Cell Death and Metabolic Stress in Gymnodinium catenatum Induced by Allelopathy

Allelopathy between phytoplankton species can promote cellular stress and programmed cell death (PCD). The raphidophyte Chattonella marina var. marina, and the dinoflagellates Margalefidinium polykrikoides and Gymnodinium impudicum have allelopathic effects on Gymnodinium catenatum; however, the physiological mechanisms are unknown. We evaluated whether the allelopathic effect promotes cellular stress and activates PCD in G. catenatum. Cultures of G. catenatum were exposed to cell-free media of C. marina var. marina, M. polykrikoides and G. impudicum. The mortality, superoxide radical (O2●−) production, thiobarbituric acid reactive substances (TBARS) levels, superoxide dismutase (SOD) activity, protein content, and caspase-3 activity were quantified. Mortality (between 57 and 79%) was registered in G. catenatum after exposure to cell-free media of the three species. The maximal O2●− production occurred with C. marina var. marina cell-free media. The highest TBARS levels and SOD activity in G. catenatum were recorded with cell-free media from G. impudicum. The highest protein content was recorded with cell-free media from M. polykrikoides. All cell-free media caused an increase in the activity of caspase-3. These results indicate that the allelopathic effect in G. catenatum promotes cell stress and caspase-3 activation, as a signal for the induction of programmed cell death.


Introduction
The succession among phytoplankton species during harmful algal bloom (HAB) events is complex, and the mechanisms of bloom-species selection and how some species dominate over others is not clear [1][2][3][4]. In allelopathic interactions, specific chemical compounds (allelochemicals) produced by one species can induce damage or benefit another species [5]. Some chemical signals between co-existing phytoplankton groups induce programmed cell death (PCD) as a selective strategy in intraspecies competition [6][7][8].

Allelopathy Experiments
Cell-free media from the three species caused mortality in G. catenatum (Figure 3). The highest mortality was found when G. catenatum was exposed to the largest volumes (50 and 75 mL) of cell-free media ( Figure 3A). When 75 mL from C. marina var. marina cell-free media was added, there was 79% of mortality in G. catenatum after 72 h. With the same volume (75 mL of cell-free medium) of M. polykrikoides in the same period (72 h), 74% of death in G. catenatum cells was observed, while G. impudicum caused 57% mortality at 72 h ( Figure 3B,C). Conversely, 50 mL of cell-free medium from G. impudicum caused 65 % of mortality in G. catenatum cells at 72 h, while when cells were exposed to 75 mL of cell-free media, the mortality in G. catenatum cells was lower (62 and 57 % at 48 and 72 h, respectively) compared to the mortality caused with a volume of 50 mL (49 and 65%). The cell abundance of G. catenatum cultures exposed to cell-free media from C. marina var. marina, M. polykrikoides and G. impudicum decreased in less than 72 h ( Figure 3A-C). Cell-free media (75 mL) from C. marina var. marina caused a maximum decrease from 500 to 184 ± 12 cells mL −1 in G. catenatum after 72 h of exposure; with the same volume of cell-free media from M. polykrikoides, a decrease from 500 to 224 ± 21 cells mL −1 occurred. When 75 mL of cell-free media of G. impudicum was added, the highest decrease in G. catenatum cells occurred at 48 h from 500 to 214 ± 44 cells mL −1 ; however, a slight increase to 287 ± 75 cells mL −1 was observed at 72 h, suggesting a recovering process in the cell growth. In the control treatment with their own cell-free media, cell abundance of G. catenatum increased from 500 to 887 ± 8 cells mL −1 from 0 to 72 h. In the control treatment with GSe media, the cell abundance of G. catenatum was similar to that reported in the growth phase 500 to 899 ± 12 cells mL −1 from time 0 to 72 h (data not shown by the similarity of the results). volume of cell-free media from M. polykrikoides, a decrease from 500 to 224 ± 21 cells mL −1 occurred. When 75 mL of cell-free media of G. impudicum was added, the highest decrease in G. catenatum cells occurred at 48 h from 500 to 214 ± 44 cells mL −1 ; however, a slight increase to 287 ± 75 cells mL −1 was observed at 72 h, suggesting a recovering process in the cell growth. In the control treatment with their own cell-free media, cell abundance of G. catenatum increased from 500 to 887 ± 8 cells mL −1 from 0 to 72 h. In the control treatment with GSe media, the cell abundance of G. catenatum was similar to that reported in the growth phase 500 to 899 ± 12 cells mL −1 from time 0 to 72 h (data not shown by the similarity of the results).

