Comprehensive phenotyping and transcriptome profiling to study nanotoxicity in C. elegans

Introduction

The use of engineered nanoparticles has increased enormously over the last decade, and the nanotechnology industry has grown from a 10-billion-dollar enterprise in 2012 to being valued over one trillion dollars in 2015 (Gao et al., 2011). However, the potential impacts of these nanoparticles on environment and animals have not been fully characterized and further research is warranted. Nanoparticles are defined as particulate matter ranging from 1 to 100 nm in size with properties not exhibited by their larger bulk counterparts (Khanna et al., 2015; Capco & Chen, 2014; Maynard, 2011; Djurišić et al., 2015). The reactivity of nanoparticles depends on their size, charge, dose as well as the chemical composition of their coating (Medina et al., 2007). For instance, the surface area of smaller nanoparticles is larger compared to their larger counterparts, meaning they are more reactive and hence have a larger propensity of being toxic (Oberdörster, 2010).

Caenorhabditis elegans is a free-living soil nematode reaching approximately one millimeter in length in adult stage and has a relatively simple life cycle that can be grown on solid (i.e., nematode growth medium, NGM) or liquid media (i.e., S-medium) (Brenner, 1974; Lewis & Fleming, 1995). The small size and relatively cheap maintenance cycle of C. elegans make the nematode very amenable for various phenotype screening. C. elegans has been well-established as an in vivo model for testing the effects of heavy metals and novel anthelminthics (Kaletta & Hengartner, 2006; Ruiz-Lancheros et al., 2011). In terms of conservation of genes and biological pathways with humans, C. elegans shares 60 to 80% gene homology and possesses 12 of the 17 known signal transduction pathways (Kaletta & Hengartner, 2006; National Research Council, 2000).

To validate our use of C. elegans as a model for nanotoxicity, as this bacterivore worm constantly interacts with microbes in nature, which are ingested through the pharynx, the main potential route of exposure to nanoparticles is consequently oral (Pluskota et al., 2009). Similarly, human exposure to nanoparticles is also mostly through an oral route of entry, as nanoparticles are added to food in significant amount; the most prevalent ones being Ag, SiO₂, TiO₂ and ZnO (Fröhlich & Roblegg, 2016). For example, it is estimated a 70 kg individual ingests 126 mg of Ag nanoparticles per day in Europe (Dekkers et al., 2011). The nematode worm C. elegans, as a model organism, is thus valid for studying nanotoxicity in higher eukaryotic organisms such as humans. Additionally, a second route of exposure is through the worm’s vulval slit, where nanoparticles interfere with vulval cells and spermatecae (Scharf, Gührs & Von Mikecz, 2016). However, two routes of exposure to nanoparticles that cannot be studied in C. elegans are respiratory and dermal absorption, which are prevalent routes of exposure to nanoparticles for humans (Fröhlich & Roblegg, 2016).

The fast phenotyping of C. elegans can be coupled with transcriptome profiling (i.e., gene expression microarray or RNAseq) to study underlying molecular mechanisms. For instance, using gene expression microarray, Rocheleau et al. found that C. elegans exposed to nano-TiO₂ showed increased expression of the glutathione S-transferase gene gst-3 and the cytochrome P450 gene cyp33-11; while the oxidative stress response, as measured by the stress resistance regulator scl-1, showed increased expression after exposure to both nano- and bulk-sized TiO₂ (Rocheleau et al., 2015). In addition, the expression of pod-2, a reproduction-related gene, was decreased in a concentration-dependent manner with nano-TiO₂ exposure (Rocheleau et al., 2015). Based on the same technology, Starnes et al. identified that five lysosomal pathway related genes, including genes encoding the cysteine proteases cpr-1 and cpr-2, were changed significantly after exposure to silver (Ag) nanoparticles (Starnes et al., 2016). To the best of our knowledge, no transcriptome profiling has been reported to investigate SiO₂ nanoparticles in C. elegans. The main objective of the current study is to develop and to evaluate a C. elegans-based animal model to study nanotoxicity by integrating comprehensive phenotyping and transcriptome profiling. We selected Ag and five metal oxide nanoparticles (SiO₂, TiO₂, CuO, Al₂O₃ and CeO₂), and measured four endpoints (locomotion velocity, growth, reproduction and neurotoxicity) in C. elegans after exposure to these nanoparticles. Worms that exhibited the most significant effects were subjected to RNAseq to identify the affected biological processes and pathways. Hence, we offer a novel perspective to study nanoparticle toxicity using the soil nematode C. elegans.

