First report of anatoxin-a producing cyanobacteria in Australia illustrates need to regularly up-date monitoring strategies in a shifting global distribution

Routine monitoring of toxic cyanobacteria depends on up-to-date epidemiological information about their distribution. In Australia, anatoxin producing cyanobacteria are not regularly tested for and thought to be rare if not absent from the continent. Our study investigated the presence of anatoxin-a (ATX-a) producing cyanobacteria in surface water samples (n = 226 from 67 sampling locations) collected from 2010 to 2017 across the state of Victoria, Australia. We (1) detected the presence and distribution of anaC (anatoxin-a synthetase C) gene sequences previously associated with various cyanobacteria, including Cuspidothrix issatschenkoi, Aphanizomenon sp., D. circinale, Anabaena sp., and Oscillatoria sp., from 31 sampling locations, and (2) determined the concentration of ATX-a in samples tested using ELISA, in two instances detected at >4 µg · L−1. These data present the first confirmation of ATX-a producers in Australia. Our study indicates that ATX-a should be included in regular testing of cyanobacterial blooms in Australia and highlights the importance of regular investigation of the distributions of toxic cyanobacteria worldwide, particularly amid the known expanding distribution of many cyanobacterial taxa in a period of increased eutrophication and rising surface water temperatures.

anaC gene nested PCR. Amplification of the cyanobacterial 16S rRNA and internal spike control assays prior to anaC nested PCR were found to be positive for all samples. For anaC gene screening, DNA extracts from each of the 226 samples were tested by nested PCR using published anaC general primer pairs (Table 1) 22 known to specifically target ATX-a producing cyanobacteria. In the primary PCR process, no PCR products except for the positive control were observed on a 1.5% agarose gel electrophoresis, indicating low prevalence of anaC target gene in our samples. Subsequent nested PCR step produced a total of 68 amplicons (from 31 sampling locations across the state of VIC, Australia) from 226 samples with an expected 366 bp size on a 1.5% agarose gel electrophoresis unit (Fig. 2). Out of the total 68 amplicons, 20 amplicons were observed as very strong bands on the agarose gel electrophoresis. The remaining amplicons, i.e., 4/68 and 44/68 were observed as strong and faint bands, respectively (see Supplementary Table S1). The majority of the amplicons (i.e., 59/68, 86.7%) in this study were generated from 82 samples, corresponding to a recent (2016/2017) major cyanobacterial bloom event spreading across 23 different locations along the Murray River and associated waterways (including rivers, creeks, lakes, canals and reservoirs) bordering the states of New South Wales (NSW) and VIC, Australia (Fig. 1b). The remaining nine anaC (i.e., 9/68, 13.2%) positive amplicons produced faint bands on agarose gel electrophoresis, and were represented by non-Murray bloom samples (i.e., n = 144) collected prior to 2016 (Fig. 1a).
To identify potential ATX-a producing cyanobacteria within this recent bloom, cyanobacteria community abundance was determined for each sample by observation under light microscope. anaC gene sequencing and phylogenetic analysis. To confirm the identity of anaC nested PCR products, the 24 very strong and strong amplicons were sequenced by bi-directional (capillary) sequencing.
Twenty three amplicons produced high quality (>97%) unambiguous sequence data (analysed using Geneious ® Version 10) and the remaining amplicon was removed from further analysis. A BLASTn search of the NCBI non-redundant database of each of the 23 sequenced amplicons showed 90-100% nucleotide identity to several previously annotated cyanobacteria anaC genes from uncultured cyanobacteria (Accession Numbers: KX096813,  KT246302, (Table S1).
www.nature.com/scientificreports www.nature.com/scientificreports/ All 23 anaC sequences obtained from this study along with other peer reviewed anaC sequences from GenBank were used to construct a 340 bp Maximum Likelihood (ML) tree (Fig. 3). A previously described ATX-a producing Phormidium autumnale CAWBG618 (Accession Number: KX016036) with a smaller anaC gene fragment (308 bp) was excluded from the phylogenetic analyses. The overall topology of the anaC gene ML tree recognizes three distinct groups with >98% bootstrap support. The majority of the sequences (19/23) derived from the current study were identical and clustered with a larger group of anaC sequences of many labelled in GenBank as "Uncultured cyanobacterium", Cu. issatschenkoi and Aphanizomenon sp. The remaining four anaC sequences from this study clustered with two different smaller groups including several anaC GenBank sequences of Anabaena sp., D. circinale and Oscillatoria sp. (see Fig. 3 for Accession Numbers). Small genotypic variations in anaC genes among different ATX producing cyanobacteria have resulted in the formation of subgroups within the three major groups.

