Ecophysiological Aspects and sxt Genes Expression Underlying Induced Chemical Defense in STX-Producing Raphidiopsis raciborskii (Cyanobacteria) against the Zooplankter Daphnia gessneri

Cyanobacteria stand out among phytoplankton when they form massive blooms and produce toxins. Because cyanotoxin genes date to the origin of metazoans, the hypothesis that cyanotoxins function as a defense against herbivory is still debated. Although their primary cellular function might vary, these metabolites could have evolved as an anti-predator response. Here we evaluated the physiological and molecular responses of a saxitoxin-producing Raphidiopsis raciborskii to infochemicals released by the grazer Daphnia gessneri. Induced chemical defenses were evidenced in R. raciborskii as a significant increase in the transcription level of sxt genes, followed by an increase in saxitoxin content when exposed to predator cues. Moreover, cyanobacterial growth decreased, and no significant effects on photosynthesis or morphology were observed. Overall, the induced defense response was accompanied by a trade-off between toxin production and growth. These results shed light on the mechanisms underlying zooplankton–cyanobacteria interactions in aquatic food webs. The widespread occurrence of the cyanobacterium R. raciborskii in freshwater bodies has been attributed to its phenotypic plasticity. Assessing the potential of this species to thrive over interaction filters such as zooplankton grazing pressure can enhance our understanding of its adaptive success.


Introduction
Predator-prey interactions foster the evolution of defensive adaptations [1]. A series of traits that deter predation by zooplankton have been described in phytoplankton, including the formation of colonies and spines, and/or the production of harmful metabolites [2,3]. In addition to the direct effect provided by active foraging, zooplankton also release infochemicals that induce functional responses in phytoplankton species, such as increased toxin production [4][5][6][7] and colony formation or trichome stretching, which in turn may limit grazing pressure [8][9][10][11]. These infochemicals are also referred to as kairomones or predator cues, which are chemical signals released by predators that indicate a threat and hypothesis (ODH) predicts that toxin cell content should increase in response to threat cues, to maintain population fitness [7]. According to Pavia et al. [47], the ODH assumes that both chemical and morphological defenses are costly; therefore, natural selection will act to optimize their benefit-cost ratio. This leads to the prediction that there should be a trade-off with energy allocation to chemical vs. morphological defenses.
The molecular basis of grazer-induced defenses in phytoplankton is poorly understood. A study of the expression of target genes can provide insight into defensive phenotypes, and aid exploration of the adaptive traits mediating anti-predator responses. The conservation of sxt gene clusters in different cyanobacterial species indicates that STXs may play an adaptive role in STX-producing cyanobacteria [48]. No study to date has characterized changes in the expression of genes related to STX synthesis upon the exposure of cyanobacteria to zooplankton, nor in the more general context of predatorinduced defenses. Here, we aimed to assess the effects of chemical cues released by the neotropical daphnid Daphnia gessneri on morphological and ecophysiological traits and sxt gene expression in the R. raciborskii strain T3. We tested the hypothesis that predator infochemicals promote a defense response (at the molecular and physiological levels) in the cyanobacterium but at an associated cost, as assumed by the ODH (see [49,50]).

Predator-Induced Chemical Defense
D. gessneri infochemicals significantly enhanced the total STXs production in R. raciborskii T3 over time (Figure 1; Supplementary Table S1). On the 6th day of incubation with infochemicals, the cellular amount of STXs expressed as STX biovolume quota was 0.63 ± 0.12 µg mm −3 . This amount was twice that measured in the control (0.31 ± 0.10 µg mm −3 ; Bonferroni's test, p < 0.001), which indicated that a chemically induced defense response was displayed by the cyanobacterium (Figure 1). In addition, neoSTX and STX content increased significantly in response to the infochemicals, which resulted in a higher cellular relative toxicity (infochemicals = 2.92 ± 0.53 µg STX eq mm −3 ; control = 1.41 ± 0.47 µg STX eq mm −3 ; Bonferroni's test, p < 0.001) (Figure 2; Supplementary Table S2). examined the early responses of STX-producing cyanobacteria to chemical signals released by grazers.
The evolution of inducible defenses appears to be favored over constitutive defenses because grazing pressure by zooplankton varies on both temporal and spatial scales [45]. Defenses induced by the presence or action of predators have lower costs compared with constitutive defenses [46]. If predation can be deterred by STXs, the optimal defense hypothesis (ODH) predicts that toxin cell content should increase in response to threat cues, to maintain population fitness [7]. According to Pavia et al. [47], the ODH assumes that both chemical and morphological defenses are costly; therefore, natural selection will act to optimize their benefit-cost ratio. This leads to the prediction that there should be a trade-off with energy allocation to chemical vs. morphological defenses.
The molecular basis of grazer-induced defenses in phytoplankton is poorly understood. A study of the expression of target genes can provide insight into defensive phenotypes, and aid exploration of the adaptive traits mediating anti-predator responses. The conservation of sxt gene clusters in different cyanobacterial species indicates that STXs may play an adaptive role in STX-producing cyanobacteria [48]. No study to date has characterized changes in the expression of genes related to STX synthesis upon the exposure of cyanobacteria to zooplankton, nor in the more general context of predatorinduced defenses. Here, we aimed to assess the effects of chemical cues released by the neotropical daphnid Daphnia gessneri on morphological and ecophysiological traits and sxt gene expression in the R. raciborskii strain T3. We tested the hypothesis that predator infochemicals promote a defense response (at the molecular and physiological levels) in the cyanobacterium but at an associated cost, as assumed by the ODH (see [49,50]).