Thiobarbituric Acid Reactive Substances (TBARs) Levels
Lipid peroxidation in G. catenatum was higher when exposed to 25 and 50 mL volumes at 48 and 72 h of cell-free media from all species ( Figure 4D-F). When G. catenatum was exposed to cell-free media from the raphidophyte at 24 h ( Figure 4D   shown as mean ± SD. Letters represent significant differences among treatments (different volumes of cell-free media) with respect to the respective control (p < 0.05, n = 3).

Protein Content
Total protein content in G. catenatum was significantly different among the different treatments when adding cell-free media from the three species ( Figure  With 75 mL of G. impudicum cell-free medium, the protein content in G. catenatum was 0.67 ± 0.05 mg mL −1 proteins 10 −4 cells, higher than when adding 25 and 50 mL (~0.47 ± 0.05 mg mL −1 proteins 10 −4 cells); statistical differences between the control and 75 mL of cell-free media compared to 25 and 50 mL of cell-free media treatments were found (ANOVA, F 3,8 = 3.53, (p < 0.05)).
Correlation analyses of caspase-3 activity with molecules related to oxidative stress and the total protein content in G. catenatum exposed to cell-free filtrates of C. marina var. marina, M. polykrikoides and G. impudicum are shown in Table 2. The cell-free media from C. marina var. marina showed strong negative correlations between caspase-3 and O 2 •− production in G. catenatum at 48 and 72 h (r = −0.796, r = −0.707, p < 0.05), respectively; TBARS levels had a negative correlation (r = −0.927) at 72 h (p < 0.05); the SOD activity during 24 and 48 h had a negative correlation (r = −0.852 and r = −0.733, respectively) (p < 0.05); the protein content also presented a negative correlation at 24 and 48 h (r = −0.731 and r = −0.739, respectively) (p < 0.05). With M. polykrikoides cell-free media, strong significant negative correlations were found in G. catenatum between caspase-3 activity and TBARs levels (r = −0.923) at 72 h (p < 0.05); with the protein content a negative correlation at 24 h and 72 h (r = −0.709 and r = −0.751, respectively), was observed (p < 0.05). In addition, when G. catenatum was exposed to the cell-free filtrates of G. impudicum, caspase-3 activity had a positive correlation with TBARS levels at 24 h (r = 0.783) and at 48 h had a negative correlation (r = −0.709), while at 72 h there was a strong significant positive correlation (r = 0.927) (p < 0.05) and a negative correlation with the protein content (r = −0.737) after 72 h of exposure. Correlation analyses of caspase-3 activity with molecules related to oxidative stre and the total protein content in G. catenatum exposed to cell-free filtrates of C. marina va marina, M. polykrikoides and G. impudicum are shown in Table 2 Table 2. Correlation between caspase-3-like activity and the oxidative stress indicators in Gymnodinium catenatum exposed to allelopathic cell-free media of Chattonella marina var. marina, Margalefidinium polykrikoides and Gymnodinium impudicum. •− , Superoxide radical production; TBARS, thiobarbituric acid reactive substances; SOD, superoxide dismutase activity, total protein concentration. * Marked correlation are significance at p < 0.05 (n = 12).

