Microcystis Chemotype Diversity in the Alimentary Tract of Bigheaded Carp

Most cyanobacterial organisms included in the genus Microcystis can produce a wide repertoire of secondary metabolites. In the mid-2010s, summer cyanobacterial blooms of Microcystis sp. occurred regularly in Lake Balaton. During this period, we investigated how the alimentary tract of filter-feeding bigheaded carps could deliver different chemotypes of viable cyanobacteria with specific peptide patterns. Twenty-five Microcystis strains were isolated from pelagic plankton samples (14 samples) and the hindguts of bigheaded carp (11 samples), and three bloom samples were collected from the scums of cyanobacterial blooms. An LC-MS/MS-based untargeted approach was used to analyze peptide patterns, which identified 36 anabaenopeptin, 17 microginin, and 13 microcystin variants. Heat map clustering visualization was used to compare the identified chemotypes. A lack of separation was observed in peptide patterns of Microcystis that originated from hindguts, water samples, and bloom-samples. Except for 13 peptides, all other congeners were detected from the viable and cultivated chemotypes of bigheaded carp. This finding suggests that the alimentary tract of bigheaded carps is not simply an extreme habitat, but may also supply the cyanobacterial strains that represent the pelagic chemotypes.


Introduction
Most cyanobacterial species can produce a wide range of secondary metabolites with diversified biological activities. The unique cyanobacterium-specific secondary metabolites originating from variable biosynthetic pathways show great chemical diversity and are common across cyanobacterial taxa. Many of these compounds that are of interest in several scientific fields (pharmacology, toxicology, protection. Moreover, direct use of phosphorus by this cyanobacterial species has been detected in fish guts during passage [47]. In the mid-2010s, summer cyanobacterial blooms of Microcystis spp. occurred on a regular basis in Lake Balaton, the largest lake in Central Europe ( Figure 1). The main objective of the present work was to investigate Microcystis chemotypes within these waterbodies in the period between 2013-2016. More specifically, we aimed to: (i) Identify new and well known congeners of the cyanobacteria peptide family using a LC-MS-based untargeted approach; and (ii) determine the pattern, abundance, and distribution of Microcystis chemotypes among the pelagial, bloom area, and in the alimentary tracts of filter-feeding bigheaded carps ( Figure 2)-as this fish species represent a massive stock in the lake recently.
Toxins 2019, 11, x FOR PEER REVIEW 3 of 29 In the mid-2010s, summer cyanobacterial blooms of Microcystis spp. occurred on a regular basis in Lake Balaton, the largest lake in Central Europe (Figure 1). The main objective of the present work was to investigate Microcystis chemotypes within these waterbodies in the period between 2013-2016. More specifically, we aimed to: (i) Identify new and well known congeners of the cyanobacteria peptide family using a LC-MS-based untargeted approach; and (ii) determine the pattern, abundance, and distribution of Microcystis chemotypes among the pelagial, bloom area, and in the alimentary tracts of filter-feeding bigheaded carps ( Figure 2)-as this fish species represent a massive stock in the lake recently.   In the mid-2010s, summer cyanobacterial blooms of Microcystis spp. occurred on a regular basis in Lake Balaton, the largest lake in Central Europe ( Figure 1). The main objective of the present work was to investigate Microcystis chemotypes within these waterbodies in the period between 2013-2016. More specifically, we aimed to: (i) Identify new and well known congeners of the cyanobacteria peptide family using a LC-MS-based untargeted approach; and (ii) determine the pattern, abundance, and distribution of Microcystis chemotypes among the pelagial, bloom area, and in the alimentary tracts of filter-feeding bigheaded carps ( Figure 2)-as this fish species represent a massive stock in the lake recently.

Morphology-Based Identification of Cyanobacterial Bloom Causing Organisms, and the Isolated Strains
Microcystis cells were spherical with a 5-8 µm diameter. Aggregated cells were organized into colonies with very narrow and diffluent colorless mucilage. Light blue-green protoplasts appeared to be light-brown due to the optical effects of gas vesicles which were observed in the cells. Colony sizes ranged from microscopic to macroscopic. The macroscopic colony layer showed a pale green color in the natural blooms and cultures. Morphological features were identical to the characteristics of Microcystis flos-aquae [48]. Based on morphological criteria of the colinies, the dominant morphotype in the bloom samples was identified as Microcystis flos-aquae.