Predator-Induced Chemical Defense
D. gessneri infochemicals significantly enhanced the total STXs production in R. raciborskii T3 over time ( Figure 1; Supplementary Table S1). On the 6th day of incubation with infochemicals, the cellular amount of STXs expressed as STX biovolume quota was 0.63 ± 0.12 µg mm −3 . This amount was twice that measured in the control (0.31 ± 0.10 µg mm −3 ; Bonferroni's test, p < 0.001), which indicated that a chemically induced defense response was displayed by the cyanobacterium (Figure 1). In addition, neoSTX and STX content increased significantly in response to the infochemicals, which resulted in a higher cellular relative toxicity (infochemicals = 2.92 ± 0.53 µg STXeq mm −3 ; control = 1.41 ± 0.47 µg STXeq mm −3 ; Bonferroni's test, p < 0.001) (Figure 2; Supplementary Table S2). Total saxitoxin quota by biovolume of Raphidiopsis raciborskii T3 exposed to Daphnia gessneri infochemicals (gray bars) and under control conditions (white bars). Significant differences (***) = Bonferroni's test, p < 0.001. Total saxitoxin quota by biovolume of Raphidiopsis raciborskii T3 exposed to Daphnia gessneri infochemicals (gray bars) and under control conditions (white bars). Significant differences (***) = Bonferroni's test, p < 0.001. There was no significant effect of kairomones on the total pool of dissolved STXs ( Figure 3). Predator cues resulted in an STX release rate of 0.20 ± 0.04 µg L −1 d −1 , while under control conditions, the release rate was 0.16 ± 0.03 µg L −1 d −1 (data not shown). Concentrations of (A) intracellular and (B) extracellular saxitoxin in Raphidiopsis raciborskii T3 under control conditions (black bars) and exposed to Daphnia gessneri infochemicals (gray bars). No significant differences were detected between the two conditions. Overall, the STX biovolume content, relative biovolume toxicity and volumetric STX concentration increased over time (two-way RM ANOVA, p < 0.01) (Figures 1-3). The latter also reflected the increase in cell concentration in both experimental conditions. First-order rate kinetics showed different patterns in toxin production/growth in R. raciborskii T3. Under control conditions, STX production was coupled to growth (1:1 µstx/µg ratio) (Figure 4). In contrast, predator cues caused the mean value of the µstx/µg ratio to be significantly greater than 1 (Student's t-test; T = 4.055, p < 0.05), indicating that the specific rate of STX production was faster than the specific growth rate, resulting in increased cell toxin accumulation ( Figure 1). There was no significant effect of kairomones on the total pool of dissolved STXs ( Figure 3). Predator cues resulted in an STX release rate of 0.20 ± 0.04 µg L −1 d −1 , while under control conditions, the release rate was 0.16 ± 0.03 µg L −1 d −1 (data not shown). There was no significant effect of kairomones on the total pool of dissolved STXs ( Figure 3). Predator cues resulted in an STX release rate of 0.20 ± 0.04 µg L −1 d −1 , while under control conditions, the release rate was 0.16 ± 0.03 µg L −1 d −1 (data not shown). Concentrations of (A) intracellular and (B) extracellular saxitoxin in Raphidiopsis raciborskii T3 under control conditions (black bars) and exposed to Daphnia gessneri infochemicals (gray bars). No significant differences were detected between the two conditions. Overall, the STX biovolume content, relative biovolume toxicity and volumetric STX concentration increased over time (two-way RM ANOVA, p < 0.01) (Figures 1-3). The latter also reflected the increase in cell concentration in both experimental conditions. First-order rate kinetics showed different patterns in toxin production/growth in R. raciborskii T3. Under control conditions, STX production was coupled to growth (1:1 µstx/µg ratio) (Figure 4). In contrast, predator cues caused the mean value of the µstx/µg ratio to be significantly greater than 1 (Student's t-test; T = 4.055, p < 0.05), indicating that the specific rate of STX production was faster than the specific growth rate, resulting in increased cell toxin accumulation ( Figure 1).

Figure 3.
Concentrations of (A) intracellular and (B) extracellular saxitoxin in Raphidiopsis raciborskii T3 under control conditions (black bars) and exposed to Daphnia gessneri infochemicals (gray bars). No significant differences were detected between the two conditions. Overall, the STX biovolume content, relative biovolume toxicity and volumetric STX concentration increased over time (two-way RM ANOVA, p < 0.01) (Figures 1-3). The latter also reflected the increase in cell concentration in both experimental conditions. First-order rate kinetics showed different patterns in toxin production/growth in R. raciborskii T3. Under control conditions, STX production was coupled to growth (1:1 µ stx /µ g ratio) ( Figure 4). In contrast, predator cues caused the mean value of the µ stx /µ g ratio to be significantly greater than 1 (Student's t-test; T = 4.055, p < 0.05), indicating that the specific rate of STX production was faster than the specific growth rate, resulting in increased cell toxin accumulation ( Figure 1).