Discussion
Allelopathy in Gymnodinium catenatum via cell-free media promotes oxidative stress, inducing the activation of caspase-3-like protein, involved in apoptosis processes. Evidence suggests a relationship between oxidative stress and caspase-3 activity with the growth phases. The allelopathic effect of cell-free cultures from C. marina var. marina caused the maximum O 2 •− production in G. catenatum; the highest TBARS levels in G. catenatum were determined with cell-free media from M. polykrikoides and G. impudicum. Cell-free media of G. impudicum caused maximum SOD activity. The cell-free media of C. marina var. marina and M. polykrikoides caused the lowest SOD activity. The protein content in G. catenatum due to allelopathic effect was similar when exposed to cell-free media of C. marina var. marina, M. polykrikoides and G. impudicum. Caspase-3 activity was highest in G. catenatum with the cell-free media from all species. Strong positive and negative correlations were recorded between the caspase-3 activity and O 2 •− production, TBARS levels, SOD activity and protein content in G. catenatum, due to the allelopathic effect of C. marina var. marina, M. polykrikoides and G. impudicum cell-free media.
Growth rates of dinoflagellates species vary among strain and culture conditions. The average growth rate recorded for G. catenatum of 0.57 div day −1 was lower compared to values reported by Band-Schmidt et al. [10] for other strains from Mexico,~0.77 div day −1 , and similar to those reported by Bravo and Anderson [62] with 0.56 div day −1 in strains from Spain (Table 3). In this study, M. polykrikoides registered an average growth rate of 0.59 div day −1 , similar to the values (0.56 div day −1 ) reported by Yamatogi et al. [63] for strains from Japan, while Kim et al. [64] reported lower growth rates (0.35 div day −1 ) also in a Japanese strain. Recently, Aquino-Cruz et al. [23] reported a growth rate of 0.41 div day −1 for M. polykrikoides from a strain isolated from the coasts of Mexico. Chattonella marina var. marina had a growth rate of 0.43 div day −1 , similar to the growth rate reported by Marshall and Hallegraeff [65], with 0.56 div day −1 for an Australian strain, and higher than those reported by Band-Schmidt et al. [61] of 0.30 div day −1 in a Mexican strain. For G. impudicum, a growth rate of 0.48 div day −1 was recorded, higher than a strain from Korea reported by Oh et al. [15] with 0.37 div day −1 . Such differences and similarities in growth rates of strains of the same species can be due to multiple variables, such as the culture medium, photoperiod, salinity, temperature conditions or even geographical origin of the strains. in the last two species [24,27,66]. In this research, C. marina var. marina and M. polykrikoides presented the highest O 2 •− production values; however, there were no significant differences between C. marina var. marina and G. impudicum, and between M. polykrikoides and G. catenatum. These results suggest that other species of phytoplankton also produce high amounts of reactive oxygen species. In other studies, a high O 2 •− production in G. catenatum was reported to be 59.7 ± 15.2 CCU per cell of O 2 •− production, and total O 2 •− measured in 300 µL of culture 8.0 ± 0.1 TCU × 10 4 , concluding that Gymnodiniales dinoflagellates can be potentially toxic due to O 2 •− production [66]. In the phytoflagellate aggregations, ROS are generated as signaling agents, decreasing their production during the decay of cultures; they also are considered precursors of allelopathic effects [67,68].
Lipid peroxidation can be taken as an indicator of oxidative damage in lipids. Aquino-Cruz et al. [24] reported for Chattonella spp. and M. polykrikoides maximum TBARS values during the EEP, whereas in this study TBARS concentration increased towards the DP, especially in M. polykrikoides. In this study, the analyses were carried out between the EP and DP, while Aquino-Cruz et al. [24] evaluated TBARS levels only up to the EP. The results of this study can be attributed to the fact that as cells age, lipid peroxidation and mortality increase [60,69], since culture senescence is also associated with higher SOD activity. The outcome of oxidative damage is reflected in the production and integrity of proteins [69]. In this study, all analyzed species had the highest protein content in younger cultures. These results suggest a relationship in the oxidative stress as a consequence of a decreased or increased antioxidant enzyme SOD activity and glutathione. This also was reported in cyanobacteria species, Aphanizomenon ovalisporum and Microcystis aeruginosa. Cell extracts, and pure toxins microcystin and cylindrospermopsin increased the antioxidant activity in the green algae Chlorella vulgaris [70]; this activity can potentially act as a conservative strategy similar to that of the Antarctic cyanobacterium Nostoc commune, which possesses two isoforms of SOD and catalase to endure various stress conditions [71]. In higher plants, the reduced glutathione (GHS) also regulates water status and prevents chlorophyll degradation under biotic stress [72].
The caspase-3 activity in strains of C. marina var. marina, M. polykrikoides, G. impudicum and G. catenatum suggest that M. polykrikoides and G. catenatum have a shorter growth curve and a higher signal of caspase-3 activity. The maximum caspase-3 activity occurs during the DP; therefore, signaling programmed death is activated towards the end of the growth curve, probably due to a decrease of nutrients in the culture medium. Programmed cell death by nutrient decrease in phytoplankton cultures was reported in laboratory conditions in the coccolithophore Emiliana huxyleyi [73], and the diatom Thalassiosira pseudonana [74]. Nutrients may contribute, in the natural environment, to the regulating mechanism of PCD in the dinoflagellates Karenia brevis and Prorocentrum donghaiense, as a survival strategy [49,75,76].
Nutrients can affect allelopathy [77,78]. Nutrient depletion increases the toxicity of allelochemicals and their production [79]. The addition of nutrients can end allelopathy [80] and promote greater co-occurrence between phytoplankton species [36]. In this study, nutrient concentrations were not analyzed, but according to the experimental design suggested for allelopathic studies [81,82] and by Legrand et al. [34], all the experiments were carried out under optimal nutrient conditions for both the donor species of the allelopathic effect (C. marina var. marina, M. polykrikoides, and G. impudicum) and the acceptor species (G. catenatum). In addition, maximum exposure time (72 h) in allelopathic experiments was short, compared to the 18 or 26 days needed to research the DP in the tested species; therefore, our results can be assumed to be due to cell-free media and not nutrient-depletion media.
