Conserved Molecular Responses to Arsenite Exposure in Drosophila melanogaster

Abstract

Arsenic exposure is a pervasive global health threat strongly associated with increased risk of morbidities such as diabetes, cardiovascular disease, and cancer. Despite extensive studies describing the dangers of arsenic exposure, the molecular initiating events that link arsenic to chronic disease onset and progression remain poorly defined. To address this knowledge gap, we combined time-resolved transcriptomic and metabolomic profiling of adult Drosophila melanogaster exposed to sodium (meta) arsenite (NaAsO₂). We uncovered coordinated, dose-dependent shifts in gene expression and metabolite abundance that activate canonical detoxification pathways and mirror arsenic-associated disease signatures in humans. Notably, flies rapidly upregulated heatshock and xenobiotic response gene networks, followed by biomarkers characteristic of diabetic states (elevated glucose, lactate, and methylglyoxal, for example). These findings reveal conserved molecular pathways that couple arsenic exposure to metabolic dysfunction and establish Drosophila as a powerful whole-organism model for identifying early biomarkers and mechanistic drivers of arsenic-induced disease phenotypes.

INTRODUCTION

Over 200 million people worldwide are exposed to arsenic through multiple environmental and dietary sources, posing a substantial global public-health concern (Organization, 2022, Podgorski and Berg, 2020). Arsenic contamination is especially severe in the heavily populated regions of Bangladesh and West Bengal, where naturally high concentrations occur in soil and groundwater (Chowdhury et al., 2000, Mazumder et al., 2010, Ivy et al., 2023). Similarly, arsenic contamination is a major health hazard for the western and midwestern regions of the United States, as well as parts of northern New England, where groundwater in millions of domestic wells exceeds the World Health Organization (WHO) drinking water guideline of 10 μg/L (Organization, 2022, Ayotte et al., 2017).

In addition to drinking water, the global food-supply system distributes arsenic exposure far beyond these regions, adding to overall intake and increasing cumulative exposure risk. Rice (Oryza sativa L.) is particularly vulnerable to arsenic contamination, often containing concentrations up to ten times higher than other domesticated crops such as wheat or corn (Su et al., 2010, Upadhyay et al., 2019). Consequently, populations that rely on rice as a staple food may consume significant quantities of arsenic, as do infants and young children fed rice-based baby formula and baby foods (Signes-Pastor et al., 2016, Parker et al., 2022, Wang et al., 2025). The risk of arsenic exposure similarly extends to populations that consume large amounts of seafood, as some marine species, such as shellfish, bioaccumulate arsenic at levels that can substantially exceed those found in most terrestrial foods, in some cases approaching or exceeding levels associated with adverse health outcomes (Taylor et al., 2017). Importantly, these dietary exposures add to arsenic intake from drinking water and other sources, increasing cumulative exposure risk. Together, these diverse and overlapping exposure routes complicate efforts to mitigate arsenic risk at its source and underscore the need to understand how arsenic perturbs conserved biological pathways. A deeper understanding of the molecular responses to arsenic exposure is therefore critical for identifying biomarkers and molecular initiating events that precede the development of disease phenotypes associated with arsenic exposure.

Arsenic exposure results in pleiotropic health effects, as both acute and chronic exposures affect multiple organ systems and increase the risk of serious disease. The mechanisms underlying acute arsenic toxicity are relatively well defined and include disruption of mitochondrial oxidative phosphorylation, leading to decreased ATP production, increased reactive oxygen species, and activation of cell death pathways (Hughes, 2002). In contrast, chronic arsenic exposure, which more commonly occurs through ingestion of contaminated water, food, or air, is associated with a broad spectrum of long-term health outcomes, including disorders of the skin, lungs, liver, gastrointestinal tract, cardiovascular system, kidneys, and nervous system (Jomova et al., 2011, Singh et al., 2011, Tchounwou et al., 2019). Moreover, prolonged exposure significantly raises the risk of developing diabetes mellitus as well as cancers of the skin, liver, lungs, urinary bladder, and kidney (Lai et al., 1994, Ratnaike, 2003, Sharma and Sohn, 2009, Hughes, 2002, Singh et al., 2015). Thus, while the cellular consequences of acute arsenic toxicity are well characterized, a deeper understanding of the molecular initiating events and adaptive responses that link arsenic exposure to chronic disease onset and progression remains a critical unmet need. Given the widespread nature of arsenic exposure and its diverse health impacts, new genetic models are essential for rapidly dissecting the molecular mechanisms underlying systemic arsenic toxicity in a cost-efficient and ethical manner.

The fruit fly Drosophila melanogaster is an important genetic model to investigate the biological effects of arsenic exposure on animals, including humans, with previous studies examining how arsenic disrupts development, neurobiology, metabolism, and fertility (Goldstein and Babich, 1989, Ramos-Morales and Rodriguez-Arnaiz, 1995, Adebambo et al., 2024, Anushree et al., 2023, Anushree et al., 2022, Rizki et al., 2002, Rizki et al., 2006, Bahadorani and Hilliker, 2009). In addition, transcriptomic and population genetic studies in the fly have identified phylogenetically conserved mechanisms that influence arsenic sensitivity, underscoring the power of using Drosophila to identify and characterize the molecular mechanisms that control toxicity responses (Muñiz Ortiz et al., 2009, Muñiz Ortiz et al., 2011, Rand, 2010, Rand et al., 2023, Sykiotis and Bohmann, 2008, Vincent and Tanguay, 1982, Chaturvedi et al., 2025). However, much of this work has focused on individual genes or single exposure time points, leaving the temporal dynamics of arsenic-induced changes in gene expression and metabolite levels—particularly during the earliest stages of exposure—poorly defined. Because these immediate molecular responses likely represent initiating events that precede overt pathology, resolving how transcriptional and metabolic programs evolve over time is essential for understanding how arsenic exposure progresses toward disease states. To address this gap, we applied a multi-omic approach to track how the molecular response to arsenic evolves over time. Using semi-targeted UHPLC-MS-based metabolomics and RNA-seq, we comprehensively characterized transcriptomic and metabolic changes in Drosophila following NaAsO₂ exposure across a 48-hour time course.

This integrated analysis revealed dose- and time-dependent shifts in gene expression and metabolite abundance that mirror patterns observed in mammalian systems and align with molecular signatures previously associated with arsenic-related human disease, indicating that these responses are evolutionarily conserved. Notably, we observed enrichment of differentially expressed genes linked to diabetes mellitus across sexes and exposure levels, even during this short-term exposure paradigm. Consistent with these transcriptional changes, metabolomic profiling uncovered disruptions in carbohydrate metabolism and purine degradation that resemble metabolic features commonly associated with diabetic states. Together, these findings highlight conserved molecular responses that emerge rapidly following arsenite exposure and underscore the value of Drosophila as a model for identifying early biomarkers and molecular initiating events associated with chronic arsenic-related disease, rather than overt disease phenotypes themselves.

METHODS

Drosophila Genetics and Husbandry

Fly stocks were maintained on Bloomington Drosophila Stock Center (BDSC) media (https://bdsc.indiana.edu/information/recipes/bloomfood.html) at 25°C. Unless noted, all studies described herein used Oregon-R (RRID:BDSC_25211). Flybase (https://flybase.org/) was utilized as a resource throughout these studies (Öztürk-Çolak et al., 2024).

