Cyanolichens are highly specialized, stable symbioses between heterotrophic fungi (mycobionts), mainly Ascomycota, and photosynthetic diazotrophic cyanobacteria (cyanobionts). Symbiotic genotypes of Nostoc (Nostocales, Cyanobacteria) occur in a wide variety of cyanolichens, either as the primary photobionts in bipartite lichens or as accessory photobionts together with green algae in tripartite lichens. Nostoc symbionts are especially common in Peltigeralean lichens (Peltigerales, Ascomycota), and for example all species of Collema, Leptogium, Nephroma, Peltigera, Pseudocyphellaria, and Sticta associate with Nostoc (Rikkinen, 2002, 2009). Species of Peltigera and Nephroma are common and sometimes abundant in many temperate and cool regions of the world. Species of Peltigera can be grouped into eight monophyletic sections: Polydactylon, Chloropeltigera, Peltidea, Horizontales, Peltigera, Retifoveata, Phlebia, and Hydrothyriae, with the last three groups being monotypic (Miadlikowska & Lutzoni, 2000). Mycobionts of cyanolichens tend to associate with a limited number of closely related Nostoc genotypes or genotype groups. Symbiotic strains of Nostoc group into two distinct phylogenetic lineages which associate with different groups of lichen-forming fungi. One group is found in bipartite species of Nephroma, Sticta and Pseudocyphellaria, while the other group associates with many species of Peltigera (reviewed in Rikkinen, 2013).
Microcystins and nodularins are small cyclic hepatotoxic peptides linked to the fatal intoxication of humans and livestock (Sivonen, 2009). They are the product of a complex biosynthetic pathway consisting of hybrid non-ribosomal peptide synthetase and polyketide synthase enzymes (reviewed in Dittmann et al., 2013). A few free-living strains of Nostoc are known to produce microcystins in aquatic environments (e.g. Sivonen et al., 1990) and also symbiotic Nostoc strains from lichens and cycads were shown to encode this pathway and produce microcystins and nodularins (Oksanen et al., 2004; Kaasalainen et al., 2009, 2012; Gehringer et al., 2012). Kaasalainen et al. (2012) found microcystins in 45 lichen specimens representing several species each of Peltigera, Nephroma, and Sticta, and one specimen of Lobaria (Peltigerales). However, microcystins are not produced in all cyanolichens and appear to be more common in some genera than others. Here we analyse and discuss the diversity of microcystin chemical structures in relation to the identity of the mycobiont and cyanobacterial microcystin synthetase and ribosomal RNA genes.
Materials and Methods
Eight hundred and three cyanolichen specimens from different parts of the world representing many different groups of lichen-forming fungi mainly from the order Peltigerales were collected in order to achieve a wide coverage of different cyanolichen groups, ecosystems, and geographical areas (Supporting Information Table S1). All specimens were screened for cyanobacterial toxins by amplifying a portion of the mcyE gene encoding an enzyme which participates in microcystin biosynthesis and by detecting toxins directly from lichen thalli (Kaasalainen et al., 2012).
Nonmetric multidimensional scaling (NMS) from the statistical package PC-ORD version 5.33 (McCune & Mefford, 2006) was used to analyse and produce graphical depictions of similarities in the microcystin variant spectra of lichen specimen extracts. Raw data on the distribution of toxin variants was extracted from table S2 of Kaasalainen et al. (2012). NMS is a nonparametric ordination technique and well suited to data that are non-normal, are on discontinuous scales, and contain a large proportion of zero values. The raw data matrix included 45 lichen specimens and 52 microcystin chemical variants. In order to reduce noise, rare microcystin variants and the lichen specimens from which nodularins or only one microcystin variant was detected were removed (lichen specimens 2, 5, 11–21, 47, 86, 90, 91, and 97 in table S2 of Kaasalainen et al., 2012) from the data set, resulting in a matrix of 34 sample units and 18 chemical variants. The NMS was performed in the ‘slow and thorough’ mode using the Correlation distance measure, 500 iterations, and random starting coordinates. A two-dimensional ordination of the data matrix was produced after determining that higher dimensional solutions did not substantially reduce stress. A randomization test showed that the observed minimum stress was clearly smaller than that expected by chance. Microcystin variants were superimposed on the resulting ordination using a joint plot, based on the correlations (r) of these variables with the two axes of the ordination. For visual clarity, the ordination was rigidly rotated to load one major microcystin variant ([Asp3, ADMAdda5]MC-LR) on the vertical axis. Variance explained was expressed by the coefficient of determination between distances in the ordination space and distances in the original microcystin variant space.
