Research Article

Received: 14 January 2014, Revised: 22 February 2014, Accepted: 22 February 2014 Published online in Wiley Online Library

(wileyonlinelibrary.com) DOI 10.1002/jat.3011

Toxicity profiles and solvent–toxicantinterference in the planarian Schmidteamediterranea after dimethylsulfoxide (DMSO)exposureAn-Sofie Stevensa, Nicky Pirottea, Michelle Plusquinb, Maxime Willemsc,d,Thomas Neyense, Tom Artoisa and Karen Smeetsa*

ABSTRACT: To investigate hydrophobic test compounds in toxicological studies, solvents like dimethylsulfoxide (DMSO) areinevitable. However, using these solvents, the interpretation of test compound-induced responses can be biased. DMSOconcentration guidelines are available, but are mostly based on acute exposures involving one specific toxicity endpoint.Hence, to avoid solvent–toxicant interference, we use multiple chronic test endpoints for additional interpretation of DMSOconcentrations and propose a statistical model to assess possible synergistic, antagonistic or additive effects of testcompounds and their solvents. In this study, the effects of both short- (1 day) and long-term (2weeks) exposures to lowDMSO concentrations (up to 1000μl l�1) were studied in the planarian Schmidtea mediterranea. We measured differentbiological levels in both fully developed and developing animals. In a long-term exposure set-up, a concentration of 500μl l�1

DMSO interfered with processes on different biological levels, e.g. behaviour, stem cell proliferation and gene expression pro-files. After short exposure times, 500μl l�1 DMSO only affected motility, whereas the most significant changes on differentparameters were observed at a concentration of 1000μl l�1 DMSO. As small sensitivity differences exist between biologicallevels and developmental stages, we advise the use of this solvent in concentrations below 500μl l�1 in this organism. Inthe second part of our study, we propose a statistical approach to account for solvent–toxicant interactions and discussfull-scale solvent toxicity studies. In conclusion, we reassessed DMSO concentration limits for different experimentalendpoints in the planarian S. mediterranea. Copyright © 2014 John Wiley & Sons, Ltd.

Additional supporting information may be found in the online version of this article at the publisher’s web-site.

Keywords: planarians; DMSO; toxicology; different biological levels; solvent–toxicant interference

*Correspondence to: Karen Smeets, Zoology: Biodiversity and Toxicology, Centrefor Environmental Sciences, Hasselt University, Agoralaan, building D, BE 3590Diepenbeek, Belgium.E-mail: karen.smeets@uhasselt.be

aZoology: Biodiversity and Toxicology, Centre for Environmental Sciences, HasseltUniversity, Agoralaan, building D, BE 3590, Diepenbeek, Belgium

bEnvironmental biology, Centre for Environmental Sciences, Hasselt University,Agoralaan, building D, BE 3590, Diepenbeek, Belgium

cLaboratory of Pharmaceutical Technology, Harelbekestraat 72 Jozef Plateaustraat22, BE 9000, Gent, Belgium

dLaboratory of Environmental Toxicology & Aquatic Toxicology, Harelbekestraat72 Jozef Plateaustraat 22, BE 9000, Gent, Belgium

eI-BioStat, Hasselt University, Agoralaan, building D, BE 3590, Diepenbeek, Belgium

1

IntroductionDimethylsulfoxide (DMSO) is a solvent frequently used in toxico-logical studies and pharmacological screenings to enhance thesolubility of hydrophobic chemicals (Castro et al., 1995).Although they are necessary to ensure experimental stability,carrier solvents often interfere with proper cellular functioningand several studies already pointed to the intrinsic toxicity ofDMSO at different biological levels. Exposure to DMSO leads,among others, to a decrease in nerve conduction velocity, tohistological abnormalities and to changes in the locomotoractivity of rodents and zebrafish (Castro et al., 1995; Cavalettiet al., 2000; Chen et al., 2011). Dimethylsulfoxide also eliminatesthe response of sensory neurons to mechanical stimulation andhas an inhibitory effect on the moving behaviour of inverte-brates (Theophilidis and Kravari, 1994; Anderson et al., 2004).At the molecular level, the neurobehavioral effects of DMSOcould be explained by its interaction with ion channels and theresulting alterations of ion currents (Jourdon et al., 1986;Nakahiro et al., 1992; Lu and Mattson, 2001; Rosenblum et al.,2001). Besides neurobehavioral defects, DMSO was found to bemutagenic in certain strains of Salmonella typhimurium andEscherichia coli used in the Ames test, but only at very high con-centrations (Hakura et al., 1993). Other effects of DMSO include

J. Appl. Toxicol. (2014) Copyright © 2014 John

an increase in membrane fluidity or structural modifications suchas perturbations of phospholipid bilayers (Henderson et al., 1975;Anchordoguy et al., 1992; Gurtovenko and Anwar, 2007).Recommendations on the use of carrier solvents in aquatic

assays are available, and limits were set on 500μl l�1 for acuteexposures and 100μl l�1 for chronic exposure levels by the U.S.Environmental Protection Agency (U.S. EPA, 1975). Guidelines

Wiley & Sons, Ltd.

A.-S. Stevens et al.

2

of the Organization for Economic Co-operation and Development(OECD, 2000) have set a maximum acceptable limit of 100μl l�1,mainly based on historical data for acute toxicity thresholds. Nev-ertheless, higher DMSO concentrations are often necessary toascertain the solubilization of the lipophilic chemicals, therebyensuring proper experimental design. This is especially problema-tic in long-term exposure set-ups, where sublethal effects of thesesolvents become more apparent. Moreover, even if these guide-lines are followed, interaction effects between carrier solventsand test compounds cannot be excluded.