Dose-response curves

A stock solution of sodium (meta)arsenite (NaAsO₂, Sigma-Aldrich, St. Louis, MO) was prepared by dissolving an accurately massed portion of NaAsO₂ into a volumetrically measured amount Milli-Q water. The solution was stirred for 15 minutes until complete dissolution and titrated to neutral pH (~7.5) with concentrated HCl. Dose-response curves were performed as previously described (Holsopple et al., 2023). Briefly, bottle cultures containing Oregon-R male and female flies were allowed to eclose for two days at 25°C, and the resulting adults were transferred to bottles containing fresh BDSC food. Following a two-day incubation at 25°C, adult male and female flies were sorted by sex and aged in vials of BDSC food for an additional 48 hours at 25°C to allow for recovery from the CO₂ anesthesia used for sorting. Flies within individual vials were then transferred without anesthesia to starvation vials (Whatman filter paper No. 1 soaked in sterile milli-Q water) for 16 hours at 25°C. Following the starvation period, flies were transferred without anesthesia to exposure vials containing liquid food [4% sucrose (m/v), 1.5% yeast extract (m/v)] and the noted concentration of NaAsO₂. The number of dead female or male flies per vial was counted in each vial at 24 and 48 hours. A total of six replicate vials were scored for each concentration. Dose-response curves were calculated using the frequentist dichotomous Hill model in Benchmark Dose Software (BMDS), published by the U.S. Environmental Protection Agency (Agency, 2023). The dose response curves between males and females were analyzed with GraphPad Prism 10.5.0 using the Dose-Response Special ECanything equation.

Sample Collection

NaAsO₂-exposed flies and unexposed controls were prepared for LC-MS and RNA-seq using the same protocol described for the dose-response curves (Holsopple et al., 2023). Briefly, following the overnight starvation, vials of 20 male or female flies were transferred into a vial containing liquid media supplemented with 0 mM, 0.25 mM, or 1.0 mM NaAsO₂. Samples were then collected at 1, 2, 4, 8, 24, and 48 hrs by individually transferring the flies from the exposure vials to empty fly culture vials, which were then immediately placed on dry ice. Once immobilized, the number of flies in individual vials was counted and subsequently transferred to a tared 2 mL screw cap tube containing 1.4 mm ceramic beads (Fisher Brand; 15-340-153). The cap was screwed onto the tube, the sample mass measured using a Radwag AS 82/220.X2 Plus analytical balance, and the closed tube immediately placed in liquid nitrogen. Frozen tubes were stored in an −80°C freezer.

RNA preparation

Briefly, bead tubes containing either control or exposed flies were transferred from the −80°C freezer into a tube rack surrounded by dry ice. 800 μL of Trizol reagent (Invitrogen; 15596018) was added to each tube and the sample was homogenized in a 4°C room for 30 secs at 6.45 m/sec using an Omni Beadruptor 24. Homogenized samples were removed from the instrument and immediately placed on ice for 2 min, then allowed to incubate for 5 min at room temperature. After incubation, 160 μL of chloroform was added to each tube and vortexed for 15 seconds. Samples were then incubated at room temperature for 15 min and subsequently centrifuged at 12,000 × g for 15 min at 4°C. Following centrifugation, 300 μL of the upper aqueous phase was transferred to an RNase-free 1.5 mL microfuge tube containing 400 μL of isopropanol, mixed by inverting, and centrifuged at 12,000 × g for 15 min at 4°C. The isopropanol was removed from each sample using a 200 μL pipette while being careful not to disturb the RNA pellet. 1 mL of 75% ethanol was added to each tube and samples were centrifuged at 12,000 × g for 15 min at 4°C. The ethanol was removed from each sample using a 200 μL pipette while being careful not to disturb the white RNA pellet. The samples were allowed to briefly air dry to completely remove the remaining ethanol. Samples were resuspended in 50 μL of RNAase free water and stored in a −80°C freezer. Resuspended RNA samples were subsequently cleaned using a Qiagen RNeasy Mini Kit per manufacturer’s instructions (Qiagen; 74104).

RNA sequencing

Libraries of these 114 samples, 57 of each sex, were generated with the Illumina TruSeq Stranded mRNA HT protocol and were loaded evenly on the NextSeq 2000 P3 platform and run for 100 cycles, with a target of ≈8.8 million 2×50 paired-end reads per sample. Any samples with < 9 million reads were topped up in a further run, with additional reads concatenated to their respective samples. Raw reads and processed counts are available on NCBI GEO at accession GSE241663. All library preparation and sequencing were performed at the IU Center for Genomics and Bioinformatics (CGB).

Transcriptomic Analysis

RNA-seq read quality was assessed with FastQC (Andrews, 2010) and MultiQC (Ewels et al., 2016); raw reads were not trimmed or filtered. Reads were quasi-aligned and quantified using Salmon v1.10.2 (Patro et al., 2017), with the Salmon index built from the D. melanogaster BDGP6.32 transcriptome and reference assembly (for decoy sequences) retrieved through Ensembl (Yates et al., 2021).

Differential expression analysis was performed with tximport (Soneson et al., 2015) and DESeq2 v1.30.1 (Love et al., 2014) running in RStudio v2023.07.999 on R v4.3.1. For count tables, including raw counts, normalized counts, and variance-stabilized log counts, significance of differentially expressed genes was assessed with the likelihood-ratio test (LRT), adjusted p ≤ 0.05. For pairwise comparisons within sex at each time point and arsenite concentration relative to its time-matched 0 mM control, the Wald test with a concatenated contrast of “[sex]_[concentration]_[time]” vs. “ [sex]_0mM_[time]” was used to identify genes with adjusted p ≤ 0.05 and an absolute log fold change (abs(LFC)) ≥ 1. All computation was performed on Indiana University’s High Performance Computing clusters using system modules and user conda environments.

Gene Set Enrichment Analysis

The list of genes that were significantly altered in 0.25 mM and 1.0 mM NaAsO2-exposed flies compared to 0 mM controls was analyzed using PAthway, Network and Gene-set Enrichment Analysis (PANGEA) (Hu et al., 2023) using the “Search Multiple Gene Lists” function. PANGEA analyses were conducted as follows: (i) the time course analysis used SLIM GO BP, FlyBase Signaling Pathway (Experimental Evidence), KEGG Pathway D.mel, and REACTOME Pathway; (ii) the disease enrichment analysis used FlyBase phenotypes for classical alleles and Disease Annotation AGR; (iii) tissue expression analysis used Preferred tissue (modEncode RNA_seq) and Expression annotation AGR. All results are presented as heatmaps with GO categories for the time point of interest rank-ordered according to adjusted p-value. PANGEA was also used to analyze the 100 genes that are differentially expressed in all four timecourses (male & female, 0.25mM & 1mM arsenite) using the parameters described above. The node graph was generated in PANGEA based on the top 20 most significantly enriched gene sets. Gene sets with the lowest amount of redundancy were selected to generate the graph.

LC-MS Sample Preparation

For metabolomic analysis, frozen samples were transferred from a −80°C freezer to a −20°C-cooled enzyme carrier caddy. 800 μL of prechilled (−20°C) 90% methanol containing 2 μg/mL succinic-d4 acid was added to each sample tube using a positive displacement pipette. The samples were homogenized using an Omni Beadruptor 24 (Omni International) at 4°C for 30 seconds at 6.45 m/sec. Once homogenized, the samples were incubated at −20°C for 2 hours. After incubation, the samples were centrifuged at 12,000 × g for 5 min at 4°C. The tubes were carefully removed from the centrifuge and 600 μL of the upper supernatant was transferred to a new labelled 1.5 mL microcentrifuge tube on dry ice. Each tube containing the newly extracted supernatant was centrifuged 12,000 × g for 5 min at 4°C. 200 μL of supernatant from each sample was transferred to a designated well in a 96 well plate on dry ice. Once the plate was fully loaded with sample supernatants, the plate was dried under a nitrogen evaporator. Once fully dried, the plate was covered with an aluminum seal and stored at −80°C. Remaining supernatant was transferred to new labelled 1.5 mL microcentrifuge tubes and dried down and stored at −80°C.

