Identification of harmful cyanobacteria in the Sacramento-San Joaquin Delta and Clear Lake, California by DNA barcoding

Harmful cyanobacterial blooms (CyanoHABs) are a serious global concern and are often associated with odorous metabolites in drinking water and toxins in aquaculture facilities and in the environment (Mankiewicz et al. 2003; Smith et al. 2008). Different types of toxins are produced from several cyanobacterial species including hepatotoxins (microcystins), cytotoxins (cylindrospermopsin), neurotoxins (anatoxin-a, antillatoxin, saxitoxins), and dermatoxins (lyngbyatoxins). These potent toxins render serious consequences to the health of ecosystems, aquatic organisms, domestic animals, and humans upon direct contact or consumption of CyanoHAB impacted water (Mankiewicz et al. 2003; Osswald et al. 2007; Puschner et al. 2010; Acuña et al. 2012).

The current study focused on molecular analysis of cyanobacterial species from two ecosystems that are ecologically and economically important in California. The Sacramento-San Joaquin Delta is a critical water supply system in Northern California, which provides drinking water to two-thirds of the California population (more than 20 million people) and irrigates 4.5 million acres of farmlands (Jassby 2008). The estuary also provides essential habitats for many anadromous, commercial, and recreational fish such as striped bass (Morone saxatilis), Chinook salmon (Oncorhynchus tshawytscha), and several endangered fish species such as the delta smelt (Hypomesus transpacificus) (Sommer et al. 2007). Blooms of the hepatotoxin-producing cyanobacterium Microcystis aeruginosa were first recorded in the Sacramento-San Joaquin Delta in 1999, and since then, cyanobacterial blooms have re-occurred and have been monitored for biomass and toxicity (Lehman et al. 2005, 2008, 2010; Spier et al. 2010). Colonial forms of M. aeruginosa are widely distributed along the 180 km of freshwater and brackish waterways of the delta that may affect indigenous invertebrates and fishes (Lehman et al. 2005, 2008, 2010; Ger et al. 2010). Other harmful cyanobacteria such as Aphanizomenon, Dolichospermum (formerly recognized as planktic Anabaena) (Wacklin et al. 2009), and Oscillatoria have been observed in the Sacramento-San Joaquin Delta, although to a lesser extent than Microcystis (Cloern and Dufford 2005; Lehman et al. 2010; Spier et al. 2010). Because algal bloom studies in the Delta have mainly focused on M. aeruginosa (Lehman et al. 2005, 2008, 2010), the occurrence, abundance, and potential role of other toxin-producing cyanobacteria to indigenous fisheries resources are largely unknown.

Clear Lake is the largest natural lake in California and provides drinking water to local communities. The lake supports recreational activities and tourism for sport fishing and water contact sports, forming an industry greater than 50 million dollars in the local county (Goldstein and Tolsdorf 1994). Although considered “impaired” in terms of hyper-eutrophication from phosphorus and sulfate overload from anthropogenic activities, the lake is used for storage of irrigation water for downstream agricultural lands. Land use such as construction of farmlands, road building, livestock grazing, logging, and firewood cutting have accelerated erosion resulting in large phosphorus inputs mostly from basins around the lake (Richerson et al. 1994). Cyanobacterial assemblages in the lake reached the highest densities from the mid 1970’s to 1990 (Horne 1975; Richerson et al. 1994; Winder et al. 2010) and were dominated by diazotrophic cyanobacteria such as Aphanizomenon, Dolichospermum, and the non-nitrogen fixing cyanobacterium Microcystis (Horne 1975; Richerson et al. 1994). Lyngbya (now known as Limnoraphis) blooms have also been recorded since 2009 (Mioni and Kudela 2011, Mioni et al. 2012). These cyanobacteria form scum on the water surface and deteriorate water quality (Smith et al. 2008; Mioni and Kudela 2011, Mioni et al. 2012).