Molecular Phylogenetic Analyses
In the phylogenetic analysis, altogether 25 ITS strain sequences were isolated from Lake Balaton and hindgut content of bigheaded carps (Hypophthalmichthys spp.). These were involved with Microcystis ITS sequences originating from six different morphospecies. All ITS sequences in this study formed a distantly related cluster with all but one Microcystis morphospecies ( Figure A1). In the phylogenetic tree, the ITS sequences of strains from the lake's water and hindgut samples were arranged into eight separated lineages, however, no clear distinction according to sample type or sampling time could be observed. Our strains showed the closest phylogenetic relation to morphospecies M. flos-aquae and M. viridis.

Identification of Peptides and Comparative Analysis of Bloom Samples and M. flos-aquae Strains
The cyanobacterial peptides identified from M. flos-aquae strains and blooms that originated from the pelagic and bigheaded carp matrix are presented in Tables 1-3. contained amino acid modification in two positions compared to FR3. The 26 Da mass increase can be explained by an amino acid exchange from threonine to leucine at position 2, and a substitution of tyrosine with homotyrosine at position 4, which is also supported by MS 2 data (Table A1).    (Table A1).  (Table A2).
Each of the partially characterized ANAs showed characteristic product ions with the 26 Da difference in the MS 2 experiences. In four cases, no amino acid or a simple derivative thereof could be assigned to amino acid position 1 based on MS/MS, but their ring structure suggested that these were compounds related to previously known ANAs. For the remaining cases, the MS analysis data were insufficient to determine all the structural elements due to the low peak intensity (Table A2).
The heatmap shows the relative abundance of given natural products in the Microcystis samples-the color is proportional to the log10 of the area under the curve values.
The most important phenomenon is the lack of peptide pattern separation for Microcystis originating from the gut, lake, and bloom-samples. The samples were separated into relatively loose clusters based on their metabolomes, spreading all three functional groups throughout these clusters. No separation by sampling year could be found either (Figures 3 and 4).   (Table A3).
The heatmap shows the relative abundance of given natural products in the Microcystis samples-the color is proportional to the log10 of the area under the curve values.
The most important phenomenon is the lack of peptide pattern separation for Microcystis originating from the gut, lake, and bloom-samples. The samples were separated into relatively loose clusters based on their metabolomes, spreading all three functional groups throughout these clusters. No separation by sampling year could be found either (Figures 3 and 4).

Classification of Oligopeptide Pattern into Chemotypes
Investigating individual Microcystis strains and samples showed a high number of different, but more or less distinct, chemotypes (with different peptide patterns) that originated from all habitats. On the other hand, more or less distinguishable chemotypes were observed among the samples, which were separated into clusters of varied densities based on the absence or co-occurrence of biosynthetic metabolite clusters. The samples either contained: (1) a variety of ANAs and a few unidentified peptides, (2) MCYs, or (3) MGs as their major compounds, and usually contained significantly lower or no detectable amounts from other peptide groups. For example, in many samples, a group containing ANAs that lack MGs and MCYs were seen to be of this chemotype. The co-occurrence of several MCY types was observed in only a few samples. This was also illustrated by the compact cluster separating MCYs from other natural products. These MCY-rich samples lacked other chief compounds of interest. Another subset of cyanobacteria contained various MGs at high concentrations, but did not contain ANAs or other peptides as major components. These compounds were all present in their producers, resulting in a relatively compact MG cluster. The bootstrapped hierarchical clustering analysis revealed the existence of two main chemotypes: MG-dominant and ANA-dominant ( Figure 5). The existence for both clusters was p < 0.05.

Classification of Oligopeptide Pattern into Chemotypes
Investigating individual Microcystis strains and samples showed a high number of different, but more or less distinct, chemotypes (with different peptide patterns) that originated from all habitats. On the other hand, more or less distinguishable chemotypes were observed among the samples, which were separated into clusters of varied densities based on the absence or co-occurrence of biosynthetic metabolite clusters. The samples either contained: (1) a variety of ANAs and a few unidentified peptides, (2) MCYs, or (3) MGs as their major compounds, and usually contained significantly lower or no detectable amounts from other peptide groups. For example, in many samples, a group containing ANAs that lack MGs and MCYs were seen to be of this chemotype. The co-occurrence of several MCY types was observed in only a few samples. This was also illustrated by the compact cluster separating MCYs from other natural products. These MCY-rich samples lacked other chief compounds of interest. Another subset of cyanobacteria contained various MGs at high concentrations, but did not contain ANAs or other peptides as major components. These compounds were all present in their producers, resulting in a relatively compact MG cluster. The bootstrapped hierarchical clustering analysis revealed the existence of two main chemotypes: MG-dominant and ANA-dominant ( Figure 5). The existence for both clusters was p < 0.05.