Ultra High-pressure Liquid Chromatography - Mass Spectrometry (UHPLC-MS)-based Metabolomics

UHPLC-MS metabolomics analyses were performed at the University of Colorado Anschutz Medical Campus, as previously described (Nemkov et al., 2019). Briefly, the analytical platform employs a Vanquish UHPLC system (Thermo Fisher Scientific, San Jose, CA, USA) coupled online to a Q Exactive mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA). The (semi)polar extracts were resolved over a Kinetex C18 column, 2.1 × 150 mm, 1.7 μm particle size (Phenomenex, Torrance, CA, USA) equipped with a guard column (SecurityGuard^™ Ultracartridge – UHPLC C18 for 2.1.0 mM ID Columns – AJO-8782 – Phenomenex, Torrance, CA, USA) using an aqueous phase (A) of water and 0.1% formic acid and a mobile phase (B) of acetonitrile and 0.1% formic acid for positive ion polarity mode, and an aqueous phase (A) of water:acetonitrile (95:5) with 1.0 mM ammonium acetate and a mobile phase (B) of acetonitrile:water (95:5) with 1.0 mM ammonium acetate for negative ion polarity mode. The Q Exactive mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA) was operated independently in positive or negative ion mode, scanning in Full MS mode (2 μscans) from 60 to 900 m/z at 70,000 resolution, with 4 kV spray voltage, 45 sheath gas, 15 auxiliary gas. Calibration was performed prior to analysis using the Pierce^™ Positive and Negative Ion Calibration Solutions (Thermo Fisher Scientific).

Statistical Analysis of Metabolomics Data

All metabolomics datasets were analyzed using Metaboanalyst 5.0 (Pang et al., 2021), with data normalized to sample mass and preprocessed using log normalization and Pareto scaling. Data generated by GC-MS analysis of individual compounds was analyzed using GraphPad Prism 10.

Data Availability

All strains and reagents are available upon request. Processed RNA-seq data is presented in Tables S2–S9 and raw data are available in NCBI Gene Expression Omnibus (GEO; GSE241663). All metabolomics data described herein are included in Tables S10 and S11.

RESULTS

NaAsO₂ exposure induces conserved transcriptional responses in adult Drosophila

To investigate dynamic molecular responses activated in Drosophila adults upon NaAsO₂ exposure, we used a combined transcriptomic and metabolomic approach to investigate time- and dose-dependent changes in gene expression and metabolite abundance. Previous studies using 48 hr NaAsO₂ exposures (Holsopple et al., 2023, Adebambo et al., 2024, Hayot et al., 2025), as well as our own analysis (Figure S1, Table S1), established LD₁₀ values for Oregon-R males and females ranging between 0.3 mM and 0.5 mM, and LD₅₀ values ranging from 0.65 mM to 1.0 mM. Based on these benchmarks, we exposed Oregon-R adult male and female flies to sublethal (0.25 mM; <LD₁₀) and lethal (1.0 mM; >LD₅₀) NaAsO₂ concentrations, aiming to capture both adaptive responses that help maintain homeostasis and pathological responses. Samples were collected at 1, 2, 4, 8, 24, and 48 hours of exposure for RNA-seq analysis (Table S2–9), and 2, 4, 8 24, and 48 hrs for semi-targeted UHPLC-MS-based metabolomic analysis (Table S10–11). For each time point, changes in gene expression and metabolite abundance were compared to sex-matched, time-matched unexposed controls.

Our initial analysis of both the transcriptomic and metabolomic data revealed dose-, time-, and sex-dependent differences in response to NaAsO₂ (Figures 1, 2, and S2; Table S12 and S13). We noted immediately that the sex-specific differences are extensive, with less than half of the DEGs shared between males and females at any given time point (Figure S2). These sex-specific changes in gene expression and metabolite levels will be addressed in detail elsewhere (Tourigny et al., In preparation). In this study, we focused on the dose- and time-dependent molecular responses shared between NaAsO₂-exposed males and females.

Figure 1. Principal Component Analysis (PCA) of RNA-seq data from adult Drosophila exposed to NaAsO₂ across a 48-hour time course.

PCA was used to analyze RNA-seq data from adult (A) male and (B) female flies exposed to control diet, 0.25 mM NaAsO₂, and 1.0 mM NaAsO₂. In both sexes (A,B), samples exposed to 1.0 mM mM NaAsO₂ for 8, 24, and 48 hrs separated clustered in a similar position along the PC1 axis. RNA-seq data are available in Table S2, S4, S6, and S8.

Figure 2. Partial Least Squares Discriminant Analysis (PLS-DA) of metabolomics data from adult Drosophila exposed to NaAsO₂ across a 48-hour time course.

Metaboanalyst 6.0 was used to conduct a PLS-DA of metabolomics data from adult (A) female and (B) male flies exposed to control diet, 0.25 mM NaAsO₂, and 1.0 mM NaAsO₂. In (A), the control samples and time points 2, 4, and 8 hr group compose two distinct clusters and are encircled by a dashed line. In (B), the 24 and 48 hr time points from both the low- and high-dose exposure cluster separately from all other sample sets. (C,D) Heatmap of Kendall’s Tau correlations of the experimental variables with principal components 1 through 5 for (C) females and (D) males. Asterisks denote increasing levels of statistical significance. Analyzed data are available in Table S10 and S11.

To compare the kinetics of transcriptional and metabolic responses to arsenite, we performed Principal Component Analysis (PCA) on the RNA-seq data and PLS-DA on the metabolomics. In the RNA-seq data, samples separated primarily along PC1 (Figure 1A–B), which was largely driven by arsenite concentration, with additional contributions from time (Figures 1C–D). By 8 hours of exposure, transcriptional profiles, particularly at 1.0 mM, had already diverged from earlier time points, with samples from 8, 24, and 48 hours clustering together along PC1. In contrast, metabolomic profiles showed little separation at 8 hours, with distinct clustering emerging only at 24 and 48 hours (Figure 2A–B). Together, these analyses indicate that arsenite induces rapid, dose-dependent transcriptional changes that precede large-scale metabolic remodeling.

Because arsenite concentration emerged as the dominant driver of transcriptional variation, we focused first on the RNA-seq data to identify dose-dependent gene expression programs that precede metabolic changes. Using PANGEA (Hu et al., 2023), we performed gene set enrichment analysis across all time points and exposure conditions (Table S14). Examination of the top 10 most significantly enriched gene sets at each time point revealed dynamic gene expression programs associated with metabolic, xenobiotic, and stress responses (Figures 3, 4, S3, S4 and Table S14). Strikingly, all datasets showed early enrichment for targets of the Heat Shock Transcription Factor (Hsf) (Clos et al., 1990, Zimarino et al., 1990, Wu et al., 1987), with “HSF-1-dependent transactivation” (R-DME-3371571) and “Regulation of the HSF1-mediated heat shock response” (R-DME-3371453) ranking among the most significantly enriched gene sets at either the 1 hr or 2 hr time point in all four datasets. Consistent with these findings, NaAsO₂ elicits induction of Heat-shock protein transcripts following as little as 1 hr of exposure (Figure 5A–B, and Table S14), highlighting the Hsf gene regulatory network as one of the earliest responders to NaAsO₂-exposure. We hypothesize that this response, although strongly induced by arsenite in our system, is unlikely to be arsenic-specific and instead represents a conserved stress response shared across diverse toxic exposures.

Figure 3. Gene Set Enrichment Analysis of DEGs in female Drosophila melanogaster exposed to 1.0 mM NaAsO₂ across a 48-hour time course.

PANGEA was used to identify significantly overrepresented gene sets among differentially expressed genes (DEGs) at each exposure time point (see Methods and Table S14). The top 10 enriched gene sets for each time point are presented with heatmaps of their rank-ordered adjusted p-value at the corresponding time (outlined heat map column), along with their significance (green gradient) or lack thereof (grey) at the other time points.

