, e1002519.

Reece-Hoyes, J.S., Deplancke, B., Shingles, J., Grove, C.A., Hope, I.A., Walhout, A.J.,2005. A compendium of Caenorhabditis elegans regulatory transcription factors: aresource for mapping transcription regulatory networks. Genome Biol. 6, R110.

Ritz, C., Streibig, J.C., 2005. Bioassay analysis using R. J. Stat. Softw. 12, 1–22.Roh, J.Y., Sim, S.J., Yi, J., Park, K., Chung, K.H., Ryu, D.Y., Choi, J., 2009. Ecotoxicity of

silver nanoparticles on the soil nematode Caenorhabditis elegans using functionalecotoxicogenomics. Environ. Sci. Technol. 43, 3933–3940.

Ruan, H.B., Singh, J.P., Li, M.D., Wu, J., Yang, X., 2013. Cracking the O-GlcNAc code in

metabolism. Trends Endocrinol. Metab. 24, 301–309.Silveira, P.P., Portella, A.K., Goldani, M.Z., Barbieri, M.A., 2007. Developmental origins

of health and disease (DOHaD). J. Pediatr. 83, 494–504.Slawson, C., Hart, G.W., 2011. O-GlcNAc signalling: implications for cancer cell biology.

Nat. Rev. Cancer 11, 678–684.Solis, G.M., Petrascheck, M., 2011. Measuring Caenorhabditis elegans life span in 96 well

microtiter plates. J. Vis. Exp. 49, 2496.Squiban, B., Belougne, J., Ewbank, J., Zugasti, O., 2012. Quantitative and automated

high-throughput genome-wide RNAi screens in C. elegans. J. Vis. Exp. 60, 3448.Vafeiadi, M., Roumeliotaki, T., Myridakis, A., Chalkiadaki, G., Fthenou, E., Dermitzaki,

E., Karachaliou, M., Sarri, K., Vassilaki, M., Stephanou, E.G., Kogevinas, M., Chatzi,L., 2016. Association of early life exposure to bisphenol A with obesity and cardio-metabolic traits in childhood. Environ. Res. 146, 379–387.

Wakil, S.J., Abu-Elheiga, L.A., 2009. Fatty acid metabolism: target for metabolic syn-drome. J. Lipid Res. 50 (Suppl), S138–S143.

Wassenaar, P.N.H., Legler, J., 2017. Systematic review and meta-analysis of early lifeexposure to di(2-ethylhexyl) phthalate and obesity related outcomes in rodents.Chemosphere 188, 174–181.

Webster, D.M., Teo, C.F., Sun, Y., Wloga, D., Gay, S., Klonowski, K.D., Wells, L., Dougan,S.T., 2009. O-GlcNAc modifications regulate cell survival and epiboly during zebra-fish development. BMC Dev. Biol. 9, 28.

Willi, R., Pfab, F., Zilker, T., Buters, J., Schalock, P., Huss-Marp, J., Todorova, A., Ring, J.,Darsow, U., 2011. Danger from the workplace: allergic contact dermatitis from thefirst exposure to isothiazolinones. Contact Dermatitis 64, 361–362.

Williams, T.M., 2007. The mechanism of action of isothiazolone biocides. PowerPlantChem. 9.

Yang, Y.R., Kim, D.H., Seo, Y.K., Park, D., Jang, H.J., Choi, S.Y., Lee, Y.H., Lee, G.H.,Nakajima, K., Taniguchi, N., Kim, J.M., Choi, E.J., Moon, H.Y., Kim, I.S., Choi, J.H.,Lee, H., Ryu, S.H., Cocco, L., Suh, P.G., 2015. Elevated O-GlcNAcylation promotescolonic inflammation and tumorigenesis by modulating NF-kappa B signaling.Oncotarget 6, 12529–12542.

Yoon, J., Cho, H.J., Lee, E., Choi, Y.J., Kim, Y.H., Lee, J.L., Lee, Y.J., Hong, S.J., 2017.Rate of humidifier and humidifier disinfectant usage in Korean children: a nationwideepidemiologic study. Environ. Res. 155, 60–63.

Zhao, L., Shah, J.A., Cai, Y., Jin, J., 2018. ‘O-GlcNAc Code’ mediated biological functionsof downstream proteins. Molecules 23.

Y. Kim and J. Choi Toxicology and Applied Pharmacology 376 (2019) 1–8

8