Lack of Phylogenetic Relationship Was Found among the Chemotypes
Although all of the isolates were identified as the M. flos-aquae morphotype by conventional microscopy techniques, several phylogenetic clusters ( Figure A1) and more distinct chemotypes (Figures 4 and 5) have also been defined. The lack of a phylogenetic relationship was found among the chemotypes. To test the statistical significance of chemical patterns and other variables (such as origin, phylogenetic position), principal component score values were subjected to the Kruskal-Wallis significance test, where the variable studied was the group-determining factor. This approach

Lack of Phylogenetic Relationship Was Found among the Chemotypes
Although all of the isolates were identified as the M. flos-aquae morphotype by conventional microscopy techniques, several phylogenetic clusters ( Figure A1) and more distinct chemotypes (Figures 4 and 5) have also been defined. The lack of a phylogenetic relationship was found among the chemotypes. To test the statistical significance of chemical patterns and other variables (such as origin, phylogenetic position), principal component score values were subjected to the Kruskal-Wallis significance test, where the variable studied was the group-determining factor. This approach for all metabolites allowed higher statistical power compared to direct statistical analyses. No significant association was found between the metabolite pattern, abundance, and genetic background. The 16S-23S ITS phylogenetical group had no significant effect on the metabolite pattern (p > 0.05), represented by one of the 28 PC dimensions in the analysis. No significant association was found between the metabolite pattern, abundance, and source of isolate. Isolate source (water, fish, or bloom) had no significant effect on the metabolite pattern (p > 0.05), represented by one of the 28 PC dimensions in the analysis.