Figure 4. Gene Set Enrichment Analysis of DEGs in male Drosophila melanogaster exposed to 1.0 mM NaAsO₂ across a 48-hour time course.

As in Figure 3, PANGEA was used to identify significantly overrepresented gene sets among differentially expressed genes (DEGs) at each exposure time point (See Methods and Table S14). The top 10 enriched gene sets for each time point are presented with heatmaps of their rank-ordered adjusted p-value at the corresponding time (outlined heat map column), along with their significance (green) or lack thereof (grey) at the other time points.

Figure 5. Up-regulated gene families across a 48 hr time course in male and female Drosophila melanogaster exposed to 1.0 mM NaAsO₂.

(A, B) Heatmaps displaying log₂ fold changes of pairwise comparisons (time-matched treatment vs. control) for significantly upregulated (adj p ≤ 0.05) genes belonging to four canonical detoxification and stress-response families: cytochrome P450s (FBgg0001222), glutathione S-transferases (FBgg0000077), heat shock proteins (FBgg0000501), and metallothioneins (FBgg0000197). Gene expression changes are presented separately for (A) females and (B) males across 6 exposure time points (1, 2, 4, 8, 24, and 48 hours). Gene names in bold red exhibit significant up-regulation in both sexes. Non-significant time points are left blank. The data reveal distinct temporal dynamics and sex-specific patterns of gene induction among the canonical detoxification response pathways. Notably, males exhibit earlier activation of heat shock proteins, with females showing more rapid upregulation of metallothioneins. These findings highlight the coordinated and sex-dependent transcriptional activation of detoxification pathways in response to arsenite exposure.

Beyond the early heat shock response, several additional gene sets associated with metabolic and stress pathways were significantly enriched during the first 8 hours of exposure (Figures 3, 4, S3, S4, and Table S14). These gene sets are primarily composed of cytochrome P450s (path:map00982; path:map00980), glutathione S-transferases (Gst; path:map00480), and metallothioneins (Mtn; GO:0042221)—all key components of arsenic and metal detoxification networks in flies and humans (Yiwen et al., 2022, Muñiz Ortiz et al., 2009, Thomas, 2009). A closer examination of the significant pairwise comparisons revealed that these three enzyme families are coordinately upregulated throughout the 48 hrs of exposure (Figure 5A–B and S5A–B), although the exact sets of genes activated at these time points varied by dose and sex.

At later time points, gene sets linked with digestion and lipid metabolism became prominent (Figures 3, 4, S3, S4, and Table S14), suggesting that intestinal gene expression networks are especially sensitive to arsenite with continuous exposure. Consistent with this, tissue-specific enrichment analysis identified the digestive system as the most strongly associated tissue across all conditions (Figure 6A and Table S15). At the 1.0 mM exposure, many of these DEGs belong to the trypsin- and chymotrypsin-like Jonah family of serine hydrolases, lysozymes, and Niemann-Pick family members involved in cholesterol trafficking—all of which were downregulated following NaAsO₂ exposure (Figure 6B–C, Table S2–S9). These findings highlight the particular impact of arsenite on digestive and absorption functions in the gut and reveal how tissue transcriptional responses evolve over time: sensing, followed by stress and detoxification responses, and finally adaptive or disease-related changes to exposure.

Figure 6. Identification of tissue-associated transcriptional responses to NaAsO₂ across a 48-hour time course.

(A) PANGEA was used to identify enrichment for tissue-associated gene sets among differentially expressed genes (DEGs) at each exposure time point (1, 2, 4, 8, 24, and 48 hours; see Table S12). DEGs from each gene set were analyzed using the Preferred tissue (modEncode RNA_seq) gene set. The top three “Preferred tissue” gene sets are displayed according to average p-value across the time points. Values in parentheses indicate the NaAsO₂ dose used in the exposure. The digestive system consistently shows the strongest enrichment across all conditions and time points. (B, C) Heatmaps displaying decreasing log₂ fold changes in expression for gene families associated with digestion and cholesterol trafficking. Gene expression changes are presented separately for (B) females and (C) males across 6 exposure time points (1, 2, 4, 8, 24, and 48 hours). Gene names in bold red exhibit significant down-regulation in both sexes. Gene families include Jonah serine hydrolases (FBgg0001076), lysozymes (FBgg0000606), and Niemann-Pick type C genes (FBhh0000396, FBhh0000397), all of which are significantly downregulated following NaAsO₂ exposure, indicating disrupted intestinal metabolism.

Dose- independent responses to NaAsO₂ encompass genes involved in lipid metabolism, chromatin architecture, and endocrine signaling

While the time-resolved analyses highlighted how transcriptional responses evolve with dose and exposure duration, we next sought to define a set of arsenite-responsive genes that are robustly induced during acute exposure and shared across experimental conditions. To do so, we identified genes that were differentially expressed in arsenite-exposed flies relative to time-matched controls in all four exposure conditions (male and female flies exposed to 0.25 mM or 1.0 mM NaAsO₂), regardless of the specific time point at which the response occurred. This analysis yielded 100 genes that define a core arsenite-responsive transcriptional program (Table S16). Importantly, this approach does not reflect an average across time points but instead captures genes that respond reproducibly at one or more stages of the acute exposure time course, consistent with their role in initiating or sustaining the arsenite response. The majority of these genes encode proteins with predicted human orthologs (72/100; Table S17), indicating that this core response reflects conserved biological processes rather than condition- or time-specific effects.

Gene set enrichment analysis of these 100 core genes using PANGEA revealed two interconnected networks that together define a canonical arsenite stress response (Figure 7 and Table S18). One network is dominated by heat shock proteins and closely related stress-response genes (Figure 7A), consistent with early activation of the Hsf pathway observed in the time-resolved analysis. The second network is enriched for glutathione S-transferases, cytochrome P450s, and metallothioneins (Figure 7B), reflecting a coordinated detoxification and metal-response program that persists across doses. In addition to these canonical detoxification genes, this network includes genes involved in one-carbon and methionine metabolism (Gnmt and MsrA) as well as multiple ecdysteroid 22-kinases, an insect-specific enzyme family linked to steroid hormone signaling and xenobiotic responses (Scanlan et al., 2020). Together, these networks define a conserved transcriptional backbone that integrates proteostasis, redox control, and metabolic adaptation in response to arsenite exposure.

Figure 7. NaAsO₂ exposure activates a conserved detoxification and stress-response network.

The 100 genes consistently differentially expressed across all NaAsO₂ exposure conditions (see Table S16) were analyzed using PANGEA (Table S18). From these, depicted gene sets were selected based on lowest adjusted p-value and minimal overlap between genes and used to generate a gene set node graph. (A,B) The resulting graphs consists of a gene set networks focused on (A) the heat shock response and (B) a series of metabolic pathways that define an integrated defense response to arsenite. Triangle nodes indicate gene sets, and circle nodes denote individual genes.

Beyond the detoxification response, five additional metabolic genes emerged from this analysis (Table S16 and S17). These include CG33502, a homolog of Nfu1 involved in Fe-S cluster assembly, and CG32669, a member of the SLC5 transporter family – implicating mitochondrial function and Na+ electrochemical gradient in the NaAsO₂ response. We also observe consistent changes in two lipid metabolism genes: (i) FASN1 was generally downregulated, (ii) Dgat2 was upregulated. These findings suggests that arsenite reduces de novo fatty acid synthesis and storage triglycerides, consistent with observations in mice and C. elegans (Zdraljevic et al., 2019, Adebayo et al., 2015). Finally, CG8745, which encodes an ethanolamine-phosphate phospho-lyase, was also induced in common–a change consistent with elevated phosphoethanolamine levels, a directly testable hypothesis (see metabolism section below).

