Reliable Screening of Dye Phototoxicity by Using a Caenorhabditis elegans Fast Bioassay

Phototoxicity consists in the capability of certain innocuous molecules to become toxic when subjected to suitable illumination. In order to discover new photoactive drugs or characterize phototoxic pollutants, it would be advantageous to use simple biological tests of phototoxicy. In this work, we present a pilot screening of 37 dyes to test for phototoxic effects in the roundworm Caenorhabditis elegans. Populations of this nematode were treated with different dyes, and subsequently exposed to 30 min of white light. Behavioral outcomes were quantified by recording the global motility using an infrared tracking device (WMicrotracker). Of the tested compounds, 17 dyes were classified as photoactive, being phloxine B, primuline, eosin Y, acridine orange and rose Bengal the most phototoxic. To assess photoactivity after uptake, compounds were retested after washing them out of the medium before light irradiation. Dye uptake into the worms was also analyzed by staining or fluorescence. All the positive drugs were incorporated by animals and produced phototoxic effects after washing. We also tested the stress response being triggered by the treatments through reporter strains. Endoplasmic reticulum stress response (hsp-4::GFP strain) was activated by 22% of phototoxic dyes, and mitochondrial stress response (hsp-6::GFP strain) was induced by 16% of phototoxic dyes. These results point to a phototoxic perturbation of the protein functionality and an oxidative stress similar to that reported in cell cultures. Our work shows for the first time the feasibility of C. elegans for running phototoxic screenings and underscores its application on photoactive drugs and environmental pollutants assessment.


Introduction
The phototoxic properties of many compounds were demonstrated long ago [1], consisting in the capability of certain molecules to be activated when subjected to suitable illumination. When this effect requires molecular oxygen it has been named photodynamic action, and allowed the development of a new cancer treatment called "photodynamic therapy" ("PDT") [2].
In a PDT treatment patients are treated with the drug and, once absorbed, the tumor area is subjected to light irradiation in order to activate drug phototoxicity and destroy cancer cells. The molecular mechanism of photoactivation is based on the generation of reactive oxygen species (ROS, and mainly singlet oxygen, 1 O 2 ), which damage cellular structures and induce cell death, either by apoptosis or necrosis [3][4][5]. Specific photoactivable drugs have been discovered and are currently used in human healthcare [6], and several common dyes have shown photodynamic activity. Some well characterized molecular structures with photoactive properties are thiazine [7][8][9], xanthene [10], acridine [11,12], and triarylmethane [13] derivatives.
At the moment, most of the work on photoactivity of dyes has been concerned with biomedical applications [14], particularly PDT of cancer [4,15,16]. However, this particular effect spreads to other fields of interest such as photosterilization of water and blood products [9,12], and environmental pollution [17][18][19][20]. It is worth to note that dyes used in industrial activities often present mutagenic, carcinogenic, and genotoxic photoactivity [15,21,22], which can result in a relevant risk for both human and environmental health.
Biological models, from microorganisms to higher organisms, have been thoroughly employed to evaluate phototoxic effects of chemicals in order to explore the potential of new drugs for PDT or to prevent environmental damage from xenobiotics. Several bioassays for the detection of photodynamic effects are currently on use, examples being Paramecium [30], Candida [31], Allium [32], Drosophila [22], Daphnia, sea urchin [10], amphibian embryos [33], cell cultures [34], erythrocyte hemolysis [35], and intradermal tests [36].
In order to assess more precisely the phototoxicity of compounds such as drugs or pollutants, the use of a small translucent whole organism seems to be advantageous for simple, rapid and cheap bioassays. One of the simplest animal models is the nematode Caenorhabditis elegans, which has a fully sequenced genome, rudimentary organs similar to those found in mammals, and highly conserved cellular signaling pathways [37]. Also, the ease to culture, growth and reproduction of this organism has helped to establish it as a very suitable biological model for a variety of assessments. The main attraction of C. elegans for use in pharmacological and toxicological screening consists in the capability to be cultured in microplates (96 or 384 well plates), with small amounts of liquid medium, where animals can be incubated with experimental compounds [38].
In a whole animal approach, treatment with toxic doses will cause death, whereas subtoxic effects will affect animal behavior. Based on this premise, we employed here a motility tracking system, previously developed in our laboratory [39], to characterize the effects of thirty seven dyes, some already known to have phototoxic properties. As far as we know, this is the first time that the C. elegans model has been used for screening of this kind of compound.