STX Gene Expression
Significant up-regulation of the expression of two STX-related genes, sxtI (two-way RM ANOVA, F(1,4) = 8.588; p < 0.05) and sxtU (two-way RM ANOVA, F(1,4) = 153.2; p < 0.001), was detected in R. raciborskii T3 grown in the presence of infochemicals compared with the control ( Figure 5; Supplementary Table S3). In the presence of Daphnia infochemicals, sxtI expression was significantly higher than the control at the 6th day of incubation (Bonferroni's test, p < 0.0001) ( Figure 5A), which was also when a significant increase in STX production was observed (see Figure 1). Transcript levels increased over time (F(3,12) = 8.172; p < 0.01), and a significant interaction between both factors (time × treatment, F(3,12) = 10.13; p < 0.01) was observed. Transcript abundances of sxtU were higher in the treatment than in the control on the 2nd and 4th days (F(1,4) = 153.2; p < 0.001) (Bonferroni's test, p < 0.0001) ( Figure 5B), four days before an increase in STX content was detected. There was also a significant interaction in the transcript levels of this gene between both factors (time × treatment, F(3,12) = 85.65; p < 0.0001). The abundance of sxtU transcripts decreased gradually over time, and on the 6th day, it was similar under both treatment and control conditions (F(3,12) = 137.7; p < 0.0001).  . Ratio between specific total saxitoxin production rate (µ stx ) and specific growth rate (µ g ) obtained from first-order rate kinetics (log phase) for Raphidiopsis raciborskii T3 grown under control conditions (CTRL) and with Daphnia gessneri infochemicals (INFO). The dashed line indicates a 1:1 relationship between µ stx and µ g . (*) = t-test, p < 0.05.

STX Gene Expression
Significant up-regulation of the expression of two STX-related genes, sxtI (two-way RM ANOVA, F (1,4) = 8.588; p < 0.05) and sxtU (two-way RM ANOVA, F (1,4) = 153.2; p < 0.001), was detected in R. raciborskii T3 grown in the presence of infochemicals compared with the control ( Figure 5; Supplementary Table S3). In the presence of Daphnia infochemicals, sxtI expression was significantly higher than the control at the 6th day of incubation (Bonferroni's test, p < 0.0001) ( Figure 5A), which was also when a significant increase in STX production was observed (see Figure 1). Transcript levels increased over time (F (3,12) = 8.172; p < 0.01), and a significant interaction between both factors (time × treatment, F (3,12) = 10.13; p < 0.01) was observed. Transcript abundances of sxtU were higher in the treatment than in the control on the 2nd and 4th days (F (1,4) = 153.2; p < 0.001) (Bonferroni's test, p < 0.0001) ( Figure 5B), four days before an increase in STX content was detected. There was also a significant interaction in the transcript levels of this gene between both factors (time × treatment, F (3,12) = 85.65; p < 0.0001). The abundance of sxtU transcripts decreased gradually over time, and on the 6th day, it was similar under both treatment and control conditions (F (3,12) = 137.7; p < 0.0001).

STX Gene Expression
Significant up-regulation of the expression of two STX-related genes, sxtI (two-way RM ANOVA, F(1,4) = 8.588; p < 0.05) and sxtU (two-way RM ANOVA, F(1,4) = 153.2; p < 0.001), was detected in R. raciborskii T3 grown in the presence of infochemicals compared with the control ( Figure 5; Supplementary Table S3). In the presence of Daphnia infochemicals, sxtI expression was significantly higher than the control at the 6th day of incubation (Bonferroni's test, p < 0.0001) ( Figure 5A), which was also when a significant increase in STX production was observed (see Figure 1). Transcript levels increased over time (F(3,12) = 8.172; p < 0.01), and a significant interaction between both factors (time × treatment, F(3,12) = 10.13; p < 0.01) was observed. Transcript abundances of sxtU were higher in the treatment than in the control on the 2nd and 4th days (F(1,4) = 153.2; p < 0.001) (Bonferroni's test, p < 0.0001) ( Figure 5B), four days before an increase in STX content was detected. There was also a significant interaction in the transcript levels of this gene between both factors (time × treatment, F(3,12) = 85.65; p < 0.0001). The abundance of sxtU transcripts decreased gradually over time, and on the 6th day, it was similar under both treatment and control conditions (F(3,12) = 137.7; p < 0.0001).