Discussion
The detected Microcystis flos-aquae blooms were unexpected but not unique, because at the beginning of the 20th century, Microcystis blooms were observed in Lake Balaton localizing to small areas [54]. In addition, from the middle of the century, mainly nitrogen fixing filamentous cyanobacteria species like Aphanizomenon and Anabaena (Dolichospermum) caused this phenomenon in the lake. In Lake Balaton, the species Cylindrospermopsis raciborskii was identified in the 1970s and initiated whole-lake-area sized water blooms in 1982 and 1994, which adversely affected the tourism and economy of the area [55]. A number of comprehensive water management measures at the end of the 1980s were aimed at curbing eutrophication such as the drainage of communal sewage from the coastal zone. At the same time, the scale of agricultural activity decreased, resulting in lower nutrient loads to the lake. From this period until the present report, there has been no cyanobacterial blooming in Lake Balaton. The reappearance of Microcystis blooms in the lake, which were observed in this study, suggests that external nitrogen loads may have initiated the multiplication of non-diazotrophic cyanobacteria. In our present study, 11 Microcystis strains were isolated from gut samples, 14 from the pelagic plankton from Lake Balaton, and three collected Microcystis bloom samples.
Taxonomic classification of Microcystis is difficult. The combination of microscopic observations with molecular data can be the most adequate method for identifying isolates [56]. Most species of this genera have been described via their morphological characteristics [57,58], however, colony variance can be huge, and the external qualities of many populations overlap the limiting specifications [59,60]. Therefore, it is difficult to find the differences between traditional species [61].
The taxonomic position of the isolated strains was confirmed by phylogenetic analysis. Our strains showed the closest phylogenetic relation to morphospecies M. flos-aquae and M. viridis ( Figure A1). Several papers have been published on the correlation between Microcystis morphotypes and genotypes [62,63]. The heterogeneity of this genus between different regions has also been well documented using further genetic markers [16,[18][19][20]22,23,29]. Neilan et al. [64] defined genetic similarity and noted that the species had no specific phylogeographic structure, which was in accordance with work described by Bittencourt-Oliveira et al., (2001) [17] where Microcystis strains did not show any distinct phylogenetic pattern.
In our work, we identified a strong mucous envelope characteristic of the studied Microcystis morphotype. This can explain how this cyanobacterial species survive in the alimentary tract of bigheaded carps. Exopolysaccharides (EPS) can be crucial for cellular attachment, adhesion, and survival. This highly hydrated layer provides protection to cyanobacterial cells against desiccation, toxic agents, or the digestive enzymes of other organisms. This role of the EPS has been confirmed in several studies [65].
Thirty-six ANA and 17 MG variants were identified from the isolated strains (Tables 1 and 2). New MCY congeners are more rarely identified, perhaps because this has been the most investigated cyanobacterial peptide family. However, bioactive peptide families like MGs and ANAs are receiving growing attention. Four known and 32-to the best of our knowledge-previously unknown ANA variants were fully or partially identified in our analysis. Several partly identified peptide fragments were detected (Tables 4 and A4), which were clustered into this family using heat map analysis. ANA F, OSC B, and C are inhibitors of protein phosphatases (PP). N-MeHty and the positively charged Arg are crucial parts of molecules relating to this activity [66]. ANAs were also found to be active toward proteinase enzymes such as trypsin, chymotrypsin, elastase, and carboxypeptidase A [8,67,68]. The relaxing activity of rat aortic preparations was detected by treatments with ANA B and ANA 906 [69]. ANA B and ANA F, the most frequently found ANAs, were shown to inhibit the growth of many M. aeruginosa strains by inducing the lytic cycle in cyanobacteria [70,71]. Taking into account the published effects of these metabolites, its ecological roles might be important [72].
Two known, and 15-to the best of our knowledge-previously unknown MG variants were identified in our analysis. Several partly identified peptid fragments were also detected as part of this family.
MGs are a 40-member group of linear and nonribosomal peptides, which have been detected and purified from several bloom-forming cyanobacteria isolates. The number of congeners is growing [2,3,7]. These are built by an α-hydroxy-β-amino derivate of decanoic or octanoic acid, which is rarely chlorinated at its terminal methyl group with three to five additional amino acids in the molecules [4,7,73,74]. MGs are zinc metalloprotease inhibitors (e.g., angiotensin-converting enzyme), and aminoproteinases that bind their α-hydroxy-β-amino residue to the zinc at the active site of the enzyme. Our findings indicate that these substances are important candidates for treating hypertension [72]. The patchy distribution of oligopeptide patterns in cyanobacterial populations enables classifying isolates into several oligopeptide-based chemotypes [14,75]. It is important to note that mainly ANA and MG dominant strains were detected from Lake Balaton in this study, but 10 strains from the alimentary tract were MCY producing.
The distinguishable chemotypes we found in our analysis were separated into clusters of varied density based on the absence or co-occurrence of biosynthetic metabolite clusters, similar to several other studies that have investigated Microcystis and other cyanobacterial peptide producers such as Planktothrix sp., Nostoc sp., etc. [14,[75][76][77]. These genera often possess variable chemotypes with the ability to produce different peptide families in natural assemblages [3]. In the Microcystis isolates originating from natural populations, four chemotypes were characterized based on the fact that they contained a few or several main peptides, but in many cases, the appearance of several different peptides belonging to different biosynthetic clusters has been observed.
From Lake Balaton, primarily ANA-and MG-dominated strains were detected, with the observation that many lesser-known or new congeners appeared from both groups of peptides. In addition, several partly identified peptide fragments were detected during the analysis whose metabolites seemed to belong to the metabolism (synthesis or degradation) of the two main groups identified as suggested by the heat map visualization. Focusing on the bloom samples from 2014-2016, it is important to note that the naturally collected material of all three samples represent mixed matrices, and each of them contained several chemotypes and/or genotypes of the genus Microcystis. These samples belonged to different chemotype clusters. The bloom samples from the same sampling site showed different bioactive peptide patterns. While the 2014 sample was mainly MG-dominated, the 2015 bloom sample was rather ANA-rich. The 2016 bloom community contained mostly non-identified peptides ( Figure A2). Considering all the identified peptides in our samples, we found that the isolated and identified chemotypes originating from the gut and pelagic sample could be involved in the Microcystis community that built the bloom phenomenon.