Material and Methods

Results

The pooled average locomotion velocity for C. elegans grown under control conditions was 150.1 µm/s across all treatments, indicating that worms are active after growth in S-medium for 72 h at 21 °C. In comparison, the average locomotion velocity of worms decreased to 91.5 µm/s in the presence of 10 µg/ml Ag nanoparticles and further reduced to 44.2 µm/s in the presence of 50 µg/ml Ag nanoparticles, which was statistically significant for each concentration (P = 0.0001 and P < 0.0001, respectively) (Fig. 1A). Additionally, the difference in the reduction levels by the concentrations (>50%) indicated a dose-dependent decrease in average locomotion velocity in response to increasing Ag nanoparticle concentration. The second nanoparticle to influence the worms’ average locomotion velocity was SiO_2, which caused significant decreases at 10 or 50 µg/ml doses in comparison with control (P < 0.0001 for both concentrations). Surprisingly, it appeared that TiO₂ nanoparticles increased C. elegans average locomotion velocity at 10 µg/ml (P = 0.0257), although this could be a statistical artifact due to sampling effect based on the data distribution (Fig. 1C). Al₂O₃-_,TiO₂- and CuO-treated C. elegans showed decreases in average locomotion velocity when tested at 50 µg/ml (P = 0.0196, for Al₂O₃ and P < 0.005 for both TiO₂ and CuO), displaying locomotion velocities of 104.4, 114.8 and 112.2 µm/s, respectively (Figs. 1C–1E). CeO₂ nanoparticles had no effect at any of the concentrations tested (P > 0.05, for both concentrations).

For the growth inhibition (body length) assay, Ag nanoparticles had the greatest effect. The pooled average body length of worms grown in the control S-medium after a three-day incubation at 21 °C was 1199.6 µm, indicating that the worms grew efficiently in this medium. As with the locomotion velocity endpoint assay, a concentration-dependent decrease in body length was observed when worms were exposed to different concentrations of Ag nanoparticles, with average body lengths of 1017.8 µm under 10 µg/ml and 859.3 µm under 50 µg/ml, respectively (P < 0.0001 for both concentrations) (Fig. 2A). Exposure to SiO₂ nanoparticles induced significant decrease at 10 µg/ml (P < 0.0001) (Fig. 2B), but no further significant decrease was observed at 50 µg/ml. Exposure to CuO nanoparticles also resulted in a concentration-dependent decrease in body length, leading to an average body length of 1055.3 and 923.1 µm, respectively (P < 0.0001 for both concentrations) (Fig. 2F). For CeO₂ and Al₂O₃ nanoparticles, significant decreases in body length were observed at 10 µg/ml compared to control worms (P < 0.0001, Figs. 2D and 2E). For TiO₂ nanoparticles, a significant difference was observed only at 50 µg/ml, in which the exposed worms showed an average body length of 1130.8 µm (P = 0.001) (Fig. 2C). In summary, these results show Ag and SiO₂ nanoparticles have similar toxicity on C. elegans, although the effect appears to be concentration-dependent for Ag nanoparticles whereas there is likely a threshold effect for SiO₂ nanoparticles.

