Low Temperature and Cold Stress Significantly Increase Saxitoxins (STXs) and Expression of STX Biosynthesis Genes sxtA4 and sxtG in the Dinoflagellate Alexandrium catenella

Toxic dinoflagellate Alexandrium spp. produce saxitoxins (STXs), whose biosynthesis pathway is affected by temperature. However, the link between the regulation of the relevant genes and STXs’ accumulation and temperature is insufficiently understood. In the present study, we evaluated the effects of temperature on cellular STXs and the expression of two core STX biosynthesis genes (sxtA4 and sxtG) in the toxic dinoflagellate Alexandrium catenella Alex03 isolated from Korean waters. We analyzed the growth rate, toxin profiles, and gene responses in cells exposed to different temperatures, including long-term adaptation (12, 16, and 20 °C) and cold and heat stresses. Temperature significantly affected the growth of A. catenella, with optimal growth (0.49 division/day) at 16 °C and the largest cell size (30.5 µm) at 12 °C. High concentration of STXs eq were detected in cells cultured at 16 °C (86.3 fmol/cell) and exposed to cold stress at 20→12 °C (96.6 fmol/cell) compared to those at 20 °C and exposed to heat stress. Quantitative real-time PCR (qRT-PCR) revealed significant gene expression changes of sxtA4 in cells cultured at 16 °C (1.8-fold) and cold shock at 20→16 °C (9.9-fold). In addition, sxtG was significantly induced in cells exposed to cold shocks (20→16 °C; 19.5-fold) and heat stress (12→20 °C; 25.6-fold). Principal component analysis (PCA) revealed that low temperature (12 and 16 °C) and cold stress were positively related with STXs’ production and gene expression levels. These results suggest that temperature may affect the toxicity and regulation of STX biosynthesis genes in dinoflagellates.


Introduction
Dinoflagellates are unicellular micro-eukaryotes that mostly inhabit marine water, with some observed in freshwater. They have evolved to adapt to different environments with a variety of morphological diversity and trophic modes [1]. Approximately half of the dinoflagellates are photosynthetic; thus, as a primary producer, they play an important role in aquatic ecosystems [2]. Some dinoflagellates, however, are responsible for harmful algal blooms (HABs), causing considerable damages to marine environments and aquaculture, and even human health [3][4][5].
In addition, certain dinoflagellates (e.g., Alexandrium catenella, Gymnodinium catenatum, Karenia brevis, and Prorocentrum minimum) can produce biotoxins, such as neurotoxins and hepatotoxins, and these compounds can be accumulated in shellfish via filter feeding [6][7][8]. Thus, toxic dinoflagellates can cause shellfish poisoning when humans consume contaminated shellfish. Approximately 2000 cases of shellfish poisoning are reported annually around the world, resulting in serious illness or even death [9]. There are four typical are considered to play the most important role in STX biosynthesis [48]; however, the relationship between STX toxicity and response of the core genes to environmental factors is insufficiently understood.
In the present study, we evaluated the effects of temperature on cellular STXs' accumulation and expression of the two core STX biosynthesis genes, sxtA4 and sxtG, in the toxic dinoflagellate Alexandrium catenella (Alex03). Then, we compared the relationship between temperature and STX production through transcriptional regulation. The test species occurs predominantly in Korean coasts and is suspected to cause PSP in marine aquaculture thorough molecular phylogeny and laboratory culture researches [49,50].

Effects of Temperature on Cell Growth and Size
Alexandrium catenella Alex03 was cultured at 20 • C, and was successively adapted to different temperatures (16 • C and 12 • C). Cells cultured at lower temperatures were healthy and not much different compared to those cultured at 20 • C. Cell growth patterns in the three test temperatures showed sigmoidal cell growth at 16 • C but relatively slow growth at 12 • C and 20 • C ( Figure 1A). Comparisons at the exponential phase (day 18) revealed that cell number at 16 • C was much higher than that at 12 • C (2.9-times at 16 • C) and 20 • C (4.7-times at 16 • C). Hence, optimal cell growth was recorded at 16 • C, with the maximum cell density of 5.9 × 10 3 cells/mL, of which growth rate was calculated to be 0.49 division/day. [41,42]. Of these genes, sxtA and sxtG are involved in the initial steps of STX biosynthesis and have been investigated at transcriptomic and genomic levels [45,47]. To date, these genes are considered to play the most important role in STX biosynthesis [48]; however, the relationship between STX toxicity and response of the core genes to environmental factors is insufficiently understood.
In the present study, we evaluated the effects of temperature on cellular STXs' accumulation and expression of the two core STX biosynthesis genes, sxtA4 and sxtG, in the toxic dinoflagellate Alexandrium catenella (Alex03). Then, we compared the relationship between temperature and STX production through transcriptional regulation. The test species occurs predominantly in Korean coasts and is suspected to cause PSP in marine aquaculture thorough molecular phylogeny and laboratory culture researches [49,50].