Two genes with annotated roles in chromatin organization and DNA replication also emerged from this analysis (Tables S16 and S17). We observed consistent induction of His2B:CG33872, which encodes a histone H2B variant, and timeless, a gene with established roles in circadian regulation and DNA replication fork stability (Vosshall et al., 1994, Sehgal et al., 1994, Vipat and Moiseeva, 2024). Although arsenite exposure has previously been linked to circadian disruption in Drosophila (Adebambo et al., 2024), we did not observe coordinated changes in other core clock genes. This suggests that altered timeless expression may reflect replication-associated or chromatin-linked stress responses, rather than direct perturbation of circadian timing. However, given that these inferences are based on transcriptional changes alone, additional functional studies will be required to define the precise role of these factors in the arsenite response.

We also identified two genes encoding secreted proteins, Arc1 and Limostatin, that were consistently upregulated across all exposure conditions (Figure 7B and Table S16). Arc1 has been implicated in lipid metabolism and intercellular communication (Mattaliano et al., 2007, Mosher et al., 2015), while Limostatin suppresses insulin secretion in response to nutrient availability (Alfa et al., 2015). The induction of these genes suggests that arsenite exposure engages endocrine and inter-organ signaling pathways, a conclusion that is further supported by the accompanying metabolomic evidence of disrupted carbohydrate metabolism (see below). Together, these findings raise the possibility that arsenite alters hormonal and metabolic signaling early during exposure, although direct effects on endocrine function will require targeted physiological validation.

NaAsO₂ exposed flies exhibit transcriptomic profiles indicative of arsenic-associated human disease states

Given that most arsenite-induced DEGs encode proteins conserved in humans, we reanalyzed the RNAseq data using the Phenotype gene set options in PANGEA, which draws on disease annotations from the Alliance of Genome Resources and phenotypes from FlyBase, to determine whether transcriptional changes in flies correspond to gene expression patterns associated with human arsenic-related diseases. This analysis was used to assess whether early transcriptional responses to acute arsenite exposure in flies overlap with gene sets that have been associated with arsenic-related diseases in humans, rather than to infer disease phenotypes directly. Focusing on the 8-hour exposure datasets, we observed significant enrichment of fly DEGs in multiple disease-associated gene sets, including cataracts, renal failure, peripheral vascular disease, hyperlipidemia, pancreatitis, and cardiovascular disorders (Figures 8 and Table S19). Importantly, these disease associations are known to arise from chronic arsenic exposure in humans (Jomova et al., 2011), and their appearance at early time points in flies suggests that arsenite rapidly engages molecular pathways that are implicated in disease risk.

Figure 8. Disease phenotype enrichment among DEGs in female and male Drosophila melanogaster exposed to NaAsO₂.

Gene enrichment analysis was performed using PANGEA to identify disease-associated gene sets among differentially expressed genes (DEGs) in female and male flies exposed to either (A, B) 0.25 mM or (C, D) 1.0 mM NaAsO₂ across a 48-hour time course (see Table S19). Disease annotations were derived from the Alliance of Genome Resources (AGR) and FlyBase phenotype databases. Heatmaps display the top 20 enriched disease phenotypes at the 8 hr time point, rank ordered by significance (indicated by −log₁₀ p-value), as well as the significance value for those gene sets at all other timepoints. Gene sets associated with Type 1 and 2 diabetes are highlighted in red. The entire analysis for all timepoints can be found in Table S19. Note that high-dose exposure (1mM) resulted in broader and more sustained enrichment of disease phenotypes compared to the 0.25 mM condition.

Although disease associations varied by exposure level and sex, one pattern was consistent: genes linked to diabetes mellitus—both Type 1 and Type 2—were enriched across all datasets (Figure 8 and Table S19]). Notably, Type 1 diabetes–associated genes were significantly enriched in females after just 1 hour of exposure to 1.0 mM NaAsO₂ (Figure 8 and Table S19), one of the few disease-related sets reaching significance at this early time point (Table S19). This early enrichment coincided with induction of Limostatin, a secreted peptide known to suppress insulin secretion in flies (Alfa et al., 2015), which was upregulated across exposure conditions (Table S16). Limostatin is functionally analogous to somatostatin in mammals, a peptide hormone that negatively regulates insulin release during nutrient deprivation (Rorsman and Huising, 2018). Additionally, the insulin receptor (InR) was upregulated in males at multiple timepoints (Tables S3, S7), consistent with previous observations that increased InR expression correlates with reduced insulin signaling in flies (Puig and Tjian, 2005). Together, these transcriptional changes are consistent with perturbations to endocrine and glucose-regulatory pathways previously implicated in arsenic-associated metabolic dysfunction in human populations (Navas-Acien et al., 2008, Rahimi Kakavandi et al., 2023, Grau-Pérez et al., 2017).

NaAsO₂ exposure induces metabolic defects indicative of a diabetic state

Having established that NaAsO₂ exposure induces rapid transcriptional changes in conserved gene networks, including pathways linked to endocrine signaling and glucose homeostasis, we next turned to the metabolomics data to assess the functional metabolic consequences of these responses. Although earlier analyses indicated that metabolic remodeling follows initial gene expression changes, examining the metabolite profiles in greater detail allowed us to identify which metabolic pathways are most sensitive to arsenite exposure. This analysis revealed an early and pronounced disruption of central carbon metabolism, with pathways governing carbohydrate utilization emerging as a prominent feature of arsenite-induced metabolic stress. Among the top 25 significantly altered metabolites, nearly half were associated with carbohydrate metabolism (Figure 9A–B; Tables S20 and S21). Glucose and lactate levels increased significantly between 4 and 8 hours of exposure (Figure 9A–B; Figure S5A–D; Tables S20 and S21), consistent with early disruption of glycolytic metabolism. Pentose sugars (ribose/ribulose) also accumulated markedly (Figure 9A–B; Tables S20 and S21). These changes were accompanied by elevated levels of methylglyoxal (Figure 9A–B; Figure S5E–F), a reactive byproduct of glycolysis whose accumulation is consistent with increased glycolytic flux (de Bari, 2020). Because lactate and methylglyoxal are established biomarkers of dysregulated glucose metabolism in humans (Moraru et al., 2018, Schalkwijk and Stehouwer, 2020, Lu et al., 2024), these findings support the conclusion that glycolytic dysfunction is a central component of arsenite-induced metabolic stress and align with the transcriptional signatures observed above.

Figure 9. The metabolome of male and female Drosophila changes in response to dose and exposure duration.

Heatmaps showing top 25 significantly changed metabolites for (A) male and (B) female adult Drosophila exposed to both 0.25 mM and 1.0 mM NaAsO₂ across time (see Tables S10 and S11). The 12 metabolites that are in common between the sexes are highlighted in bold. Each column represents the scaled average metabolite abundance for the six biological replicates analyzed at each time point. Heatmaps were generated using Metaboanalyst 6.0. Statistical significance determined using an ANOVA test. Relative abundance is visualized using Pareto scaling, with elevated levels in red and decreased levels in dark blue.

Beyond carbohydrate metabolism, NaAsO₂ exposure disrupted several interconnected metabolic pathways associated with oxidative and xenobiotic stress responses. These changes converge on processes that support cellular redox balance and antioxidant capacity, consistent with arsenite-induced oxidative burden.

1. Methionine and folate/one carbon metabolism: Multiple intermediates linked to one-carbon metabolism—including methionine, L-methionine-S-oxide, and S-adenosylmethionine (SAM)—were among the most significantly altered metabolites following NaAsO₂ exposure (Figure 9A–B; Tables S20 and S21). These metabolites participate in the SAM cycle, which serves as the primary cellular source of methyl groups and contributes to the generation of NADP(H) and glutathione through downstream one-carbon flux (Ducker and Rabinowitz, 2017). Perturbation of this pathway is therefore likely to have broad consequences for redox homeostasis and detoxification capacity. Although the mechanistic basis for these changes remains to be resolved, altered one-carbon metabolism provides a plausible metabolic link between arsenite exposure and increased demand for antioxidant and xenobiotic defense systems.