ATX-a concentration in cyanobacterial bloom samples.
To confirm the presence of anatoxin producers in these water samples, we undertook direct ATX-a detection in 40 (32 anaC nested PCR positive and 8 negative) samples corresponding to a large recent (2016-2017) cyanobacterial bloom extending over 500 kms across the Murray River and associated waterways, Australia. All samples were analysed in duplicate using the Abraxis ® ELISA kit. A total of 29 of 32 nested PCR positive samples showed ATX-a concentrations between 0.15-1.6 µg · L −1 , with two samples exceeding >4 µg · L −1 and one sample showing <0.15 µg · L −1 . Of the eight negative nested PCR samples, seven samples showed zero or <0.15 µg · L −1 (the minimum reading for the test) of ATX-a and one sample had 0.3 µg · L −1 of ATX-a (Table S1). A significant difference (p < 0.01, Welch Two sample t-test) was observed between ATX-a ELISA concentrations of nested PCR-positive and negative samples (Fig. 4a). Additionally, the relationship between observed nested PCR band intensities (i.e., very strong, strong, weak and negative) and ATX-a ELISA values was also found to be highly significant (p < 0.000003 and p < 0.02, Tukey's honest significant difference (HSD) test and ANOVA) (Fig. 4b). All PCR positive groups showed greater  Table 1. List of PCR primers 22 used in this study to amplify anatoxin-a synthetase C (anaC) gene.

Discussion
We examined the presence of ATX-a producing cyanobacteria in 226 samples from 67 locations ( Fig. 1) across the state of VIC, Australia by microscopy, nested PCR and ELISA assays. Based on the microscopy data, we detected several potential/known ATX-a producing cyanobacteria in many samples (Table S1). Nested PCR testing of these same samples detected the presence of anaC, a gene required for ATX-a synthesis, at overall sample prevalence of 30.1% (68/226 samples tested) representing 46.2% of sample locations (31/67 sample sites: Figs 1, 2 and Table S1).
Bi-directional sequencing confirmed the identity of the nested PCR amplicons from our samples as the anaC gene. The phylogenetic clustering by unrooted ML analysis clustered these sequences into three distinct groups, of which the majority of anaC sequences most likely represent ATX-a producing C. issatschenkoi along with Aphanizomenon sp., Oscillatoria sp., Anabaena sp. and D. circinale (Table S1 and Fig. 3). Overall, our findings (corroborated by microscopy, nested PCR, sequencing and phylogenetic analysis) agree with previous reports of common ATX-a producing cyanobacteria in many parts of Europe 20,28,33,43,44 , Asia 41,45,46 and New Zealand 27,30,31 .
Direct toxin detection (using the Abraxis ® anatoxin-a ELISA kit) confirmed the presence of ATX-a in all but one of the anaC positive samples ( Fig. 4 and Table S1). It is not clear whether these samples represent significant health risks as currently there are no guideline values documented for ATX-a toxin and its congeners in Australia and/or by World Health Organisation (WHO). None of our samples showed toxin levels exceeding the New Zealand provisional maximum value for ATX-a in drinking water (i.e., 6 µg · L −1 ) 47 . Two exceeded a provisional guideline value (i.e., 3 µg · L −1 ) for ATX-a in drinking water outlined by the province of Oregon, USA 47 , however it should be noted that these samples were taken from raw water (i.e. pre-treatment) and not product water (post-treatment). ATX-a ELISA assay was performed on recently collected (2016-2017) samples (i.e., from the Murray River and associated waterways) only. All samples from 2010-2015 were tested by anaC nested PCR only. Positive anaC detection in these samples is indicative of their potential to produce ATX-a, but actual toxin production has not been confirmed.
Although the ELISA assay quantified ATX-a production in our samples, ELISA values did not strongly correlate with the total number/bio-volume of potential/known ATX-a producing cyanobacteria strains identified in our microscopy analysis (Table S1). The possible reasons for this are, (1) the ATX-a producing cells in our samples declined rapidly prior to our sampling, but the ATX-a did not, (2) despite planktonic cyanobacteria being known as the major ATX-a producers in surface waters, the primary source in some samples may be ATX-a producing benthic cyanobacteria species 48,49 , or (3) these samples contain other major species of ATX-a producing cyanobacteria that are yet to be identified. Of these possibilities, the latter seems most likely as all samples were transported under cold storage and processed immediately upon receipt and yet none contained large quantities of a known ATX-a producer on microscopic examination. Additionally, while microscopy could not clearly identify a major ATX-a producer at high abundance, associated anaC genes were readily detectable by nested PCR in all ATX-a positive samples, indicating against the primary toxin source being a benthic species not present in www.nature.com/scientificreports www.nature.com/scientificreports/ the surface water grabs. These possible explanations, however, must be evaluated through further exploration. Additionally, further studies using more accurate quantitative methods for direct toxin detection and quantification (i.e., HPLC-MS / GC-MS) 50,51 are needed.
Additional investigation is needed to identify and isolate the specific source of the ATX-a producing cyanobacteria (although phylogenetic clustering implicates a Cuspidothrix/Aphanizomenon/Dolichospermum/ Anabaena and Oscillatoria sp.) in our samples, and its current and historical distribution in Australia. Although previous studies in Australia [10][11][12][13] have reported the presence of potential ATX-a producers [e.g., Cu. issatschenkoi (previously Aphanizomenon issatschenkoi)] by microscopy, their toxigenicity was not confirmed. Phylogenetic analyses using 16S rRNA, rpoC1 or Phycocyanin intergenic spacer (PC-IGS) marker genes on isolated ATX-a producing strains may provide further insights to the biogeography of ATX-a producers in Australia as has been previously studied for CYN producing R. raciborskii (see review by Wilk-Woźniak et al. 9 ).