To measure the effects of the various nanoparticles on C. elegans reproduction, we incubated five L4-young adults in S-medium supplemented with E. coli OP50 and the respective nanoparticles for four days (96 h) at 21 °C. It was expected that each worm would lay approximately 300 eggs in that time span, so that the total number of progeny per 5 worms would be near 1,500 under control conditions (Sulston & Hodgkin, 1988; Sonowal et al., 2017). The average control value of progeny produced by 5 worms under our experimental conditions was 1,288 after 96 h. We found that most nanoparticles reduced the number of progeny produced by C. elegans. At 10 µg/ml, Ag nanoparticles decreased the brood size of C. elegans to around 37%, which was statistically significant (P < 0.0001) (Fig. 3A). The effect was even more pronounced at 50 µg/ml, as Ag nanoparticles decreased the number of progeny to 33% of the control value, suggesting that these nanoparticles do indeed decrease C. elegans brood size (P < 0.0001) (Fig. 3A). On the other hand, SiO₂ nanoparticles decreased brood size substantially at both 10 µg/ml and 50 µg/ml (P < 0.0001) (Fig. 3B), indicating that SiO₂ nanoparticles are potent inhibitors of C. elegans reproduction in concentrations ranging in µg/ml. In contrast, TiO₂ nanoparticles, known to inhibit C. elegans reproduction, reduced brood size to about 80% of the control value in our testing concentration range (Fig. 3C). CeO₂ nanoparticles inhibited C. elegans reproduction as well. The brood size decreased to 55% of the control-treated value at a concentration of 10 µg/ml (P = 0.0002) (Fig. 3D). Interestingly, at 50 µg/ml, the decrease was not as pronounced, equating to 89% of the control-treated value (P > 0.05) (Fig. 3D). This observation may be due to the aggregation of CeO₂ at higher concentrations. Al₂O₃ nanoparticles did not show statistically significant effects on C. elegans under our conditions (P > 0.05) (Fig. 3E), whereas CuO nanoparticles decreased the brood size value to 83% and 71% of the control value, at 10 µg/ml and 50 µg/ml, respectively (Fig. 3F).

Both Ag and SiO₂ nanoparticles showed no significant impact of the number of head thrashes based on a population of worms in 60 s at 10 µg/ml versus control worms (Figs. 4A and 4B). SiO₂ nanoparticles showed a slight effect on the neurotoxicity assay at 50 µg/ml (P = 0.0033), whereas Ag nanoparticles had no such effect on neurotoxicity. Similar results were observed for TiO₂ and nanoparticles at both concentrations tested (Fig. 4C). In contrast, CeO₂, Al₂O₃ and CuO nanoparticles showed significant effects on neurotoxicity at 10 µg/ml under our conditions, as determined by a significant decrease in the number of head thrashes, and this trend was conserved at 50 µg/ml (P < 0.0001 for both CeO₂concentrations, P = 0.0087 and P < 0.0001 for Al₂O₃ at 10 µg/ml and 50 µg/ml, respectively and P = 0.0002 for CuO at 10 µg/ml), although the effect on neurotoxicity of CuO nanoparticles at 50 µg/ml was not significant (P > 0.05) (Figs. 4D and 4F).

Based on the phenotyping results, we further performed RNAseq analysis on C. elegans exposed to Ag and SiO₂(10 µg/ml) as these two nanoparticles demonstrated the most outstanding effect on majority of the parameters (locomotion velocity, growth and reproduction). For each sample (5 replicates for the Ag, SiO2 and control groups), around 20 million reads were obtained. Approximately 97% of reads were mapped to the worm’s genome and a total of 18,861 gene sequences were identified, using a minimal total read count of 3 across samples. Sample distribution by principal component analysis (PCA) is shown in Fig. S1. Differentially expressed genes (DEGs) based on edgeR between control and the toxicity groups (2,648 DEGs in the Ag group and 1,087 DEGs in the SiO2 group) are shown in Data S1 (FDR < 0.05).