Cyanobacteria are traditionally classified and identified by microscopic analysis of morphological characters such as shape and size of vegetative cells, heterocytes, akinetes, presence/absence of sheath, and morphology of terminal cell. This task is challenging even for a well-experienced taxonomist due to significant phenotypic changes that may occur in natural assemblages and morphological transformation upon cultivation in the laboratory environment (Palinska et al. 1996). Comprehensive morphological identification combined with molecular characteristics have been reported for cyanobacteria found in Nordic countries belonging to the order Nostocales such as Anabaena, Aphanizomenon, Dolichospermum, Trichormus, and Nostoc (Rajaniemi et al. 2005; Wacklin et al. 2009). Genetic relationships have been characterized among Chroococcales (Cyanobium, Synechocystis, and Synechococcus), Oscillatoriales (Leptolyngbya, Microcoleus, Phormidium, and Romeria), and Nostocales (Nostoc and Nodularia) in Portuguese estuaries (Lopes et al. 2012). These studies have greatly enriched the cyanobacterial database by linking genetic information and morphological features to facilitate species identification.

DNA barcoding is a taxonomic identification method that relies on the use of standardized species-specific DNA regions known as “barcodes” (Hebert et al. 2003). Species identification by DNA barcodes provides a rapid and specific detection tool for various organisms such as mammals (Murphy et al. 2001), birds (Khan et al. 2010), amphibians (SanMauro et al. 2005), and fish (Kochzius et al. 2010). Because each organism possesses unique gene sequences, DNA barcoding offers an accurate identification of known species and leads to the discovery of unique organisms with discrete genetic profiles. DNA barcoding has been employed for assessment of cyanobacterial assemblages (Betournay et al. 2007; Lopez-Legentil et al. 2011) and genetic diversity of diatoms and dinoflagellates (Litaker et al. 2007; Lin et al. 2009; Moniz and Kaczmarska 2010). DNA barcoding has also been used to analyze changes in bacterial community composition potentially affecting biotic interactions due to Microcystis blooms (Cheng et al. 2011).

Over the last decades, the species composition of recurring blooms in the Sacramento-San Joaquin Delta and Clear Lake has been assessed by traditional morphological taxonomy. As morphological identification is not always conclusive, molecular analysis such as sequencing of species-specific regions followed by phylogenetic analysis is a widely applied technique for obtaining precise taxonomic classification of biological specimens (Robertson et al. 2001; Casamatta et al. 2005; Rajaniemi et al. 2005; Ezhilarasi and Anand 2009; Lopes et al. 2012). Our goal in the current study is to facilitate the accurate identification of dominant cyanobacterial species from two water bodies in California impacted by seasonal CyanoHABs by traditional taxonomic identification combined with molecular techniques.

Using a combination of similarity search (BLASTN) and phylogenetic analysis, several species of cyanobacteria were identified in Clear Lake including Aphanizomenon flos-aquae, Dolichospermum (formerly Anabaena) lemmermannii, Dolichospermum spp., Limnoraphis (formerly Lyngbya) robusta, Microcystis aeruginosa, and Synechococcus spp. In contrast, fewer species of cyanobacteria were observed in the Sacramento-San Joaquin Delta that included A. flos-aquae, Dolichospermum spp., and Synechococcus spp. Other species of Microcystis were not observed in the samples examined from the Sacramento-San Joaquin Delta and Clear Lake by molecular analysis. Of particular importance is the species-specific identification of A. flos-aquae, D. lemmermannii, and Lm. robusta in Clear Lake; to date, they (especially D. lemmermannii, and Lm. robusta) have been identified only to the genus level by morphological analysis. In addition, DNA barcoding identified a small nondescript unicellular cyanobacterium as Synechococcus spp. from both Clear Lake and the Sacramento-San Joaquin Delta. Further analysis such as isolation, microscopy, and sequencing are necessary to determine their species-specific identification. In addition to cyanobacteria, other bacteria were also present in the samples examined by molecular techniques. For example, Gram-positive (i.e. Paenibacillus alvei and Bacillus pumilus) and Gram-negative (i.e. alpha proteobacteria and Flexibacteria) bacteria were also found in the bloom samples examined from Clear Lake. While these bacteria can tolerate various ranges of environmental conditions and are expected to be present in the water column, their potential role in algal blooms have not been extensively studied.