Allelopathy decreases algal growth, damages cell membranes and causes high mortality via lysis [33,83,84]. Mortality above 76 % associated with cells lysis caused by cell-free media, and cultures with and without cell contact was reported for G. catenatum [18,38]. Morphological damages were reported in vegetative cells of Alexandrium pacificum caused by algal allelochemicals [85]. Cochlodinium germinatum causes high mortality via lysis in the microalgae Prorocentrum micans, Akashiwo sanguinea, Karlodinium veneficum, and Rhodomona salina [86]. Lysis and temporary cyst formed in Scrippsiella trochoidea by allelopathic effects caused from cell-free media from Karenia mikimotoi, Alexandrium tamarense and Chrysochromulina polylepis [35,87]. Results from this study suggest that when the allelopathic effect in G. catenatum is more intense (i.e., higher mortality), cellular stress signals detected are lower. Greater O 2 •− production in G. catenatum was caused by lower volumes (25 and 50 mL) of cell-free media from C. marina var. marina, M. polykrikoides and G. impudicum (Figure 4A-C).
Lu et al. [88] reported the activity of allelochemicals, finding the maximum production of ROS in the cyanobacterium Microcystis aeruginosa treated with the allelochemical phenol pyrogalic acid at 48 h, which was 2 times higher than what was recorded at 216 h. In this study, O 2 •− production and the TBARS lipid peroxidation were consistently higher in G. catenatum when exposed to the lower volumes of cell-free filtrates. The SOD activity in G. catenatum depends on the cell-free media of the species from which it was obtained and on the exposure time ( Figure 4G-I). Hong et al. [89] described the growth and the SOD activity in M. aeruginosa after 4 h of exposure to the allelochemical ethyl 2methyl acetoacetate (EMA) isolated from the reed Phragmites communis at concentrations from 0.24 to 4 mg L −1 . This suggests that the highest O 2 •− production decreased the cytoplasmic SOD activity, and the antioxidant defense may cause growth inhibition in M. aeruginosa from initial exposure to EMA [89]. It is probable that the decrease in SOD activity observed in G. catenatum in treatments with cell-free media from C. marina var. marina, M. polykrikoides and G. impudicum, which had the highest mortality, could be explained by a similar mechanism of action when maximum O 2 •− production is related to the exposure time, although not necessarily to higher doses of cell-free culture media.
The allelopathic effect associated with oxidative stress in phytoplankton species affects their photosynthetic capacity, potentially causing a decrease in the photochemical performance of photosystem II, which in turn increases the permeability of the membrane due to the oxidation of fatty acids [90][91][92]. During the interaction with the larger volumes (75 mL) of cell-free media from all donor species of the allelopathic effect, a high concentration of total proteins was recorded in G. catenatum. These results are consistent with those of Zhang et al. [93] who reported that higher ROS production promotes lower protein content in the dinoflagellate Heterosigma akashiwo exposed to 1.0 µg mL −1 of prodigiosin, an algaecide from bacterial origin, even when with the highest concentration treatment the protein content was similar to the controls.
Allelochemicals and algicides increase oxidative-stress-activating pathways related to programmed cell death [73,83]. In this study, treatment of G. catenatum cell-free media from C. marina var. marina, M. polykrikoides and G. impudicum promoted caspase-3 activity proportional to dose-time. Similar results were reported for other phytoplankton groups exposed to allelopathic or algicidal compounds. Linoleic acid promotes the caspase-3 activity as a result of oxidative stress in Karenia mikimotoi [94]. Polybrominated diphenyl ethers induce oxidative stress, activating programmed cell death signals in the diatom Thalassiosira pseudonana [95]. During blooms of Peridinium gutunense, CO 2 limitation triggers a ROS-PCD cascade reaction [83]. Also, in M. polykrikoides, exposure to the algicide copper sulfate and oxidizing chlorine activates a gene related with a metacaspase, a type of protease analogous to caspases [96].
In this study, caspase-3 activity was correlated to O 2 •− production, TBARS levels, SOD activity and total protein in G. catenatum exposed to cell-free filtrates from C. marina var. marina, M. polykrikoides and G. impudicum (Table 2). This supports the hypothesis of oxidative stress and caspase-3 activation related to programmed cell death caused by allelopathy [8,60,83,97]. Therefore, understanding the ecological importance of programmed cell death and the relationship to chemical signaling (e.g., allelopathy) is important to understanding microscale phenomena that are reflected in the phytoplankton community [97].
Although there are no field studies of the allelopathic effect in G. catenatum, the continuous dominance in cell abundance of C. marina var. marina and M. polykrikoides on G.
catenatum when these species coexist during blooms has been reported [13][14][15]17,28,[98][99][100]. Chattonella spp. and M. polykrikoides were reported to promote allelopathy, inhibit growth, deform cells and cause lysis in chlorophytes, diatoms and dinoflagellates [67,101,102]. Similarly, an allelopathic effect, associated with growth inhibition, higher number of cellchains, loss of flagella, cell deformation, swelling, prominent nucleus, rupture of cell membrane, lysis and formation of temporary cysts of C. marina on G. catenatum under laboratory conditions was reported by Fernández-Herrera et al. [38]. In addition, the allelopathic effect, including growth inhibition, cell chain fragmentation, rounded cells, loss of flagella, cell damage and lysis of M. polykrikoides and G. impudicum on G. catenatum was reported by Band-Schmidt et al. [18]. Results from this study could suggest a similar effect of C. marina var. marina, M. polykrikoides and G. impudicum on G. catenatum, potentially via allelochemicals extruded to the culture media.
The challenge for future studies is to elucidate the allelochemicals responsible for the dominance of phytoplankton species coexisting with G. catenatum. Bidle [8,60] proposed that programmed cell death acts as an ancestral survival strategy in the internal machinery of phytoplankton species in response to abiotic and biotic factors. Understanding the type of programmed cell death related to allelopathy can contribute to comprehend the dynamics, duration and species succession during blooms, as well as potential strategies in G. catenatum to survive the dominance of allelopathic species. Our results suggest that the allelopathic activity of C. marina var. marina, M. polykrikoides and G. impudicum in G. catenatum activates multiple oxidative stress mechanisms.

