Chemotyping of terrestrial Nostoc -like isolates from alkali grassland areas by non-targeted peptide analysis

The Nostoc genus is a well-known heterocytous, filamentous cyanobacterium which can be found all over the world. The size of terrestrial and/or freshwater colonies can be microscopic and macroscopic as well. In addi-tion, Nostoc species are one of the most common photosynthetic cyanobacterial partners in symbiotic interac- tions. Terrestrial cyanobacterial colonies were collected and isolated in this study from various alkali grassland habitats (Great Hungarian Plain). Altogether 133 colonies were isolated from the 65 collected samples. The peptide patterns of the Nostoc -like strains were examined using HPLC-ESI-MS/MS and 41 peptides were identified from 45 isolated Nostoc -like strains; these compounds belonged to 4 different peptide classes. Twelve nostoginin/microginin, 16 anabaenopeptin, 12 banyaside/suomilide variants were identified. 37% of our isolated Nostoc -like strains produced some of the peptide metabolites we tested. These strains showed distinct chemotypes according to their peptide patterns, and can be divided into 4 groups based on their metabolisms. Strains either contained: (1) nostoginins/microginins, (2) anabaenopeptins, (3) anabaenopeptins and banyasides or (4) banyasides as major compounds. Banyasides were present in many of our strains and showed very high intensity in some cases. A number of previously unknown banyaside variants have been identified.


Introduction
Cyanobacteria are a highly diverse group of photosynthetic gramnegative prokaryotes. They can be found in a wide range of habitats (freshwater, marine and terrestrial ecosystems) and are able to adapt to special circumstances such as high UV levels, extreme temperatures and pH conditions [1][2][3][4]. Their organization can be unicellular, colonial or filamentous and they can be spherical, rod and spiral shaped. Filamentous species are divided into two types, homocytous (=undifferentiated, e.g., Oscillatoria) and heterocytous (= differentiated, having heterocytes, e.g., Nostoc) [5]. Cyanobacteria are divided into five major sub-sections based on their morphological and physiological characteristics [6], but the groups are not supported by phylogenetic studies. Although several new genera and species within family Nostocaceae have been described in the last years including the reclassification of some Nostoc isolates, the taxonomic identification of Nostoc-like strains is still controversial due to the incongruences between morphological characters and phylogenetic data [7,8].
Nostoc is a genus of heterocytous, filamentous cyanobacteria. They can be found all over the world in almost all circumstances; the size of the colonies can be microscopic as well as macroscopic [1,9]. As a nitrogen-fixing cyanobacterium, Nostoc plays a role in the maintenance of soil fertility and has been used as "green manure" in many countries. In tropical and subtropical regions, rice paddies are characterized by the mass presence of Nostoc species when they are flooded [9]. They significantly contribute to building-up soil fertility and increase rice yields. Nostoc species are one of the most common photosynthetic partners in symbiotic interactions [1,9]. Symbiotic Nostoc strains can occur in liverwort, hornwort [10,11], ferns, cycads, angiosperms and lichens [12]. Nostoc flagelliforme has great economic value. Its medicinal value has been known since ancient times, and it has been used as food delicacy for about 2000 years. Nostoc commune is consumed as human food in many countries, because it is rich in fibers and proteins [13,14].
Since the 1990s, the number of publications dealing with MCYs and other cyanopeptides has greatly increased. Over the past few decades, many new cyanopeptides have been identified, but only a small part of them belong to the MCYs [18].
Grasslands are the essential elements of conventional landscapes and are very important in the perspective of biodiversity conservation. The size and species richness of natural grasslands have decreased in Europe as a result of agricultural intensification, urbanization and decline in the extensive traditional land use practices. In spite of the considerable changes in land use regimes in Central-Europe extensive stands of alkali grasslands have remained in a good nature conservation status. These areas are utilized as pastures since their soil is unsuitable for arable crops due to their low-grade soil quality and fluctuating water regime [43]. Terrestrial cyanobacteria can be found in nearly every alkali habitat. Under continental conditions, these areas are covered by water during springtime but dried out until the second half of summer, and the salt content varies from moderate to high. In Europe, these habitats cover approx. 209,152 ha, which is almost entirely (> 98%) located in Hungary. Colonies of Nostoc species appear at the beginning of spring, when the soil surface is flooded with water. Swollen colonies form mucous layers which then dry out during summer ( Fig. 1) [44].
Beyond the well-known cyanobacterial toxin-screening approaches in this discovery work we focused on the appearance of the rarely investigated peptides also. The bioactivity, toxicology and natural consequences of the cyanobacterial peptide metabolites are in focus nowadays but for understanding the roles of these compounds it is obviously essential to study the real natural producers. In our chemotyping study we analyzed the peptide pattern of the identified Nostoc samples originated from a selected area with non-conventional semiterrestrial habitats where these organisms are dominant with characteristic mats. Terrestrial cyanobacterial colonies were collected and isolated from different habitats of alkali grassland (Great Hungarian Plain). The main objective of this work was the: (i) determination of the main chemotypes of the Nostoc species occurring on Hungarian alkaline grasslands; and (ii) identification of well-known and new cyanotoxins and bioactive cyanopeptides produced by Nostoc species with non-targeted LC-MS based method.