2. Oxidative Stress Markers: Consistent with elevated oxidative stress, levels of 5-oxoproline—an established marker of oxidative burden—were increased following arsenite exposure (Figure 9A–B; Tables S20 and S21) (Van Der Pol et al., 2017, Pederzolli et al., 2007, Pederzolli et al., 2010). In addition, cystine levels increased whereas ascorbate levels decreased in flies exposed to 1.0 mM NaAsO₂, particularly at later time points (8, 24, and 48 hours; Figure 10A–F; Tables S20 and S21). This reciprocal pattern is consistent with disruption of the cystine/cysteine redox cycle and increased utilization of ascorbate as an antioxidant buffer. Together, these changes indicate sustained oxidative stress and suggest a shift toward reliance on non-enzymatic antioxidant systems during prolonged arsenite exposure.

3. Purine metabolism: Inosine and xanthine levels increased (Figure 9A–B). Xanthine showed particularly large increases at 24 and 48 hours, ranking among the most significantly altered metabolites (Figure 9 and 10C–F and Tables S20 and S21). Importantly, Drosophila purine metabolism was previously linked to antioxidant defense, as mutations in the rosy gene, which encodes the Drosophila homolog of xanthine dehydrogenase, render flies sensitive to oxidative stress (Hilliker et al., 1992). Thus, elevated xanthine levels likely reflect a metabolic response to increased oxidative burden. We also note these changes potentially intersect with folate/one-carbon metabolism, as purine biosynthesis requires two one-carbon units from 10-formyl-tetrahydrofolate (Ducker and Rabinowitz, 2017).

Figure 10. Comparison of metabolite levels between Drosophila exposed to 1.0 mM NaAsO₂ and unexposed controls.

Volcano plots comparing relative metabolite levels between NaAsO₂-treated and untreated flies for the following conditions: (A) Female flies exposed for 8 hours. (B) Male flies exposed for 8 hours. (C) Female flies exposed for 24 hours. (D) Male flies exposed for 24 hours. (E) Female flies exposed for 48 hours. (F) Male flies exposed for 48 hours. For all exposures, relative levels are determined by the average value of n=6 samples each containing 20 flies. Vertical dashed line represents a log₂ fold change (FC) of 2 and horizontal dashed line represents −log₁₀ p-value of 1.

In addition to the metabolic changes associated with oxidative stress, phosphoethanolamine was consistently elevated in arsenite-exposed flies across doses and time points (Figure 9A–B; Tables S20 and S21). This observation is notable because CG8745, which is predicted to encode an ethanolamine-phosphate phospho-lyase, is among the 100 genes induced under all exposure conditions (Tables S16 and S17). Although the functional consequences of these changes remain to be determined, the coordinated elevation of phosphoethanolamine and induction of a gene linked to its catabolism suggests that phosphoethanolamine metabolism responds robustly to arsenite exposure and may represent a sensitive indicator of arsenite-induced metabolic stress.

Taken together, the metabolomic data indicate that NaAsO₂ exposure perturbs central carbon metabolism, elevates reactive metabolic byproducts, and engages oxidative stress–associated pathways. When considered alongside the transcriptional changes in genes involved in endocrine and metabolic regulation, these findings support the use of Drosophila as a tractable whole-organism system for dissecting conserved metabolic responses to arsenic exposure and for identifying early molecular features associated with arsenite-induced metabolic dysfunction.

DISCUSSION

This study provides a comprehensive, time-resolved multi-omic analysis of arsenite exposure in Drosophila melanogaster, revealing how molecular responses evolve from the earliest stages of exposure through the onset of lethality. Unlike previous studies that focused on individual genes, isolated pathways, or single exposure time points (Muñiz Ortiz et al., 2009, Chaturvedi et al., 2025), our integrated transcriptomic and metabolomic approach resolves the temporal ordering of molecular responses across dose and time. This resolution enables identification of early, coordinated activation of conserved stress, detoxification, endocrine, and metabolic programs that precede large-scale metabolic remodeling. Consistent with this early pathway engagement, disease-associated gene set enrichment analyses indicate that acute arsenite exposure in flies engages molecular pathways implicated in arsenic-associated chronic diseases in humans. Together, these findings highlight the translational utility of the fly model for identifying early molecular signatures and initiating events relevant to arsenic-associated disease risk.

Importantly, among the 100 genes consistently induced across all exposure conditions, 72 encode proteins with predicted human orthologs (Table S17), indicating that the core arsenite-responsive transcriptional program in Drosophila is largely evolutionarily conserved. Notably, this level of conservation is consistent with previous observations that ~70% of adversity- and disease-linked genes are shared among distantly related species (Colbourne et al., 2022, Wangler et al., 2017). Rather than defining discrete regulatory networks, these conserved genes converge on a small number of recurring functional themes, including oxidative stress defense, xenobiotic metabolism, and endocrine-linked metabolic regulation—processes that are broadly implicated in arsenic toxicity in humans. For example, multiple glutathione S-transferases, cytochrome P450 enzymes, and metallothioneins were consistently induced, reflecting a conserved detoxification and redox-buffering response to arsenite across species. In addition, conserved regulators of metabolic signaling, including Limostatin and the insulin receptor (InR), were consistently altered, suggesting that arsenite exposure engages endocrine pathways linked to glucose regulation early in the response. Together, these findings support the use of Drosophila as a whole-organism model for identifying conserved early molecular responses and biomarkers associated with arsenic exposure. In this context, arsenic serves as a well-characterized exemplar, illustrating how time-resolved multi-omic analysis in Drosophila can reveal conserved early molecular responses relevant to environmental health.

Both RNA-seq and semi-targeted metabolomic analyses revealed coordinated molecular responses to NaAsO₂ exposure, with particularly strong disruption of carbohydrate metabolism. Within hours of exposure, levels of glucose, lactate, and methylglyoxal increased markedly, coincident with transcriptional changes in endocrine regulators such as Limostatin, a peptide that suppresses insulin secretion in flies (Alfa et al., 2015). Similar alterations in insulin signaling and glycolytic control have been associated with chronic arsenic exposure in human populations (Navas-Acien et al., 2008, Lai et al., 1994, Grau-Pérez et al., 2017). The rapid onset of these metabolic changes highlights the sensitivity of carbohydrate metabolism to arsenite exposure and supports the existence of a conserved axis linking arsenic toxicity to metabolic dysregulation.

Another key advance from this work is the temporal ordering of arsenite-responsive transcriptional programs. Our data suggest that the coordinate activation of three transcription factors drives many of the initial responses. First, we observed rapid induction of heat-shock protein genes, which are canonical targets of the Heat Shock Factor (Hsf) family of transcription factors (Wu et al., 1987, Clos et al., 1990). This result is consistent with the known role of Hsf in protecting cells against proteotoxic insults from a variety of environmental stresses, including heavy metals and toxins (Akerfelt et al., 2010). This initial response associated with Hsf is followed by upregulation of Phase I xenobiotic and oxidative stress response genes, including glutathione S-transferases and cytochrome P450s—targets of Cap ‘n’ collar (Cnc), the Drosophila homolog of Nrf2 (Sykiotis and Bohmann, 2008, Misra et al., 2013). Similar to mammalian Nrf2, Cnc is controlled by its cytoplasmic inhibitor Keap1, which suppresses Nrf2 transcriptional activity and targets this transcription factor for degradation by the proteasome (Baird and Yamamoto, 2020, Sykiotis and Bohmann, 2008). Oxidative stress and exposure to electrophiles disrupts sulfhydryl bonds within Keap1, releasing Cnc/Nrf2 and activating the downstream transcriptional program (Baird and Yamamoto, 2020). In this regard, studies from human cells demonstrated that inorganic arsenic indeed disrupts critical sulfhydryl bonds within Keap1, resulting in Nrf2 activation (Lau et al., 2013). Thus, arsenite potentially activates this coordinated transcriptional network directly in the fly by changing the structure of Keap1, which requires no transcriptional or translational control. Finally, the early transcriptional responses driven by Hsf and Cnc/Nrf2 occur in parallel with induction of five Drosophila metallothionein genes. Increased expression of these genes likely reflects activation of the MTF-1 transcription factor, which is known to drive Mtn gene expression in response to heavy metals (Zhang et al., 2001).