Conclusions
This study suggests that an allelopathic effect caused by cell-free media of the raphidophyte Chattonella marina var. marina and the dinoflagellates Margalefidinium polykrikoides and Gymnodinium impudicum on the toxic dinoflagellate Gymnodinium catenatum, potentially via allelochemicals extruded to the culture media, promotes changes in O 2 •− production, lipid peroxidation levels, SOD activity and total protein content. Cell-free media from C. marina var. marina increase the caspase-3 activity in G. catenatum correlated with O 2 •− production, TBARs levels and SOD activity. In addition, the cell-free media of M. polykrikoides promotes an increase of caspase-3 activity in G. catenatum correlated positively and negatively with O 2 •− production and TBARS. The increase in the caspase-3 activity in G. catenatum by cell-free media of G. impudicum activity is correlated positively with TBARS levels. All cell-free media promoted a higher caspase-3 activity that was correlated positively with the protein content. These results support the hypothesis that the oxidative stress and the increase in the caspase-3 activity can induce programmed cell death in G. catenatum.
Furthermore, results from this study suggest that the increased caspase-3 activity induces programmed cell death in G. catenatum. These results suggest an effect of C. marina var. marina, M. polykrikoides and G. impudicum on G. catenatum potentially via allelochemicals extruded to the culture media.  Table 1. All strains were cultured in modified GSe medium with the addition of earth worm extract obtained by the composte of organic waste using earthworms according to Bustillos-Guzmán et al. [103]; briefly, 50 g of earth worm humus was dissolved in 500 mL of distilled water and sterilized at 121 • C for 15 min. After 24 h it was filtered twice through GFF filters and refrigerated until its use [103]. Strains were maintained at 12/12 h light/dark cycle,~150 µmol photons m −1 s −1 of irradiance at 24 ± 1 • C and 34 salinity. These culture conditions were maintained in all the experiments.