Materials and Methods
Sample collection and storage. A total of 226 samples representing cyanobacterial blooms were collected between November 2010 to April 2017 from surface water sources across the state of VIC, Australia (Fig. 1). Bloom samples (1 L surface grabs) were collected in polyethylene terephthalate (PETG) bottles. For microscopy, a 50 mL aliquot of the sample was preserved in Lugol's iodine solution. All samples were dispatched and transported on ice to the laboratory within 24 hours of collection. Upon arrival, 50 mL aliquots of each bloom sample was centrifuged at 3600 × g for 15 min and pellets were stored at −20 °C for genomic DNA extraction and ELISA analysis. www.nature.com/scientificreports www.nature.com/scientificreports/ Microscopy and cyanobacteria identification. Microscopic identification and enumeration of cyanobacteria was carried out on a Nikon ECLIPSE Ci-H550S light microscope at 200 to 600 times magnification. All taxa were identified to genus and species level using established cyanobacterial taxonomic keys 37,38 . Cell counting was performed on a calibrated Sedgewick Rafter Chamber (Pyser-SGI ® , United Kingdom) that holds 1 mL of sample over an area of 50 × 20 mm. The biovolume of each taxon was calculated from the cell count data using the "Biovolume calculator" 52 . In circumstances where a taxon was identified to a genus or family level, the mean cell/filament volume of the largest known toxic species within that genus or family was used to calculate the biovolume.
Genomic DNA extraction and amplification of anaC gene by nested PCR. Sample enrichment and genomic DNA extraction was performed using the PowerBiofilm Kit (MoBio, USA) as described elsewhere 53 . Each DNA sample was eluted in 100 µL of 10 mM Tris buffer and stored at −20 °C for PCR analysis. All DNA samples were tested for sample degradation and/or for the presence of PCR inhibitors prior to anaC nested PCR assay using cyanobacteria-specific 16S rRNA primers and an internal control as described by Baker et al. 53 . Presence of the anaC gene (required for ATX-a biosynthesis) was assessed by nested PCR using previously published anaC primer pairs (primary PCR: anxgenF/anxgenR and secondary PCR: anaC-genF/anaC-genR) 22 (Table 1) with minor modifications to their published protocols. For each step in the nested PCR assay, a 50 µL mastermix consisted of 2.5 mM MgCl 2 , 0.2 mM dNTP, 1X Green GoTaq ® Flexi Buffer (Promega, USA), 2.5U of GoTaq ® DNA Polymerase (Promega, USA) and 5% dimethyl sulfoxide (DMSO). In the primary PCR, 2 µL genomic DNA and 0.3 µM each primer was added to the mastermix and the following thermocycler protocol was used: an initial denaturation at 94 °C for 2 min followed by 30 cycles of 94 °C for 30 sec (denaturation), 52 °C for 30 sec (annealing), 72 °C for 30 sec (extension) and a final extension step at 72 °C for 5 min.
In the secondary (nested) PCR, 2 µL of PCR product generated from the first PCR reaction was added as the template to the mastermix along with 0.3 µM each of internal primer (i.e., anaC-genF and anaC-genR). The thermocycling conditions for the secondary PCR were as follows: an initial denaturation at 94 °C for 2 min followed by 30 cycles of 94 °C for 30 sec (denaturation), 58 °C for 30 sec (annealing), 72 °C for 30 sec (extension) and a final extension step at 72 °C for 5 min. All PCR amplifications were run on a BioRad T100 ™ thermal cycler. A known ATX-a producing Cu. issatchenkoi CY1941 was used as a positive control and a known negative ATX-a producer (Nostoc punctiforme PCC73102) and non-template control (nuclease free water) were used as negative controls for all PCR reactions [Non-template controls (NTC) from the primary PCR were used as carry-over NTC controls in the secondary PCR]. To determine the size of each amplicon and its specificity, all nested PCR products were subjected to electrophoresis on a 1.5% agarose gel at 90V for 35 min followed by visualisation of bands using a Gel Doc ™ XR+ System. sequencing and phylogenetic analysis of anaC gene. All nested PCR amplicons were cleaned up prior to sequencing using exoSAP-IT ® shrimp alkaline phosphatase. The exoSAP mastermix consisted of