Effects of Temperature on Cell Growth and Size
Alexandrium catenella Alex03 was cultured at 20 °C , and was successively adapted to different temperatures (16 °C and 12 °C ). Cells cultured at lower temperatures were healthy and not much different compared to those cultured at 20 °C . Cell growth patterns in the three test temperatures showed sigmoidal cell growth at 16 °C but relatively slow growth at 12 °C and 20 °C ( Figure 1A). Comparisons at the exponential phase (day 18) revealed that cell number at 16 °C was much higher than that at 12 °C (2.9-times at 16 °C ) and 20 °C (4.7-times at 16 °C ). Hence, optimal cell growth was recorded at 16 °C , with the maximum cell density of 5.9 × 10 3 cells/mL, of which growth rate was calculated to be 0.49 division/day.
The average cell size of three cultures was the largest at 12 °C (36.7 ± 0.5 µ m), followed by those at 16 °C (30.7 ± 0.3 µ m) and 20 °C (30.5 ± 0.5 µ m) ( Figure 1B). The widest distribution of cell size was recorded at 20 °C , ranging from 23.1 to 37.4 µ m.

Figure 1.
Growth curves (A) and box plots of cell size (B) of Alexandrium catenella Alex03 dependent on adapted temperature (12,16, and 20 °C ). The "*" were determined by the 1st and 99th percentiles. Morphology of A. catenella observed in bright-field microscopy (scale bar: 20 µm) (C). Cell growth was observed over a period of 28 days, and the gray background in growth curve represents the presence of an exponential phase. The size of the cells harvested on day 18, i.e., the exponential phase was analyzed.

Phylogenetic Relationships of sxtA4 and sxtG and Characterization
Partial open reading frame (ORF) sequences of sxtA4 and sxtG from A. catenella were determined to be 714 bp and 1112 bp, respectively. BLASTx search of the sxtA4 sequence of Alex03 (GenBank accession no. MW884259) showed 100% identity with that of the dinoflagellate A. fundyense (ADY62525), followed by 99.5% with A. tamarense (BCG06333) and 96.3% with A. australiense (AIY25738). The sxtG sequence (MW884258) of Alex03  (12,16, and 20 • C). The "*" were determined by the 1st and 99th percentiles. Morphology of A. catenella observed in brightfield microscopy (scale bar: 20 µm) (C). Cell growth was observed over a period of 28 days, and the gray background in growth curve represents the presence of an exponential phase. The size of the cells harvested on day 18, i.e., the exponential phase was analyzed.

Phylogenetic Relationships of sxtA4 and sxtG and Characterization
Partial open reading frame (ORF) sequences of sxtA4 and sxtG from A. catenella were determined to be 714 bp and 1112 bp, respectively. BLASTx search of the sxtA4 sequence of Alex03 (GenBank accession no. MW884259) showed 100% identity with that of the dinoflagellate A. fundyense (ADY62525), followed by 99.5% with A. tamarense (BCG06333) and 96.3% with A. australiense (AIY25738). The sxtG sequence (MW884258) of Alex03 matched 100% with the same gene of A. tamarense (AGC84356), 99.7% with A. fundyense (AGC84339.1), and 96.3% with A. insuetum (AGC84351). In silico analyses of functional domains predicted that a class II aminotransferase was present at 23 to 226 amino acid residue of sxtA4 protein, whereas no specific domain was found in sxtG of Alex03.
In addition, we investigated phylogenetic relationships of the genes sxtA4 and sxtG using broad taxon samplings from NCBI. ML trees of the deduced amino acid sequences showed that A. catenella Alex03 obviously formed one well-supported cluster, with the already known Alexandrium spp. (Figure 2). In particular, the phylogenetic tree of sxtA4 showed that the gene was clustered with A. fundyense and A. tamarense, and it formed a sister clade, which was separated into cyanobacteria and bacteria. In addition, sxtG protein also clustered into a single clade, which included other Alexandrium species (A. catenella, A. fundyense, A. insuetum) and G. catenatum. The sxtG clade was divided into two sub-clusters of fungi and the sister clades of cyanobacteria and metazoa. In addition, we investigated phylogenetic relationships of the genes sxtA4 and sxtG using broad taxon samplings from NCBI. ML trees of the deduced amino acid sequences showed that A. catenella Alex03 obviously formed one well-supported cluster, with the already known Alexandrium spp. (Figure 2). In particular, the phylogenetic tree of sxtA4 showed that the gene was clustered with A. fundyense and A. tamarense, and it formed a sister clade, which was separated into cyanobacteria and bacteria. In addition, sxtG protein also clustered into a single clade, which included other Alexandrium species (A. catenella, A. fundyense, A. insuetum) and G. catenatum. The sxtG clade was divided into two subclusters of fungi and the sister clades of cyanobacteria and metazoa.

