Genetic divergence among toxic and non-toxic cyanobacteria of the dry zone of Sri Lanka

Cyanobacteria, previously known as Blue Green Algae (BGA) are oxygenic photosynthetic gram negative prokaryotes (Whitton and Potts 2000; Olson 2006). They are widespread, extremely adaptable and successful group, colonizing in diverse ecosystems (Whitton 2012). Further, they exhibit the most diverse and complex morphology among all prokaryotic groups. Their external gross appearance depends on their external environment and is often unicellular, colonial and multicellular filamentous forms, with colours ranging from dark green, blue-green, yellow, brown to black, and rarely red (Kulasooriya 2011). Traditionally, cyanobacterial classification has been based on morphological characters, which can vary in different environmental or growth conditions and even can lose during cultivation (Castenholz and Waterbury 1989; Zapomělová et al. 2008; Hasler et al. 2011). Further, certain species cannot grow in the laboratory conditions (uncultured cyanobacteria) (Ward et al. 1995) and therefore it is difficult to classify cyanobacteria into correct taxonomic groups in botanical nomenclature. However, with the advent of new molecular biological tools and along with morphology, cyanobacteria is now being classified into five orders namely; Chroococcales, Pleurocapsales, Oscillatoriales, Nostocales and Stigonematales (Castenholz and Waterbury 1989; Castenholz 1992, 2001) which are similar to the Bergey’s five subsections (I–V) described by Stanier et al. in 1978 (Stanier et al. 1978; Rippka 1988).

Phylogeny of cyanobacteria represents their evolutionary development and history with time. Molecular data have become a popular tool for phylogenetic analysis of the cyanobacteria (Kulasooriya 2011). A number of genetic marker genes have been utilized or phylogenetic analyses of cyanobacteria including 16S rRNA (Neilan et al. 1997; Wilmotte and Herdman 2001; Wu et al. 2011), the phycocyanin operon (Neilan et al. 1995, Robertson et al. 2001, Premanandh et al. 2006), randomly amplified polymorphic DNA (RAPD) PCR (Neilan 1995), the internal transcribed spacer region (ITS) of the 16S-23S rRNA (West and Adams 1997; Boyer et al. 2001), nif genes (Henson et al. 2004) and rpoC gene (Wilson et al. 2000; Wu et al. 2011). Among them, 16S rRNA gene sequence is the most widely applied strategy for assessing cyanobacterial biodiversity in nature (Ludwig and Schleifer 1994; Ercolini 2004). Additionally, for broad phylogenetic studies, sequence data from the 16S rRNA gene are the most commonly used housekeeping genetic marker due to a number of reasons. These reasons include its existence in all bacteria (often existing as a “multigene family” or as “operons”), the function of the 16S rRNA gene has remained constant over the time and therefore random sequence changes are a more accurate measure of evolution, and the 16S rRNA gene is large enough for informatics purposes (Patel 2001). DNA sequence information for the small subunit rRNA gene acquired from cyanobacterial culture has used to investigate the presence of cyanobacteria and their abundance in environment (Rudi et al. 2000). Furthermore, the comparative analyses of 16S rRNA gene sequences provide a novel approach to investigate the difference between strain collections and natural communities (Weller et al. 1991; Ferris et al. 1996). As sequences of 16S rRNA genes are growth independent can easily be captured by PCR from small amounts of DNA extracted from laboratory cultures or natural environments (Giovannoni 1991; Rajendhran and Gunasekaran, 2011). Cyanobacterial specific primers have been developed for the 16S rRNA gene (Nübel et al. 1997; Lepère et al. 2000; Valério et al. 2009).

Cyanobacteria are well acknowledged for their ability to produce potent toxins which have been accountable for numerous livestock and human poisonings (Chorus and Bartram 1999; Kuiper-Goodman and Fitzgerald 1999; Codd et al. 2005; Dittmann and Wiegand 2006). Among them, microcystins (MCs) and cylindrospermopsins (CYNs) are considered as most predominant and potent cyanotoxins in fresh waters that cause acute and chronic illnesses in animals and humans. They are well documented for hepatotoxicity, cytotoxicity and potential carcinogenicity, preliminarily affecting to liver and kidney functions (Chorus and Bartram 1999; Codd et al. 1999; Stewart et al. 2006).