DNA amplification and sequencing and phylogenetic analyses
The mycobiont internal transcribed spacer (ITS) was amplified from the same DNA extractions used for screening for the cyanobiont mcyE gene (Kaasalainen et al., 2012), and then sequenced as described in Fedrowitz et al. (2012). All sequence data was submitted to NCBI GenBank (Supporting Information Table S2). Phylogenetic analyses were performed with MrBayes v3.1.2 (Huelsenbeck & Ronquist, 2001) using methods that are described in Supporting Information Methods S1.
Results and Discussion
The lichen specimens which contained microcystins or nodularins represented different species of Peltigera, Nephroma, and Sticta, and Lobaria (Fig. 1). Our results show that the distribution of toxin-producing cyanobacteria in lichens is not random, but is concentrated in some taxa within the Peltigeraceae (e.g. Peltigera degenii and P. membranacea; Supporting Information Fig. S1) and Nephromataceae (e.g. Nephroma parile and N. cellulosum; Supporting Information Fig. S4). All of the lichen genera and most of the lichen species from which hepatotoxins were detected also included specimens from which no toxins were detected (Figs. 1, Supporting Information Figs S1–S4). The lichen genus Peltigera had the highest frequency of toxic cyanobacteria, with toxins detected in 11% (29 specimens) of all specimens (Fig. 1). Toxins were detected in several species representing different phylogenetic lineages within the genus, including the sections Polydactylon where 19% of the specimens contained toxins (10 specimens; Fig. S1), section Peltigera where 13% of the specimens contained toxins (17 specimens; Fig. S2), section Chloropeltigera where 5% of the specimens contained toxins (one specimen), and section Horizontales where 4% of the specimens contained toxins (one specimen; Fig. S3). In addition many specimens in these four sections were found to contain the mcyE gene but no toxins were detected. None of the 43 specimens of Peltigera section Peltidea contained toxins or the mcyE gene. The percentage of toxin containing specimens in Nephroma was 7% (11 specimens; Fig. S4), in Sticta 9% (four specimens), and in Lobaria 2% (one specimen).
None of the Leptogium and Pseudocyphellaria specimens examined contained microcystins even though a large number of specimens (nearly 100) were analysed from both genera. Also, seven specimens of Stereocaulon (Lecanorales) were analysed, but neither the mcyE gene nor toxins were detected. In some cases the relatively small number of specimens available for analysis may have been the primary reason for not detecting the mcyE gene or microcystins from a particular lichen genus. However, in the case of Leptogium, Pseudocyphellaria, and Peltigera section Phlebia, the large sample sizes clearly indicate that toxin-producing cyanobacteria cannot be common in these groups. Hence, the occurrence of toxin-producing cyanobionts in lichens is not random and differs significantly between different lichen genera and species (Figs. 1, Supporting Information Figs S1–S4).
The two-dimensional NMS ordination grouped the 34 lichen specimens containing > 1 microcystin variant according to similarities in their compound spectra (Fig. 2). The two ordination axes explained 81.0% of all variation in microcystin content. After rotation, 64.2% of the variation was explained by the vertical axis, aligned with amount of microcystin variant [Asp3, ADMAdda5]MC-LR in the analysed lichen specimens (r = 0.78). The vertical axis was also positively correlated with the quantity of microcystin variants [ADMAdda5]MC-RR (r = 0.61) and [Asp3, ADMAdda5]MC-RR (r = 0.52), and negatively correlated with the amount of microcystin variant [Leu1]MC-LR (r = −0.68). The horizontal axis, representing 16.8% of all variation, was positively correlated with the amount of the microcystin variant [Leu1, ADMAdda5]MC-LR (r = 0.34), and negatively correlated with microcystin variants MC-RR (r = −0.49) and [Asp3]MC-RR (r = −0.49). The genotype groups A–E in the overlay are well supported groups in the mcyE gene tree (Kaasalainen et al., 2012).