Effect of Water Temperature on STXs
Total STXs eq for each sample was determined by summing 12 STX analogues calculated with toxicity equivalency factors (TEF). The results showed that temperature significantly changed the total STXs eq and their profiles in A. catenella ( Figure 3). STXs eq was the highest at 16 °C (86.38 fmol/cell), which was 3.6-and 1.2-times higher than that at 20
The STXs' profiles of A. catenella exposed to different temperatures and cold/heat stress were compared ( Figure 3D). Of the 12 STXs, GTX1, GTX3, and GTX4 were most dominantly detected in Alex03. Particularly, GTX1 accounts for the highest proportion of Figure 3. Comparison of saxitoxins equivalent (STXs eq) of Alexandrium catenella Alex03 in different temperatures; adaptation temperature (A), cold stress (B), and heat stress (C). The mean STXs eq (fmol/cell) for each STX analogues and total STXs eq were calculated and displayed to two decimal places. Undetected analogues were marked '-'. The proportion of each STX analogues (%) to the total STXs eq is expressed by a heatmap (D). Significant differences between the control and treated samples were determined by one-way ANOVA and highlighted as * p < 0.05, ** p < 0.01, and *** p < 0.001.

Correlation of Temperature, STXs eq and Sxt Genes Expression
PCA analysis showed a distinct partitioning of heat and cold stress samples and correlation with total STXs eq, sxtA4 and sxtG expression levels ( Figure 5). The ordination plot showed 99.19% of total variation in the data, and a linear positive relation was shown between STXs eq toxicity, sxtA4, and sxtG. PCA1 explained 92.43% of the variance, whereas PCA2 explained 6.76% of the variance. The samples cultured at 20 • C, 16→20 • C, and 12→20 • C (heat stress) were well-represented on the negative part of PCA1, with no strong relation between STXs eq, sxtA4, and sxtG. Conversely, cold stress samples (20→16 • C, 20→12 • C and 16→12 • C) and those cultured at 12 • C and 16 • C were clustered in the positive portion of PCA1. Strong affinity was observed between cold stress and STXs eq and expressional levels of the genes.

Correlation of Temperature, STXs eq and Sxt Genes Expression
PCA analysis showed a distinct partitioning of heat and cold stress samples and correlation with total STXs eq, sxtA4 and sxtG expression levels ( Figure 5). The ordination plot showed 99.19% of total variation in the data, and a linear positive relation was shown between STXs eq toxicity, sxtA4, and sxtG. PCA1 explained 92.43% of the variance, whereas PCA2 explained 6.76% of the variance. The samples cultured at 20 °C , 1620 °C , and 1220 °C (heat stress) were well-represented on the negative part of PCA1, with no strong relation between STXs eq, sxtA4, and sxtG. Conversely, cold stress samples (2016 °C , 2012 °C and 1612 °C ) and those cultured at 12 °C and 16 °C were clustered in the positive portion of PCA1. Strong affinity was observed between cold stress and STXs eq and expressional levels of the genes.

Figure 5.
Principal component analysis (PCA) of temperature parameters (adaptation temperature, cold stress, and heat stress) for saxitoxins equivalent (STXs eq) and relative expression levels of STX biosynthesis-related genes (sxtA4 and sxtG).