Description of the study sites
The study area is in the Great Hungarian Plain (East Hungary) and covers approximately 25,000 km 2 . The area is characterized by a continental climate with an annual precipitation of 538 mm and mean annual temperature of 10.4°C [45]. Even though the landscape is predominantly an agricultural one with the dominance of arable lands, large stands of alkali grasslands have been preserved in areas characterized by salt-affected soils. Nostoc species generally occur in alkali steppes (short and closed grasslands, moist in springtime and dry in summer, moderate soil salt content), open alkali grasslands (short and open grasslands, moist in springtime and dry in summer, high soil salt content) and alkali meadows (tall and closed meadows, moist habitat conditions, low soil salt content) [46] (Fig. A1). These habitats generally compose a heterogenous mosaic complex in the landscape. For the habitats present at the studied sites and their most frequent vascular plant species please refer to Table 1.

Morphology-based identification of the isolated Nostoc and other terrestrial cyanobacteria strains
Altogether 133 colonies were isolated from the 65 collected samples of dry material ( Fig. 2 and Table 1). The color of the colonies varied from light green to dark brown; a gelatinous sheath was often present but gas vesicles were rarely characteristic for the colonies. Based on microscopic identification [47,48], most of the isolates (90%) belonged to the Nostoc genus (Fig. A2), and in addition, members of order Nostocales (Calothrix sp.; Scytonema sp.; Tolypothrix sp.; Trichormus sp.) and order Oscillatoriales (Leptolyngbya sp.; Microcoleus sp.; Oscillatoria sp.; Phormidium sp.; Tychonema sp.) rarely appeared.

Molecular phylogenetic analyses and fingerprinting
In the phylogenetic analysis of the isolated and peptide-producing Nostoc strains, 16S rRNA gene sequences of our isolates (35) and reference sequences of different Nostoc and Nostoc-like morphospecies (14) were involved (Fig. 3). Our strains formed three different lineages in the phylogenetic tree (supported with high bootstrap values) separately from the reference strains of Nostoc, Aliinostoc and Desmonostoc, and many of them were only distantly related with sequences of Nostoclike strains available in the GenBank database (95-99% pairwise sequence similarity values, data not shown). No clear affiliation of these groups with the sampling site nor with the metabolic profile could be observed. Additionally, genetic heterogeneity of the isolated strains was examined with DNA fingerprint method. Short tandemly repeated repetitive (STRR) and enterobacterial repetitive intergenic consensus (ERIC) fingerprinting differentiated the isolated strains but no correlation with the result of the phylogenetic analysis were observed.