The early activation of Hsf target genes relative to Cnc/Nrf2 transcriptional networks indicates that the heat shock response represents one of the earliest cellular defenses engaged following arsenite exposure. This temporal ordering is notable given that arsenic is known to directly disrupt Keap1 function and activate Cnc/Nrf2 signaling, yet our time-resolved data reveal that Hsf-dependent transcriptional programs are mobilized more rapidly. One explanation for this pattern is that arsenite rapidly induces widespread proteotoxic stress (Jacobson et al., 2012, Guerra-Moreno et al., 2015), leading to protein misfolding and increased demand for molecular chaperones. Under such conditions, pre-existing pools of heat shock proteins become engaged in stabilizing damaged or misfolded proteins, thus relieving chaperone-mediated repression of Hsf and permitting its rapid activation (Akerfelt et al., 2010, Morimoto, 2011). This mechanism is consistent with established models in which Hsf functions as a sensor of proteostasis imbalance, responding swiftly to disruptions in protein folding homeostasis. In this framework, Hsf-driven transcription represents an early response to proteostasis disruption, while activation of Cnc/Nrf2-dependent detoxification and redox pathways reflects a coordinated but temporally distinct response to arsenite-induced oxidative and electrophilic stress.

Our time-resolved analysis also suggests temporal separation between transcriptional and metabolic responses to arsenite. Principal component analysis of the RNA-seq data shows that gene expression changes emerge rapidly, with distinct exposure clustering by dose as early as 8 hours. In contrast, clustering analysis of the metabolomic profiles reveals segregation only at later time points (Figures 1–2). This observation, however, does not imply that metabolic activity remains static at the early stages of arsenite exposure. Flux through a pathway can increase substantially without causing large shifts in steady-state metabolite concentrations if production and consumption rates rise in parallel. In the context of arsenite exposure, enhanced detoxification and antioxidant demands likely accelerate flux through folate metabolism, the methionine cycle, and glutathione biosynthesis, while homeostatic mechanisms buffer pool sizes to prevent harmful depletion or imbalance. This dynamic could explain why strong transcriptional enrichment of genes involved in glutathione metabolism coincides with relatively modest changes in relevant metabolite levels, particularly at early time points. In contrast, the accumulation of xanthine and other purine metabolites at later stages potentially reflects significant, destabilizing imbalances in cellular metabolism that occur once homeostatic mechanisms are no longer sufficient to buffer stress. Such unobserved regulation underscores the importance of integrating flux-based approaches with transcriptomic and metabolomic profiling in future studies.

While our multi-omic approach highlights the dynamic nature of the arsenic defense response, it also reveals that molecular pathways linked to chronic arsenic-associated diseases are engaged at the earliest stages of acute exposure. This observation underscores the importance of conducting whole-organism, time-resolved analyses to identify molecular initiating events that precede overt pathology and highlights a future need for tissue-specific studies that refine the mechanistic interpretation of these early molecular signatures. For example, gene set enrichment analysis suggests a key role for intestinal physiology and barrier function in the response to arsenite, pointing to the need for tissue-specific and single-cell analyses to uncover organ-level contributions. In this context, recent single-nucleus RNA-seq profiling of the Drosophila brain following NaAsO₂ exposure complements our analysis and reveals both similarities and differences in cell-specific arsenite responses (Chaturvedi et al., 2025). Notably, seven genes from our core signature (MtnA, MsrA, CG8745, tim, ple, Arc1, and CG12224) were also significantly altered in the brain dataset (Chaturvedi et al., 2025). Moreover, both datasets converge on the same detoxification gene families, including glutathione S-transferases, cytochrome P450s, UDP-glucuronosyltransferases, metallothioneins, and methionine sulfoxide reductases (Chaturvedi et al., 2025).

Taken together, our findings establish Drosophila as a uniquely powerful whole-organism model for investigating the molecular initiating events that precede onset of toxicant-induced diseases. By leveraging its genetic tractability, rapid life cycle, multiple organ systems, and capacity for sex-specific analysis, we demonstrate how the fly can illuminate evolutionarily conserved biological responses and uncover early biomarkers of adverse outcome pathways (Ankley and Edwards, 2018). As toxicology continues to evolve toward mechanistic, predictive, and ethically responsible approaches, Drosophila is poised to play a central role in bridging experimental discovery with human health relevance (Rand et al., 2023). Our study not only advances the application of multi-omic strategies in NAMs but also lays the groundwork for future efforts to integrate fly-based insights into regulatory frameworks and public health initiatives aimed at mitigating the global burden of environmental toxicants such as arsenic.

Supplementary Material

Supplement 1

Figure S1. Dose-response curves of Drosophila adult males and females exposed to NaAsO₂. Dose-response curves illustrate the measured proportions of dead (A) female and (B) male Oregon-R flies at increasing concentrations of NaAsO₂ following a 48 hr exposure. Both the best fit line and the LD₁₀, LD₂₅, and LD₅₀ were calculated using a Dichotomous Hill Model. n=6 vials at each concentration with 20 flies per vial. (A,B – see text in lower right of panel) Male and female dose response curves were statistically analyzed with a global nonlinear regression model analysis using an extra sum-of-squares F test. This analysis revealed a significant difference in the LD₂₅ and LD₅₀ values between Oregon-R male and female flies (LD₁₀ F-statistic of F(1,112) = 1.581, p = 0.2112; LD₂₅ F-statistic of F(1,112) = 5.586, p = 0.0198; LD₅₀ F-statistic of F(1,112) = 16.85, p < 0.0001). However, the hillslope of the dose response curve was similar between males and females (F-statistic of F(1,112) = 0.3659, p = 0.5465).

Figure S2. A comparison of DEG in males and females exposed to 1.0 mM NaAsO₂. Venn diagrams illustrating the overlap between up- and down-regulated DEGs in males and females at 4, 8, 24, and 48 hrs. Note that less than 50% of the DEGs at any given time point are shared between the two sexes. The 1 hr and 2 hr time points were excluded from this analysis due to the low number of DEGs.

Figure S3. Gene Set Enrichment Analysis of DEGs in female Drosophila melanogaster exposed to 0.25 mM NaAsO₂ across a 48-hour time course. As in Figure 3, PANGEA was used to identify significantly overrepresented gene sets among differentially expressed genes (DEGs) at each exposure time point (see Methods and Table S3). DEGs from each time point were simultaneously analyzed for enrichment using four annotation sources: (i) SLIM2 GO Biological Process, (ii) FlyBase signaling pathways (experimental evidence), (iii) KEGG Pathway (D. melanogaster), and (iv) REACTOME Pathway (D. melanogaster). The top 10 enriched gene sets for each time point are presented with heatmaps of their rank-ordered adjusted p-value at the corresponding time (outlined heat map column), along with their significance (green gradient) or lack thereof (grey) at the other time points.

Figure S4. Gene Set Enrichment Analysis of DEGs in male Drosophila melanogaster exposed to 0.25 mM NaAsO₂ across a 48-hour time course. As in Figure 3, PANGEA was used to identify significantly overrepresented gene sets among differentially expressed genes (DEGs) at each exposure time point (see Methods and Table S5). The top 10 enriched gene sets for each time point are presented with heatmaps of their rank-ordered adjusted p-value at the corresponding time (outlined heat map column), along with their significance (green) or lack thereof (grey) at the other time points.