Discussion
A. catenella is known for producing STXs, but its toxicity varies depending on geographical origin and environmental conditions, even for cultured strains [4,[51][52]. A. catenella Alex03 was isolated in 2017 from Jeju Island coasts, Korea. Although it had been cultured for a long time and adapted well to laboratory conditions, we found that the morphology and STXs' toxicity may not or were lightly altered. In addition, comparisons of the 28S rRNA sequences showed that A. catenella Alex03 (GenBank accession no. MW882944) shared 100% similarity with A. catenella CCAP-1119/32 (MK566200; isolated from Scotland), H5 (MK566199; Argentina), and SCCAP K-1490 (MK566199; Canada). These results suggested that Alex03 was genetically similar or identical to A. catenella distributed in coastal waters around the world. The present research firstly evaluated STXs' production of A. catenella Alex03 isolated from Korean coast and analyzed the regulation of sxt genes under different temperature conditions. Molecular phylogeny using 28S rRNA showed that A. catenella Alex03 belonged to Group I (=A. fundyense) of A. tamarense complex [53][54], of which members are known as STX-producing Alexandrium [16,55]. Alex03 produced different levels of STXs in all culture conditions. In addition, its STXs eq toxicity (14-96 fmol/cell) was comparable to those

Discussion
A. catenella is known for producing STXs, but its toxicity varies depending on geographical origin and environmental conditions, even for cultured strains [4,51,52]. A. catenella Alex03 was isolated in 2017 from Jeju Island coasts, Korea. Although it had been cultured for a long time and adapted well to laboratory conditions, we found that the morphology and STXs' toxicity may not or were lightly altered. In addition, comparisons of the 28S rRNA sequences showed that A. catenella Alex03 (GenBank accession no. MW882944) shared 100% similarity with A. catenella CCAP-1119/32 (MK566200; isolated from Scotland), H5 (MK566199; Argentina), and SCCAP K-1490 (MK566199; Canada). These results suggested that Alex03 was genetically similar or identical to A. catenella distributed in coastal waters around the world. The present research firstly evaluated STXs' production of A. catenella Alex03 isolated from Korean coast and analyzed the regulation of sxt genes under different temperature conditions. Molecular phylogeny using 28S rRNA showed that A. catenella Alex03 belonged to Group I (=A. fundyense) of A. tamarense complex [53,54], of which members are known as STX-producing Alexandrium [16,55]. Alex03 produced different levels of STXs in all culture conditions. In addition, its STXs eq toxicity (14-96 fmol/cell) was comparable to those reported previously in A. catenella (Table 1). The test strain (Alex03) showed optimal growth at 16 • C, in agreement with previous reports [56][57][58]. For examples, A. catenella CCAP-1119/27, ATTL01 and ATTL02, and BAH91 contained 2732.5 fg/cell, 5-44 fg/cell, and 9.9 fmol/cell when maintained at 15 • C. In addition, A. catenella ACC02 maintained at low temperature (10 • C) were found to be the most toxic (27.7 fmol/cells) [56]. These results reveal that the high toxicity of A. catenella may be related to optimal growth temperature at around 16 • C. Moreover, the difference in STXs' toxicity may be caused by the components of STXs, in which STX and GTX1 were much more toxic than C1 and C2 [59]. A. catenella ACT03 showed different dominant analogues depending on culturing temperatures, for example, C1 at 12 • C and GTX4 at 18-30 • C [58]. These results were well-matched with that of Alex03, in which GTX1 was dominantly analyzed at 16 • C. The STXs' toxicity of A. catenella was further compared to other toxic Alexandrium species according to cultured temperatures. It was revealed that Alexandrium sp., which belongs to the A. tamarense complex, can produce STXs at optimal growth temperature and lower than 20 • C [63,64]. For example, A. fundyense BOF and MI were most toxic (100-544 fmol/cell) at 5 • C, and GTX 1-4 and STX with high TEF were primarily detected [61]. In addition, A. tamarense contained variants of GTX 1-4 at 15 • C and 17 • C, while the highest toxicity at 12 • C was analyzed [64]. Moreover, when Alexandrium minutum was incubated over a wide temperature range (15-25 • C), GTX3-5 were predominantly detected at low temperature [20]. In contrast, A. affine and A. andersonii were cultured at high temperature (15-34 • C) without producing STXs. These results and our finding showed that the toxic Alexandrium spp. vary their STX levels and profiles depending on the strains and incubating temperatures [4,51,52]. In addition, total STXs eq and components may be affected by culture temperature, with the highest toxicity under optimal growth condition [19].