Gene set enrichment analysis (GSEA) based on Gene Ontology biological processes (BPs) showed various statistically enriched positively or negatively based on the fold changes and running enrichment scores (Data S2). The top 20 and phenotype reflecting enriched BPs are shown in Table 2 (FDR < 0.05). These BPs were related to various physiological events such as cellular and metabolic responses. Apoptotic process showed gene enrichment with positive NES in both Ag and SiO2 groups under significant levels (FDR < 0.0001). Phenotype reflecting BPs including regulation of locomotion, reproductive process and cell growth were enriched with negative NES significantly (FDR < 0.0001 for regulation of locomotion, FDR = 0.001 for regulation of reproductive process, FDR = 0.002 for regulation of cell growth in the Ag group; FDR < 0.0001 for regulation of locomotion, FDR = 0.008 for regulation of reproductive process, FDR = 0.042 for regulation of cell growth in the SiO2 group) (Fig. 5). A number of genes were commonly detected within the top five category in the phenotype reflecting enriched BPs for both Ag and SiO2 groups, including transcription factor (che-1) and MiRP K channel accessory subunit (mps-1) in regulation of locomotion, caveolin (cav-1) in regulation of reproductive process, and cyclic nucleotide-gated cation channel (tax-4) and protein let-756 (let-756) in regulation of cell growth (Data S2).

The other significantly enriched BPs were associated with various physiological events and immune defense of the organism. As both Ag and SiO₂ nanoparticles were foreign substances for the worm, we further examined the expression profiles of genes associated with innate immunity. Compared to Ag and Control, SiO₂ exposure led to significant downregulations of genes related to the innate immune response (FDR < 0.0001) (Fig. 6).

The enriched BPs by GOATOOLS were mainly related to the physiological events, of which, some were identical to those observed by the GSEA analysis (Data S2). However, the number of the BPs detected by GOATOOLS was lower compared to those detected by GSEA, which was likely related to the differences in gene inputs (DEGs with FDR < 0.05 vs cut-off free DEGs ranked by fold change) and/or the methodologies used. Nevertheless, the phenotype reflecting BPs including the regulations of locomotion and reproductive process were also enriched in the Ag group in the GOAtools analysis. However, this was not the case for the SiO2 group, which appeared to be related to lower number of DEGs in this group, compared to the Ag group.

Analysis of the RNAseq data using WormExp showed gene enrichment in newly determined terms (Data S2). Some of these terms were found associated with regulations of locomotion, reproduction and cell growth in both groups. The “glp-1 mutant” term was enriched in the Ag and SiO2 groups, which refers to the diminished reproductive capability of C. elegans (Gracida & Eckmann, 2013). In regard to development terms, “pgl-1 mutant” and “P-granule RNAi” were enriched in the Ag and SiO2 groups, respectively, linking regulation of cell growth (Knutson et al., 2017). The “wdr-23 mutant” term was also present for both nanoparticle exposure groups. This is notable as wdr-23, through the action of skn-1, is involved in proper locomotion of C. elegans (Staab et al., 2013). Finally, the other terms, including regulation by heavy metals/NPs (such as Ag), were found for both groups, further validating the nanoparticle effect. Altogether, the WormExp gene enrichment terms, obtained from the RNAseq data, appear to be in agreement with the phenotypic assay results.

Pathway enrichment analysis against the KEGG database showed various significantly enriched pathways (Data S3). The top 20 enriched biological pathways are shown for both exposures in Table 3. Ribosome, proteasome, aminoacyl-tRNA biosynthesis and RNA transport were significantly upregulated in both groups, indicating overall higher rate of protein turnover upon exposure. In contrast, biological pathways reflecting phenotypes including neuroactive ligand–receptor interaction [regulating locomotion (Kong et al., 2015)], wnt-signaling (regulating reproduction and cell growth (Inoki et al., 2006; Hernandez Gifford, 2015) and MAPK signaling (regulating reproduction and cell growth (Zhang & Liu, 2002; Andrade et al., 2014) were significantly down-regulated in both exposures (Fig. 7). Some genes were found commonly within the top 5 enriched gene category in each enriched biological pathway for both Ag and SiO₂ treatments (based on fold change and running enrichment score), which were tachykinin receptor family (tkr-2), serotonin/octopamine receptor family (ser-1) in neuroactive ligand–receptor interaction pathway, skp1 related (ubiquitin ligase complex component, skr-8, skr-10, skr-12) in wnt signaling pathway, protein ver-1 (ver-1) and heat shock protein (hsp 70) in MAPK signaling pathway (Data S3).