Growth Rates and Growth Curves Stages
Growth rates and growth curves stages were determined by triplicate for each strain in 300 mL Erlenmeyer culture flasks with 150 mL of media. Growth curves were initiated with 500 cells mL −1 and every second day a 2.0 mL sample was fixed with lugol for cell counts. Only in the case of Chattonella, cells were fixed with hepes-buffered paraformaldehyde [104]. Cells were counted on 1.0 mL Sedgwick-Rafter slide under an inverted microscope (Carl Zeiss Axio Vert. A1). Cell density was used to calculate growth rates (µ) [105] according to the formula: where, N t and N 0 are the total cells at the end of exponential phase (T t ) and start of log phase (T 0 ), respectively. The number of generations per day (tg) was calculated with the formula [106].
For each species, the early exponential (EEP), late exponential (LEP), and decline phase (DP) were determined.

Allelopathy Experiments
Monoalgal batch cultures of C. marina var. marina, M. polykrikoides and G. impudicum were inoculated at an initial cell density of 1000 cells mL −1 in 1000 mL with 500 mL of medium and maintained 6 days until early exponential phase. In exponential phase, from a volume of 30 mL, cells were removed by gentle filtration using glass GF/F filters (Whatman ® ICT, SL, Gipuzkoa, Spain). Volumes of cell-free culture media (25, 50 and 75 mL) were recovered, these media contained cell exudates of each species, and were added to cultures of G. catenatum in 300 mL flasks in the following proportions (16% of cell-free media, 69% fresh GSe media and 14% of G. catenatum cells to 25 mL treatment), (32% of cell-free media, 52% fresh GSe media and 14% of G. catenatum cells to 50 mL treatment) and (48% of cell-free media, 37% fresh GSe media and 14% of G. catenatum cells to 75 mL treatment), all flasks with 500 cells/mL in a final volume of 150 mL, the cell free media, by triplicate. As a control, G. catenatum cultures were inoculated only with the GSe media and a control with their own culture medium filtrate. After 24, 48 and 72 h, from each treatment, a 2.0 mL sample was fixed with lugol for cell counts. For the determination of caspase-3-like activity, a second sample of 2.0 mL was centrifuged at 3000 rpm, the culture medium was removed, and the cell pellet was frozen at −80 • C until analyzed. The remaining volume (~146 mL) was collected in three Falcon tubes of 50 mL and placed on ice at −4 • C to be analyzed immediately. Samples from all three phases of the growth curve, on days 4, 14 and 18 for G. catenatum and M. polykrikoides, and on days 4, 24 and 26 for C. marina var. marina and G. impudicum, were analyzed separately.

Mortality
With the abundance data, the percentage of mortality (PD) was calculated as an indicator of the allelopathic effect, with the equation described by Fistarol et al. [35], where the number of cells of G. catenatum exposed to cell-free media (D) was related to the number of cells registered in the control (N cont) at the time of exposure to the cell-free filtrates (24, 48 and 72 h).

Superoxide Radical (O 2 •− ) Production
Production of O 2 •− was analyzed through the reduction of ferricytochrome C. The remaining volume (~146 mL) of samples was centrifuged at 2000 rpm at 24 • C; the cell pellet was recovered and re-suspended in 5.0 mL of GSe medium (cell homogenate). Then, 250 µL of the cell homogenate was transferred to a 1.75 mL conical microcentrifuge tube (Fisherbrand TM ) and kept on ice. Cells were lysed by vortexing for 2 min. Krebs buffer (0.11 NaCl, 4.7 mM KCl, 12 nM MgSO 4 , 12 nM NaH 2 PO 4 , 25 mM NaHCO 4 , and 1 mM glucose) and cytochrome-C (15 µM) were added. Tubes were capped and incubated at 37 ± 1 • C during 15 min in a shaking water bath (Polyscience, Niles, IL, USA). Nethylmaleimide (3 nM) was added, and the homogenate was shaken to stop the reaction. Tubes were centrifuged at 3000 rpm, 4 • C for 10 min. Supernatant was transferred to polystyrene disposable cuvettes (Fisherbrand TM ), and absorbance was recorded at 560 nm in a spectrophotometer (Beckman Coulter DU 800, Fullerton, CA, USA). A blank without homogenate was included for each sample. Superoxide radical production was calculated, according to the following formula [107]: •− = Abs (Sample − Blank)/21 nmol/L cm = nmoles O 2 •− /min mL (4)