To avoid under- or overestimations of the toxicity of tested com-pounds, it is important to exclude the effects of DMSO itself on theparameters under study and to identify solvent–toxicant interfer-ence, leading to synergistic, additive or antagonistic effects. Up untilnow, the effects of DMSO were predominantly examined in short-term (a fewminutes or hours) test procedures and in vitro identifiedmolecular or cellular events were not always linked to the in vivo ef-fects, observed on organism level. However, as sublethal effects ofsolvents are more likely to cause problems in chronic studies, abetter understanding of the low level, long-term DMSO exposuresis needed. To characterize the stressor–response relationships andto assess acceptable concentration limits, effects should be moni-tored in different species, from molecular to behavioural level.

Planarians are considered as a model organism forneurotoxicological research (Best and Morita, 1991; Grebe andSchaeffer, 1991; Pagan et al., 2006; Raffa and Scott, 2008; Yuanet al., 2012; Tsushima et al., 2012; Raffa et al., 2013). They possessa well-organized nervous system, including many of the sameneuroactive compounds as vertebrates and, together with theirextensive repertoire of behaviour, they are suitable organismsfor the study of neuronal effects of toxic substances. In planarians,an exposure to DMSO alters mobility and behaviour and affectsthe levels of anti-oxidants (Pagan et al., 2006; Yuan et al., 2012).High DMSO concentrations are mutagenic according to the Amestest, and as such, DMSO-induced oxidative stress may be a precur-sor of a carcinogenic response (Hakura et al., 1993). Unfortunately,there is limited information about the effects of long-term expo-sures to low concentrations of DMSO in planarians, while this isessential knowledge for the interpretation of chronic exposureeffects of hydrophobic compounds. To have a full understandingof the toxic profile of DMSO, it is beneficial to monitor effects atdifferent biological levels simultaneously.

In this study, we evaluate in vivo the recommended DMSOconcentration limits in the planarian S.mediterranea. Thestrength of this study is the integration of responses from differ-ent biological levels. The emphasis is put on effects of low DMSOconcentrations in long-term test procedures, which is essentialinformation for the experimental design of compound studiesthat require chronic exposure times to elicit certain responses.Studying DMSO toxicity in S. mediterranea also offers an oppor-tunity to monitor DMSO-induced stem cells responses in vivo.Moreover, characterising stress in these cells can explain variousanti-stress mechanisms that are also present in other stem cells(or in cancer stem cells). This can reveal a defence strategy of or-ganisms to unfavourable circumstances.

Materials and Methods

Test Organism and Experimental Design

Asexual strains of the freshwater planarian S. mediterranea (Baguñà,1973; Benazzi et al., 1975) were maintained in water that was first

Copyright © 2014 Johnwileyonlinelibrary.com/journal/jat

deionized and then distilled. The following nutrients were added:1.0mMMgSO4 (VWR, Radnor, PA, USA), 1.6mMNaCl, 1.0mM CaCl2,0.1mM MgCl2, 0.1mM KCl and 1.2mM NaHCO3 (Acros Organics,Geel, Belgium). The animals were continuously kept in the dark ata room temperature of 20 °C and were fed once a week with beefliver. Experiments were performed in both fully developed anddeveloping animals to account for developmental toxicity ofDMSO. To synchronize their physiological state, the worms werecut transversally at respectively a week (developed animals) andimmediately before exposure (developing animals). During theexperiments, the animals were exposed to DMSO (99.9%, Merck,Darmstadt, Germany) in Petri dishes containing 20ml of medium.Developed animals were exposed for 1day and 2weeks to 0, 500and 1000μl l�1 DMSO and developing animals were exposed for2weeks to 0, 100, 500, 1000 and 5000μl l�1 DMSO. All animals werestarved for 1week before the experiments.

Mitotic Activity of Stem Cells

The mitotic activity of the stem cells was determined by immuno-staining with Histone H3 antibody (Millipore, Darmstadt, Germany),performed as previously described (Plusquin et al., 2012a).

After exposure, the worms were treated for 5min on ice withfive-eighths Holtfreter (Armstrong and Malacinski, 1989) solutioncontaining 2% HCl (VWR, Radnor, PA, USA) to remove the mucuslayer. The samples were fixed in Carnoy’s fixative (Yoneyamaet al., 2003) for 2 h (on ice) and were rinsed in 100% methanol(VWR, Radnor, PA, USA) during 1 h and bleached overnight atroom temperature in 6% H2O2 (VWR, Radnor, PA, USA) (in 100%methanol). Subsequently, the worms were rehydrated througha graded series of methanol/phosphate buffered saline-triton(PBST) (PBS tablets; VWR, Radnor, PA, USA) (Triton X-100 proanalyse, VWR, Radnor, PA, USA) washes (75%, 50%, 25%, 0%) for10min each, where after non-specific binding sites were blockedin bovine serum albumin (Sigma-Aldrich, St. Louis, MO, USA)(PBST/BSA) (0.1% Triton X-100 and 0.1 gml�1 BSA) for 3 h. Ani-mals were incubated at 4 °C for 44 h with a primary antibody[anti-phospho-Histone H3 (Ser10), biotin conjugate, Millipore,catalogue number: 16–189] 1:600 diluted in PBST/BSA. Theanimals were rinsed repeatedly for 1 h in PBST and incubated inPBST/BSA for 7 h. Then, the animals were incubated with a sec-ondary antibody (goat anti-rabbit IgG rhodamine conjugated,Millipore, catalogue number: 12–510), 1:500 diluted in PBST/BSA for 16 h. Afterwards, animals were rinsed repeatedly for30min in PBST and mounted in glycerol.