Aphanizomenon flos-aquae was the dominant species from the 2011 blooms in the Sacramento-San Joaquin Delta as determined by DNA barcoding and by morphological identification (Tables 1, 2). In addition, quantitative (q) PCR assays that we designed for A. flos-aquae and M. aeruginosa based on the obtained sequences in this study revealed more than 2 order of magnitude greater for the A. flos-aquae during the 2011 bloom season (unpublished data). While previously observed in less frequency and distribution, blooms of A. flos-aquae have not been reported in the Sacramento-San Joaquin Delta in the past decade to the best of our knowledge. Pioneering studies on algal blooms in the Delta showed M. aeruginosa as the dominant species in recurring blooms (Lehman et al. 2005, 2008, 2010). Although the cause(s) in the shifts of cyanobacterial assemblages, at least during the duration of this study, in the upper San Francisco Estuary remains unclear, continuous monitoring of dominant and potentially toxin-producing cyanobacteria including M. aeruginosa and A. flos-aquae is warranted in terms of mitigating their adverse impacts to aquatic organisms and conservation of water quality. Water temperature and other physicochemical factors have been associated with the emergence of A. flos-aquae (Cloern and Dufford 2005) and other cyanobacterial species during the 2011 blooms in the Sacramento-San Joaquin Delta (Mioni et al. 2012). A. flos-aquae is a diazotrophic cyanobacterium which produce endotoxins such as anatoxin-a, saxitoxins, and cylindrospermopsin (Sivonen and Jones 1999; Castle and Rogers 2009).

A species identification system using a combination of BLASTN search and phylogenetic analysis for 16S rDNA is a powerful method; however this approach has limitations when used to analyze taxa with nearly identical sequences or to classify unknown sequences. For example, Lm. robusta, Lm. cryptovaginata, and Ly. majuscula share identical 16S rDNA sequences (1.5 kb) with a few base pairs differences as observed from 6 clones (SWMP11-12) amplified from Clear Lake samples (Jüttner and Watson 2007; Guiry and Guiry 2012). Although the phylogenetic tree (Figure 2) implies that these clones are most closely related to Lm. cryptovaginata, SWMP11-12 is most likely Lm. robusta by virtue of their characteristic morphological features (Komárek 2003, Komárek et al. 2013; Rejmánková et al. 2011). While Ly. majuscula is also placed in the same clade, SWMP11-12 is unlikely Ly. majuscula as it is a marine species (Jones et al. 2011). Lm. robusta blooms in Lake Atitlan, Guatemala was first reported in 2008 by (Rejmánková et al. 2011). This cyanobacterium formed dense patches covering approximately 40% of Lake Atitlan surface during the peak of the blooms although cyanotoxin production from this species remains unclear (Rejmánková et al. 2011). Cylindrospermopsin and saxitoxins were detected during the Atitlan bloom in 2009, but the concentrations remained low (12 and 58 ng g^-1 from the freeze dried specimen) (Rejmánková et al. 2011; Komárek et al. 2013). Lm. robusta found in Clear Lake, California is unlikely introduced from Lake Atitlan as there are no reports documenting the introduction of this species across the two water bodies. Further analysis such as comparison of variable gene region will provide a better understanding of the relationship between the same species found in two distant locations. Initial findings showed that environmental factors such as water temperature and nutrient concentrations may affect the growth and abundance of Lm. robusta and other emerging cyanobacteria in Clear Lake during the 2011 blooms (Mioni et al. 2012).

The classification of clone SWMP11-06 in the phylogenetic tree is debatable (Figure 1). The genotype is placed in a clade of Dolichospermum, however, the reference strain A. flos-aquae strain 1tu37s13, which was morphologically identified by Rajaniemi et al. (2005), is also placed in the same group. The values of posterior probability are relatively low for the branching (< 0.70), precluding a conclusive identification of the genotype.

SWMP11-13 and -14 from Clear Lake showed similarity to Lm. robusta by BLASTN search but interestingly these sequences are distinct from Lm. robusta, Lm. cryptovaginata, or any other sequences in the NCBI-GenBank database (16S rDNA Pairwise% Identity: 96.8%). Stackebrandt and Goebel (1994) suggested a cutoff value of 97.5% (or higher) for acceptable similarity values for species identification using the 16S rDNA sequence. These two genotypes are tentatively designated as Limnoraphis sp. as placed in the phylogenetic tree; further analysis is necessary for species-specific identification (Figure 2).

