Toxicant Induced Changes on Delayed Fluorescence Decay Kinetics of Cyanobacteria and Green Algae: A Rapid and Sensitive Biotest

Algal tests have developed into routine tools for testing toxicity of pollutants in aquatic environments. Meanwhile, in addition to algal growth rates, an increasing number of fluorescence based methods are used for rapid and sensitive toxicity measures. The present study stresses the suitability of delayed fluorescence (DF) as a promising parameter for biotests. DF is based on the recombination fluorescence at the reaction centre of photosystem II, which is emitted only by photosynthetically active cells. We analyzed the effects of three chemicals (3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU), 3,5 Dichlorophenol (3,5 DCP) and copper) on the shape of the DF decay kinetics for potential use in phytoplankton toxicity tests. The short incubation tests were done with four phytoplankton species, with special emphasis on the cyanobacterium Microcystis aeruginosa. All species exhibited a high sensitivity to DCMU, but cyanobacteria were more affected by copper and less by 3,5 DCP than the tested green algae. Analyses of changes in the DF decay curve in response to the added chemicals indicated the feasibility of the DF decay approach as a rapid and sensitive testing tool.


Introduction
The introduction of chemicals and pollutants in the aquatic environment by anthropogenic activities has become increasingly significant in the past decades indicating the need for an increased monitoring effort of water quality [1]. Appropriate assessment of water pollution in various habitats requires, in addition to knowledge on types and origins of the pollutants, proper analytical methods and understanding of the transport and fate of the pollutants as well as precise quantification of adverse effects on organisms. Identification and quantification of toxins/pollutants in aquatic ecosystems are often complicated by synergistic effects between two or more non-toxic compounds or by indirect effects of the contaminants [2]. Bioassays, which expose the indicator organisms directly to the target solution could circumvent this analytical maze, and in the presence of a biohazard, would react in a manner that allows proper monitoring. Although such an approach does not allow identifying a given chemical substance, it resembles a pre-screening test for the presence of a pollutant above a potentially harmful threshold [3].
Today, automated biomonitoring systems encompass an array of advanced electronic systems, specifically designed to measure changes in physiological or behavioural responses in fish or invertebrates [4]. Another well-known toxicity test is based on the swimming behaviour of the zooplankter Daphnia magna [5].
Algae are particularly suitable for bio-testing because of their high abundance and sensitivity to environmental pollution in aquatic systems [6]. However, standard growth rate measurements in the presence of potential hazards require exposure of up to 72 h and, therefore, unsuitable for rapid responses (e.g. [7]). A much faster alternative to algal growth is the non-invasive measurement of the in vivo fluorescence which facilitates monitoring of activity changes by affecting photosystem II (PS II) [8]. For example, pulse amplitude modulation (PAM) of prompt fluorescence (PF) is a well-established method for detecting impact of chemicals on algal activity (e.g. [9][10][11][12]).
A promising alternative to PF is delayed fluorescence (DF). DF, which was first observed by Strehler and Arnold [13], represents a recombination fluorescence at the reaction centres of PS II that can be measured as a very weak fluorescence signal, when photosynthetically active cells are transferred from light to the dark. For green algae, it has been shown that DF intensity, the integral over DF decay kinetic, represents a sensitive and reliable parameter for toxicity tests [14][15][16][17]. In addition, different chemicals alter the DF decay kinetics in a characteristic manner that renders its use in algal toxicity tests more efficient. [14,17]. DF occurs because electrons flow back from the electron transport chain (ETC) to reaction centre P680 until the thylakoid membrane is completely discharged. In contrast to PF with decay times in the range of ns, temporal behaviour of DF is determined by electron holes and the liberation of electrons from the electron traps within the ETC. This liberation, which is a thermally activated process, is relatively slow (up to minutes) compared to PF [18]. The shape of the DF decay kinetic, however, is influenced by the redox states of the sources of holes (redox active tyrosine residue Y Z , oxygen evolving complex, and the positive charged inner site of the thylakoid membrane) and sources of electrons (components of the ETC: Q A , Q B , Plastoquinone (PQ), Cyt b6/f , P 700 , Ferredoxin and the negative charged outside of the thylakoid membrane) [19]. While the fast decay of DF depends on the state of P680 [20], the slower decay is influenced by the redox state of components of the ETC and from the S2 and S3 states of the oxygen evolving complex [21].
In this study, we used a quantitative approach based on changes in the shape of the DF decay curves upon exposure to chemicals. We measured toxicity effects on the DF decay in the presence of the herbicide 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU, Lancester Synthesis; UK), 3,5 Dichlorophenol (3,5 DCP) a standard substance in biotests [22,23] and copper, a common industrial and agricultural pollutant, which has a high potential to affect photosynthetic activity as well as other cell metabolism pathways [24][25][26]. We tested the performance of the DF approach with cyanobacteria: Microcystis aeruginosa and Aphanizomenon flosaquae, often representing the dominant primary producers in aquatic systems [27][28][29]. To better generalize our results, we have also used two common green algae: Desmodesmus subspicatus and Scenedesmus obliquus.