Growth, Morphology and Photosynthetic Parameters
R. raciborskii T3 displayed a significant decrease in biovolume (two-way RM ANOVA, p < 0.001) and specific growth rate when exposed to Daphnia-conditioned medium compared with the control (µcontrol = 0.47 ± 0.15 day −1 ; µinfochemicals = 0.24 ± 0.04 day −1 ; Student's t-test, p < 0.05) ( Figure 6A; Supplementary Table S4). No differences were observed in chlorophyll-a concentrations ( Figure 6B). Additionally, growth in the presence of predator infochemicals did not affect photosynthesis, as estimated by the light-harvesting efficiency (α), relative PSII quantum yield (Fv'/Fm'), light saturation parameter (Ik) and maximum electron transport rate (ETRmax) ( Table 1). However, a significant increase in the photosynthetic parameters ETRmax (two-way RM ANOVA, p < 0.05) and Ik (two-way RM ANOVA, p < 0.001) over time was observed (Table 1). Table 1. Photosynthetic parameters of Raphidiopsis raciborskii T3 exposed to Daphnia gessneri infochemicals (INFO) and under control conditions (CTRL). Yield-relative PSII quantum yield-at a saturation pulse of 36 PAR (photosynthetically active radiation), ETRmax-maximum electron transport rate, Ik-light saturation parameter, and alpha (α)-lightharvesting efficiency. SD = standard deviation. Values with the same letter are not significantly different (Bonferroni's test, p < 0.05). The irradiation curves of R. raciborskii T3 (maximum PSII quantum yield) indicated no effect of predator infochemicals (Figure 7). Under both control and treatment conditions, the cells were sensitive to light intensities above 64 µmol photons m −2 s −1 , and R. raciborskii T3 displayed a significant decrease in photosynthetic efficiency when exposed to increased light pulses (two-way RM ANOVA, p < 0.0001) (Figure 7). Additionally, growth in the presence of predator infochemicals did not affect photosynthesis, as estimated by the light-harvesting efficiency (α), relative PSII quantum yield (Fv'/Fm'), light saturation parameter (Ik) and maximum electron transport rate (ETR max ) ( Table 1). However, a significant increase in the photosynthetic parameters ETR max (twoway RM ANOVA, p < 0.05) and Ik (two-way RM ANOVA, p < 0.001) over time was observed (Table 1). Table 1. Photosynthetic parameters of Raphidiopsis raciborskii T3 exposed to Daphnia gessneri infochemicals (INFO) and under control conditions (CTRL). Yield-relative PSII quantum yield-at a saturation pulse of 36 PAR (photosynthetically active radiation), ETR max -maximum electron transport rate, Ik-light saturation parameter, and alpha (α)-light-harvesting efficiency. SD = standard deviation. Values with the same letter are not significantly different (Bonferroni's test, p < 0.05). The irradiation curves of R. raciborskii T3 (maximum PSII quantum yield) indicated no effect of predator infochemicals (Figure 7). Under both control and treatment conditions, the cells were sensitive to light intensities above 64 µmol photons m −2 s −1 , and R. raciborskii T3 displayed a significant decrease in photosynthetic efficiency when exposed to increased light pulses (two-way RM ANOVA, p < 0.0001) (Figure 7).  Figure 7. Maximum PSII quantum yield derived from the photosynthesis irradiation curve of Raphidiopsis raciborskii T3 control (A) and exposed (B) to Daphnia gessneri infochemicals for 6 days. Significant differences (* p < 0.05; ** p < 0.01; **** p < 0.0001) after Bonferroni's test.
The morphology of R. raciborskii T3 did not change (Dunnett's test, p > 0.05) when exposed to infochemicals. Trichomes ranged from an initial length of 74

Discussion
Infochemicals released by D. gessneri significantly reduced growth but enhanced sxt gene expression and toxin production in R. raciborskii T3, evidencing a predator-induced chemical defense. We observed these effects using a density of cladocerans near to those naturally occurring, and we detected an approximately two-fold increase in sxt gene transcript levels and cellular STX quotas of the cyanobacterium in response to zooplankton alarm cues.
Several studies have characterized the physiological responses underlying predatorinduced chemical defenses in microcystin-producing cyanobacteria [5,[51][52][53], including gene expression [45,54]. In contrast, an increase in STX production mediated by predator cues has only been reported in eukaryotic marine dinoflagellates. For example, Selander et al. [7] found that there was a more than two-fold increase in STX production by Alexandrium minutum when exposed to kairomones of the copepod Acartia tonsa. Similar findings were obtained for STX-producing A. tamarense and A. fundyense exposed to different predators [55][56][57]. Despite the effect that the C:N:P ratio has on STX production as suggested by the stoichiometric hypothesis [58], the cellular content of STX can increase in response to predator cues, regardless of nutrient conditions [59].
R. raciborskii T3 has sxtM/sxtF genes that encode multidrug and toxic compound extrusion (MATE) transport proteins [60]. However, no effect on STX secretion by this cyanobacterium was observed, despite the increased biovolume toxin content it displayed following exposure to Daphnia infochemicals. In fact, increasing intracellular toxin content is an effective cyanobacterial defense against grazing pressure by zooplankton as opposed to simply releasing these metabolites into the surroundings. Feeding is thought to be the most important route of exposure to toxins in aquatic systems [61]. This assumption is based on the harmful acute effects and life-history impairments of zooplankton, observed when they actively feed on STX-producing cyanobacteria [16,17,19,62,63].
According to Yang et al. [64], the kairomone-induced increase in STX production is mediated by chemoreception and does not involve mechanical damage. This defensive response does not depend on active grazing and must be assessed by exposing harmful Figure 7. Maximum PSII quantum yield derived from the photosynthesis irradiation curve of Raphidiopsis raciborskii T3 control (A) and exposed (B) to Daphnia gessneri infochemicals for 6 days. Significant differences (* p < 0.05; ** p < 0.01; **** p < 0.0001) after Bonferroni's test.
The morphology of R. raciborskii T3 did not change (Dunnett's test, p > 0.05) when exposed to infochemicals. Trichomes ranged from an initial length of 74.91 ± 32.54 µm to 75.05 ± 40.44 µm at day 6 in the presence of Daphnia infochemicals, and from 74.91 ± 32.54 µm to 79.69 ± 42.79 µm under control conditions. Similarly, no changes were detected in trichome thickness, with mean widths of 2.63 ± 0.52 µm and 2.53 ± 0.43 µm under control and infochemical conditions, respectively.