Sri Lanka has a wide range of topographic and climatic discrepancies which leads the occurrence of diverse ecological niches providing excellent growth conditions for varied cyanobacteria. Cyanobacterial blooms have been recorded from different parts of the country (Jayatissa et al. 2006; Magana-Arachchi et al. 2009) which could be due to intensive agricultural practices (pesticides and fertilizers), domestic and industrial effluents, cattle domestication and inadequate management of watersheds (Wimalawansa and Wimalawansa 2014, 2015). It was evident that since early eighties, a dense dark green bloom of Spirulina sp. and Microcystis aeruginosa co-exists in the Beira Lake (6.9294°N, 79.85 42°E). Later, it was recorded that Beira Lake was nourished with other cyanobacterial species such as Synechocystis, Lyngbya, Anabaena, Synechococcus, Limnothrix, Calothrix, Pseudanabaena and Leptolyngbya along with potentially toxic M. aeruginosa (Magana-Arachchi et al. 2011). Further, Microcystis spp has been recorded from Kotmale reservoir (7°03′39″N, 80°35′50″E) in 1991. The bloom had a serious threat on operational activities but disappeared gradually with the influx of water during the following rainy season. The outbreak of M. aeruginosa in the Kotmale reservoir was an indication of Nitrogen (N) and Phosphorus (P) loading into the reservoir (Silva and Wijeyaratne 1999). Also a scum of Anabaena and Aphanizomenon was recorded from Parakrama Samudra reservoir (7.8996°N, 80.9705°E) in 1993 and disappeared with the release of water (Silva and Wijeyaratne 1999). Reports from Silva and Samaradiwakara (2005) showed the presence of toxigenic cyanobacteria in the Kandy Lake (7.2912°N, 80.6421°E) and their dominance in most of the reservoirs of the Mahaweli river basin (Silva and Wijeyaratne 1999). Further, Silva (2003) and Liyanage et al. (2010) recorded the occurrence of toxic M. aeruginosa in Kandy Lake. In the dry zone, C. raciborskii was recorded as the dominant cyanobacterial species in Kala wewa, Tissa wewa, Nuwara wewa and Jaya ganga in Anuradhapura District. Aphanizomenon spp., Merismopedia spp., Chroococcus spp., which were comparatively moderate to low in numbers were also observed in all the four tanks along with M. aeruginosa, Anabaena spp., Phormidium spp., Microcystis spp., Limnothrix spp., Pseudanabaena spp., Arthrospira spp., Gloeocapsa spp. and Planktothrix spp. (Magana-Arachchi and Liyanage 2012). This C. raciborskii dominancy was also observed by Zoysa and Weerasinghe (2016) in Nuwara wewa and Tissa wewa reservoirs. Sethunge and Manage (2010a, b) recorded the dominancy of toxic bloom forming M. aeruginosa in the dry zone reservoirs in the country. Further, C. raciborskii was recorded from Lake Gregory (6.9563°N, 80.7803°E) which is located in the upland wet zone during the driest period in 2011 (Perera et al. 2012). Though there were number of wide spread morphological incidences of cyano-blooms recorded in Sri Lanka, the availability of molecular data for this episode is still narrow. Further, the dry zone of Sri Lanka has a vast number of reservoirs which are actively or passively serve as drinking water reservoirs providing water for urban and rural communities. Also these reservoirs are used for irrigation and other recreational activities and are also of an aesthetic value. As a developing country, Sri Lanka is gravely concerned on increasing reports of serious health problems resulting from unsafe drinking water especially in the dry zone of Sri Lanka (Dissanayake and Weerasooriya 1987; Dissanayake 1996, 2005; Rango et al. 2015). Therefore, it is important to investigate the water quality from cyanobacterial aspect that could deteriorate water quality. However, these dry zone reservoirs have not been genetically assessed either for any cyanobacterial diversity or cyanotoxin producers.

Therefore, in this study, we attempted to employ partial sequences of 16S rRNA gene to explore genetic divergence and phylogenetic relationships among cyanobacteria isolated from reservoirs and well waters in the dry zone of Sri Lanka and also to identify possible toxin producers that promote potential health impacts to human and other water dependents. For that, reservoirs and well waters in dry zone were selected and isolated 45 cyanobacterial isolates and their level of diversity were assed to the classical five orders of cyanobacteria identified previously (Castenholz and Waterbury 1989). Further, attempted to determine the phylogenetic position of uncultured cyanobacterial isolates was also described.