Identification of peptides and comparative analysis of Nostoc-like strains
The peptide patterns of the Nostoc-like isolates were examined using HPLC-ESI-MS/MS, and the identified compounds are summarized in Table 2. Forty-one peptides were identified from 45 isolated Nostoc-like strains, and these compounds belong to 4 different peptide classes.
Twelve compounds were identified as NG/MG variants. These 4-6 amino acid long linear peptides are characterized by an N-terminal octanoic acid derivative (NGs) or decanoic acid derivative (MGs).
Compounds 6 and 8 were identified as NG BN 741 [49] (Fig. 4) and MG SD 755 [50]. Compounds 9 and 11 showed an identical fragmentation pattern to the previous ones, so the 16 Da mass difference in the protonated molecular mass indicates the presence of a hydroxy-tyrosine amino acid at the last position. Compounds 10 and 12 were characterized as analogues of NG BN 741 with amino-hydroxy-decanoic acid (Ahda) at position 1, and in the case of compound 12, position 5 was occupied by hydroxy-tyrosine. Structures of compounds 4, 5, and 7 were derived from NG BN 741. All these variants incorporated phenylalanine at position 5 but position 1 was occupied by amino-hydroxyoctanoic acid (Ahoa), methyl-Ahoa, and Ahda, respectively. Compound 3 exhibited a 16 Da mass decrease compared to compound 4, which can be explained with a substitution from valine to proline at position 2 and an exchange from methyl-isoleucine to isoleucine at position 3. Compounds 1 and 2 were identified as the shortened variants of compounds 8 and 10; in both cases the last tyrosine unit is absent (Tables 2 and A1).
Sixteen ABP variants were present in our samples, which can be divided into 3 groups based on the structure of the side chain (positon 1). In the cases of seven ABPs (13, 14, 16, 18, 21, 22 and 24) isoleucine was present in position 1. Compounds 13 and 14 were identified as schizopeptin (SCP) 791 and ABP 807 [51]. The structure of compound 16 was deduced from MS 3 data according to the fragmentation scheme described by Mayumi et al. [61], and the Ile-CO-[Lys-Val-Hph-MeAsn-Hph] amino acid sequence was postulated. The remaining four ABPs were identified as the congeners of 16. Homophenylalanine at position 4 was found to be methylated in 18 and 22, and ethylated in 21 and 24. The 16 Da mass difference between 18 and 22 indicated an amino acid exchange from homophenylalanine to homotyrosine at position 6, and the 14 Da mass increase of 24 compared to 21 suggested a change from valine to isoleucine at position 3. In four ABPs phenylalanine was found in position 1. Compounds 17, 19, and 20 were previously described by Sanz et al. [52] and 15 was characterized as a de-methylated form of 17 (alanine at position 5). The remaining five compounds contained tyrosine as a side chain, and four of these were previously described. Compound 23 showed a similar ring fragmentation pattern to SCP 791 with a 26 Da mass difference, which comes from an exchange from phenylalanine to methyl-homophenylalanine at position 6. Compounds 25-28 showed very similar fragmentation patterns and were identified as ABP KB899 [53], ABP MM913 [54], ABP 915 [55] and ABP G [56]  were only partially defined due to insufficient data. According to the MS/MS product ions, modifications (addition or loss of methyl-, ethyl-, propyl-group) which can be observed in these compounds affected the glucose and/or the hexanoic acid unit in all cases (Tables 2 and A3).
Compound 41 was separated from the previous classes because of its very different MS/MS spectra. Although the MS/MS data set was very poor, sufficient information was obtained for the structural determination and on the basis of the data published by Liaimer et al. [62], 41 was identified as nostopeptolide A1/A3 (Fig. 4, Tables 2 and A4).  Table 1 List of the sampling sites and typical habitats and characteristic vascular plant species present on the studied sites.

Distribution of oligopeptides and determination of chemotypes
Only 37% of our isolated Nostoc-like strains produced some of the peptide metabolites we tested. These strains showed distinct chemotypes according to their peptide patterns. The strains can be divided into 4 groups based on their natural product pattern. Strains either contained: (1) NGs/MGs, (2) ABPs, (3) ABPs and BNSs or (4) BNSs as major compounds (Fig. 5). While strains containing NG/MG or ABP only formed compact clusters, strains containing BNSs only could be divided into two subgroups, according to BNS subtypes: strains either contained BNS A and its congeners or BNS B and its variants (Fig. 5). Despite the fact that ABPs were present in several strains, the total intensity of these peptides was only about half of that of the NGs/MGs or BNSs. Furthermore, NGs/MGs were present in relatively few samples, but their total intensity was almost the same as that of BNSs. Although different phylogenetic clusters were defined, no correlation was found among the chemotypes and the genotypes.