Discussion
Infochemicals released by D. gessneri significantly reduced growth but enhanced sxt gene expression and toxin production in R. raciborskii T3, evidencing a predator-induced chemical defense. We observed these effects using a density of cladocerans near to those naturally occurring, and we detected an approximately two-fold increase in sxt gene transcript levels and cellular STX quotas of the cyanobacterium in response to zooplankton alarm cues.
Several studies have characterized the physiological responses underlying predatorinduced chemical defenses in microcystin-producing cyanobacteria [5,[51][52][53], including gene expression [45,54]. In contrast, an increase in STX production mediated by predator cues has only been reported in eukaryotic marine dinoflagellates. For example, Selander et al. [7] found that there was a more than two-fold increase in STX production by Alexandrium minutum when exposed to kairomones of the copepod Acartia tonsa. Similar findings were obtained for STX-producing A. tamarense and A. fundyense exposed to different predators [55][56][57]. Despite the effect that the C:N:P ratio has on STX production as suggested by the stoichiometric hypothesis [58], the cellular content of STX can increase in response to predator cues, regardless of nutrient conditions [59].
R. raciborskii T3 has sxtM/sxtF genes that encode multidrug and toxic compound extrusion (MATE) transport proteins [60]. However, no effect on STX secretion by this cyanobacterium was observed, despite the increased biovolume toxin content it displayed following exposure to Daphnia infochemicals. In fact, increasing intracellular toxin content is an effective cyanobacterial defense against grazing pressure by zooplankton as opposed to simply releasing these metabolites into the surroundings. Feeding is thought to be the most important route of exposure to toxins in aquatic systems [61]. This assumption is based on the harmful acute effects and life-history impairments of zooplankton, observed when they actively feed on STX-producing cyanobacteria [16,17,19,62,63].
According to Yang et al. [64], the kairomone-induced increase in STX production is mediated by chemoreception and does not involve mechanical damage. This defensive response does not depend on active grazing and must be assessed by exposing harmful algae or cyanobacteria to predator kairomones. Experiments involving direct exposure would lead to selective grazing on less toxic cells, which might favor cells with a higher mean toxin content; these cells may be indistinguishable from those in which toxin production had been induced [7].
In our study, two representative genes of the sxt cluster, sxtI and sxtU, showed increased transcript levels in R. raciborskii cells exposed to D. gessneri infochemicals. This response preceded the observed increase in STX cellular quotas. sxtU encodes a dehydrogenase that reduces the terminal aldehyde group of the STX precursor, whereas sxtI encodes a carbamoyl transferase that catalyzes carbamoyl transfer from carbamoyl phosphate onto the free hydroxyl at C-13, forming STX [60]. These genes are reliable markers for assessing STX synthesis because they encode enzymes that participate in key steps of STX formation. Other studies have successfully used these sxt molecular markers as reporters to assess STX biosynthesis in R. raciborskii [65,66]. Yang et al. [64] were the first to demonstrate an increased cellular toxin content and associated transcript profile of a STX-producing dinoflagellate exposed to zooplankton cues, although no genes of the sxt cluster were identified. Later, Wohlrab et al. [67] assessed sxtA gene expression coupled with STX production in A. fundyense exposed to predators. Despite a two-fold increase in the STX cell quota, no changes were observed at the transcript level. The authors suggested that sxtA might have been overexpressed early at the onset of the dinoflagellate defense response prior to the observed physiological changes. This assumption is confirmed in our findings, where we demonstrated that an increase in sxt transcript levels preceded the increase in the STX cellular quota during the cyanobacterial response to predator cues.
Molecular events involved in induced chemical defenses in cyanobacteria are still poorly explored, especially in STX-producing species. Although the presence of genes related to toxin synthesis precedes the existence of metazoans [68], these toxins might have evolved to diversify their functions and represent an adaptive response against predators [23]. This is exemplified by the predator-mediated induction of STX production coupled to the increased resistance to grazers in A. minutum [7]. Thus, it is likely that the potential adverse effects of STXs on zooplankton have undergone positive selection more recently, as has been suggested for MCs by Rzymski et al. [69].
In addition to the increased toxin production induced in response to predators, the growth of R. raciborskii T3 decreased, and biovolume concentrations were low under these conditions. Overall, the imbalance in growth and STX production observed after exposure to zooplankton infochemicals indicates the presence of a trade-off. According to photosynthetic parameters, R. raciborskii cells were in a normal physiological state, but growth was still reduced. It is likely that the potential threat indicated by the infochemicals caused the cells to allocate energy toward toxin production instead of cell division. Indeed, studies examining the effects of environmental stressors on R. raciborskii physiology have suggested that there is a trade-off between growth and toxin production [33]. Different stress conditions can activate similar physiological responses as cyanotoxin production. Studies with both STX-and CYN-producing R. raciborskii strains have demonstrated this imbalance between growth and toxin production in response to abiotic stress (e.g., nutrient limitation, light and temperature stress) [30,[70][71][72]. However, our study is the first to report the response of the cyanobacterium R. raciborskii to biotic stress caused by predation. Additionally, costs are not only related to the production of well-described toxins, but also to other unknown metabolites. For example, Blossom et al. [73] studied the costs of toxicity in STX-producing Alexandrium and did not find any growth reduction related to toxin production; instead, growth reduction was related to lytic activity, which was associated with other poorly known metabolites.