Culture Conditions for Cyanobacteria and Green Algae Strains
We used two different cyanobacteria species and two green algae species. The green algae Desmodesmus subspicatus Hegewald et Schmidt (SAG 86.81), Scenedesmus obliquus (Turpin) Kütz. (SAG 276-1) and the cyanobacterium Aphanizomenon flos-aquae (L.) Ralfs (SAG 31.87) were obtained from the Culture collection of Algae, Göttingen. Further, we tested two strains of Microcystis aeruginosa Kütz. Both strains grow unicellular, but differ in their toxicity. While HUB018 lacks Microcystin-LR, HUBW333 contains a high level of this toxin [30,31]. Cyanobacteria strains were cultured in Z-medium [32], and green-algae were kept in WC-medium [33]. All strains grew semi-continuously by dilution every third day in 1L clear glass bottles (Schott). Cultures were illuminated for 24 h (Osram Lumilux cool daylight, 55 mmol m 22 s 21 ) and continuously aerated with sterile filtered air (Sartorius, Midisart 2000) at 21.561uC. Chemicals for all media were obtained from Merck.

Chemical Effects on Delayed Fluorescence (DF)
System set up. DF measurements were performed with a custom made DF decay kinetic instrument [16]. In brief, the lightadapted phytoplankton suspension is pumped in a closed cycle, from the culture vessel into a dark emission-cell passing on its way an excitation cell where it is illuminated by a red light LED (55 mmol m 22 s 21 ). The decay of the DF signal (680-750 nm) is measured for 90 s by a Channel-photomultiplier (R1463, Hamamatsu, Japan) in the photon counting mode upon the pump stops. Temperature of the double-jacket excitation-cell was kept at constant 21.5uC via an external temperature controller (Thermo Haake C10).
Measurements. Three days ahead of each experiment, fresh growth medium was inoculated with cultures. On the day of the experiment, cultures were diluted with fresh medium (without EDTA, to avoid complexation of ions) to a DF start value between 8000 and 12000 (counts s 21 ), corresponding to ca. 50-70 mg Chl a L 21 . 50-ml subsamples were exposed to the following chemicals: DCMU, dissolved in Ethanol, Copper (as Cu 2 SO 4 , Merck, dissolved in Aqua dest), and 3,5 Dichlorophenol (3,5 DCP; ABCR GmbH+CoKG, Germany) dissolved in Dimethyl Sulfoxide yielding final concentrations in a range of 10 210 to 10 26 M for DCMU, 10 27 to 10 25 M for copper, and 10 27 M -2.5* 10 25 M for 3,5 DCP. Addition of each solvent to a subsample of the phytoplankton cultures served as the respective controls. Additional concentrations of all chemicals were prepared for the doseresponse curves with M. aeruginosa HUB018 and D. subspicatus to reach the saturation impact on changes in the DF decay kinetic. This was done to determine the concentration in which the halfmaximal effect could be measured (EC 50 ).
After adding chemicals, samples were incubated for 20 min under irradiation (55 mmol m 22 s 21 ) with the same light sources as used for algal cultivation. DF decay kinetic of each sample was measured for 90 s in five replicates. In order to compare rapid and possible long-term effects of exposure with the test toxins, DF decay kinetics of M. aeruginosa HUB018 and D. subspicatus were measured for selected toxin concentrations, which were in the range of the short term experiments, after 24 h of incubation. All experiments were done in triplicates.
Analysis of the DF signal. Starting with the control (algal suspension+solvent), each algal sample was measured five times and the last three decay kinetics were averaged for further calculations. To account for possible variability of subsamples and comparability between experiments, the counts of the target sample (t i ) at each measured time point (incl. those from the control, c i ) were normalized to the start value (c 1 ) of the control. After normalization, the decay curve of the target sample (t in Fig. 1) was subtracted from that of the control leading to the curve of residuals (c i 2 t i ) ( Fig. 1, inlet). Then, the absolute value of the differences was summed up as DDF given in equation 1.
Accordingly, DDF quantifies all changes in the shape of the DF decay curve in comparison to that of the control (Fig. 1). Obtained DDF values were compared for the tested phytoplankton species as a function of the concentration of the tested toxin via the doseresponse curves.
In addition, DDF was related to changes in the DF-integral for M. aeruginosa and D. subspicatus for short time (30 min) and 24 h incubations. The DF-integral for each DF decay kinetic was calculated using OriginPro8. Hereby, effects on the integral are defined as percent decrease of the integral of the treatment compared to the integral of the respective control. They are given as DFI 30min for the short-time tests and DFI 24h after 24 h of incubation.