Phylogeny in the context of chemodiversity
Species identification is very difficult in the case of Nostoc-like genera, due to the limited morphological taxonomic characters and the morphological changes that occur during cultivation [63,64]. The introduction of genetic markers has raised more and more problems, leading to multiple strains being renamed several times, and resulting in the frequent use of "Nostoc sp." as identification. For example, Nostoc sp. strain PCC 7120 was originally named as Nostoc muscorum, then was moved to the Anabaena genus by Rippka et al. [6], and later, based on molecular data (DNA-DNA hybridization [65] and repetitive (STRR) DNA sequences [66]), it was renamed as Nostoc sp. strain PCC 7120. As a result of the polyphasic approach, heterogeneity within the Nostoc genus was reported in different studies [67][68][69] and clusters of Nostoclike cyanobacteria was reported later [70][71][72][73]. Desmonostoc [71] and Aliinostoc [73] represent phylogenetically distant clades, although   based on morphological characters these genera are nearly indistinguishable from the members of Nostoc sensu stricto due to the presence of huge amount of morphological plasticity within Nostoc-like taxa [93]. 16S rRNA gene sequence analysis revealed that our strains are related to genera Nostoc, Aliinostoc and Desmonostoc but formed distinct lineages from them, which further extended the cryptic diversity within Nostoc. Solely based on 16S rRNA gene-based phylogeny, our strains are supposed to belong to a new genus (or several new genera), but the formal description is not given here, since this study focused on the produced peptides not on taxonomy.
Although our strains can be divided into multiple phylogenetic groups, these groups were not formed reflecting geographical origin or metabolic properties. However, regarding the diversity of the produced peptide type, the main phylogenetic groups showed differences. One group contained only ABP and BNS/SUO producing strains, while another was formed by strains able to produce ABP, NG/MG (and/) or BNS/SUO. In these two clusters only 15% of the peptide-producing strains were able to produce two different types simultaneously. Almost half (46%) of the sequenced peptide-producing strains belonged to a third phylogenetic group and many of them were able to produce several peptide types, the strain which produced all four types also belonged to this group. The observation that strains did not clustered on the phylogenetic tree based on their isolation source nor according their chemotype can be explained by the fact that the dispersal of several free-living microorganisms is not limited [74]. Similar findings regarding geographic distribution have been published earlier in the case of Microcystis and freshwater Synechococcus [75,76]. The production of different cyanobacterial oligopeptides has been extensively studied worldwide [23,52,[77][78][79][80][81][82], and the lack of correlation between chemotypes and genotypes was found in a similar study on Microcystis in Spanish reservoirs [82]. No connection was found between the result of the fingerprinting and the phylogenetic analysis. In addition, the result of the STRR and ERIC fingerprinting showed no strict correlation with  Table 2 for details). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) the origin of the strains nor the produced peptide type metabolites, but most of the ABP producer strains clustered together (Fig. A3).