Additionally, a trade-off response was also confirmed through the first-order kinetics rate. The calculated ratio between the rates of toxin production and cell division indicated that induced chemical defense in R. raciborskii T3 led to a greater increase in the STX production rate than in growth. The production of other cyanotoxins, such as microcystins, is considered a constitutive process that is directly coupled to cell division [74,75]. Cells can achieve an up to three-fold variation in the toxin quota depending on their growth stage [76]. However, the cellular STX content depends not only on the cell division rate but also on environmental factors [33].
Generally, photosynthesis is thought to be disrupted when the cellular growth of phytoplankton is reduced, as the photosynthetic machinery is key for biomass acquisition. In our study, although infochemicals have negatively affected R. raciborskii T3 growth, no photosynthetic impairments were observed. However, a significant increase in the ETR max and Ik parameters over incubation time occurred, indicating enhanced light acquisition by R. raciborskii, which suggests that self-shading was caused by the accumulation of cells.
We examined photosynthetic parameters to assess early changes in cyanobacterial physiology, as little is known about how predator-induced defenses in phytoplankton might affect photosynthesis. Savic et al. [77] evaluated the photosynthetic response of toxic and nontoxic Microcystis strains exposed to D. magna kairomones. Decreases in photosynthetic activity and chlorophyll-a content in the toxic strain were observed in response to Daphnia. Savic et al. [78] reported that zooplankton cues did not affect M. aeruginosa photosynthetic activity under a similar experimental set-up, which was probably explained by the fact that only ETR max was examined. In our study, the finding that zooplankton infochemicals had no effect on photosynthesis was based on a more comprehensive set of photosynthetic parameters (i.e., Ik, ETR max , α and φ m ), in addition to the commonly used chlorophyll-a concentration, ETR and PSII quantum yield.
We also characterized trichome length and thickness, which are proxies of induced morphological defense; however, no significant changes were observed. Several studies have examined the effects of predator kairomones on cyanobacterial filament morphology, and most studies have reported an increase in width [11,79]; however, less is known about the effect on trichome length [11,80]. Filamentous morphology acts as a functional trait that prevents ingestion by clogging the predator filtration apparatus, and some microalgae can adjust their length in response to predator cues to minimize losses associated with zooplankton grazing pressure [81][82][83][84].
Defenses induced under variable grazing threat/attack are considered more energetically efficient than constitutive ones because they enable organisms to maximize benefits and reduce the costs associated with the investment in the constant expression of defensive strategies [85,86]. Numerous studies have attempted to estimate the costs and benefits of defense mechanisms in phytoplankton; Pančić and Kiørboe [87] argue that the costs associated with induced defenses have largely been undetermined. They also assume that if there is no cost to predator-resistance trait expression, all species would evolve toward a state of equal defense, and the community would not be structured by predation. This claim is consistent with the ODH, which is based on the fundamental assumption that the benefits of defense outweigh the costs, especially under high predation pressure [49]. Although the ODH was originally proposed for plant defenses, this hypothesis also predicts defensive responses in unicellular organisms, including phytoplankton, where phenotypically plastic organisms can express defensive traits upon exposure to a threat. However, intraspecific variability in these responses indicates that the associated costs are genotype-specific [53]. Thus, the adaptive evolution of defensive traits can be mediated by both Mendelian (e.g., genotype-specific responses) and non-Mendelian (e.g., phenotypic plasticity) evolving mechanisms.
In general, the predominance of R. raciborskii in water bodies has been attributed to its phenotypic plasticity and its dispersal capacity [26,27]. Assessing the ability of this species to cope with diverse ecological interaction filters (e.g., competition, predation and parasitism) [88] could provide insight into the complex factors that contribute to its adaptive success. Our results point out the potential of a STX-producing strain of the invasive (opportunistic) cyanobacterial species Raphidiopsis raciborskii in responding to grazer cues with an increase in toxin production at different levels. As cyanotoxins are also regarded as anti-grazing chemicals, describing an induced chemical defense and other underlying responses in this species adds more information into the knowledge of adaptive traits (e.g., grazing resistance) which allow its wide occurrence and dominance in several lakes. Additionally, regarding the ecological relevance of our experimental model, R. raciborskii has been reported co-occurring with D. gessneri, in addition to other Daphnia species, in several water bodies [89][90][91].
In conclusion, D. gessneri infochemicals induced the up-regulation of sxt genes and led to a subsequent increase in toxin content. On the other hand, R. raciborskii reduced its growth. These results support the hypothesis that the anti-predator defenses of R. raciborskii may be costly.
Zooplankton consisted of the neotropical species Daphnia gessneri isolated from Barra do Braúna Reservoir (Minas Gerais, Brazil) in 2018. Individuals were kept in RT medium [94] enriched with commercial 0.1% (~2.25 mg C L −1 ) humic extract (Microbe-lift ® Amazon Black & Soft Water Conditioner, USA) at an initial pH 7.6, 24 ± 1 • C, 50 µmol photons m −2 s −1 and a 12/12-hour dark-light cycle. Animals were fed a cell suspension of the green algae Selenastrum capricornutum at a final concentration of 500 µg C L −1 once every two days. The algae population was dominated by unicells (mean length and width dimensions: 13.80 ± 1.56 × 3.18 ± 0.66 µm).