Cell growth and STX toxicity of Alexandrium varies depending on their growth stages and culture conditions, such as salinity, nutrients, and/or temperature [5,17,27,[66][67][68] In general, high STXs content were detected under exponential phase of optimal growth conditions [7]. This was in accordance with our results that the highest toxicity was recorded in A. catenella Alex03 cultured at 16 • C. Perhaps, when cells divide rapidly in exponential phase, cell size may decrease with increase in the division rate. Hence, cell division decreases cellular contents, thereby decreasing their cellular toxicity.
Interestingly, we found that the STXs eq/cell of Alex03 was the highest at the optimum temperature 16 • C (maximum growth rate among three tests), and its mean cell size was much smaller than that at 12 • C. The escalation in total STXs eq at 16 • C was resulted by an increase in STX analogues at high TEF, particularly GTX1 [59]. The explanation that whether the toxicity of the cells is due to an increase in cell volume or STX production is controversial [37,67]. Previous results and our data, however, suggest that Alex03 should synthesize cellular STXs (e.g., GXT1) at low temperature, which supports optimal growth.
STXs eq and profiles of Alexandrium are partially explained by their biosynthesis gene regulation. As noted previously, sxtA and sxtG seem to be involved in early stage because they are considered as core genes linked to the initial process of STX production [28,45,47]. Thus, upregulation of the genes indicates the higher catalyzing rate of malonyl-CoA and arginine with increased Intermediate A' and Intermediate B' [34,69,70]. After this process, STX is then converted to GTX1-6 and C1-4 by other tailoring enzymes that are encoded by sxt cluster genes, resulting in STXs' accumulation in toxic microorganisms [34]. In the present study, significant upregulations of both sxtA4 and sxtG levels and STXs' content were observed in A. catenella Alex03 exposed to cold shock, which showed the correlation of transcriptional responses and STX biosynthesis. This was also supported statistically by PCA, showing a positive relationship between sxtA4 and sxtG expression levels and total STXs eq at low temperature and cold shock but not with heat shock. Similar to our results, sxtA expression levels and total STXs' concentration decreased in cyanobacteria when exposed to 10 mM NaCl for 5 h [71]. Additionally, Geffroy et al. [72] suggested that transcriptional levels of sxtA were related to STXs' content of toxic dinoflagellate A. minutum. In addition, both sxtA4 and sxtG were positively correlated with total STXs' production of A. minutum AmKB02 in different nutrient conditions [73]. On the other hand, results of weak correlation between mRNA levels of sxtA and sxtG and intracellular STXs have been reported in Mediterranean A. minutum [28]. The result indicates that STXs' synthesis genes may be regulated by post-transcriptional regulation, suggesting the inconsistency between mRNA copies and its protein abundance [28,74]. However, since they significantly correlated in stoichiometric amounts [75,76], upregulations of STXs' synthesis genes can lead to increase of the related proteins.
In contrast to cold temperature, we detected low toxicity and concentrations of diverse STX analogues from A. catenella Alex03 cultured at 20 • C and heat stress. Similarly, low toxins were measured at temperature higher than the optimum growth temperature; for example, 36.5 STXs eq µg/L at 23 • C and 83.3 STXs eq µg/L at 30 • C in the cyanobacterium Aphanizomenon gracile [19]. In addition, when A. catenella ACC02 was cultured at 10 to 16 • C, the lowest toxin levels (3.46 fmol STXs eq/cell) were measured at 16 • C [56]. In the present research, the concentrations of GTX1 and GTX4 and sxtA4 expression levels decreased, while GTX3, STX, and neoSTX increased when A. catenella was exposed to heat stress. As mentioned above, downregulation of sxtA4 and sxtG may reduce biosynthesis and accumulation of STXs [34,69]. sxtG mRNA levels increased in heat stress samples, and our results complemented the results of transcriptional responses and STX analogues. These results show that temperature regulates sxtA4 expression levels, thereby affecting STXs eq/cell and profiles.
STX biosynthesis and modification are accomplished by many sxt enzymes, and their activity should be affected by temperature, resulting in different toxin levels [77]. For example, as one of the tailoring enzymes, N-sulfotransferase (sxtN) transfers the sulfate group (3 -phospho-adenosine-5 -phosphosulfate) into GTX2/3 in the dinoflagellate Gymnodinium catenatum [78]. In addition, the two putative sulfotransferases sxtN and sxtSUL produce the sulfated C-toxins in the cyanobacterium Anabaena circinalis [79,80]. These represent that sxtN is responsible for transferring STX into C-toxins (C1-4) and GTX-toxins (GTX1-4), changing the STXs profile of toxin-producing species [70]. Thus, sulfotransferase activity may alter STX profile of Alex03, and it was dependent on temperature, which can be predicted by the coefficient Q 10 [81]. These results suggest that the sulfotransferase activity varies depending on the temperature, and may be more active at high temperature and heat shock in the toxic dinoflagellates Alexandrium.