The animals were examined with fluorescence microscopyperformed with a Nikon Eclipse 80i (Nikon Instruments, Melville,NY, USA). The total number of mitotic neoblasts was normalizedto the body size of the animals (cfr. body area). Before the startof the colouring, three photos were taken of each animal at themoment it stretches its body. Of these three photos, the averagebody size was calculated, using Image J (1.44p, National Institutesof Health, Bethesda, MD, USA), for normalization afterwards.

Motility

The motility test was modified from the locomotion activity set-up of Raffa et al. (2001). The experimental set-up was blindedand randomized. At each time point, individual animals wereplaced in a Petri dish containing medium or DMSO solutions,which then was placed above a grid (squares of 0.25 cm2). A coldlight source was placed 18 cm above the animals. After an

J. Appl. Toxicol. (2014)Wiley & Sons, Ltd.

DMSO toxicity profile in the planarian S. mediterranea

acclimation period of 1min, the number of lines crossed perminute was counted during 10 consecutive minutes.

Gene Expression

RNA was isolated using a phenol–chloroform extraction procedure(Chomczynski and Sacchi, 2006) and was precipitated withNa-acetate and 70% ethanol (VWR, Radnor, PA, USA). RNA concen-trations were assessed on the NanoDrop ND-1000 spectrophoto-meter (NanoDrop Technologies, Wilmington, DE, USA). GenomicDNA was removed with the Turbo DNA free kit (Ambion, Austin,TX, USA). cDNA was synthesized using the high capacity cDNAreverse transcription kit with RNase inhibitor (Applied Biosystems)according to the manufacturer’s instructions.

Real-Time PCR was performed in an optical 96-well plate usingthe ABI PRISM 7900 (Applied Biosystems, Foster City, CA, USA) underuniversal cycling conditions. SYBR Green (Applied Biosystems,Foster City, CA, USA) chemistry-based real-time PCR was carriedout. PCR primers were designed using Primer3 software(Untergrasser et al., 2012). The selection of potential reference geneswas based on Plusquin et al. (2012b), after which the most stablereference genes during DMSO exposure were determined byNormfinder (MOMA, Aarhus, Denmark) and geNorm (GhentUniversity Hospital, Ghent, Belgium) analysis. Gene expression wasperformed with MIQE guidelines taken into account (Bustin et al.,2009). Details of the procedure are given in supplemental Table S1.

Statistical Analysis

Groups were statistically compared using one-way ANOVA.Normality was tested according to the normality test ofCramer-von Mises and Anderson-Darlin. If the assumptions ofnormality were not met, a transformation of the data set wasapplied (Log, Square root, 1/× and ex). Multiple comparison test-ing was performed based on the Tukey–Kramer method. A non-parametric Kruskal–Wallis test was performed when data werenot normally distributed after transformations. P-values less than0.05 were considered significant, significance levels of 0.1 levelwere also reported to empower observed patterns. The statisti-cal analyses were performed using SAS 9.2 (SAS Institute, Cary,NC, USA) and Microsoft Excel.

Figure 1. Motility in response to DMSO. Box plots of the planarian locomoapart, measured over a period of 10 minutes). DMSO-exposed worms (500 anwith an exposure time of (a) 1 day and (b) 2 weeks and (c) 2 weeks in develothe median of minimum 5 biological repeats. Whiskers represent 5th and 95errors of the control group are: (a) 7.34 lines/min ± 1.61, (b) 14.30 lines/minthe corresponding control group, are indicated with stars: ***: p < 0.01; **: p

J. Appl. Toxicol. (2014) Copyright © 2014 John

ResultsIn the present study, we used the planarian S. mediterranea toevaluate the toxicity of DMSO in vivo in a set-up with lowexposure concentrations for both short- (1 day) and long-term(2weeks) exposures. In addition, toxic responses were studiedat different biological levels simultaneously, to improve ourunderstanding of the mechanisms underlying DMSO toxicity.

Impact of DMSO on Behaviour

The neurotoxic effects of chemicals are often reflected in behav-ioural changes that can be monitored in time (Raffa and Scott,2008). We observed an increased motility of the worms, after1 day of exposure to 500 and 1000μl l�1 DMSO (Fig. 1A). Inter-estingly, these motility effects were restored to the control levelafter 2weeks of exposure (Fig. 1B).Because dividing cells are more sensitive to external insults,

we also measured the behavioural effects of DMSO in develop-ing animals, i.e. animals that were forced to regenerate duringexposure. These worms were not able to recover after 2weeksof exposure to 500 and 1000μl l�1 DMSO and displayed anincreased motility, indicating a higher sensitivity of the developingnervous system to DMSO (Fig. 1C).

Impact of DMSO on Stem Cell Proliferation

Because cells with stem-like characteristics have been identifiedin several tumours, stem cells are frequently used in cancer re-search to explore the modes of action of carcinogens and/orchemotherapeutics (Clarke et al., 2006; Visvader and Lindeman,2008). Therefore, in vivo stem cell proliferation was followed ina low-exposure set-up. After 1 day of exposure, 1000μl l�1 DMSOsignificantly elevated the neoblast division, compared with theunexposed group (Fig. 2). After 2weeks of exposure to 1000μl l�1

DMSO, a slight, although not significant, decrease in the prolifera-tive activity of neoblasts was noticed (Fig. 2).If the organisms were exposed during development, stem cell

divisions showed an increase after 2 weeks exposure to 500μl l�1

DMSO (Fig. 3).

tor velocity (fraction of lines crossed per minute, grid lines spaced 5 mmd 1000 μl l-1 DMSO) are compared to non-exposed worms (motility = 1),ping worms. The bold line indicates the mean and the thin line indicatesth percentiles, closed circles represent outliers. The mean and standard± 1.85 and (c) 14,02 lines/min ± 0.41. Significant effects, compared to< 0.05.

Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jat

3

Figure 2. Neoblast divisions in response to DMSO. Mitotic divisionsper mm2 after 1 day and 2 weeks exposure to 0, 500 and 1000 μl l-1 DMSO.The number of mitotic cells was normalised against the total body area ofthe worms. The values indicated in the graphs are the average ± se ofminimum 4 biological repeats. Significant effects as compared to thecorresponding control group are indicated with stars: **: p < 0.05.

Figure 3. Neoblast divisions in response to DMSO. Mitotic divisionsper mm2 after 2 weeks exposure of developing worms to 0, 100, 500,1000 and 5000 μl l-1 DMSO. The number of mitotic cells was normalisedagainst the total body area of the worms. The values indicated in thegraphs are the average ± se of minimum 3 biological repeats. Resultswere obtained from two separate experiments, from which the controlgroups are combined. Significant effects as compared to the controlgroup, are indicated with stars: *: p < 0.1.

Table 1. Transcript levels of stem cell, cancer-relatedand neuronal genes in response to dimethylsulfoxide(DMSO) in developing animals. Representation of thechanges in gene expression of cancer-related (msh2, egfr1,foxo and ndk), stem cell (pcna, smedinx-11 and mcm2) andneuronal (pc2) genes, expressed relative to the control group(0μl l�1 DMSO) after 2weeks exposure to 0, 1000 and 5000μll�1 DMSO. The values indicated in the table are the average±SE of a minimum of three biological replicates. Significant ef-fects (as compared with the non-exposed worms per group):

p< 0.05; p< 0.1

A.-S. Stevens et al.

4

Impact of DMSO on Gene Expression Profiles

Toxicogenomic approaches provide insights in the underlyingmechanisms of toxic responses and predict toxicity when appar-ent physiological signs are still absent. The effects of solvents ongene expression profiles are therefore detrimental for thecorrect interpretation of compound-induced transcriptomeresponses. Hence, we quantified the expression of somefrequently measured cancer-related genes (msh2, egfr1, foxoand ndk), stem cell-specific genes (pcna, smedinx-11 andmcm2) and a neuronal gene (pc2) under different DMSOexposure set-ups (Table 1). One day exposure to 1000μl l�1

DMSO caused a decrease in pc2 transcript levels as well as anincrease in the expression of ndk and mcm2. One day exposureto 500μl l�1 did not show any significant changes on theselected transcripts (data not shown). Exposing developing ani-mals for 2weeks to 100 and 500μl l�1 DMSO only influencedthe expression of a few individual genes: 100μl l�1 DMSOinhibited the expression of pcna and 500μl l�1 DMSO inducedthe expression of the cancer-related genes msh2 and egfr1 (datanot shown). Increasing the DMSO concentration to 1000μl l�1

induced altered gene expression profiles in one-half of themeasured genes (Table 1). However, 5000μl l�1 DMSO had lessimpact on the expression profile and only changed the expres-sion of foxo (Table 1).

Copyright © 2014 Johnwileyonlinelibrary.com/journal/jat

Detecting Solvent–Toxicant Interference: A Statistical Approach

In the previous paragraphs, appropriate thresholds for bothshort- and long-term DMSO exposures were determined in anintegrated experimental design. These thresholds are definedin a single exposure set-up and cannot exclude synergistic, addi-tive or antagonistic effects between the solvent and the testcompound. Therefore, to identify this possible bias, which iscompound-specific, we statistically investigated the compoundand solvent concentrations as being categorical, and a two-way ANOVA was applied to find possible toxic interactions (Fig. 4a).Unfortunately, because the compound will not dissolve, therewere no data available for the cases in which the concentrationof the solvent is zero, hence the experimental set-up had to bereconsidered in function of this drawback (Fig. 4B and Fig. 5).

By testing if there is an interaction effect, we attempted toinvestigate in what way the differences in toxicity, caused byaltering compound concentrations, depend on the solvent con-centrations and vice versa. When the interaction effect can bediscarded, the solvent’s toxic effect changes when the solventconcentrations increase, but independently of the compoundconcentrations, which means that more straightforward guide-lines can be proposed by simply identifying one (or a few)threshold value(s). When the interaction is significant, thresholdscan possibly still be identified, but this becomes more complexas compound–solvent combinations have to be analysed. Sucha full-scale investigation of the solvent’s toxicity (Fig. 4B and

J. Appl. Toxicol. (2014)Wiley & Sons, Ltd.

a) Theoretical Full Scale Design

Solvent Concentration

Com

poun

d C

once

ntra

tion

0

1

2

3

4

0 1 2 3 4

b) Practical Full Scale Design

Solvent Concentration

Solvent Concentration

Solvent Concentration

Solvent Concentration

Com

poun

d C

once

ntra

tion

0

1

2

3

4

0 1 2 3 4

c) Small Scale Design 1

Com

poun

d C

once

ntra

tion

0

1

2

3

4

0 1 2 3 4

d) Small Scale Design 2

Com

poun

d C

once

ntra

tion

0

1

2

3

4

0 1 2 3 4

e) Small Scale Design 3

Com

poun

d C

once

ntra

tion

0

1

2

3

4

0 1 2 3 4

Figure 4. Experimental design proposals. Black dots: available data, open dots: unavailable but unnecessary data, grey dots: unavailable but nec-essary data (missing data).

Practical Full Scale Example

0 1 2 3 4

1.0

1.5

2.0

2.5

3.0

3.5

4.0

0

1

2

3

4

Solvent Concentration

Com

poun

d C

once

ntra

tion

Tox

icity

Mea

sure

Figure 5. Practical Full Scale Example.