Although M. aeruginosa was detected at different sampling sites and times in the Sacramento-San Joaquin Delta by microscopic observation, samples that were examined by molecular analysis did not detect this species. This result may be due to the following reasons: 1) M. aeruginosa was lacking in the samples examined for molecular analysis due to its colonial nature and heterogeneity across subsamples, and 2) inhibition by other abundant cyanobacteria precluding the amplification of M. aeruginosa in the samples that were PCR tested. It is important to note that A. flos-aquae, instead of the historically recurring M. aeruginosa, dominated the blooms in the Sacramento-San Joaquin Delta during the duration of these studies. Although the field samples that we chose for molecular analyses were based on morphologic microscopic analysis (qualitative and quantitative), the DNA fragments of the expected algal species were not obtained from the subsamples. Another potential explanation may be due to the small number of clones that were analyzed that may not represent the wide variety of cyanobacterial species present in the blooms. Analyzing more clones from appropriate field samples using emerging sequencing technologies would probably yield a greater number of sequences from potentially toxin-producing cyanobacteria with a sample such as CL3(6) from Clear Lake that showed more diverse bacterial species (Table 2). Another ideal approach is to use parallel algal samples for morphologic taxonomic identification and molecular analyses to better understand the cyanobacterial composition in the Sacramento-San Joaquin Delta and Clear Lake.

Most of the cyanobacterial sequences deposited in NCBI-GenBank database originate from geographically distant locations such as Portugal, Japan, India, and Nordic countries (Robertson et al. 2001; Rajaniemi et al. 2005; Ezhilarasi and Anand 2009; Lopes et al. 2012). Despite the distant origins of species-specific sequences, we were able to successfully identify the taxonomic classification of the clones based on the 16S rDNA. Using the 16S rDNA for classification and identification of cyanobacteria is widely accepted because 1) the gene is present in all bacterial genomes, and 2) the frequency of sequence variations and insertions in this gene serves as a molecular clock and reflects evolutional history, allowing the distinction of a broad range of taxonomic groups and identification of individual species (Casamatta et al. 2005; Janda and Abbott 2007). Although the resolution of the 16S rDNA for specific identification remains debatable due to the high degree of their sequence conservation (Janda and Abbott 2007), the gene has been a reliable barcode providing identification to the genus level, in some cases to specific species level as we have demonstrated in this study. We attempted to use the ITS region in our analysis, however, the sequences were not suitable for alignments due to their high variability when compared with the reference strains from other geographic locations (unpublished observation). Another gene involved in nitrogen fixation, nif H, has been used as an alternate barcode for classification and identification of cyanobacteria by providing better resolution for species identification (Zehr et al. 1997). However, the nif H gene is not appropriate for analyzing complex algal assemblages by DNA barcoding as non-diazotrophic cyanobacteria such as M. aeruginosa do not possess this gene in their genome.

Microscopic observation coupled with DNA barcoding effectively identified cyanobacterial species in the Sacramento-San Joaquin Delta and in Clear Lake. For the first time in Northern California, this tiered approach provided species-specific identification of dominant species in the blooms including Microcystis aeruginosa, Aphanizomenon flos-aquae, Dolichospermum (formerly Anabaena) lemmermannii, Dolichospermum spp., Limnoraphis (formerly Lyngbya) robusta, Limnoraphis spp., and Synechococcus spp. The precise identification using DNA barcoding provides two important ecological implications in these water bodies. First, we have identified A. flos-aquae as the new dominant species in the Sacramento-San Joaquin Delta during the course of this study, an apparent shift from M. aeruginosa that have dominated the recurring blooms at the delta in the past decade. Second, DNA barcoding documented the first occurrence of Lm. robusta in North America. To date, this harmful cyanobacterium has only been reported from Lake Atitlan in Guatemala where the climate is different from that in California (Komárek et al. 2013). It is important to understand the factors affecting the emergence of Lm. robusta in California and the potential link promoting the growth of the cyanobacterium between the two geographically distant water bodies. Lastly, the identification of prokaryote assemblages by DNA barcoding will enhance the current cyanobacterial monitoring efforts by allowing us to develop specific quantitative PCR (qPCR) assays using the sequences obtained in this study. We are currently validating the reliability and reproducibility of the qPCR tests for estimating the abundance of key cyanobacterial species with potential toxin production. Assessment of cyanobacterial assemblages using the interdisciplinary approach (i.e. DNA barcoding and qPCR supported by morphological identification) will aid in formulating effective mitigation measures by addressing the specific identity of cyanobacteria, their corresponding physiological features, and determining the effects of fundamental environmental factors on species-specific toxicity.

TK, DVB, SJT are members of the Aquatic Health Program at UC Davis, School of Veterinary Medicine: http://www.vetmed.ucdavis.edu/aquatic_health/index.cfm.