Growth Inhibition Experiments
Growth inhibition experiments were carried out with non-toxic M. aeruginosa HUB018 and D. subspicatus to compare the impact of copper, DCMU and 3,5 DCP on growth and DF kinetic. Precultures were grown at 21.5uC in a temperature-and light- As growth parameter, we determined the content of Chl a (mg L 21 ) at the beginning of each experiment and after 1, 2, 3, 6 and 9 days. The standard period of three days (ISO 8692), had to be prolonged because of delay in growth of HUB018. Chl a fluorescence as biomass indicator is a commonly used surrogate parameter for cell number in toxicity tests [34], but several characteristics should be taken into account. Due to temperature and light history dependency as well as dial periodicity [35,36,37], it is therefore important to keep this parameters as well as  [38]. Tested chemical compounds were evaluated in four replicates for each concentration, and all experiments were repeated twice.
Growth of the cultures was calculated as increase of Chl a concentration over time. Inhibition was expressed as percent growth of each treatment relative to the respective control (100%) for each measuring time point. To test if DF is a suitable parameter, linear correlation between growth inhibition on days 3 and 9 and the DF parameters DDF, as well as the DFI 30min and DFI 24h for M. aeruginosa HUB018 and D. subspicatus were calculated.

Statistical Analysis of the Species Specific Sensitivity
Analysis of species-specific sensitivity at low concentrations was done with linear regression of the four lowest concentrations of each chemical and strain. Slopes of regressions were compared with a t-test and Bonferroni-Holm correction (copper, df = 22, p,0.05; DCMU, df = 22, p,0.05; 3,5 DCP, df = 24, p,0.05).

Results
The normalized DF decay curves of control suspensions of the five tested phytoplankton species revealed pronounced differences ( Fig. 2A). Particularly in cyanobacteria, a transient peak could be observed with variable, species-specific heights and timing. In contrast, decay curves of green algae did not show any peak at any of the experimental conditions applied.
Changes in DF decay kinetics are exemplified in Microcystis HUB018 (Fig. 2B), where the control DF decay kinetic showed a clear transient peak with a maximum at 25 s the peak was lower after exposure to the highest concentration of 3,5 DCP (peak height control: 0.3; 3,5 DCP: 0.21). Copper suppressed the transient peak nearly completely and, after DCMU addition, no transient peak occurred. In addition, the DF decay during the first seconds was affected differently. While DCMU caused the strongest decline in DF (DF signal at 1 sec = 0.9; at 2 sec = 0.3), copper and 3,5 DCP did not steepen the initial DF decay.
All tested phytoplankton species showed a measureable DDF response to the tested chemicals. Cyanobacteria and green algae were highly sensitive against DCMU, but differed strongly in their sensitivity towards copper and 3,5 DCP ( Fig. 3A and 3 B).
Both Microcystis strains showed a higher copper sensitivity than A. flos-aquae, D. subspicatus and S. obliquus (p,0.001 for all), whose sensitivity did not differ significantly from each other. Whereas A. flos-aquae was affected most by low DCMU concentrations (t-test of b, df = 22, p,0.05), S. obliquus was the most DCMU resistant species, the other tested strains did not differ significantly in their sensitivity. In contrast, D. subspicatus was influenced the strongest by low concentrations of 3,5 DCP (t-test of b, df = 24, p,0.05), followed by S. obliquus and A. flos-aquae. However, both Microcystis strains were significantly less affected by 3,5 DCP.   The degree of inhibition detected by our DF-based biotest was compared with that of algal growth inhibition tests in a series of growth inhibition experiments, which were performed in parallel to DF-assays. For this comparison we had chosen Microcystis strain HUB018 and D. subspicatus because of their relatively high sensitivity to copper and 3,5 DCP, respectively (see dose-response curves in Figs. 3).
Due to their sensitivity to the added chemicals (M. aeruginosa HUB018 to DCMU and copper, D. subspicatus to DCMU and 3,5 DCP), both tested algae showed a clear relationship between DDF as well as DFI 30min and DFI 24h and the growth inhibition at day 3 and 9, respectively (with R 2 values .0.9 for the linear regression; Fig. 4). There was one exception for M. aeruginosa tested against copper using the DFI 30min (R 2 .0.71). Small effects on DF and growth could be either seen for 3,5 DCP added to M. aeruginosa and for copper added to D. subspicatus (Fig. 4) caused a weaker relation. Especially, the correlation between the DF parameters and growth inhibition on day 3 were small (R 2 = 0.12 for D. subspicatus, DDF of copper; R 2 = 0.07, DFI 24h HUB018, 3,5 DCP), but still a high correlation between DDF and growth inhibition was reached for D. subspicatus at day 9 (R 2 of 0.87, DDF of copper). Our results suggest that DDF based on changes in the shape of the DF decay kinetic yield similarly good results as changes of the DF integral even after 24 h.