Thiobarbituric Acid Reactive Substances (TBARS) Levels
Hydroperoxides and aldehydes resulting from lipid peroxidation in the sample were analyzed by the reaction of 2-thiobarbirutic acid (TBA) to form malondialdehyde (MDA), following the method of Persky et al. [108] and Zenteno-Savín et al. [109]. In a 1.7 mL conical microcentrifuge tube (Fisherbrand TM ), 500 µL of the cell homogenate was incubated at 37 • C in a shaking water bath (Polyscience). Tubes were placed on ice and in a solution of trichloroacetic acid (TCA 20 %), and HCl (1.0 M) was added to stop the reaction, followed by the addition of TBA 1%, and vortexed. Samples were incubated at 90 • C in a shaking water bath for 10 min, followed by 1 min on ice and centrifuged at 3000 rpm (1500× g) for 10 min at 4 • C. The absorbance of the supernatant was recorded in a spectrophotometer (Beckman Coulter DU 800, Fullerton, CA, USA) at 530 nm. Calculations for TBARS concentration in cells were done adjusting the values to a standard curve of 1,1,3,3-tetraethoxypropane (TEP), in concentrations that ranged from 0 to 5 mmol 250 µL −1 , and the result was expressed in nmol 10 −4 cells −1 .

Superoxide Dismutase (SOD) Activity
Superoxide dismutase activity was determined using the xanthine/xanthine oxidase system as O 2 •− generator, when it reacts with nitroblue tetrazolium (NBT), reducing it and producing formazan. This chemical species can be detected by spectrophotometry, when SOD inhibits the reduction of NBT [110].

Total Protein
The amount of total proteins was determined by the Bradford method [111]. This method is based on the reaction of Coomassie brilliant blue (Bio-Rad ® inc, Hercules, CA, USA.) with the basic amino acid residues, especially arginine. Phosphate buffer (0.1M) and EDTA (60 mM) were added to an aliquot of 250 µL the cell homogenate; samples were mixed in a vortex for 2 min, centrifuged at 3000 rpm for 15 min at 4 • C and the supernatant was recovered. In a 96-well microplate, samples and the standard curve with bovine serum albumin (BSA) at concentrations from 0.005 to 0.2 mg mL −1 of distilled water were mixed with the colorant and the microplate was shaken gently for 30 s. The microplate was covered and incubated at 25 • C. The absorbance was determined at 620 nm in a microplate reader (Thermo-Scientific).
Results were expressed in mg mL −1 proteins 10 −4 cells.

Caspase-3 Activity
Apoptotic activity was determined using the Enzchek Caspase-3 Assay Kit #2 (Invitrogen). In triplicate, cell homogenates were centrifuged at 3000 rpm in 2.0 mL microcentrifuge tubes (Eppendorf®), the culture medium was removed, and the cell pellet was frozen at −80 • C. Each pellet was resuspended in the lysis buffer, stirred during 2 min in a vortex, frozen and thawed twice; cell pellet lysates were centrifuged at 5000 rpm, during 5 min at 4 • C. The supernatant was recovered and transferred to 96-well microplates, Z-DEVD-R110 (5 mM) substrate in solution with 2x reaction buffer added to all wells. Samples were incubated for 2 h in darkness. The fluorescent signal of rhodamine (R-110) whit substrate-enzyme Z-DEVD-R110 was subsequently determined using a microplate reader (Ex 485 nm; Em: 520 nm). Lysis buffer (1X) and GSe medium were used as a negative control. The moles produced in the reactions with the activity of caspase-3 were calculated in relative fluorescent units (RUF) h −1 mg −1 protein. Prior to adding the substrate, cell pellets were incubated for 30 min with reversible caspase inhibitor Ac-DEVD-CHO (5 mM) to confirm that the observed fluorescence corresponded to the activity of the caspase-3 proteases. [112,113].

Statistics Analysis
Kolmogorov-Smirnov, Shapiro-Wilk normality tests and Levene test homoscedasticity were performed on all data. For growth phases and allelopathic experiments, a one-way analysis of variance (ANOVA) (p < 0.05) with a post hoc Tukey HSD (Honest Significant Difference) were applied. A Pearson correlation analysis was performed to evaluate the relationship between PCD and ROS. All statistical analyses were done using Statistica StatSoft ® (Tulsa, OK, USA) software.