Altogether, 13 MCY congeners were also identified in this work from phytoplankton and digestive tract strains. There is no doubt that MCYs are the most harmful and notorious family from the described cyanobacterial peptides. All of the detected MCY forms are already known, and no new MCY variant was identified in the present analysis. Although only a few MCY variants have been identified near the large number of MG and ANA, these peptide-producers display a separate cluster in our analysis. It is especially worth paying attention to this group because MCY is the most common toxin produced by cyanobacteria in waters [78,79], and can also cause death, illness, complications, and damage in humans, animals, and plant organs.
In the several cases where Microcystis and Planktothrix oligopeptide patterns have been investigated worldwide [12][13][14]75,[80][81][82], and in our local area [83][84][85][86], the numbers of genotypes have been identified with the help of phylogenetic markers [64,87]. The lack of correlation between the Microcystis chemotypes and phylogenetic genotypes found in cases similar to our present study suggest that the synthesis of bioactive peptides is not phylogenetically conserved in this genera. This has also been the conclusion of recent work [88] where Microcystis chemotypes were researched in Spanish freshwater and reservoirs. The findings in our study are consistent with the statement that the distribution of oligopeptide production abilities does not correlate with morphospecies, phylogenies based on commonly used molecular markers, or the geographical origin of the isolated organisms [64,89].
Bigheaded carps were introduced into Lake Balaton (Hungary) in 1972 and were stocked until 1983 [90][91][92][93][94]. These filter-feeder fish species can consume almost all algal and cyanobacterial taxa from ambient water, but the ingested algae are only partially utilized [47]. In fact, a fraction of the consumed phytoplankton cells or colonies (e.g., Microcystis sp., diatoms, volvocalean, and chlorococcalean green algae) may stay alive after passing through the digestive tract of fish as they are protected either by a mucilaginous envelope or by a thick, cellulose-based cell wall [47,95]. In the present study, 11 Microcystis strains were isolated from gut samples and identified as Microcystis flos-aquae with a characteristic mucilage envelope that can be the main protecting layer against the extreme gut environment.
In the scientific literature, there is contradictory information on the abundance and composition of cyanobacteria in the alimentary tract of bigheaded carps. On one hand, a study by Ye et al. [96] found cyanobacteria to be predominant in the gut microbiota of bigheaded carp living in different North American rivers, while Li et al. [97] reported the abundance of cyanobacteria was typically low in the intestines of both silver and bighead carp inhabiting Wuhu Lake (China). Based on the gut content metagenome analysis, Microcystis was identified as the most abundant cyanobacterial genus detected in the gut of bigheaded carp in Lake Balaton [98].
Beyond the outer polysaccharide layer, it is worth noting the detected peptides in this work, mainly the large number of ANAs and MGs (Tables 1 and 2) and the above discussed bioactivity. Digestive enzymes' activity in bigheaded carp species have been investigated in several studies [99,100]. Phosphatases and proteases are principal groups of enzymes for the fish species [101]. "The rapid excretion rate of silver carp would require quick digestion and nutrient uptake of foods to support high growth rates" [102]. Several microorganisms that can play a role in digestion and be responsible for higher levels of the above-mentioned digestive enzymes have been identified in the gut of silver carp [96]. Our suggestion is that the detected MGs and ANAs, as potent protease inhibitors, could modify the digestive capacity by binding directly to the enzymes. However, it is known that most oligopeptides stay in the producing cyanobacteria cell and are only released via cell lysis following cell death, and thus, would only provide protection for the surviving cyanobacterial cells in the gut.
While the traditional approaches of toxin and/or bioactive metabolite research of cyanobacteria have mainly focused on individual peptides, exploring their effects or biosynthesis, our chemotyping study with non-targeted analysis investigated the occurrence of various peptides in Microcystis strains that originated from bloom, pelagic plankton samples, and from the gut of a notorious invasive fish species. Except for 13 peptides, all other congeners were detected from viable and cultivated chemotypes originating from bigheaded carps. This finding suggests that the alimentary tract of bigheaded carps is not only a special habitat, but also a supplier for strains that represent the pelagic chemotypes and can initiate blooms in the waterbody. This potentially malicious feature can come from the ability of this fish species to filter plankton efficiently, but a few organisms such as the peptide-producing mucilaginous enveloped cyanobacterial species M. flos-aquae are digested improperly or not at all in the digestive system. In addition, several studies have noted that the toxicity of cyanobacteria remained unaffected or even increased after defecation. Kolmakov et al. [103] demonstrated that the physiology of the investigated cyanobacterial species were not suppressed by passing through fish intestines, but rather enhanced when they returned to the water. Kolar et al. [35] also noted that some Microcystis cells were not eliminated by the digestive processes of fish species. Lewin et al. [46] suspected that Microcystis could survive, and even use the phosphorus in fish guts as nutrients [46].
Wide time interval evacuation rates have been estimated for silver carp at different water temperatures [104]. This is why it is not easy to calculate the retention of viable Microcystis cells in the gut. Although it is worth raising the opportunity that bigheaded carps carrying cyanobacterial chemotypes in their guts from one habitat can invade new areas, and that the viable cyanobacterial cells may be released by defecation from fish [32,33].
In our chemotyping study, Microcystis strains isolated from the invasive non-native bigheaded carps and their peptide patterns were compared to pelagic and bloom material strains. Our results draw attention to the fact that bigheaded carps not only carry and spread viable, mucilaginous envelope-covered Microcystis cells from their alimentary tracts, but harmful cyanobacterial strains can also be found among them according to the chemotypes.