According to the climatic distribution in the country, the Anuradhapura, Kurunegala and Girandurukotte area of Badulla districts belong to the dry zone of Sri Lanka. These localities have a number of reservoirs which are used for drinking, irrigation and other recreational activities and are also of an aesthetic value.

Assigning uncultured cyanobacteria into particular genera either by morphology or biochemically and understanding their toxicity is a difficult task. Thus the study utilized molecular tools to overcome the above constrains. The phylogenetic relationship arose from 16S rRNA gene sequence comparison supported the traditional classification of cyanobacteria which was based on morphological characters. All eleven clusters clearly demonstrated five cyanobacterial orders with few exceptions and therefore it showed the suitability of 16S rRNA gene for taxonomic differentiation. This was also supported in the work done by Nelissen et al. (1992), Li et al. (2001), Wilmotte and Herdman (2001). According to the results, out of 45 isolates, 24 uncultured cyanobacteria were able to place on their taxonomic positions up to order level with 16S rRNA sequences. Among them, 13 isolates were positioned in cluster A which composed of Chroococcales cyanobacteria and further to Synechococcus cyanobacteria. Uncultured cyanobacterium AN2, GK14, GK15, and AT3 gathered with toxic Synechococcus sp. with >98% genetic similarity while sub cluster 1a, 2a and 3a clustered with toxic M. aeruginosa PCC 7806 (Tillett et al. 2001) with >90% sequence similarity. Therefore, Nuwara wewa (AN2), Tissa wewa (AT3, AT2 and AT4), Nika wewa (KN1) and well waters in Girandurukotte (GK9, GK3, GK6, GK2, GK7, GK8, GK5, GK14, and GK15) contained potential MC producing cyanobacterial strains which belonged to order Chroococcales.

Though Radiocystis species were also known to produce MC (Vieira et al. 2003; Lombardo et al. 2006), no sequence of MC producing Radiocystis strain has been deposited in the NCBI database and therefore toxin producing ability of Radiocystis could not be determined. However, Radiocystis sp. A2 formed a tight cluster (99% similarity) with Radiocystis sp. JJ30-3 in which the toxicity was unknown. Therefore, Radiocystis sp. A2 which was isolated from a Nuwara wewa reservoir may or may not have the genetic potential to produce MC. Similarly, Chroococcidiopsis sp. SAG 2025 in which the toxicity unknown clustered with GK11 and HM1 with high BT support (98%). Further, Leptolyngbya, Phormidium and Oscillatoria were polyphyletic and this polyphyletic nature of order Oscillatoriales was also reported previously by other studies (Nelissen et al. 1992; Garcia-Pichel et al. 2001). Similarly, Chroococcales cyanobacterial isolates (in cluster A, J and K) also reflected polyphyletic nature which was also been reported by the previous authors (Nelissen et al. 1992; Garcia-Pichel et al. 2001).

Hapalosiphon sp. BZ-3-1 was known to produce MC (Prinsep et al. 1992) and therefore Hapalosiphon sp. HM11 and HM14 have a potential to produce MC since they shared >97% similarity with 94% BT support. Those Hapalosiphon isolates HM11 and HM14 were isolated from well waters. Interestingly, cluster I clustered all Nostocales cyanobacteria reflected a monophyletic behaviour. The monophyletic nature of the heterocysts forming cyanobacteria in order Nostocales was reported previously by Nelissen et al. (1992), Wilmotte and Herdman (2001) and Lyra et al. (2001). Considering cluster 2i, Nostoc sp. HM7 clustered with highly toxic MC producing Nostoc sp. IO-102-I (Oksanen et al. 2004) and hepatotoxic Nostoc sp. 152 (Lyra et al. 2001) sharing >97% sequence similarity and also with non-toxic Nostoc sp. CCAP 1453/28 sharing >97% similarity. Therefore, Nostoc sp. HM7 isolated from a well water in Girandurukotte may or may not have the potential to produce hepatotoxins. Raphidiopsis curvata AJ1 isolate (in cluster 4i) which was isolated from Jaya ganga reservoir, clustered with Raphidiopsis curvata CHAB1150 which is a CYN producer (Jiang et al. 2012) sharing >98% similarity and therefore Raphidiopsis curvata AJ1 has a potential to produce CYN. Moreover, all C. raciborskii spp. irrespective to their toxicity formed a tight cluster sharing 100% sequence similarity with 89% significant BT support. Among the species, C. raciborskii LEGE 051 was known to produce CYN (Moreira et al. 2011) and therefore C. raciborskii AN3 isolated from Nuwara wewa may or may not produce CYN. The high similarity of DNA sequences of Raphidiopsis and Cylindrospermopsis highlighted that the strains of these two genera were closely and phylogenetically related supporting the conclusions drawn by Stucken et al. (2010) and Wu et al. (2011). However, the isolates of Raphidiopsis and Cylindrospermopsis distributed randomly in the cluster I irrespective to their toxicity, suggesting that no clear correspondence was shown between toxicity and phylogeny. This was also supportive with the cluster 2i where both highly toxic and non-toxic Nostoc sp. IO-102-I and Nostoc sp. CCAP 1453/28 clustered sharing 100% similarity.