Cell culture and Adaptation
The strain Alex03 (formerly known as LIMS-PS-2645) of A. catenella was obtained from the Marine Bio Resource Information System (MBRIS) of Korea Institute of Ocean Science & Technology (KIOST, Jangmok, Korea). It was isolated from the Korean South sea and cultured and maintained in f/2 medium without silicate [82] at 20 • C and 65 µmol photons/m 2 /s of photon flux density under a 12:12 h light-dark cycle.
For experiments related to temperature, cells cultured at 20 • C were adapted to each test temperature. In specific, we included low temperature cultures while gradually lowering the temperature (by 0.1 • C per day) using the standard cultures of 20 • C. Finally, three temperatures (12,16, and 20 • C) were chosen to simulate the water temperature from March to May in the southern coasts of Korea (10-23 • C), because toxic Alexandrium blooms and PSP outbreaks were mostly reported during that season [49]. Each temperatureadapted culture in the exponential phase was sub-cultured in fresh f/2 medium for at least 6 months.

Design for Temperature Experiments
At the start of the experiment, each strain was inoculated at a density of 165.8 ± 6.3 cells/mL using fresh f/2 medium adapted to each temperature. The samples were either incubated at the same temperature or transferred to another incubator at different temperature for stress induction. To induce cold stress, the culture flasks incubated at 20 • C were transferred to 16 • C (20→16 • C) and 12 • C (20→12 • C), and from 16 • C to 12 • C (16→12 • C). Moreover, the cultures maintained at 12 • C were shifted to 16 • C (12→16 • C) or 20 • C (12→20 • C), and from 16 • C into 20 • C for inducing thermal stress. The samples were exposed to different temperatures for 72 h on day 15 and harvested on day 18.

Cell Density and Size Measurements
Cell counts were estimated for examining growth patterns and determined every two days using a plankton-counting chamber (Matsunami Glass, Osaka, Japan) under a light microscope (Carl Zeiss Axioskop, Oberkochen, Germany). All experiments were independently performed in triplicate. Individual cells in each culture were counted using Auto T4 CellometerTM (Nexcelom Biosciences, Lawrence, MA, USA) and were then used to calculate the average cell size.

DNA and RNA Extraction and cDNA Synthesis
Total RNA was extracted using 200 mL of A. catenella cultures with an initial density of 6.0 × 10 3 cells/mL. In brief, the cells were harvested by centrifugation at 2000× g for 5 min at 4 • C, and the pellets were dissolved in 1 mL of TRIzol (Invitrogen, Carlsbad, CA, USA). To prevent RNA degradation, the samples were immediately frozen using liquid nitrogen and stored at −80 • C until RNA extraction.
For RNA extraction, the samples were physically lysed by freeze-thawing in liquid nitrogen and homogenized using zirconium beads (diameter 0.1 mm) using a Mini-bead beater (BioSpec Products Inc., Bartlesville, OK, USA). The samples were additionally purified using the Mini Spin Columns of the RNeasy Mini Kit (Qiagen, Valencia, CA, USA) as per the manufacturer's instructions. After analyzing the RNA quality and quantity using Agilent 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA), cDNA was synthesized using a TOPscript™ cDNA Synthesis Kit with random hexamer and oligo(dT) 18 (Enzynomics, Daejeon, Korea) for gene cloning and gene expressional experiments. cDNA templates were diluted five times with nuclease-free water for molecular experiments.

Cloning of sxtA4, sxtG and 28S rRNA
Partial sequences of sxtA4 and sxtG were retrieved from A. catenella EST data (157,342,442 sequence reads, 15.5 Gb) obtained in our laboratory, wherein the DNA sequences were determined using the Illumina Hiseq 2500 sequencing platform. These sequences were used to design primers for gene cloning. Nested PCR was performed using specific primer pairs ( Table 2) and genomic DNA and cDNA templates. The reaction conditions for PCR were as follows: pre-denaturation at 95 • C for 5 min; followed by 35 cycles of 95 • C for 30 s, 55 • C for 30 s, 72 • C for 1 min and final extension at 72 • C for 10 min. PCR products were purified and cloned using the TOPcloner TM TA kit (Enzynomics Inc., Daejeon, Korea). Each colony was used as template for PCR amplification and subjected to DNA sequencing.