Chemical diversity of Nostoc-like isolates
A little more than a third of our isolates produced some kind of peptide-type metabolite(s). In this study, 41 components were identified, which were classified into 4 peptide classes. Twelve NG/MG, 16 ABP, 12 BNS/SUO and 1 NPL variants were found in the extracts of our strains ( Table 2). Despite the fact that MCY and nodularin producing Nostoc species have been described in many cases [83][84][85][86][87] neither MCY nor nodularin variants were detected in the studied strains. Interestingly, our strains showed high similarity to Nostoc strains from Gen-Bank, such as TAU-MAC 0799 or TAU-MAC 0899, which also did not produce any MCY variants [88]. We could not detect a connection between the sample location and the produced peptide metabolites. However, different chemotype groups can be distinguished at the level of individual strains, based on their peptide metabolite composition and content (Figs. 5 and 6).
Two known and -to the best of our knowledge -10 new NG/MG variants were identified in our strains ( Table 2). NGs/MGs are linear nonribosomal peptides with an α-hydroxy-β-amino derivate of octanoic or decanoic acid. The presence of tyrosine or phenylalanine amino acids at the C-terminal end is common. The number of known variants can be around 40-50 and still growing. These peptides are inhibitors of zinc metalloproteases and aminoproteinases and may be important candidates for the treatment of hypertension [89].
Nine previously described and -to the best of our knowledge -7 new ABP variants were found in our analysis ( Table 2). These hexapeptides are characterized by the formation of a ring structure between amino acids at position 2 and position 6. The one-amino acid side chain is attached to the lysine at position 2 via an ureido group. All the amino acid positions except for 2 are variable, and homo-amino acids and methylated amino acids are characteristic at positions 4 and 5 [16]. The numbers of known ABP variants has risen sharply in the last 15 years; Welker and von Döhren [16] [91], and in some cases calmodulinactivated brain phosphodiesterase and sarcoplasmic reticulum Ca 2+ -ATPase inhibitory effects have been reported [92,93]. ABP B-and ABP 906-triggered relaxing activity has been detected of norepinephrineinduced contractions of rat aortic preparations [94].
The BNS/SUO is a little known group of cyanobacterial secondary metabolites ( Table 2). These glycopeptides are structurally similar to aeruginosins, as they contain an arginine derivative (Aaep) at position 4 [16]. The characteristic bicyclononane moiety in position 3 is linked to an α-glucose which seems to be the only variable part. Positions 1 and 2 are occupied by methylglyceric acid sulfate and leucine [57,58]. The formerly known variants were found to be trypsin and thrombin inhibitors [57]. BNSs were present in many of our strains, which showed very high intensity in some cases, and a number of previously unknown BNS variants have been identified. In a germination test, Sonkoly et al. [44] showed that the BNS-rich extract of BGSD.2012 strain influenced (inhibited) the germination of characteristic species of alkali grasslands.
In cyanobacteria, NPLs were identified from the terrestrial strain Nostoc sp. GSV 224 by Golakoti et al. [60]. In the last few years, several new NPL variants and structurally similar nostoweipeptin variants have been identified, and they often occur in symbiotic Nostoc strains [62,95]. Each of these cyclic peptides is characterized by non-proteinogenic amino acids such as 4-methylproline or 4-hydroxyproline. Liaimer et al. [62,96] found that nostopeptolides and other polyketide compounds are involved in the life cycle regulation and transition from motile stage to vegetative growth of Nostoc punctiforme. Moreover NPLs and nostoweipeptins were identified as antitoxins against MCYs; they inhibited the MCY-induced apoptosis of HEK293 hepatocytes by blocking the organic anion-transporters OATP1B1/B3 which are responsible for MCY uptake [95].
The exact function of many of these metabolites is still unknown. During the examination of symbiotic Nostoc species, Liaimer et al. [11] found that some metabolites (e.g. MCYs) may have a role in the development of symbiotic relationships. In addition, they suggested the possibility that certain peptides may play an important role in communication, nutrient binding, or protection (due to their antibiotic activity), as they have appeared in a more significant amount in the medium than intracellularly [11]. These substances can have a significant impact on the species composition of natural communities [44]. On the other hand, these and similar cyanobacterial metabolites may have therapeutic significance [97][98][99][100][101]. According to our results, metabolite content of some Nostoc species can vary and present at high concentration values, which is consistent with the results of other experiments [11,52]. Comparing the data with our previous study [102] it can be seen that the studied terrestrial Nostoc spp. contained less diverse and fewer amount of peptide metabolites compared to other freshwater species (e.g. Microcystis sp.).
Although there are no exact comparative studies between terrestrial and a pelagic cyanobacteria (from the same genus) about the peptide producing ability/peptide content neither in the Nostoc nor in the other genus, our results suggest that these studied terrestrial Nostoc spp. contained less diverse and fewer amount of peptide metabolites compared to other freshwater species (e.g. Microcystis sp., Planktothrix sp.) [103][104][105][106].
As stated above, the exact ecophysiological function of these peptides is not known, but it is worth emphasizing that our isolates are mostly derived from characteristic plant communities. Given that the cyanotoxin-plant interaction in freshwater has been intensively researched in recent decades [107][108][109][110][111], and some preliminary findings [44] indicate that some terrestrial plant species are sensitive to some bioactive peptides, this raises the possibility of their role as an allelopathic metabolite at terrestrial ecosystems also. However, it cannot be excluded that these bioactive peptides play a role in control of consumers, parasites and/or in intra-and interspecific competition, as is believed for freshwater ecosystems.