DMSO toxicity profile in the planarian S. mediterranea

5

Fig. 5) is time-consuming, labour intensive and difficult to under-take, therefore the solvent concentration is taken as low aspossible, to investigate the toxic effect of a compound with

J. Appl. Toxicol. (2014) Copyright © 2014 John

altering concentrations. Although the absence of a large datasetwill hamper the usability and generalization of the results, thereare a few possibilities, but with differing statistical validity: (1) asa typical experiment focuses on the toxicity of the compound,frequently the only data available is coming from a series ofcompound concentrations in combination with a non-changingsolvent concentration (Fig. 4C). As a control group, the solvent’stoxicity, when not combined with a compound, is mostly used(Fig. 4C). In this set-up, interaction effects cannot be investigatedas only one solvent concentration is used, but a simple one-wayANOVA can be applied. By using pair-wise testing, differences be-tween different compound concentrations and between com-pound concentrations and the control group can be furtherinvestigated. However, nothing can be said about the solvent’stoxicity and (2) often a (0,0)-concentration is used as an extracontrol group (Fig. 4D). This frequently used approach is notvalid, when looking at it statistically. To analyse this correctly, atwo-way ANOVA could be used, but data for a lot of groups (thegroups with 0 concentrations for the solvent) would then bemissing, making the results not trustworthy. When looking at itas a one-way ANOVA problem where you have the treatmentgroups, the first control group with only the solvent (xi,0) andthe second control (0,0), the interpretation becomes problem-atic, as when e.g. pairwise comparisons indicate that there is adifference between (xi, yj: i,j≠ 0) and (0,0), it is impossible toknow if that difference is caused by the compound’s and/or

Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jat

A.-S. Stevens et al.

6

solvent’s effect. Furthermore, if for example the pairwise testsindicate that (x1,x1) and (0,x1) do not differ, but (x1,x1) and(0,0) do, it is still impossible to state that the compound’s effectsis the sole reason for the difference between the latter, as a possibleinteraction effect is not tested for here. (3) A more reasonable andfeasible solution is to design a small experiment such that a two-way ANOVA can still be done. For example, when in the course ofthe experiment, data for groups (xi,0),…, (xi,yV), are collected, addi-tionally, data for groups (xj,yk) and (xj, yl) can be sampled, such thattogether with previously collected data for (xi,yk) and (xi, yl), a 2×2two-way ANOVA can be used to investigate if there is an interactioneffect and, if that can be discarded, if there is a solvent toxic effect(Fig. 4E). A side note that should be made here is, that, owing tothe limited number of study groups, the results provided by thisset-up are prone to errors, e.g. when no significant interaction effectcan be quantified, effects can still exist outside of the framework ofthe study. This comment can be formulated though for each exper-iment, but owing to the framework here being critically limited, itbecomes even more important to take into account.

DiscussionThe use of organic solvents such as DMSO is inevitable to solubi-lize lipophilic compounds. However, care should be taken as theintrinsic toxicity of acute DMSO exposures has been reported indifferent organisms (Cavaletti et al., 2000; Lu and Mattson, 2001;Anderson et al., 2004; Chen et al., 2011). Less well studied are theeffects of long-term exposures to low concentrations of DMSO,which is essential information to evaluate compounds thatpredominantly exert toxic effects of interest after longerexposure times, e.g. carcinogenic compounds. Also, while manystudies focus on adverse effects of DMSO in differentiated cells,less information is available about DMSO-induced stem cellresponses. In the present study, we evaluated the toxicity ofDMSO in vivo in S. mediterranea in a set-up with low exposureconcentrations for both short- and long- term exposures. Westudied toxic responses at different biological levels simulta-neously to explore the mechanisms underlying DMSO toxicity,paying special attention to stem cell behaviour.

In vertebrates, changes in motility were reported after acuteDMSO exposures, ranging from a few minutes up to 2 h, inconcentrations between 500 and 64 0000μl l�1 DMSO (Castroet al., 1995; Sackerman et al., 2010). In previous studies onplanarians, an exposure of 1 h to 5000–10 0000μl l�1 DMSOproduced significant, concentration-dependent changes in thespontaneous locomotor velocity (pLMV), whereas an exposureof 8min to 1000μl l�1 DMSO did not cause any behavioural ortoxic effects (Pagan et al., 2006; Yuan et al., 2012). We also ob-served an increased motility of the organisms after an exposureof 1 day to low concentrations of DMSO (500 and 1000μl l�1). Incontrast to other studies, this effect was restored to the controllevel after 2weeks of exposure, which can be as a result of theenormous regenerative capacity of these worms (Plusquinet al., 2012a). Forcing the animals to regenerate during DMSOexposure made them unable to recover after 2weeks ofexposure. This indicates a higher sensitivity of the nervous sys-tem during development. So, care should be taken when usingthis carrier solvent for long-term exposures in behavioural orneurological studies, as the recommended maximum DMSOconcentrations can still influence commonly used behaviouralparameters such as motility. To avoid misinterpretation of the

Copyright © 2014 Johnwileyonlinelibrary.com/journal/jat

tested compounds, we advise to keep DMSO concentrationsbelow 500μl l�1 during these types of experiments.