Sample Collection and Initial Sample Processing
Bigheaded carp were collected from Lake Balaton (Hungary), which is the largest lake in Central Europe. Its surface area is 596 km 2 , while the average water depth is about 3 m [105]. Bigheaded carps and water samples were collected from the lake in April, May, June, September, and October in 2013.
The local fishery company (Balaton Fish Management Non-Profit Ltd., Hungary) had a permit to harvest fish by nets (including bigheaded carp) from Lake Balaton in 2013 (permit reg. no.: 2013/N000001, issued by the Fisheries Authority of Somogy County, Hungary). The fishery company provided samples to the researchers from their commercial catches. After receiving the samples, researchers of the Balaton Limnological Institute (Center for Ecological Research, Hungarian Academy of Sciences) transported them to the laboratory within 30 min. The Institute has a permit for the delivery and use of fish for scientific purposes (permit reg. no.: VE-I-001/01890-3/2013, issued in 22 August 2013 by the Food-Security and Animal Health Directorate, Governmental Office of Veszprém County, Hungary). Gut content samples were collected aseptically, as described by Görgényi et al. [47]. Subsamples were taken (cc. 5 g) and stored in sterile Eppendorf tubes at 4 • C until laboratory processing, all done within 24 h.
Water samples for chemical and biological analyses were collected by immersion from the upper water layer at the beginning, one-third, two-third, and ending points of each transect (Figure 6a). Water samples for chemical and biological analyses were collected by immersion from the upper w Cyanobacterial bloom samples were collected from blooming waters of Microcystis morphotypes during the summer season (July-August) from 2014 to 2016. Twenty-five isolated strains (11 from the hindgut of bigheaded carps and 14 from free living pelagic plankton) and three collected bloom samples were analyzed in this study. Their origin and localization are shown on the map in Figure 6.
Gut content and water samples were incubated in nitrate containing Allen medium [106] for 5 days and the visible Microcystis colonies were collected and inoculated in nitrate-containing medium at 26 °C under continuous illumination (100 lux m −2 s −1 ) for a week. Prior to the molecular analyses, the collected bloom samples and the cyanobacterial strains were studied using light microscopy.