Conclusions
In this study, we demonstrated the peptide metabolite-producing ability of terrestrial nitrogen-fixing cyanobacteria from different sites of the special alkaline habitats of Hungary. The isolated Nostoc-like strains could be classified into different chemotype groups based on their metabolic pattern. A total of 41 peptide-type metabolites were identified which belonged to 4 different peptide families. None of the wellstudied cyanotoxins (e.g. MCYs) were found, and many previously unknown biologically active peptide metabolites were identified. Several analyses, in which cyanobacterial strains collected from environmental or bloom/mats samples were targeted, showed negative results if focusing on "just" the well-known toxin families such as like MCYs, nodularins, anatoxins, saxitoxins and β-N-methylamino-L-alanine variants. However other studies emphasized that field and laboratory bioactivity/toxicity cannot be explained (just) with the mentioned cyanotoxins and encourage to get to know new agents. The exact role of these lesser-known or unknown metabolites remains to be clarified, and they can be significant from the medical point of view also, as they may be important in the development of potential pharmaceutically active ingredients. Our results also draw attention to the use of untargeted test methods. Among our achievements, we highlighted the significance of banyaside producers within the Nostocaceae family which were represented in high number in our investigated area and moreover appeared as a prominent chemotype. These glycopeptides which are structurally similar to aeruginosins and found to be protease inhibitors are much more water-soluble than several other peptides because of the sugar moiety. On the basis of the occurrence and the intensity, this peptide family needs more research and further investigation in the future.

Sample collection and initial sample processing
Terrestrial cyanobacterial samples were collected from different parts (13 sites) of the Great Hungarian Plain in 2013 ( Fig. 2 and Table 1). Dry samples were washed 3 times with sterile distilled water and incubated in nitrate-free Allen medium [112] at 26°C under continuous illumination (100 lx m −2 s −1 ). After 1 week, samples were spread onto 1.5% w/v agar-containing nitrate-free Allen plates and incubated for an additional week. Morphologically different, separate colonies were transferred into 96 wells microplates, each well containing 200 μL nitrate-free Allen medium. Colonies were streaked on plates 3 times and macro-/microscopic heterogeneity were checked. Unialgal isolates were cultured in 1 L nitrate-free Allen containing Erlenmeyer flasks for 2-3 weeks; cells were then harvested by centrifugation (4500g; J-10 rotor of Beckman Avanti J-25, Beckman Instruments, Palo Alto, CA, USA) and lyophilized (Christ Alpha 1-2 LD plus, Martin Christ Gefriertrocknungsanlagen GmbH, Osterode am Harz; Germany). 25 mg of each sample was extracted with 80% v/v aqueous methanol. After centrifugation (21000g; J-18 rotor of Beckman Avanti J-25; Beckman Instruments, Palo Alto, CA, USA) the supernatant was analyzed by HPLC-ESI-MS/MS in positive ion mode.

Statistics
Metabolite abundance data were filtered (values below the threshold value of 1.0 × 10^2 were discarded), and log10 transformed.
The transformed data were used to create a distance matrix using canberra`distance in R 3.6.0, which was used to hierarchical cluster the data using`Ward.D2`method in both directions -among samples as well as among metabolites [114].