Because stem cells can provide information about the modesof action of carcinogens and/or chemotherapeutics, stem cellbehaviour was characterized in a low-exposure set-up. In previ-ous studies on both differentiated and undifferentiated cells,concentrations up to 20 000μl l�1 DMSO inhibited cell prolifera-tion after minimum 1 day exposure (Liu et al., 2001; Eter andSpitznas, 2002; Jasmin et al., 2010; Wang et al., 2012). In ourset-up of low DMSO exposures (1000μl l�1 DMSO), we foundthat the neoblast division was elevated after 1 day exposure,an effect that was also observed by Blom et al. (1998) after anacute exposure of a human breast cancer cell line to the sameDMSO concentration. Similar to the previous reported behav-ioural responses (motility), the effect on stem cell proliferationdisappeared in the organisms after 2weeks exposure,suggesting again that the high, stem cell-driven regenerativecapacity of these organisms, could ascertain recovery from toxicinsults. Also, this effect appeared stronger in developing organ-isms. Regenerating animals exposed to 500μl l�1 DMSO showedincreased stem cell divisions after 2 weeks exposure. The stemcell division rate of these animals can thus have an impact ontheir DMSO sensitivity and as this proliferation rate can vary,among other things, between different developmental stagesand environments, it is advisable to keep the DMSOconcentration below 500μl l�1 for long-term exposures andbelow 1000μl l�1 DMSO for acute exposures, in order to excludeDMSO-induced effects on proliferation. This recommendation isslightly higher than the maximum solvent concentrationsrecommended by the US EPA (1975) and the OECD (2000) guide-lines, which are more generalized and not solvent or organismspecific. Also, we analysed the activity of stem cells, known toform elaborated defence systems to cope with severe stresscircumstances. Given that DMSO toxicity is strongly related tothe experimental set-up, it is therefore advisable to run a controlmeasurement with DMSO alone in every new experiment, toexclude confounding effects.

The observed increase in cell proliferation after 1 day expo-sure to 1000μl l�1 DMSO was supported by an increase in theexpression of ndk and mcm2, respectively, a cancer-related andstem cell-specific gene (see Table 1). At the same time, a de-crease in pc2 transcripts levels was noticed. It is remarkable that,while gene expression profiles are considered to be highly sen-sitive to changes in environmental conditions, the expressionof only a few genes (pcna, msh2 and egfr1) was influenced in de-veloping animals after 2 weeks exposure to 100 and 500μl l�1

DMSO. The strongest effects on the expression profiles of the se-lected genes were observed after exposure to 1000μl l�1 DMSO,whereas motility and cell proliferation were already significantlydisturbed at 500μl l�1 DMSO. However, our gene-specific ap-proach only reveals that there are no changes in these selectedtranscripts and, although this was not within the scope of thisstudy, other genes could be affected. As the effects of solventson gene expression profiles are detrimental for the correct inter-pretation of compound-induced transcriptome responses, asolvent concentration of 100μl l�1 DMSO may therefore alreadybe too high (Turner et al., 2012). In addition, we want to empha-sise the importance of appropriate reference gene selection inthis context, as stable reference genes for data normalizationare essential in the experimental design of research. Normaliza-tion minimizes inherent technical or experimentally inducedand sample-specific variation, permitting accurate quantification

J. Appl. Toxicol. (2014)Wiley & Sons, Ltd.

DMSO toxicity profile in the planarian S. mediterranea

of biological changes (Derveaux et al., 2010). Genes that werestable in S. mediterranea during cadmium and chromium stressby Plusquin et al. (2012b), were not stable under DMSO exposureand unsuited to normalize the gene expression data of thisexperiment. Therefore, choosing suitable reference genes hasto be considered whenever new experiments using DMSO assolvent start. The reference genes selected to be the most stablein our exposure set-up were cystatin (cys), β-actin (β-act) andglyceraldehyde 3-phosphate dehydrogenase (gapdh).

By defining thresholds in a single exposure set-up synergistic,additive or antagonistic effects between DMSO and the testedcompound cannot be excluded. These interactions can be esti-mated statistically by investigating the compound and solventconcentrations as being categorical and applying a two-wayANOVA. In this regard, the only truly correct way is to perform afull-scale solvent’s toxicity study, in which each compound con-centration tested is solubilized in different concentrations ofDMSO. In many cases, this is not feasible so a valuable alterna-tive to test for interactions is to design a small experiment, inwhich two compound concentrations are solubilized in two dif-ferent DMSO concentrations. This experiment is relatively easyto perform and non-interfering solvent concentrations can beassessed statistically for specific experimental set-ups. Not onlythe number of groups is important, but also the amount ofobservations per group has to be considered carefully, as thisdepends on a number of factors such as the number of pairwisecomparisons, etc. A detailed overview is given by Kutner et al.(2005). Note also that in the statistical analyses, the parametricANOVA was considered. However, when even after possible datatransformations [e.g. log- or box-Cox transformations, see Kutneret al. (2005), Chapter 18] the model assumptions of an ANOVA arestill not met, non-parametric solutions can be used, such as aKruskal–Wallis test as a non-parametric counterpart for a one-way ANOVA (Kruskal and Wallis, 1952) or a two-way ANOVA on theranks instead of the (transformed) data (Lehmann, 2006).

In summary, we can conclude that it is not convenient to definegeneral DMSO thresholds for different experimental set-ups, evenwithin one organism. Both too high and too low solvent concentra-tion limits can hamper experiments, leading to erroneous interpre-tation of test compound responses or undue difficulties inexposure design, respectively. However, an integrated approachmeasuring DMSO toxicity at different levels simultaneously is neces-sary, as the compiled toxicity on different levels has gained attentionin toxicology research. In this regard, 500μl l�1 DMSO could be con-sidered an appropriate solvent threshold for both short-and long-term DMSO exposures in S. mediterranea, but small differences inthe sensitivity of specific biological levels and developmental stagesexist. In addition, to control for interactions between DMSO andhydrophobic test compounds, synergistic, additive or antagonisticeffects have to be identified statistically. Depending on the scopeof the study, this requires some additional experiments.

Acknowledgements

The authors wish to thank Professor Dr Oné R. Pagán for hisexpertise and advice and Natascha Steffanie and RiaVanderspikken for their skilful technical assistance.

7

Conflict of InterestThe Authors did not report any conflict of interest.