Identification of Cyanobacterial Peptides
The optimal ESI ionization parameters were as follows: heater temperature, 250 °C; sheath gas, N2; flow rate, 10 arbitrary units (arb); aux gas flow rate, 5 arb; spray voltage, 5 kV; capillary Water samples for chemical and biological analyses were collected by immersion from the upper water layer at the beginning, one-third, two-third, and ending points of each transect (Figure 6a). Cyanobacterial bloom samples were collected from blooming waters of Microcystis morphotypes during the summer season (July-August) from 2014 to 2016. Twenty-five isolated strains (11 from the hindgut of bigheaded carps and 14 from free living pelagic plankton) and three collected bloom samples were analyzed in this study. Their origin and localization are shown on the map in Figure 6.
Gut content and water samples were incubated in nitrate containing Allen medium [106] for 5 days and the visible Microcystis colonies were collected and inoculated in nitrate-containing medium at 26 °C under continuous illumination (100 lux m −2 s −1 ) for a week. Prior to the molecular analyses, the collected bloom samples and the cyanobacterial strains were studied using light microscopy.

Identification of Cyanobacterial Peptides
The optimal ESI ionization parameters were as follows: heater temperature, 250 °C; sheath gas, N2; flow rate, 10 arbitrary units (arb); aux gas flow rate, 5 arb; spray voltage, 5 kV; capillary ; hindgut content of bigheaded carp: Water samples for chemical and biological analyses were collected by immersion from the upper water layer at the beginning, one-third, two-third, and ending points of each transect (Figure 6a). Cyanobacterial bloom samples were collected from blooming waters of Microcystis morphotypes during the summer season (July-August) from 2014 to 2016. Twenty-five isolated strains (11 from the hindgut of bigheaded carps and 14 from free living pelagic plankton) and three collected bloom samples were analyzed in this study. Their origin and localization are shown on the map in Figure 6.
Gut content and water samples were incubated in nitrate containing Allen medium [106] for 5 days and the visible Microcystis colonies were collected and inoculated in nitrate-containing medium at 26 °C under continuous illumination (100 lux m −2 s −1 ) for a week. Prior to the molecular analyses, the collected bloom samples and the cyanobacterial strains were studied using light microscopy.

Identification of Cyanobacte
The optimal ESI ionizati N2; flow rate, 10 arbitrary u

Identification of Cyanobacterial Pepti
The optimal ESI ionization param N2; flow rate, 10 arbitrary units (arb Water samples for chemical and biological analyses were collected by immersion from the upper water layer at the beginning, one-third, two-third, and ending points of each transect (Figure 6a). Cyanobacterial bloom samples were collected from blooming waters of Microcystis morphotypes during the summer season (July-August) from 2014 to 2016. Twenty-five isolated strains (11 from the hindgut of bigheaded carps and 14 from free living pelagic plankton) and three collected bloom samples were analyzed in this study. Their origin and localization are shown on the map in Figure 6.
Gut content and water samples were incubated in nitrate containing Allen medium [106] for 5 days and the visible Microcystis colonies were collected and inoculated in nitrate-containing medium at 26 °C under continuous illumination (100 lux m −2 s −1 ) for a week. Prior to the molecular analyses, the collected bloom samples and the cyanobacterial strains were studied using light microscopy.

Identification of Cyanobacterial Peptides
The optimal ESI ionization parameters were as follows: heater temperature, 250 °C; sheath gas, N2; flow rate, 10 arbitrary units (arb); aux gas flow rate, 5 arb; spray voltage, 5 kV; capillary , and collected bloom samples: Water samples for chemical and biological analyses were collected by immersion from the upper water layer at the beginning, one-third, two-third, and ending points of each transect (Figure 6a). Cyanobacterial bloom samples were collected from blooming waters of Microcystis morphotypes during the summer season (July-August) from 2014 to 2016. Twenty-five isolated strains (11 from the hindgut of bigheaded carps and 14 from free living pelagic plankton) and three collected bloom samples were analyzed in this study. Their origin and localization are shown on the map in Figure 6.
Gut content and water samples were incubated in nitrate containing Allen medium [106] for 5 days and the visible Microcystis colonies were collected and inoculated in nitrate-containing medium at 26 °C under continuous illumination (100 lux m −2 s −1 ) for a week. Prior to the molecular analyses, the collected bloom samples and the cyanobacterial strains were studied using light microscopy.

Identification of Cyanobacterial Peptides
The optimal ESI ionization parameters were as follows: heater temperature, 250 °C; sheath gas, N2; flow rate, 10 arbitrary units (arb); aux gas flow rate, 5 arb; spray voltage, 5 kV; capillary . Cyanobacterial bloom samples were collected from blooming waters of Microcystis morphotypes during the summer season (July-August) from 2014 to 2016. Twenty-five isolated strains (11 from the hindgut of bigheaded carps and 14 from free living pelagic plankton) and three collected bloom samples were analyzed in this study. Their origin and localization are shown on the map in Figure 6.
Gut content and water samples were incubated in nitrate containing Allen medium [106] for 5 days and the visible Microcystis colonies were collected and inoculated in nitrate-containing medium at 26 • C under continuous illumination (100 lux m −2 s −1 ) for a week. Prior to the molecular analyses, the collected bloom samples and the cyanobacterial strains were studied using light microscopy.