Gene Characterization and Phylogenetic Analysis
The test species was identified by comparing the 28S rRNA sequence with that of other dinoflagellates obtained from GenBank and related researches [51]. The correct frames of two sxt genes were determined using BLASTx. The deduced amino acid sequences of sxtA4 and sxtG were analyzed using the PROSITE-ExPASy Bioinformatics Resource portal (http://prosite.expasy.org/, accessed 4 January 2021) and the Interpro online tool (https: //www.ebi.ac.uk/interpro/, accessed 4 January 2021) to identify the conserved domains.
Phylogenetic analyses were performed with sxt amino acid sequences of A. catenella Alex03 and related organisms. Each sequence matrix was properly aligned using Clustal X 2.0 (Conway Institute UCD Dublin, Dublin, Ireland) [86], and the final alignment of the sxtA4 and sxtG dataset consisted of 23 and 35 taxa and contained 236 and 256 amino acids, respectively. Phylogenetic tree derived from each dataset were constructed using the maximum likelihood (ML) algorithm, and the best-fitted model (LG+G) in MEGA X [87]. A bootstrap consensus tree inferred from 1000 replicates was used for taxa analysis.

Quantitative Real-Time PCR
Quantitative real-time PCR (qRT-PCR) was performed using the TOPreal qPCR 2× PreMIX SYBR Green Kit (Enzynomics Inc., Daejeon, Korea) in a CFX96 Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). The reaction mixtures were prepared as follows: each of forward and reverse primer (10 pmol/µL; 1 µL), diluted cDNA (2 µL), TOPreal qPCR 2 × PreMIX buffer (SYBR Green with high ROX, 10 µL), and distilled water (6 µL). The reaction was performed as follows: 4 min at 50 • C; 10 min at 95 • C; followed by 40 cycles of 10 s at 95 • C, 15 s at 60 • C, and 15 s at 72 • C. Each reaction was performed in triplicate to calculate the mean value. Specificity of amplification was identified by the melting curve generated by heating the sample from 65 • C to 95 • C. Primer efficiency was calculated from the standard curve using the threshold cycle (Ct) values for a 10-fold dilution series of the cDNA. Among reference genes, α tubulin (TUA) was used as internal controls for data normalization. Results were analyzed using the Student's t-test in SPSS software (Version 19.0; IBM Corp., Armonk, NY, USA).

High-Pressure Liquid Chromatography-Fluorescence Detection (HPLC-FLD)
We harvested A. catenella from 100 mL of culture by centrifugation at 2000× g for 10 min. The pellets were immediately resuspended in 0.01 M HCl (pH 3.0) and homogenized using a bead crusher (Taitec Corporation, Nishikata, Japan). The homogenized samples were boiled at 95 • C for 5 min. All samples were filtered through a 0.2-micrometer GVS syringe filter (GVS, Bologna, Italia) to remove cell debris.
Standard solutions including GTX1-4,6, C1-2, dcGTX2-3, STX, neoSTX, and dcSTX purchased from National Research Council Canada (NRC; Halifax, NS, Canada). STXs' analysis was performed using a HPLC-FLD system (Waters, Milford, MA, USA) and the post-column method with a slight modification of the process described by Rey et al. [88]. The fluorescence detector operated at an excitation wavelength of 330 nm and emission was scanned at 390 nm. STX analogues were separated using a Hypercarb ® column (150 mm × 2.1 mm i.d., 5 µm; Thermo Scientific, Madrid, Spain) and quantified using a 5-point calibration curve prepared using the reference standards. The concentrations of total STXs were calculated as STXs equivalent per cells (STXs eq fmol/cell) by referring to the TEF. Each STX analog has a different TEF value and the total STXs eq calculation was based on FAO/WHO [59]. The TEF sets the toxicity of STX to 1 and calculates the relative toxicity of the other derivatives.