Discussion

As the use of nanoparticles has increased dramatically in recent years, there is a growing concern regarding their potential impact to environment and human health. In this study, we have systematically evaluated a C. elegans-based animal model for nanotoxicity assessment. Our results have shown that Ag and SiO₂ have the most potent toxic effect on locomotion velocity and growth, as well as reproduction (brood size), but not on neurotoxicity. In this model, the transcriptome profile is concordant with the phenotype characteristic for both exposures (Fig. 8).

The top 20 GO BPs identified by the GSEA were related to various physiological events in the Ag and SiO2 toxicities. The exposure to both nanoparticles downregulated multiple regulatory biological processes, including regulation of locomotion, regulation of reproduction and regulation of cell growth, which was consistent with the phenotype profiling of our study.

Dysfunction of the enriched genes including transcription factor che-1 (che-1) and MiRP K channel accessory subunit (mps-1) (regulation of locomotion), caveolin (cav-1) (regulation of reproductive process), and cyclic nucleotide-gated cation channel (tax-4) and protein let-756 (let-756) (regulation of cell growth), have been previously shown to hinder the worm’s biological events (Uchida et al., 2003; Bianchi et al., 2003; Scheel et al., 1999; Komatsu et al., 1996; Roubin et al., 1999). Inactivation of che-1 (mediating chemotaxis through ASE neurons) and of mps-1 (a voltage-gated pore-forming potassium subunit) impairs neuronal activities such as chemotaxis and locomotion (Uchida et al., 2003; Bianchi et al., 2003). Caveolin-1 (cav-1), identified in the adult germ line and highly expressed in eggs, is required for Ras/MAP-kinase-dependent progression (Scheel et al., 1999). Inhibition of tax-4 and let-756 genes hinders chemosensation and results in larval arrest, respectively (Komatsu et al., 1996; Roubin et al., 1999).

Although biological processes reflecting phenotypes were similar between both toxicities in our study, SiO₂ nanoparticles induced a remarkable downregulation pattern in innate immune response, compared to Ag and Control. In particular, several C-type lectins, which are known to be important components in innate immunity (Mayer, Raulf & Lepenies, 2017), appeared to be exclusively downregulated by SiO₂ nanoparticles. The subject of nanoparticle exposure and the effects on immune system has been an active research area (Boraschi et al., 2017). Exposure to nanoparticles has been linked to changes in the immune response such as inflammation, hypersensitivity and immunosuppression and has been shown to induce such responses through antigen-presenting cells in humans, highlighting the interaction between nanoparticles and the innate immune response (Alsaleh & Brown, 2018). Biocoronas, formed by the interaction of the nanoparticle surface with proteins and lipids, are highly reactive immunologically and have recently gained the attention of regulatory agencies (Shannahan, 2017).

The top 20 significant KEGG pathways identified in both exposure studies are similar to the findings based on GO BPs. In particular, the regulatory biological pathways linked to phenotypes, including neuroactive ligand–receptor interaction (relates to locomotion Inoki et al., 2016), wnt and MAPK signaling pathways (relates to reproduction and cell growth (Inoki et al., 2016; Hernandez Gifford, 2015; Zhang & Liu, 2002; Andrade et al., 2014)), were found to be downregulated in both experiments. The pathway interaction analysis by ClueGO showed that wnt signaling pathway was interacting with tgf-beta pathway which was also enriched with significantly negative NES in both toxicities. These signaling pathways are known to interact with each other and control adult tissue homeostasis (Warner, Greene & Pisano, 2005). The downregulation of genes involved in neuroactive ligand receptor interaction likely to be responsible for the changes in locomotion.