Zooplankton Filtrate to Obtain Infochemicals
Prior to experiments, D. gessneri clones were starved in fresh RT medium to empty their guts and remove any superficial contamination. Cultures were established with adults at a populational density of 60 ind L −1 . The animals were fed with a cell suspension of the green algae S. capricornutum at 500 µg C L −1 and incubated for 96 h. Thereafter, animals were removed using a cup of plankton (60-µm mesh size), and the zooplankton medium was sterilized by filtration through a 0.22-µm pore membrane. The filtrate (zooplankton pre-conditioned medium) was used to prepare the ASM-1 culture medium, hereafter referred to as infochemicals. As a control, RT medium with the chlorophycean as food suspension at a similar concentration provided for the treatment with Daphnia wasalso incubated for 96 h, and the medium was then filtered through a 0.22-µm pore membrane. The filtrate was also used instead of water to prepare ASM-1 and transferred together with the infochemical-rich medium to the predator-induced defense experiment.

Predator-Induced Defenses Experiment
Zooplankton-conditioned medium was used to evaluate whether R. raciborskii cells respond to predator infochemicals. Biomass inocula of R. raciborskii T3 at 20 mm 3 L −1 (10 5 cells mL −1 ) were prepared in 1-L Erlenmeyer flasks filled with 600 mL of infochemicalrich or control culture medium (n = 3). Batch cultures were incubated over 6 days at 24 ± 1 • C and 50 µmol photons m −2 s −1 under a 12/12-h dark-light cycle. Samples were collected every two days to evaluate growth, morphological changes, photosynthetic activity, toxin content and the expression of sxt genes.

Growth, Morphology and Photosynthetic Measurements
Samples were harvested and preserved with 1% acetic Lugol's solution for trichome count in a Fuchs-Rosenthal chamber under an optical microscope (Olympus BX51). The results are expressed in trichomes mL −1 . Trichome density was converted to biovolume (mm 3 L −1 ) based on the mean trichome volume (µm 3 ) [95,96].
The specific growth rate was calculated during the exponential growth phase using Equation (1): where N 2 = final biovolume; N 1 = initial biovolume; t 2 = final time; and t 1 = initial time.
Additionally, trichome length and width (µm) at t initial and t final in both culture conditions were measured to examine morphological variation. Samples were analyzed under an Olympus BX51 light microscope (250× and 400× magnification) coupled to an image-capture system (Canon t6i). Trichomes were measured using the CellˆB software package for image acquisition.
Photosynthetic activity was evaluated using a PHYTO-PAM fluorometer (Heinz Walz GmbH, Germany) equipped with a PHYTO-EDF detection unit for measuring cyanobacterial fluorescence. Saturation pulses (36 µmol photons m −2 /s −1 ) were applied, and fluorescence data were converted to chlorophyll-a concentration (µg L −1 ) and photosynthetic yield (relative Fv'/Fm'). Light curves were performed with dark-acclimated culture samples (F 0 ) with 10-s intervals between pulses. Light intensity varied from 16 to 764 µmol photons m −2 /s −1 . From these curves, the values of the maximum electron transport rate (ETR max ), light-harvesting efficiency (α) and light saturation parameter (Ik) were obtained as a function of irradiance. Additionally, the maximum effective quantum efficiency of PSII was calculated as φ m = [(Fm − F)/Fm], where F is the fluorescence of the dark-adapted sample and Fm is the maximum dark-adapted fluorescence.

STX Analyses and Toxicity Equivalency Calculation
Aliquots were taken from R. raciborskii cultures, and cells were harvested by centrifugation (7200× g, 10 min, 4 • C). The supernatant was filtered (0.45-µm pore membrane) for toxin measurements in the dissolved fraction, and the pellet was used to measure the intracellular toxin content. Both samples were frozen at -20 • C, freeze-dried and then subjected to STX extraction. For cellular toxin extraction, 0.5 M acetic acid solution was added to the dried cells in 3 cycles of 1 h each. At each cycle, the cell extract was centrifuged as previously mentioned, and the supernatant was reserved. For dissolved STX recovery, filtered samples were freeze-dried and then resuspended in a 0.5 M acetic acid solution. Both dissolved and cellular fractions were filtered in a 0.22-µm syringe filter and stored in 1.5-mL vials prior to chromatographic analysis.
The analysis was performed on a Shimadzu Class VP liquid chromatography system with a fluorescence detector (RF-10A XL). A reversed-phase C18 column (Lichrospher ® ; 150 mm × 4.6 mm; 5 µm-Merck) and a 20-µL loop injector were used. Chromatographic analyses were performed according to the post-derivatization method [97] using a mobile phase of 2 mM sodium 1-heptane sulfonate in 30 mM ammonium phosphate and 5% acetonitrile at a flow rate of 0.8 mL min −1 . STXs were detected at an excitation wavelength of 330 nm and an emission wavelength of 400 nm.
For quantification of toxins, standard solutions of STX and neoSTX from the National Research Council (NRC)-Institute of Marine Biosciences (Canada) were used. Quantified STXs were expressed as volumetric intra-and extracellular concentrations (µg L −1 ), and cellular toxin amounts as STXs quota per biovolume (µg mm −3 ).
Total relative cellular toxicity (saxitoxin equivalents, STX eq ) was expressed as biovolume toxicity µg STX eq mm −3 ) according to the TEF described for each STX analog (in this study, neoSTX and STX) in FAO/WHO [98] and using Equation (2): where C i is the concentration of the individual toxin analog and its assigned TEF i .