Animal culture and strains
For the phototoxicity, uptake assays, and for qRT-PCR experiments, synchronized populations of Caenorhabditis elegans SS104 strain (glp-4 [bn2], a temperature sensitive sterile mutant) were cultured at non permissive temperature (25°C) in Nematode Growth Medium as described previously [40]. On the 1 st day of adult stage, animals were washed twice in saline solution (modified K-medium: 52 mM NaCl and 32 mM KCl [41] + 0.01% Triton X-100) and transferred to flat bottomed 96-well microplates (Greiner Bio-one Cat. 655101). An average of 50 animals was used per well with a final volume of 100 μl.

Behavioral phototoxicity measurement
One hour after pipetting, basal motility of worms within the wells was assessed using an infrared tracking device (WMicrotracker, Designplus SRL, Argentina) [39]. This basal activity was recorded to normalize any subsequent activity variations to that initial activity, eliminating differences between wells due to population size. After basal measurement, dyes were added to the culture medium at final concentrations of 0.1, 1, 10 or 100 μM and incubated for 1 h before light irradiation (where applied). Dye photoactivation was carried out exposing the microplates for 30 min under a fluorescent white light source (2700°K white fluorescent lamp R7s 20W, Sica, Argentina) at 10 mW/cm 2 of intensity. An additional water IR-filter (3-cm wide) was used to avoid heating, as previously reported [20].
Locomotor activity was tracked continuously in darkness during the incubation with the dye and for 4 h after dye activation. At least 4 replicate wells were used for each experiment, and the reported concentration was repeated independently three times, unless otherwise mentioned in the text.

Chemical treatment
The dyes employed in this study, as well as their characteristics and known properties, are shown in Table 1. After a preliminary determination, compounds were used at 100 μM concentration (unless otherwise indicated).

Statistical analysis
In order to quantify the toxic properties of the compounds in worms, a parameter called Vitality Rate (VR) was calculated as the ratio between dye treated worms and contol animals in the same assay. Based on previous experience on tracking C. elegans, where 10% variations in locomotor activity are common over continous 30 min periods of activity tracking (S1 Table), an activity drop below 0.8 of the control population level (significantly different from 1, p < 0.05 ONE sample t-test) was set to be considered a positive toxic effect.

Dye uptake
Worms were exposed to dyes as described previously, with the difference that the culture medium with the chemical was removed from the medium after the 1 h incubation, and worms were washed inside microplates 4 times with saline solution (modified K-medium). After this treatment, the dye accumulated inside the animal body was photoactivated for 30 min, and locomotor activity was recorded for 4 hours as described previously. In parallel, 3 replicate wells were observed using bright-field and fluorescence microscopy under blue or green excitation light with the purpose of visually assessing dye uptake. The accumulation of dyes was arbitrarily quantified by the intensity of fluorescence or staining using a 4 color scale. Drug response pathway determination The previously indicated GFP reporter strains were cultured in NGM plates until adulthood, and treated as described above for the behavioral assessment experiments. GFP expression was determined by visual inspection and imaged using a fluorescence stereoscope (OLYMPUS model: SZX-ILLK100) 4 h after dye photoactivation. In addition, RNA samples were prepared from a population of adult worms (SS104 strain) treated with 1 μM rose Bengal, our positive control dye, and hsp-4 plus hsp-6 genes were quantified by qRT-PCR using the protocol reported previously by Buzzi et al [42]. The ama-1 gene was employed as a constitutively expressed internal control. Primers used for real time PCR are as follows: for hsp-4, hsp-4F 5'GCAGATGATCAAGCCCAAAAAG3'and hsp4R 5'GCGATTTGAGTTTTCATCTGATA GG3'; for hsp-6, hsp-6F 5'GGACAAACCAAAGGGACATG3' and hsp-6R 5'AACGAATG CTCCAACCTGAG3'; for ama-1, ama-1F 5'CCTACGATGTATCGAGGCAAA3' and ama-1R 5'CCTCCCTCCGGTGTAATAATG3'. All experiments were replicated 3 times.