The genes within the top 5 enriched gene category in the indicated pathways, commonly observed in both toxicities, were tachykinin receptor family (tkr-2), serotonin/octopamine receptor family (ser-1) (neuroactive ligand–receptor interaction pathway), skp1 related (ubiquitin ligase complex component, skr-8, skr-10, skr-12) (wnt signaling pathway), and protein ver-1 (ver-1) and heat shock protein (hsp 70) (MAPK signaling pathway). The proteins encoded by tachykinin receptor and ser-1 genes regulate locomotion via mediate neurotransmission and indirect modulation of neuromuscular circuits, respectively (Pennefather, 2004; Dernovici et al., 2007). The proteins encoded by skp1 related genes (such as skr-8 and skr-10) are known to be a core element of SCF ubiquitin-ligase complexes and involved in posterior body morphogenesis, embryonic and larval development, and cell proliferation in C. elegans (Nayak et al., 2002). Putative vascular endothelial growth factor receptors (VERs) of C. elegans and Hsp70 chaperones act in the PVF-1 signalling pathway for ray 1 positioning (in the male worms) and mediate protein folding, influencing various regulatory proteins, respectively (Dalpe et al., 2013; Mayer & Bukau, 2005). Overall, these genes may play significant roles on locomotion, reproduction and cell growth in response to nanotoxicity.

Apart from the many similarities observed between the two nanoparticle effects, Ag and SiO₂ also showed opposite effects on some of the top 20 enriched biological pathways. Spliceosome, mRNA surveillance and endocytosis pathways were enriched with positive NES in Ag, but with negative NES in SiO2, despite no obvious phenotypic differences were observed in our studies.

Comparison of the transcriptomics changes during Ag and SiO₂ exposures with the findings in previous studies on metal toxicities indicated that metallothionein-2 (mtl-2), a commonly observed responsive gene to the metal toxicities (conserved in C. elegans and mammals) (Caito et al., 2012; Cui et al., 2007; Roh, Lee & Choi, 2006; Kumar et al., 2015; Anbalagan et al., 2012), was up-regulated in these groups (FDR =0.002 for Ag, FDR = 0.043 for SiO2), which further confirmed the effectiveness of the model.

Ag and SiO₂nanoparticles have been shown to affect locomotion velocity in C. elegans N2, as previously described (Jung et al., 2015). In addition, Ag and SiO₂ have been shown to reduce brood size, according to several studies (Kleiven et al., 2018; Wu et al., 2013; Pluskota et al., 2009). In contrast, we did not observe a significant decrease in brood size to exposure to 10 and 50 µg/ml of TiO₂ nanoparticles, as observed by Wu et al. (2013), indicating differences in our conditions, as these authors saw a slight decrease in progeny production when using these nanoparticles in the µg/l concentration range. The effects of the other nanoparticles used in this study (i.e., CuO, Al₂O₃, and CeO₂) are relatively unknown based on the literature. In terms of neurotoxicity experiments, as determined by the number of head thrashes per minute in our study, our results differ from published results (Wu et al., 2013; Pluskota et al., 2009; Piechulek & Von Mikecz, 2018) which show that Ag and SiO₂ are neurotoxic at lower concentrations than the ones used in this study, although SiO₂ nanoparticles were found to have a modest effect at 50 µg/ml on neurotoxicity under our conditions. Indeed, we noticed that many neuron system-related BPs or pathways were enriched with negative NES in both Ag and SiO2 groups appear to have limited influence on head thrashing in our analysis. Additionally, our study is different compared to the indicated studies, as we only looked at the effect of nanoparticles incubated from the L1 stage to the adult day one stage. We observed that CeO₂, Al₂O₃ and CuO nanoparticles had significant neurotoxicity at 10 µg/ml, as determined by the decrease in number of head thrashes per minute, which is a novel observation. The reported differences in this study may be because the Ag and SiO₂ nanoparticles used in our study were confirmed to have a spherical shape under the manufacturer’s test conditions (Table 1), whereas the others (TiO₂, CeO₂, Al₂O₃ and CuO) were unconfirmed to adopt any shape at all. Further study is required to elucidate the answer to this question. It was shown that co-feeding nanoparticles with E. coli OP50 in S-medium leads to the uptake of these nanoparticles through the pharynx and absorption through the gut (Piechulek, Berwanger & Von Mikecz, 2019). We speculate this to be true as well under our conditions. Silica (SiO₂) nanoparticles were shown to inhibit the peptide transporter OPT-2/PEP-2, present on the apical layer of the intestinal membrane in C. elegans (Piechulek, Berwanger & Von Mikecz, 2019). Inhibition of the OPT-2/PEP-2 transporter leads to the accumulation of silica nanoparticles in gut granules, indicating they are taken up within the organism. Fang-Yen et al. showed that particles with a diameter range of 0.5 µm to 3 µm are taken up by the pharynx (Fang-Yen, Avery & Samuel, 2009). We propose that this size range is circumvented when nanoparticles are co-fed with E. coli OP50 to gain access to the gut for absorption.