Exonuclease I, FastAP ™ Thermosensitive Alkaline Phosphatase and 5 µL of nested PCR template. Reactions were incubated in a BioRad T100 ™ thermal cycler at 37 °C for 30 min followed by 85 °C for 15 min. The purified nested PCR products were then subjected to bi-directional automated sequencing (BigDye ® Terminator version www.nature.com/scientificreports www.nature.com/scientificreports/ 3.1 chemistry, Applied Biosystems, USA) using the internal primers anaC-genF and anaC-genR. The quality of each sequence was assessed using the program Geneious ® Version 10 and sequence identity was confirmed by a BLASTn search of the NCBI (National Centre for Biotechnology Information) non-redundant sequence database. Verified sequences were trimmed and aligned along with published anaC sequences from NCBI and adjusted manually using the programs MUSCLE 54 and Mesquite, respectively 55 . An unrooted Maximum Likelihood (ML) tree based on the Tamura-Nei model 56 with 1000 bootstrap replicates was constructed using the program MEGA7 57 . The phylogenetic tree was viewed and edited for publication using Adobe Illustrator CC 2017.
Anatoxin-a analysis using ELISA. This work focused on recently collected (2016-2017) anaC-positive (n = 32) and representative negative (n = 8) samples identified by nested PCR to reduce the potential for false negative ELISA results due to the physical degradation of the cyanotoxin during storage 58 . Direct testing for total ATX-a (i.e., intracellular and extracellular) toxin was performed in duplicate using the ATX-a ELISA kit (Abraxis ® , USA) as per the manufacturer's protocol (available at https://www.abraxiskits.com/wp-content/ uploads/2016/06/Anatoxin-a-ELISA-rev053116.pdf). Briefly, samples were freeze/thawed three times to lyse the cells. Prior to testing, the enzyme conjugate and antibody were reconstituted with 3 mL of conjugate and antibody diluent, respectively. Wash buffer (5X) and sample diluent (10X) were diluted at a ratio of 1:5 and 1:10, respectively, using deionised water before use. For each assay, 50 µL of the standard solutions (i.e., standard 0 to standard 5), control and samples were loaded into a 96-well microtitre plate, followed by the addition of 50 µL each of the reconstituted enzyme and antibody solution. After 60 minutes incubation at room temperature, plates were washed four times using 250 µL of 1X diluted wash buffer for each well and each washing step. After drying the plates, 100 µL of substrate colour solution was added to individual wells and contents were incubated for 20-30 minutes at room temperature. Following the addition of stop solution (100 µL), absorbances were read at 450 nm on a Synergy ™ hybrid multi-mode microplate reader (BioTek, USA). A standard curve was then constructed by plotting the %B/B 0 for each of the standards (where B is the mean absorbance value for each of the standards 1 to 5, B 0 is the mean absorbance reading of standard 0) on the y-axis versus the corresponding anatoxin-a concentration on the x-axis. Finally, %B/B 0 of each sample (where B is the mean absorbance value for each of the sample, B 0 is the mean absorbance reading of standard 0) was interpolated on the standard curve to calculate the quantity of total ATX-a in each of the 40 samples. The measurement range for ATX-a ELISA kit stated by manufacturers is from 0.15-5 µg · L −1 . An ATX-a concentration below 0.15 µg · L −1 was considered to be negative or zero µg · L −1 .

statistical analysis
Statistical analyses were computed and displayed using R software 59 . Welch's two sample t-test was used to compare log-transformed ATX-a ELISA concentrations between nested PCR-positive and negative samples. Further, the relationship between nested PCR band intensities and log-transformed ELISA concentrations was compared using ANOVA followed with Tukey's honest significant difference (HSD) test. Significance was accepted for p < 0.05.

Data Availability
A subset of nucleotide sequences representing anaC sequences derived from this study are available from NCBI GenBank (https://www.ncbi.nlm.nih.gov/nucleotide) under Accession Numbers MG836989 -MG836992. Additional Table S1 attached as Supplementary Information.