J. Appl. Toxicol. (2014) Copyright © 2014 John

FundingThis work was supported by a PhD grant for An-Sofie Stevens fromIWT (Agentschap voor Innovatie door Wetenschap enTechnologie) (no. 101442). Experiments were financed by HasseltUniversity BOF-financing (Bijzonder Onderzoeksfonds: BOF08G01)and Hasselt University tUL-impulsfinancing (IMPF2PR). MaximeWillems was funded by an OZM grant by IWT (no. 100631).

ReferencesAnchordoguy TJ, Carpenter JF, Crowe JH, Crowe LM. 1992. Temperature-

dependent perturbation of phospholipid bilayers by dimethylsulfoxide.Biochim. Biophys. Acta 1104: 117–122.

Anderson GL, Cole RD, Williams PL. 2004. Assessing behavioral toxicitywith Caenorhabditis elegans. Environ. Toxicol. Chem. 23: 1235–1240.

Armstrong MA, Malacinski GM. 1989. Developmental Biology of the Axo-lotl. Oxford University Press: New York, USA.

Baguñà J. 1973. Estudios citotaxonómicos, ecológicos e histofisiología dela regulación morfogenética durante el crecimiento y la regeneraciónen la raza asexuada de la planaria Dugesia mediterranea. PHD thesis,University of Barcelona: Barcelona.

Benazzi M, Baguná J, Ballester R, Puccinelli I, Papa RD. 1975. Further contri-bution to the taxonomy of the « Dugesia Lugubris-Polychroa Group »with Description of Dugesia Mediterranea N.SP. (Tricladida, Paludicola).Boll. Zool. 42: 81–89.

Best JB, MoritaM. 1991. Toxicology of planarians.Hydrobiologia 227: 375–383.Blom A, Ekman E, Johannisson A, Norrgren L, Pesonen M. 1998. Effects of

xenoestrogenic environmental pollutants on the proliferation of ahuman breast cancer cell line (MCF-7). Arch. Environ. Contam. Toxicol.34: 306–310.

Bustin SA, Benes V, Garson JA, Hellemans J, Huggett J, Kubista M, MuellerR, Nolan T, Pfaffl MW, Shipley GL, Vandesompele J, Wittwer CT. 2009.The MIQE guidelines: Minimum information for publication of quan-titative real-time PCR experiments. Clin. Chem. 55: 611–622.

Castro CA, Hogan JB, Benson KA, Shehata CW, Landauer MR. 1995.Behavioral effects of vehicles: DMSO, ethanol, Tween-20, Tween-80,and emulphor-620. Pharmacol. Biochem. Behav. 50: 521–526.

Cavaletti G, Oggioni N, Sala F, Pezzoni G, Cavalletti E, Marmiroli P,Petruccioli MG, Frattola L, Tredici G. 2000. Effect on the peripheralnervous system of systemically administered dimethylsulfoxide inthe rat: a neurophysiological and pathological study. Toxicol. Lett.118: 103–107.

Chen TH, Wang YH, Wu YH. 2011. Developmental exposures to ethanolor dimethylsulfoxide at low concentrations alter locomotor activityin larval zebrafish: Implications for behavioral toxicity bioassays.Aquat. Toxicol. 102: 162–166.

Chomczynski P, Sacchi N. 2006. The single-step method of RNA isolationby acid guanidinium thiocyanate-phenol-chloroform extraction:twenty-something years on. Nat. Protoc. 1: 581–585.

Clarke MF, Dick JE, Dirks PB, Eaves CJ, Jamieson CHM, Jones DL, VisvaderJ, Weissman IL, Wahl GM. 2006. Cancer stem cells—Perspectives oncurrent status and future directions: AACR workshop on cancer stemcells. Cancer Res. 66: 9339–9344.

Derveaux S, Vandesompele J, Hellemans J. 2010. How to do successfulgene expression analysis using real-time PCR. Methods 50: 227–230.

Eter N, Spitznas M. 2002. DMSO mimics inhibitory effect of thalidomideon choriocapillary endothelial cell proliferation in culture. Br. J.Ophthalmol. 86: 1303–1305.

Grebe E, Schaeffer DJ. 1991. Neuobehavioral toxicity of cadmium sulfateto the planarian Dugesia dorotocephala. Bull. Environ. Contam. Toxicol.46: 727–730.

Gurtovenko AA, Anwar J. 2007. Modulating the structure and propertiesof cell membranes: The molecular mechanism of action of dimethylsulfoxide. J. Phys. Chem. B 111: 10453–10460.

Hakura A, Mochida H, Yamatsu K. 1993. Dimethyl sulfoxide (DMSO) is muta-genic for bacterial mutagenicity tester strains.Mutat. Res. 303: 127–133.

Hellemans J, Mortier G, De Paepe A, Speleman F, Vandesompele J. 2007.qBase relative quantification framework and software for manage-ment and automated analysis of real-time quantitative PCR data. Ge-nome Biol. 8: R19.

Henderson TR, Henderson RF, York JL. 1975. Effects of dimethyl sulfoxideon subunit proteins. Ann. N. Y. Acad. Sci. 243: 38–53.

Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jat

A.-S. Stevens et al.

8

Jasmin, Spray DC, de Carvalho ACC, Mendez-Otero R 2010. Chemical in-duction of cardiac differentiation in P19 embryonal carcinoma stemcells. Stem Cells Dev. 19: 403–411.

Jourdon P, Berwald-Netter Y, Dubois J-M. 1986. Effects ofdimethylsulfoxide on membrane currents of neuroblastoma×gliomahybrid cell. Biochim. Biophys. Acta 856: 399–402.

Kruskal WH, Wallis WA. 1952. Use of ranks in one-criterion variance anal-ysis. J. Am. Stat. Assoc. 47: 583–621.