Phototoxicity assessment in adult Caenorhabditis elegans
To perform phototoxicity experiments we developed a stepped protocol, where all animals are subjected to the compounds in the dark and then either exposed to the light (Fig 1A) or kept in darkness as control (Fig 1B). Their motility was recorded with the infrared tracking method for 1 h before exposure to light and 4 h after in order to record the behavioral response. Firstly, two well characterized dyes were tested, Luxol fast blue (not phototoxic) and rose Bengal (a photoactive molecule). Neither caused an effect without light irradiation (Fig 1B).
In contrast, when white light irradiation (10 mW/cm 2 , 30 min) is applied (Fig 1A), a significant reduction in locomotor activity is observed in the positive control (rose Bengal), and it remains low even after 4 h. Unexpectedly, an immediate response is also observed in control animals and with a negative control (Luxol fast blue) just after light irradiation, possibly as a transient response to light. Since this short term response is magnified in animals treated with phototoxic molecules, we decided to divide biological effects in "immediate phototoxicity" and "late phototoxicity".
We then selected a set of 37 dyes to test phototoxic effects in C. elegans cultures. Of these, 16 have been already reported to present effects in other organisms (Table 1). Of the assessed list of 37 compounds, 16 presented immediate phototoxic effects and 8 late phototoxic effects ( Table 2, Fig 2A and 2B). In order of decreasing phototoxicity, the top 5 dyes were phloxine B > primuline > eosin Y > acridine orange > rose Bengal. Many of these dyes have been previously reported as phototoxic, consistent with our results (Table 1).
In order to discard any synthetic effects between SS104 and a particular chemical, we test top 12 dyes in wild type (N2) young adult animals. As shown in S2 Table, although some sensitivity differences are observed, 100% of compounds manifest similar phototoxic behavior response.
Phototoxic molecules are acting directly from within the animal body Since we measured phototoxicity without washing the compound from the medium, we asked whether the toxic effects were mainly caused by dye molecules in the culture medium or by those absorbed inside the animal. In order to clarify this point we replicated the experimental setup for 15 positive phototoxic molecules, washing the medium previous to light irradiation. Interestingly, phototoxic activity of all retested compounds remained after washout (Fig 3). Moreover, an increase in the phototoxic effect is observed in most compounds after washout.
We hypothesize this increment could be attributed to higher light exposure of the worm, since the dye in the medium can be absorbing some light.
Also, in order to confirm that the compounds did in fact enter the animals, we carried out observations under the stereoscope to assess the staining or fluorescence due to dye uptake. Almost all phototoxic compounds displayed positive staining or fluorescence in living animals, demonstrating the permeability of this animal model to the dyes (Fig 4).  Stress response genes are turned on in response to some phototoxic molecules in worms Since it has been reported that phototoxic compounds produce oxidative stress and mitochondrial damage, we studied whether or not the compound were mainly active in any particular cellular compartment. For this purpose two mitochondrial-specific stress response proteins (hsp-6::GFP and hsp-60::GFP) and one endoplasmic reticulum stress response protein (hsp-4:: GFP) were used. Of the 16 phototoxic compounds tested (acridine orange was rejected due to its intrinsic green fluorescence), 10 were shown to stress one or both compartments while the other 6 compounds did not seem to be active in perturbing either the mitochondria or the endoplasmic reticulum (Fig 5B).
Finally, to give more support to the idea that our finding on the GFP reporter lines reflected what was happening in the phototoxicity screening, a qRT-PCR was performed on 1 μM rose Bengal exposed glp-4 worms (Fig 5C). At this low concentration, hsp-4 expression was increased 3.8 fold with respect to control (n = 3, p<0.05) while hsp-6 expression remained unaffected. No transcription was found under dark dye treatment (no photoactivation). These results confirm the response pattern observed in the reporter lines.
Correlation between our screening results and previously reported phototoxicity of compounds A set intersection between our worm experiments and published data is shown in Fig 6. When percentage of false detections is analyzed we found a rate of 8.1% of false negatives (dyes reported as phototoxic, but not detected in our experiments) and 2.7% of false positives (dyes reported as non phototoxic but detected in our system) compared to literature reports. These differences could be attributed to variability in the uptake, efficiency for ROS production, experimental protocol, as well as specific sensitivity and threshold to oxidative stress for distinct type of cells and organisms.