Statistics
Identified peptides were integrated using targeted peak search in mzMine 2.11 [108]. Thereafter, raw metabolite abundances were scaled and centered separately for each feature in R 3.5.0 [109]. The dataset was hierarchically clustered in both dimensions (samples, metabolites) using the Minkowski distance as the distance measure and Ward's method. The order of appearance on the presented heatmap's axes followed that from the clustering. The color strength was proportional to the log10 transformed raw (non-scaled) metabolite abundances, while the color hue was a function of metabolite class: ANA, red; MCY, blue; MG, red; and other peptides, magenta.
The presence of chemotypes was shown by bootstrapped hierarchical clustering analysis of the cyanobacterial lines' scaled and centered natural compound abundance values using the Minkowski distance and Ward's hierarchical clustering, bootstrapping N = 1 × 10 6 . The calculation was done with the 'pvclust' package in R 3.5.2 [110]. To test the statistical significance of chemical patterns and other variables (such as origin, taxonomic position), principal component score values were subjected to the Kruskal-Wallis significance test, with the variable studied being the group-determining factor. This approach allowed for much more statistical power than that of the direct statistical analyses for all metabolites.

Phylogenetic Analysis
To explore the phylogenetic relationships between Microcystis strains, amplification and sequence analysis of the 16S-23S internal transcribed spacer region was carried out. DNA amplification was performed by PCR using primers MITS-F (5 -AAGGGAGACCTAATTCVGGT-3 ) and MITS-R (5 -TTGCGGTCYTCTTTTTTGGC-3 ) [20] in a 2720 Thermal Cycler (Applied Biosystems, Foster City, CA, USA) with the following temperature protocol: Initial denaturation at 95 • C for 5 min, followed by 30 amplification cycles of 30 s at 94 • C, 30 s at 55 • C, and 30 s at 72 • C, followed by a final extension at 72 • C for 3 min. The PCR reaction mixture contained 200 µM of each deoxynucleoside triphosphate, 1 U of LC Taq DNA Polymerase (recombinant) (Fermentas, Vilnius, Lithuania), 1× Taq buffer with (NH 4 ) 2 SO 4 (Fermentas, Vilnius, Lithuania), 2 mM MgCl 2 , 0.3 µM of each primer, and about 20 ng of genomic DNA template in a total volume of 50 µL. PCR products were checked on a 1% agarose gel stained with Eco Safe DNA dye (Pacific Image Electronics, New Taipei City, Taiwan), and visualized using UV excitation.
Sequence analysis of the obtained PCR products was accomplished by Sanger sequencing at LGC Genomics (Queens Road, Teddington, Middlesex, UK), using the MITS-F primer.
The phylogenetic dendrogram of Microcystis-related strains was constructed using MEGA7: Molecular Evolutionary Genetics Analysis version 7.0 for bigger datasets [111] software. The evolutionary history was inferred by using the maximum likelihood method based on the Jukes-Cantor model [112]. The tree with the highest log likelihood (−716.14) is shown. The percentage of trees in which the associated taxa clustered together is shown next to the branches. Initial tree(s) for the heuristic search were obtained automatically by applying the Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using the maximum composite likelihood (MCL) approach, and then by selecting the topology with a superior log likelihood value. A discrete Gamma distribution was used to model the evolutionary rate differences among the sites (five categories (+G, parameter = 0.5472)). The rate variation model allowed for some sites to be evolutionarily invariable ([+I], 43.72% sites). The tree was drawn to scale, with branch lengths measured in the number of substitutions per site. The analysis involved 75 nucleotide sequences. All positions containing gaps and missing data were eliminated. A total of 239 positions were in the final dataset [111].

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Table A1. Product ion data for the microginin type peptides. Masses are given in Dalton rounded to the nearest integer. X 1 indicates core fragment of Ahda after abstraction of the side chain (m = 58 Da). Leucine and isoleucine cannot be distinguished from the LC-MS/MS data, these amino acids have been deduced from the nearest literary results.  Amino Acid Sequence X 4 -X 5 -X 6 -X 7 -X 1 -X 2 X 3 -X 4 -X 5 -X 6 -H 2 O X 4 -X 5 -X 6 -X 7 X 4 -X 5 -X 6 X 7 -X 1 -X 2 -X 3 -X 4 -NH 2 X 7 -X 1 -X 2 -X 3 -X 4 X 1 -X 2 -X 3 -X 4 X 4 -X 5   Figure A1. Phylogenetic dendrogram of the studied Microcystis strains. The dendrogram was based on the 16S-23S internal transcribed spacer region sequences and the evolutionary history was inferred using the maximum likelihood method based on the Jukes-Cantor model.