Application of First-Order Rate Kinetics to Assess Total STX Production
The specific growth rate (µ g ) and the specific total STX production rate (µ stx ) were calculated during the exponential growth phase according to simple first-order rate kinetics using either biovolume concentration (mm 3 L −1 ) or volumetric total intracellular STX data (µg L −1 ), respectively. Both specific rates are reported in units per day (d -l ).
The ratio between µ stx and µ g was calculated to assess different patterns of toxin production coupled to the growth cycle and subsequent changes in the toxin cell (or biovolume) quota (Q tox ) as described in Orr et al. [74,75], such that: Equation (3) describes the condition where the intracellular µ stx between t 0 and t n is slower than µ g , resulting in a lower Q tox .
Equation (4) describes the condition where the intracellular µ stx between t 0 and t n is a function of µ g , resulting in a constant Q tox (1:1 growth-toxin relationship).
Equation (5) describes the condition where the intracellular µ stx between t 0 and t n is higher than µ g , resulting in an increased Q tox .

Expression of Genes Involved in STX Biosynthesis
Transcript levels of the sxtI and sxtU genes were measured. Although most steps in the biosynthesis pathway of STX are putative, sequence similarity analysis indicates that the sxtU gene encodes a dehydrogenase that reduces the terminal aldehyde group of the STX precursor [60], and the sxtI gene encodes a carbamoyltransferase that catalyzes carbamoyl transfer from carbamoyl phosphate onto the free hydroxyl at C-13, forming STX. We chose these genes based on previous studies that reported a coherent response pattern of sxtI and sxtU and the other genes of the sxt cluster, and a coherent response of sxtI and sxtU expression and STX levels [65,66]. The reference gene used to calculate the relative expression of the target genes was rpoC1, which encodes the RNA polymerase gamma subunit [99].
Samples of 50 mL were taken from R. raciborskii cultures, and cells were harvested by centrifugation (1500× g, 15 min, 4 • C). TRIzol ® (300 µL) was added to the pellets, and the suspensions were immediately stored at -80 • C. RNA extraction was performed using the Direct-zolTM RNA Miniprep Plus (Zymo Research ® ) extraction kit per the manufacturer's instructions (including a DNAse digestion step). RNA samples were suspended in RNAsefree water, and the nucleic acid quality and concentration were determined using the RNA HS Assay kit (Life Technologies) in a Qubit fluorometer (Thermo Fisher Scientific). To check for DNA contamination, PCR was carried out using 16S rDNA primers and purified RNA samples as templates. cDNA synthesis was performed with the GoScript Reverse Transcriptase enzyme (Promega) per the manufacturer's instructions. The cDNA products were quantified in a Nanodrop, diluted 1:10 and tested for amplification with specific primers for rpoC1 [99].
The obtained cDNA was used in qPCR to quantify the relative abundance of sxtU and sxtI transcripts. For each sample, triplicate reactions were run on a QuantStudio 3 Real-Time PCR system (Thermo Fisher Scientific), with a preincubation at 95 • C for 10 min and 40 cycles of amplification at 95 • C for 15 s and 60 • C for 1 min. A melting curve was obtained by incubation at 95 • C for 15 s, 60 • C for 1 min and 95 • C for 15 s. Negative controls for each primer set were included in which nuclease-free water was used instead of cDNA template. The amplification efficiency of each primer set was calculated from standard curves using serial dilutions of cDNA, according to the equation E = 10 (−1/slope) (Supplementary Figure S1 and Table S5). Transcript levels of the target sxt genes were normalized to the reference rpoC1 gene, and the relative change in transcript levels was calculated using the ∆∆CT method of relative quantification [101].

Statistical Analysis
All data were checked for normality and homoscedasticity of variances. Specific growth rate data and the ratio between specific toxin production and growth rate were analyzed by Student's T-test to test for significant effects of zooplankton cues. Variation in the growth, total STXs biovolume quota, biovolume relative toxicity, total STXs pool size and sxt gene expression over incubation time were verified using repeated measures of two-way ANOVA with a post-hoc Bonferroni's test. Morphological trait variation was analyzed using one-way ANOVA with a post-hoc Dunnett's test using the morphological trait at the initial time as a control condition. The ∆∆CT data were log-transformed, and two-way RM ANOVA was used to compare the infochemical-rich samples with the control samples on each day. All analyses were performed, and graphs were produced using GraphPad Prism 7.0 software.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/10.3 390/toxins13060406/s1, Table S1: Results of the two-way ANOVA for differences in total saxitoxins biovolume quota of Raphidiopsis raciborskii T3 exposed (infochemicals) and not exposed (control) to Daphnia gessneri over 6 days of incubation. Table S2: Results of the post hoc Bonferroni's comparison test for differences in biovolume toxicity relative to STX analogues produced by Raphidiopsis raciborskii T3 exposed (infochemicals) and not exposed (control) to Daphnia gessneri over 6 days of incubation. Table S3: Results of the two-way ANOVA for differences in the relative expression of sxtI and sxtU by Raphidiopsis raciborskii T3 exposed (infochemicals) and not exposed (control) to Daphnia gessneri over 6 days of incubation. Table S4: Results of the two-way ANOVA for differences in the biovolume growth curves and chlorophyll-a of Raphidiopsis raciborskii T3 exposed (infochemicals) and not exposed (control) to Daphnia gessneri over 6 days of incubation. Table S5: Efficiencies and standard curve parameters obtained by real-time qPCR analysis for the cyanobacterial rpoC1-, sxtIand sxtU-specific primer sets. Figure S1: Standard curve and its linear equation for cyanobacterial rpoC1-, sxtIand sxtU-specific primers.