Kutner MH, Nachtsheim CJ, Neter J, Li W. 2005. Applied Linear StatisticalModels. McGraw-Hill/Irwin: New York.

Lehmann EL. 2006. Nonparametrics: Statistical Methods Based on Ranks.Springer: New York.

Liu C-H, Chu IM, Hwang SM. 2001. Enhanced expression of various exog-enous genes in recombinant Chinese hamster ovary cells in presenceof dimethyl sulfoxide. Biotechnol. Lett. 23: 1641–1645.

Lu C, Mattson MP. 2001. Dimethyl sulfoxide suppresses NMDA- andAMPA-induced ion currents and calcium influx and protects againstexcitotoxic death in hippocampal neurons. Exp. Neurol. 170: 180–185.

NakahiroM, ArakawaO, Narahashi T, Ukai S, Kato Y, Nishinuma K, NishimuraT. 1992. Dimethyl sulfoxide (DMSO) blocks GABA-induced current in ratdorsal root ganglion neurons. Neurosci. Lett. 138: 5–8.

OECD. 2000. Guidance document on aquatic toxicity testing of difficultsubstances and mixtures. OECD Environmental Health and SafetyPublications 23: 20–25.

Pagan OR, Rowlands AL, Urban KR. 2006. Toxicity and behavioral effectsof dimethylsulfoxide in planaria. Neurosci. Lett. 407: 274–278.

Plusquin M, Stevens AS, Van Belleghem F, Degheselle O, Van Roten A,Vroonen J, Blust R, Cuypers A, Artois T, Smeets K. 2012a. Physiologicaland molecular characterisation of cadmium stress in Schmidteamediterranea. Int. J. Dev. Biol. 56: 183–191.

Plusquin M, DeGheselle O, Cuypers A, Geerdens E, Van Roten A, Artois T,Smeets K. 2012b. Reference genes for qPCR assays in toxicmetal and salin-ity stress in two flatworm model organisms. Ecotoxicology 21: 475–484.

Raffa RB, Danah J, Tallarida CS, Zimmerman C, Gill G, Baron SJ, Rawls SM.2013. Potential of a planarian model to study certain aspects of anti-Parkinsonism drugs. Adv. Parkinson Dis. 2: 70–74.

Raffa RB, Holland LJ, Schulingkamp RJ. 2001. Quantitative assessment of do-pamine D2 antagonist activity using invertebrate (Planaria) locomotionas a functional endpoint. J. Pharmacol. Toxicol. Methods 45: 223–226.

Raffa RB, Scott MR. 2008. Planaria: A Model for Drug Action and Abuse.Landes Bioscience: Austin, Texas, USA.

Rosenblum WI, Wei EP, Kontos HA. 2001. Dimethylsulfoxide and ethanol,commonly used diluents, prevent dilation of pial arterioles byopeners of KATP ion channels. Eur. J. Pharmacol. 430: 101–106.

Copyright © 2014 Johnwileyonlinelibrary.com/journal/jat

Sackerman J, Donegan JJ, Cunningham CS, Nguyen NN, Lawless K, LongA, Benno RH, Gould GG. 2010. Zebrafish behavior in novel environ-ments: effects of acute exposure to anxiolytic compounds and choiceof Danio rerio Line. Int. J. Comp. Psychol. 23: 43–61.

Theophilidis G, Kravari K. 1994. Dimethylsulfoxide (DMSO) eliminatesthe response of the sensory neurones of an insect mechanore-ceptor, the femoral chordotonal organ of Locusta migratoria, butblocks conduction of their sensory axons at much higherconcentrations: A possible mechanism of analgesia. Neurosci. Lett.181: 91–94.

Tsushima J, Nishimura K, Tashiro N, Takata K, Ashihara E, Yoshimoto K,Ariga H, Agata K, Kitamura Y. 2012. Protective effect of planarianDJ-1 against 6-hydroxydopamine-induced neurotoxicity. Neurosci.Res. 74: 277–283.

Turner C, Sawle A, Fenske M, Cossinsy A. 2012. Implications of the solventvehicles dimethylformamide and dimethylsulfoxide for establishingtranscriptomic endpoints in the zebrafish embryo toxicity test. Envi-ron. Toxicol. Chem. 31: 593–604.

U.S. EPA. 1975. Methods for Acute Toxicity Tests with Fish,Macroinvertebrates and Amphibians. EPA-660/3-75-009.

Untergrasser A, Cutcutache I, Koressaar T, Ye J, Faircloth BC, Remm M,Rozen SG. 2012. Primer3 - new capabilities and interfaces. NucleicAcids Res. 40: e115.

Visvader JE, Lindeman GJ. 2008. Cancer stem cells in solid tumours:accumulating evidence and unresolved questions. Nat. Rev. Cancer8: 755–768.

Wang C-C, Lin S-Y, Lai Y-H, Liu Y-J, Hsu Y-L, Chen JJW. 2012. Dimethylsulfoxide promotes the multiple functions of the tumor suppressorHLJ1 through activator protein-1 activation in NSCLC cells. PloS ONE7: e33772.

Yoneyama M, Kitayama T, Taniura H, Yoneda Y. 2003. Immersion fixationwith Carnoy solution for conventional immunohistochemical detec-tion of particular N-methyl-D-aspartate receptor subunits in murinehippocampus. J. Neurosci. Res. 73: 416–426.

Yuan Z, Zhao B, Zhang Y. 2012. Effects of dimethylsulfoxide on behaviorand antioxidant enzymes response of planarian Dugesia japonica.Toxicol. Ind. Health 28: 449–457.

Supporting InformationAdditional supporting information may be found in the onlineversion of this article at the publisher’s web-site.

J. Appl. Toxicol. (2014)Wiley & Sons, Ltd.