According to the results obtained in this study, it was evident that 16S rRNA gene is an effective tool in construing phylogenetic relationships between different genera within order level. This phenomenon was also supportive with the previous studies done by several authors. Also, the 16S rRNA gene has been useful in identifying and classifying strains that belong to a single clade (Palinska et al. 1996; Otsuka et al. 1999). For examples, a study done by Nelissen et al. (1992) using 16S rRNA sequences of five strains of Pseudanabaena, concluded that they were nearly identical, and hence confirmed them as a single monophyletic taxon. Further, a phylogeny study by Honda et al. 1999 using 16S rRNA sequence data, clustered members of the genus Synechococcus and concluded that the strains were closely related.

Though 16S rRNA sequence analysis is relatively useful in determining evolutionary relationships among organisms at the genus level; it is not that helpful in differentiating at species or sub species levels due to the fact that 16S molecules is almost constant (with the total variation of about 200 bp for a mean length of 1500 bp (Rainey et al. 1996; Moreira and Philippe 2010) and thus different genes cannot be easily separated by size. Additionally, even though the 16S rDNA sequence consists of hyper variable and extremely informative regions for close relationships, it is often not divergent adequately to give good separation in close relations (Fox et al. 1992; Normand et al. 1996; Turner 1997). Further, in comparison to most of the protein-encoding genes, with time this sequence is more susceptible to changes. This is due to the fact that protein-encoding genes are more tolerable to variations in amino acid codon sequences (silent mutations in the third nucleotide in a codon) than rRNA genes. Moreover, the level of degeneracy in the 16S rRNA gene is much less than in protein-coding genes (Vinuesa et al. 1998). Due to its conservative nature, the 16S rRNA gene sequence is lacking the needed divergence for comparing species within generic clades (Bolch et al. 1999).

According to the phylogeny analysis, reservoirs and well waters in Sri Lanka consisted rich cyanobacterial diversity which also have a potential to produce cyanotoxins. Since 16S rRNA phylogeny failed to reflect a strong image on toxicity in isolated strains, it is suitable to use multi-gene analysis for a better resolution (Tanabe et al. 2007; Wu et al. 2011).

Based on 16S rRNA results obtained from the present study, it was clearly acknowledged the significant diversity in cyanobacteria with potential microcystin (MC) and cylindrospermopsin (CYN) producers within the dry zone in the country. Among 45 isolates, 20 isolates were classified as unidentified cyanobacteria and considered as novel cyanobacterial genera. Among 25 identified isolates, seven isolates were belonging to order Chroococcales, one to order Pleurocapsales, seven to order Oscillatoriales, six isolates to order Nostocales and four isolates to the order Stigonematales. Of 25 identified isolates, seven isolates were identified up to species level.

With 16S rRNA Phylogenetic results, eleven clusters clearly demonstrated five cyanobacterial orders with more than 90% similarity irrespective to their toxicity which showed the suitability of 16S rRNA gene for taxonomic differentiation. Sixteen isolates had the potential to produce MC and two isolates to produce CYN. Twenty uncultured cyanobacterial isolates were able to place on their taxonomic positions up to order level. Findings of the study confirm the rich cyanobacterial diversity and the divergence among the potential cyanotoxin producers in the dry zone water bodies and is an indication regarding the numerous health impacts faced by the people living in the dry zone of Sri Lanka.

Therefore, the study underlines a need of continuous monitoring programmes, the efficient management of water bodies and correct toxicity assessment by accurate identification of problematic cyanobacterial species is necessary for the waters used as sources of drinking water.