The observed effects in the various experiments can be either nanoparticle-specific or compound-/element-specific. Nanoparticles, ranging from 1 to 100 nm in size, are larger than their elemental constituents, which are metal cations under our experimental conditions. We reason that the observed effects in the various experiments and RNAseq analysis are due mainly to nanoparticles, although nanoparticles, such as Ag, release positively-charged ions upon incubation in liquid media and the proportion of released cations is small (Lekamge et al., 2018). In addition, metal oxide nanoparticles tend to release ions into liquid medium depending on the cationic metal charge (Simeone & Costa, 2019). For example, where Z is the oxidation number of the metal cation, oxides of nanoparticle cations with Z ≤ 2 dissolve more than 10%, whereas this fraction is reduced to less than 1% for nanoparticle oxides of cations with Z >3 (Simeone & Costa, 2019). According to this relationship, we would expect that less than 1% of the metal oxide nanoparticles used in our study to be dissolved, except for CuO nanoparticles which has a metal cation Z value of 2. As evidence for our reasoning, the WormExp enrichment analysis identified the “Ag NPs” term (Data S2).

To address whether the nanoparticles used in our study affected E. coli OP50 growth and thus C. elegans feeding, we incubated E. coli OP50 with 50 µg/ml of each nanoparticle in S-medium (Fig. S2). Compared to E. coli OP50 alone, the treated E. coli OP50, with the highest concentration of nanoparticles (50 µg/ml), demonstrated only slight growth defects, as determined by measuring bacterial density (Fig. S2). Antimicrobial nanoparticles, such as Ag nanoparticles, are antimicrobial as they interact with bacterial membranes and proteins through the released metal cations (Sondi & Salopek-Sondi, 2004). However, this effect seems to be minimal under our experimental conditions (Fig. S2), likely due to the low dissolution rate of Ag⁺ cations (Lekamge et al., 2018). Ag nanoparticles had only a major effect on bacterial density after five days compared to the other nanoparticles tested (Fig. S2). The same was observed for CuO nanoparticles (Fig. S2). Therefore, a constant source of E. coli OP50 food, as determined by bacterial density, was available during the course of the various experiments.

Conclusions

The aim of this study was to evaluate the effects of various nanoparticles on C. elegans using standard phenotyping assays and characterize transcriptomics changes of the worms exposed to selected nanoparticles (which showed the toxic effects for the majority of the parameters tested). With all these observations, we provide a novel angle to study the toxicity of nanoparticles on organisms, by exploring the mode of action of Ag and five metal oxide nanoparticles on different life history endpoints in C. elegans. To the best of our knowledge, this is the first study that integrates phenotype screening with RNAseq to investigate nanotoxicity in intact animals using C. elegans. Our RNAseq data not only confirmed positive enrichment of apoptotic process as reported in the literature (McShan, Ray & Yu, 2014; Clement & Jarrett, 1994; Kim et al., 2015), it also revealed that toxicities induced by both nanoparticles have down-regulated genes in multiple important regulatory biological processes and pathways, with opposite effects on innate immunity.

Supplemental Information