Statistical Analysis and Principal Component Analysis (PCA)
Statistical analyses were performed using the SPSS statistical package (Version 19.0; IBM Corp., Armonk, NY, USA). Mean and standard errors were calculated for each treatment, and significant differences were determined with one-way analysis of variance (ANOVA), followed by the Student-Newman-Keuls multiple comparisons test. Probability (p) values of one-way ANOVA tests were indicated as * p < 0.05, ** p < 0.01, and *** p < 0.001. p < 0.05 was considered as statistically significant.
In addition, principal component analysis (PCA) was performed to evaluate the relationship among tested variables, such as water temperatures, cell size, and STX biosynthesis (total STXs eq and relative expression of sxtA4 and sxtG) using the Paleontological Statistics package (Past v.4.03; Natural History Museum, Blindern, Norway).

Conclusions
This work first reported STXs' toxicity and contents in the toxic dinoflagellate A. catenella Alex03 under different temperatures and the possible relationships with the regulation of two core genes sxtA4 and sxtG. The optimal temperature for Alex03 was 16 • C, exhibiting the highest growth rate and cellular STXs' content. When Alex03 was exposed to low temperatures and cold stress, the total STXs eq increased notably with increased GTX1 and sxtA4 and sxtG transcriptional levels ( Figure 6). Conversely, thermal shock lowered STXs eq/cell, but a wider variety of STX analogues were detected with downregulation of sxtA4. Statistical analysis demonstrated that sxtA4 expression was correlated with STXs eq in low temperature and cold shock samples, whereas sxtG was sensitive to temperature changes. These results were in accordance with the results of field monitoring data, i.e., most PST contamination of shellfish by A. catenella occurred in spring and autumn (8.2-15 • C) in Korean coasts [89]. In addition, resting cysts of the Alexandrium spp. germinated in November at 19.1-19.6 • C [50], and optimal growth temperature for A. catenella was recorded at 10-20 • C. Based on these field data and the results, we could conclude that temperature affected the STXs and the regulation of their biosynthesis genes to the highest level at optimal and cold temperatures in A. catenella. STX biosynthesis and modification involves many enzymes and their functional gene regulations, and thus, further researches are necessary to understand the whole transcriptional responses of STX synthesis genes in the future.
catenella Alex03 under different temperatures and the possible relationships with the regulation of two core genes sxtA4 and sxtG. The optimal temperature for Alex03 was 16 °C , exhibiting the highest growth rate and cellular STXs' content. When Alex03 was exposed to low temperatures and cold stress, the total STXs eq increased notably with increased GTX1 and sxtA4 and sxtG transcriptional levels ( Figure 6). Conversely, thermal shock lowered STXs eq/cell, but a wider variety of STX analogues were detected with downregulation of sxtA4. Statistical analysis demonstrated that sxtA4 expression was correlated with STXs eq in low temperature and cold shock samples, whereas sxtG was sensitive to temperature changes. These results were in accordance with the results of field monitoring data, i.e., most PST contamination of shellfish by A. catenella occurred in spring and autumn (8.2-15 °C ) in Korean coasts [89]. In addition, resting cysts of the Alexandrium spp. germinated in November at 19.1-19.6 °C [50], and optimal growth temperature for A. catenella was recorded at 10-20 °C . Based on these field data and the results, we could conclude that temperature affected the STXs and the regulation of their biosynthesis genes to the highest level at optimal and cold temperatures in A. catenella. STX biosynthesis and modification involves many enzymes and their functional gene regulations, and thus, further researches are necessary to understand the whole transcriptional responses of STX synthesis genes in the future. Figure 6. A schematic representation of the putative saxitoxin metabolic pathway in Alexandrium catenella. The transcriptional levels of the core genes sxtA4 and sxtG regulate STX biosynthesis and are altered by water temperature. Cold stress upregulated both sxtA4 and sxtG, leading to higher STX production, whereas heat stress downregulated sxtA4 to lower the toxicity. Various analogues were analyzed. Figure 6. A schematic representation of the putative saxitoxin metabolic pathway in Alexandrium catenella. The transcriptional levels of the core genes sxtA4 and sxtG regulate STX biosynthesis and are altered by water temperature. Cold stress upregulated both sxtA4 and sxtG, leading to higher STX production, whereas heat stress downregulated sxtA4 to lower the toxicity. Various analogues were analyzed.

Data Availability Statement:
The data presented in this study are available on request from the corresponding authors. In addition, the data that support the findings of this study are openly available in GenBank with the accession numbers MW882944, MW884258, and MW884259.