Phylogenetic analysis
For phylogenetic analysis, DNA extracts were made from the strains. Dried cells (ca. 10 mg) were incubated at 65°C with lysis buffer containing 100 mM TrisHCl (pH 8.0), 1.5 M NaCl, 50 mM EDTA (pH 8.0), 1% w/v SDS. After centrifugation (21000g; 5 min) the aqueous phase was mixed in 1:1 with TE saturated phenol three times. The aqueous phase was mixed in 1:1 with chloroform and after centrifugation was transferred into clean Eppendorf tubes. DNA was precipitated 1:3 with 96% chilled-ethanol and stored at −20°C for at least 1 h. After centrifugation (21000g, 20 min, 4°C) the DNA pellet was washed with 70% v/v chilled ethanol and dried in vacuum. The dried pellet was dissolved in nuclease-free water and stored at −20°C. DNA concentration was spectrophotometrically measured.
For the PCR, 2× DreamTaq green PCR Master Mix (containing DreamTaq DNA polymerase in DreamTaq Green buffer, dATP, dCTP, dGTP and dTTP, 0.4 mM each, and 4 mM MgCl 2 ) and 16S rRNA gene primers CYA359F (GGG GAA TYT TCC GCA ATG GG) [115] and 1492R (TAC GGY TAC CTT GTT ACG AC) [116] were used. A total reaction volume of 25 μL contained 50 ng of DNA and 0.1 mM of each primer. The cycles were as follows: initial denaturation at 98°C for 3 min; 32 cycles of 94°C for 30 s, 52°C for 45 s, and 72°C for 1 min; and a final extension at 72°C for 10 min. Amplification was verified by agarose gel electrophoresis.
PCR products were purified with Bio Basic EZ-10 Spin Column PCR Purification Kit. Sequencing was carried out by Base Clear B.V. (Leiden, The Netherlands). Sequences were checked with the Chromas 2.6.4 software for the errors of automatic base-calling, then the sequences were subjected to an online similarity search by BLAST at the GenBank of the National Center for Biotechnology Information web site. The closest related reference strains (authentic strains, if that was possible) for phylogenetic analysis were mainly selected based on the data available on taxonomically accepted species names at the AlgaeBase website [117]. Sequences were aligned with the SINA Alignment Service [118]. Phylogenetic analysis (including the search for the best-fit model) was done using MEGA7 software [119]. Obtained sequences were submitted to GenBank under the accession numbers MN643555-MN643589.

Fingerprinting of the isolated strains
For the fingerprint analysis PCR with primers corresponding to the short tandemly repeated repetitive (STRR) and enterobacterial repetitive intergenic consensus (ERIC) sequences were made [120]. For the STRR primers, the cycles were as follows: initial denaturation at 95°C for 6 min; 30 cycles of 94°C for 1 min, 56°C for 1 min, and 65°C for 5 min; and a final extension at 65°C for 16 min.
For the ERIC primers, the cycles were as follows: initial denaturation at 95°C for 7 min; 30 cycles of 94°C for 1 min, 52°C for 1 min, and at 65°C for 8 min; and a final extension at 65°C for 16 min [121]. After the reaction, 5 μl of amplified DNA was separated on agarose gels, stained with ethidium bromide and recorded with UVITECH Gel Documentation System (Cambridge, UK). Gel photos were analyzed with UVI-BandMap version 11.9 software.

Declaration of competing interest
The authors declare no conflict of interest. No conflicts, informed consent, or human or animal rights are applicable to this study.

Acknowledgments
This work was supported by the Hungarian scientific grants NKFIH K119647 to GV and co-supported by the EFOP-3.6.1-16-2016-00022 and GINOP-2.3.3-15-2016-00021 projects BD and OV were supported by the Bolyai János Research Scholarship of the Hungarian Academy of Sciences and COST Action ES1105 (CYANOCOST) and ES1408 (EUALGAE)

Table A1
Product ion data for the nostoginin/microginin type peptides. Masses are given in Dalton rounded to the nearest integer. X1′ indicates core fragment of Ahoa/Ahda after abstraction of the side chain (m = 58 Da). Abbreviation: NG -nostoginin; MG -microginin; nd -not detected.    Table A4 Product ion data of nostopeptolide A1/A3. Masses are given in Dalton rounded to the nearest integer.