Discussion
Simple animal models are becoming useful to perform in vivo drug discovery experiments and testing for biological activity. In this work we propose the application of the nematode C. elegans as a reliable model for assessing phototoxicity of dyes as prototypical compounds and the utility of a simple behavioural measurement (such as global locomotor activity) as a direct readout for toxic and sub-toxic effects.
Common photoactivity experiments using cell cultures are based on the incubation with compounds, followed by washing, light exposure and measurement of viability by staining or colorimetric methods. In this work we start the experiments without dye washing. Since similar results were observed after washing of dyes, we conclude that this faster approach can be used in a rapid screening of molecules, previous or complementary to cell culture measurements. At the biological level, photoactive molecules appear to be permeating the animal cuticle or digestive tract, acting from inside the animal body (as shown in uptake experiments, Figs 4 and 5) and resulting in a reduced motility response.
It is worth to note that we were able to measure activation of the stress response machinery in at least 50% of positive compounds (Fig 5). Damage in intracellular compartments associated with protein misfolding and mitochondrial electron transport disruption are well-known mechanisms of phototoxicity action in cell cultures [3]. Our results suggest a similar mechanism of damage in worms. It might be useful to evaluate a larger set of stress-reporter strains Phototoxicity Bioassay Using Caenorhabditis elegans and, even more, the amount of ROS driven by dye photoexcitation in C. elegans in order to gain a deeper understanding of the molecular mechanisms of action. These issues are the objectives of ongoing research.
In order to compare the photodynamic action of each molecule, we can define a Phototoxic Index (PI) by analyzing the phototoxicity and intrinsic toxicity of the compound. The occurrence and degree of photosensitization could be expressed as the ratio: PI = effects with light / effect without light. Thus, the five dyes showing high immediate phototoxicity would present the following values: PI (phloxine B) = 9.81; PI (primuline) = 5.45; PI (eosin Y) = 4.2, PI (acridine orange) = 2.78, and PI (rose Bengal) = 2.19. It should be noted that the higher the PI is, the stronger the phototoxic effect.
Finally, it is important to mention that this animal model and simple protocols for biotoxicity-detection are useful not only for pharmacological testing but also in ecotoxicology assays. Regarding environmental pollution, around 10,000 types of dyes and pigments are produced annually worldwide and extensively used in textile, leather, plastic and printing industries, laboratory work, and as food, pharmaceutical and cosmetic additives. About 10-15% of the total dyes used in dyeing processes are released in wastewater [17]. Therefore, contamination by dyes [19] and the resulting phototoxicity represent a significant risk for human health and wildlife preservation. In addition, other pharmacological agents showing undesired photoactivity must be taken into account (e.g., antiinflammatory, anxiolytic, antirheumatic, antibacterial, and antiparasitic drugs), which have also revealed to be phototoxic [21,24,26,29,36,43,44]. As a logical consequence, it is increasingly necessary to evaluate the phototoxicity of possible drugs or xenobiotics to induce or prevent, respectively, biological effects through the design and development of simple, precise and cheap bioassays for oxidative stress-dependent phototoxicity studies. The present results using an automated tracking device for assessing population motility of a whole organism such as C. elegans show that this bioassay is very suitable for Stress response machinery is activated in C. elegans under phototoxic treatment. Two mitochondrial-specific Stress Response (SR) lines (hsp-6::GFP and hsp-60::GFP) and one endoplasmic reticulum SR line (hsp-4::GFP) were subjected to the phototoxicity protocol, and GFP-gene expression was observed 4 hours after dye photoactivation. A) Representative scale determination of each reporter line. B) Observed gene expression for 15 phototoxic dyes tested at 10 μM, 25 μM and 50 μM. N = 3. C) Photo-activated Rose Bengal stress response measured by qPCR. Hsp-4 was assayed as a marker of endoplasmic reticulum stress and hsp-6 was assayed as a marker of mitochondrial stress. RNA polymerase subunit gene (ama-1) was employed as internal reference gene. N = 3 ± SEM.
doi:10.1371/journal.pone.0128898.g005 Supporting Information S1 Table. Characterization of activity measurements using wild type animals (N2 strain). In order to estimate the average activity and deviation over a 4 h recording period, 8 biological replicates, independently repeated, are shown in the table. Each test was normalized to the first half hour of recording. Four wells replicates, and 50 worms per well, were used for each experiment. (DOC) S2