Discussion
The main goal of the present study was to confirm the suitability of a DF-based in vivo approach as a rapid and sensitive toxicity test for potential pollutants with cyanobacteria and green algae as indicator organisms. To achieve this goal, we focused on quantifying changes in the DF-decay curve when comparing to the respective controls. Measuring DF does not require any preparation of the algae and toxicity effects of the tested chemicals, even in low concentrations, could be obtained after 20 min of incubation. Our results indicate that DDF can be used as a rapid, reliable and sensitive phytoplankton toxicity test.
DF as a parameter to investigate toxicity effects on algae was used in the past for green algae in laboratory studies [18,21,39], but also the implementation in continuous biomonitoring stations of surface waters was suggested [40]. Further applications of DF provide a reliable tool to investigate the phytoplankton community in lakes [41], even over the seasonal aspect and vertical distribution [42] as well as online monitoring [43]. Similar findings have been obtained also for streams [44,45,46].
In the present study, the most obvious effect occurred with cyanobacteria, after addition of the potential pollutants. All three tested species displayed a prominent transient peak at about 25 to 35 sec (Fig. 2), whereas both green algae (D. subspicatus and S. obliquus) did not. In particular, changes in the transient peak, following the addition of toxin, render cyanobacteria promising candidates for our DF approach.
The occurrence of a transient peak has been described as the PSI involvement in DF decay kinetics after irradiation with longwave light (650-700 nm) [20,47,48]. However, the spectrum of the red light LED used in our DF instrument reaches its maximum at 654 nm, thus both PS are involved in the electron transport.
A more plausible explanation for this phenomenon, which is typical for cyanobacteria, is their imbalance between both photosystems. While most photoautotrophs exhibit a PSI:PSII ratio of 1, cyanobacteria exhibit a PSI:PSII ratio of up to 3 [49]. A possible reason is the involvement of their PS-ETC's enzymes in the respiratory ETC. Electrons of the respiratory ETC enter the PS-ETC at the Plastoquinone pool and leave it after the transition through the cytc553/Plastocyanin [50,51]. Confirmation of the hypothesis that the observed transient DF peak is also related to respiration sustains the applicability of cyanobacteria in toxin tests None of the species showed an overall highest sensitivity to all tested chemicals. This finding is supported by studies, which have used growth inhibition tests to compare the sensitivity in phytoplankton species, differing in their cell-wall structure or enzyme activities [52][53][54]. Wä ngberg & Blanck [55] also showed that algal phylogeny can be relevant for the measured sensitivity towards toxicants.
Among the chemicals tested, the herbicide DCMU drastically steepened the DF decay kinetic during the first seconds, which can be explained by the specific action of DCMU. DCMU blocks the electron transport between PSII and the Plastoquinone pool, therefore only PS680 + and Q A molecules contribute to the DF signal. Our measured EC 50 values for DCMU (4.8*10 28 M for HUB018 and 6.2*10 28 M for D. subspicatus) are in the range for DFI 30min (8.58*10 28 ) and DFI 24h (4.29*10 28 M) measured by Berden-Zrimec et al. [14]. Results obtained at low DCMU concentrations (Fig. 3), indicate that cyanobacteria are slightly more sensitive against DCMU than green algae.
Copper, as a necessary trace metal in cell metabolism, has the potential to disturb the photosynthetic electron transport at high concentrations [26] by blocking the enzymes of the PS' dark reaction. Further, it can also cause lipid peroxidation due to Fenton reactions leading to an intracellular increase of reactive oxygen species [24,26,[56][57][58]. Both M. aeruginosa strains revealed a higher sensitivity to copper, e.g. HUB018 showed an effect at already 10 27 M (EC 50 = 5*10 26 M), whereas both green algae were unaffected by copper up to 2.5*10 26 M. In parallel to our study of M. aeruginosa, Rojičková & Maršálek [54] showed that the cyanobacterium Synechococcus sp. exhibited higher sensitivity to copper than their investigated green algae. In contrast to our findings, however, Berden-Zrimec et al. [14] found lower EC 50 values for D. subspicatus (EC 50DFI = 3.8*10 26 M), possibly caused by the continuous illumination during the 24 hour or 72 h incubation. The prolonged period of illumination potentially leads to higher production of reactive oxygen species catalyzed by copper in the Fenton or Haber-Weis reaction [59].