Figure S5. NaAsO₂ exposure disrupts glycolytic metabolism in a dose- and time-dependent manner. Box and whiskers plots displaying normalized levels of (A, B) glucose, (C, D) lactate, and (E, F) methylglyoxal in adult flies exposed to 0 mM (control), 0.25 mM, or 1.0 mM NaAsO₂ over a 48-hour time course. Analysis conducted with Metaboanalyst 6.0. Data were normalized to sample mass and processed with Log Transformation and Pareto scaling. Statistical analysis was conducted using ANOVA with FDR correction (FDR < 0.05), followed by Fisher’s Least Significant Difference post hoc test (see Tables S20 and S21). *p < 0.05 relative to untreated time-matched control. n=6 samples per condition.

Supplement 2

Table S1. Data used to generate dose response curves in Figure s1. n=6 exposure vials per NaAsO₂ concentration; 20 animals per vial.

Supplement 3

Table S2. RNA-seq results comparing gene expression between Oregon-R Males exposed to NaAsO₂. Adult males were exposed to 0 mM and 0.25 mM NaAsO₂ for 1, 2, 4, 8, 24 or 48 hours prior to collection.

Supplement 4

Table S3. RNA-seq results comparing gene expression between Oregon-R Males exposed to NaAsO₂. Adult males were exposed to 0 mM and 0.25 mM NaAsO₂ for 1, 2, 4, 8, 24 or 48 hours prior to collection. Only genes displaying significant changes in gene expression (log₂ fold change ≥ |1| and an adjusted p-value of ≤0.05) are included in these tables.

Supplement 5

Table S4. RNA-seq results comparing gene expression between Oregon-R females exposed to NaAsO₂. Adult females were exposed to 0 mM and 0.25 mM NaAsO₂ for 1, 2, 4, 8, 24 or 48 hours prior to collection.

Supplement 6

Table S5. RNA-seq results comparing gene expression between Oregon-R females exposed to NaAsO₂. Adult females were exposed to 0 mM and 0.25 mM NaAsO₂ for 1, 2, 4, 8, 24 or 48 hours prior to collection. Only genes displaying significant changes in gene expression (log₂ fold change ≥ |1| and an adjusted p-value of ≤0.05) are included in these tables.

Supplement 7

Table S6. RNA-seq analysis comparing gene expression between groups of Oregon-R males exposed to NaAsO₂. Adult males were exposed to 0 mM and 1.0 mM NaAsO₂ for 1, 2, 4, 8, 24 or 48 hours prior to collection.

Supplement 8

Table S7. RNA-seq analysis comparing gene expression between groups of Oregon-R males exposed to NaAsO₂. Adult males were exposed to 0 mM and 1.0 mM NaAsO₂ for 1, 2, 4, 8, 24 or 48 hours prior to collection. Only genes displaying significant changes in gene expression (log₂ fold change ≥ |1| and an adjusted p-value of ≤0.05) are included in these tables.

Supplement 9

Table S8. RNA-seq analysis comparing gene expression between groups of Oregon-R females exposed to NaAsO₂. Adult females were exposed to 0 mM and 1.0 mM NaAsO₂ for 1, 2, 4, 8, 24 or 48 hours prior to collection.

Supplement 10

Table S9. RNA-seq analysis comparing gene expression between groups of Oregon-R females exposed to NaAsO₂. Adult females were exposed to 0 mM and 1.0 mM NaAsO₂ for 1, 2, 4, 8, 24 or 48 hours prior to collection. Only genes displaying significant changes in gene expression (log₂ fold change ≥ |1| and an adjusted p-value of ≤0.05) are included in these tables.

Supplement 11

Table S10. Metabolomic analysis of Oregon-R Males exposed to NaAsO₂. Adult Males were exposed to 0 mM, 0.25 mM, and 1.0 mM NaAsO₂ for 2, 4, 8, 24, or 48 hours prior to collection. n=6 samples per condition; 20 adult males per sample. Data represented as ion counts normalized to sample mass.

Supplement 12

Table S11. Metabolomic analysis of Oregon-R females exposed to NaAsO₂. Adult females were exposed to 0 mM, 0.25 mM, and 1.0 mM NaAsO₂ for 2, 4, 8, 24, or 48 hours prior to collection. n=6 samples per condition; 20 adult females per sample. Data represented as ion counts normalized to sample mass.

Supplement 13

Table S12. The number of differentially expressed genes (DEGs) in NaAsO2-exposed adult male and female flies sorted by dose and time point. The tabulated DEGs were identified as having a log2 fold change ≥ |1| and an adjusted p-value of ≤0.05. The full list of DEGs can be found in Table S3, S5, S7, and S9.

Supplement 14

Table S13. The number of significantly altered metabolites in NaAsO2-exposed adult male and female flies sorted by dose and time point. The tabulated metabolites were identified as having a log2 fold change ≥ |1| and an adjusted p-value of ≤0.01. The full list of metabolites can be found in Table S10 and S11.

Supplement 15

Table S14. PANGEA analysis of differentially expressed genes in flies exposed to either 0.25 mM or 1.0 mM NaAsO₂ for 1,2,4,8,24, and 48 hrs. DEGs from each time point were simultaneously analyzed for enrichment using four annotation sources: (i) SLIM2 GO Biological Process, (ii) FlyBase signaling pathways (experimental evidence), (iii) KEGG Pathway (D. melanogaster), and (iv) REACTOME Pathway (D. melanogaster).

Supplement 16

Table S15. PANGEA analysis of differentially expressed genes in male and female flies exposed to either 0.25 mM or 1.0 mM NaAsO₂ for 1,2,4,8,24, and 48 hrs. DEGs from each time point were simultaneously analyzed for enrichment using the Preferred tissue (modEncode RNA_seq) gene set.

Supplement 17

Table S16. Core transcriptional response to NaAsO₂ exposure shared across sexes and doses. List of 100 genes consistently differentially expressed in adult Drosophila melanogaster under all four exposure conditions (male and female flies at 0.25 mM and 1.0 mM NaAsO₂). Genes are annotated with FlyBase identifiers.

Supplement 18

Table S17. Human orthologs of conserved arsenite-responsive genes identified by DIOPT analysis. Predicted human orthologs for the 100 Drosophila melanogaster genes consistently differentially expressed across all NaAsO₂ exposure conditions (see Table S16). Orthology was assessed using the Drosophila RNAi Screening Center Integrative Ortholog Prediction Tool (DIOPT). High-confidence orthologs (DIOPT score ≥10) highlight conserved pathways involved in detoxification, stress response, metabolism, chromatin architecture, and endocrine signaling.

Supplement 19

Table S18. PANGEA analysis of 100 genes consistently differentially expressed in adult Drosophila melanogaster under all four exposure conditions (male and female flies at 0.25 mM and 1.0 mM NaAsO₂).

Supplement 20

Table S19. Disease phenotype enrichment among DEGs in female and male Drosophila melanogaster exposed to NaAsO₂. Gene enrichment analysis was performed using PANGEA to identify disease-associated gene sets among differentially expressed genes (DEGs) in female and male flies exposed to either 0.25 mM or 1.0 mM NaAsO₂ across a 48-hour time course.

Supplement 21

Table S20. Statistical analysis of male metabolomics data. Metaboanalyst 6.0 was used to analyze data in Table S12. Statistical analysis conducted using ANOVA followed by a post hoc Fisher’s Least Significant Difference test.

Supplement 22

Table S21. Statistical analysis of female metabolomics data. Metaboanalyst 6.0 was used to analyze data in Table S11. Statistical analysis conducted using ANOVA followed by a post hoc Fisher’s Least Significant Difference test.