No lichen specimens grouped in the centre of the ordination, indicating clear divisions in the occurrence of the major microcystin variants (Fig. 2). All eight Nephroma parile specimens from Scotland and Norway formed as a tight cluster in the upper part of the ordination. Also six Peltigera specimens from northern Europe were placed in the upper part of the ordination, with one Peltigera praetextata specimen grouping among the cluster of Nephroma parile specimens, which all had almost identical toxin spectra. All the earlier mentioned specimens contained microcystin variants with the amino acid ADMAdda, a demethylated and acetylated variant of the amino acid Adda (3-amino-9-methoxy-2,6,8-trimethyl-10-phenyl-4,6-decadienoic acid) peculiar to microcystins and nodularins, in the fifth position of the peptide, and the Nostoc symbionts of almost all of these lichens belonged to the same mcyE genotype group (A in Fig. 2). The remaining Peltigera specimens were placed in the lower sections of the ordination with many specimens of Peltigera degenii, P. membranacea, and P. occidentalis from different parts of the Northern Hemisphere grouping closely together. These specimens had leucine in the first position of their microcystin structure instead of the much more common alanine, and they all also shared the same mcyE genotype group (B in Fig. 2), which was also identified from one P. leucophlebia specimen that differed in its spectrum of microcystin variants. The two Sticta fuliginosa specimens from California were negatively correlated with the horizontal axis and shared the same mcyE genotype (C in Fig. 2). The Nephroma cellulosum specimen from Argentina and one Peltigera specimen from China grouped closely together and were positively correlated with the horizontal axis. The Argentinean lichen had a unique Nostoc mcyE genotype (D in Fig. 2), while the Chinese specimen had a similar mcyE genotype to those of two other Peltigera specimens (E in Fig. 2).
Microcystins and nodularins are synthesized on a large hybrid nonribosomal synthetase and polyketide synthetase enzyme complex in a programmed biosynthetic event (Moffitt & Neilan, 2001; Rouhiainen et al., 2004). The chemical structure varies and there are over 100 published structural variants of microcystins with differing toxicities (Sivonen, 2009; Neffling, 2010). It is not clear at present why microcystins are synthesized by cyanobacteria but most of this chemical variation has a genetic basis and a range of mechanisms have been linked to the chemical variation reported for microcystins (Dittmann et al.,2013). The mcyE gene is responsible for the biosynthesis of Adda and the formation of the bond between Adda and D-glutamate (Rouhiainen et al., 2004), which are essential for the toxicity of the microcystin and show very little variation between microcystin variants (Sivonen, 2009). Consequently the mcyE gene is expected to be less variable than the genes coding other more variable amino acids in the microcystin molecule. This may explain how some cyanobionts with similar or even identical mcyE genotypes produce very variable toxin structures. However, the microcystin structures detected from Nephroma cellulosum specimens and one Peltigera sp. are very similar even though the mcyE genotype groups detected from the specimens are different. Further work will be required to unravel the basis for the pronounced chemical variation in microcystin chemical structure observed from lichens (Kaasalainen et al., 2012).
The distribution of microcystin and nodularin production among cyanobacteria in general and also inside individual genera is sporadic, which is thought to be explained by multiple losses of the gene cluster, even though horizontal gene transfer has been discussed as a possible mechanism (reviewed by Dittmann et al., 2013). In this study all Nephroma and Peltigera from Scotland had similar toxin structures and also their mcyE genes belonged to the same phylogenetic group. This is somewhat surprising as Peltigera species and bipartite Nephroma species generally associate with different groups of symbiotic Nostoc (Rikkinen, 2013). The 16S rRNA gene sequences confirm that also the cyanobionts of all now studied Nephroma parile, Nephroma cellulosum, and Sticta fuliginosa specimens belong to the Nostoc lineage typically found in Nephroma-guild lichens, while the Nostoc symbionts of all Peltigera specimens belong to a different group (Kaasalainen et al., 2012).
We conclude that the distribution of different microcystin variants in different cyanolichens correlates with previously known patterns of Nostoc diversity in lichens. Similar patterns have not yet been described from aquatic bloom-forming cyanobacteria where microcystin diversity has been studied in considerable detail. The similarity of the mcyE genotypes and toxin structures detected in lichens in Scotland might even indicate lateral transfer of the microcystin synthetase genes between two different groups of symbiotic cyanobacteria. However, all cyanobacterial sequences in our study were amplified directly from lichen thalli, and therefore future studies with cultured strains are essential to confirm the mechanisms behind the patterns seen in cyanotoxins and related genes in lichen-symbiotic cyanobacteria.
The study was funded by Academy of Finland Grants 122288 to J.R. and 118637 to K.S.