In contrast to copper and DCMU, the organic and more volatile 3,5 DCP, a reference substance used in OECD guideline 201 and 221 [22,60], revealed stronger effects on green algae than on cyanobacteria. Effects of dichlorophenols on eukaryotic algae have been described earlier [14,17,61]. While Berden-Zrimec [14] reported an EC 50DFI24h of 1.38 mg L 21 for the green algae D. subspicatus (about 8.4*10 26 M), Shao et al. [62] found an EC 50 value of 23.39 mg L 21 (1.4*10 24 M) by using a luciferase marked cyanobacterium Synechocystis sp. indicating that cyanobacteria are less sensitive, a notion that validates our findings. Comparison of EC values must always be considered carefully, because of differences in endpoint parameters (growth, fluorescence) as well as incubation time with the tested toxin. In our study, we successfully used the parameter DDF taking into account changes in shape of the decay kinetic. After 20 min, all our investigated strains showed measurable effects on DDF, whereas it took Berden-Zrimec et al. [14] 30 min after adding 3,5 DCP to observe a decrease in the DF integral of D. subspicatus. However, the same Figure 4. Correlation of DF parameters and growth. Correlation between the DF parameters DDF, DFI 30min , DFI 24h with growth inhibition at day 3 (squares) and day 9 (circles) for M. aeruginosa (opened symbols) and D. subspicatus (filled symbols) as well as linear fits for M. aeruginosa (grey, dotted day 3, dashed day 9) and D. subspicatus (black, solid day 3, short dashed day 9) ) for the tested chemicals A) copper, B) DCMU, C) 3,5 DCP. doi:10.1371/journal.pone.0063127.g004 authors did not find a significant effect after adding DCMU or potassium dichromate.
Furthermore, duration of an experiment can be crucial for comparing short time DF with growth inhibition tests of M. aeruginosa HUB018 and D. subspicatus. While short-term incubations only account for acute toxicity reflected by the investigated parameters, growth inhibition experiments spanning over several generations consider chronic, accumulative or mutative effects of toxins. In our study, M. aeruginosa HUB018 and D. subspicatus showed a close correlation between DDF and growth inhibition, except for small, measurable toxicity effects (small changes in DF and low growth inhibition over the entire tested concentration spectra). Similar to other studies [52,54,63], no overall sensitive species could be detected, but it will be an exciting challenge to test the DF response to a much wider range of pollutants for both phytoplankton groups, in particular effects caused by respiratory toxins on the DF decay of Microcystis sp. Our results indicate that our DF approach is a promising tool for investigating physiological alterations caused by chemicals and pollutants, not only in green algae as has been shown earlier [14,17], but also for cyanobacteria. Furthermore, our DF test introduced in this study can easily be modified for on-line monitoring, where rapid and sensitive parameters are required to obtain a fast algal response towards a potential contamination [40]. Cyanobacteria that often dominate natural phytoplankton communities, should be included in such monitoring systems. It will be an exciting challenge to test the DF response to a much wider range of pollutants for both phytoplankton groups, in particular effects caused by respiratory toxins on the DF decay of Microcystis sp.

Conclusions
Application of our DF approach to testing toxicity in cyanobacteria provided a reliable, rapid, and sensitive tool, which yields similar results as the applied growth inhibition tests. Since the DF approach is an in vivo test, it does not require timeconsuming preparations and allows for automation for high frequency measurements. The sensitivity of cyanobacteria and green algae differs significantly to each tested chemicals. While green algae are commonly used in standard toxicity tests and biomonitoring, cyanobacteria, often dominating phytoplankton communities, have largely been neglected as test organisms for reliably monitoring of environmental disturbances by toxic substances. Adapting our DF approach to cyanobacteria for online monitoring seems to be promising for monitoring complex, natural phytoplankton communities.