The 16S rRNA gene has been used extensively for cyanobacterial identification (7, 8, 10); however, the DNA-dependent RNA polymerase (rpoC1) gene has been described as a more discriminatory marker (9). A phylogenetic examination of cyanobacterial rpoC1 gene sequence data is presented, focusing on the genus Anabaena and in particular the species A. circinalis. We report here for the first time an A. circinalis-specific PCR assay targeting the rpoC1 gene to detect this organism directly in environmental water samples.

Molecular techniques. The cyanobacterial strains used in this study (Table 1) were grown under constant light intensity for up to 14 days at 25°C in ASM-1 medium (4). Genomic DNA was extracted from reference strains as previously described (15) and from environmental samples using the InstaGene matrix (Bio-Rad) and phenol-chloroform treatment. Briefly, 10-ml environmental samples were pelleted by centrifugation and resuspended in 90% InstaGene matrix and 10% Triton X-100 to 200 μl. Following incubation at 55°C for 30 min, the cells were vortexed for 1 min, heated to 95°C for 10 min, and then centrifuged. DNA was extracted once with an equal volume of phenol-chloroform and then once with chloroform. The DNA was precipitated, resuspended in 50 μl of water, and used directly in PCRs.

Strain	Locationa	rpoC1 sequence	A. circinalis PCR	Saxitoxin productiond	A. circinalis type
Anabaena circinalis
ANA019A	Chaffey Dam, NSW	Yb	+	+	I
ANA059C	Renmark, SA	Y	+	NA	I
ANA118A	Booligal, NSW	Y	+	+	I
ANA118C	Booligal, NSW	Y	+	+	I
ANA125A	Yarrawonga, VIC	NDc	+	NA	ND
ANA150A	Burrinjuck Reservoir, NSW	Y	+	+	I
ANA175A	Collarenebri, NSW	Y	+	NA	I
ANA209F	Myponga Reservoir, SA	Y	+	NA	I
ANA301A	Riverslea, QLD	Y	+	NA	I
ANA306A	Palm Island, QLD	ND	+	−	ND
ANA311E	Geelong, VIC	Y	+	NA	II
ANA323B	Perth, WA	ND	+	+	ND
ANA332H	Yarraman, QLD	Y	+	−	II
ANA349H	Hope Valley Reservoir, SA	Y	+	NA	II
ANA350C	Craigbourne Dam, TAS	Y	+	NA	II
Anabaena flos-aquae
ANA051A	Murtho Reserve, SA	ND	−	NA
ANA076C	Iraak Creek, VIC	ND	−	NA
ANA264A	Darling Anabranch, NSW	Y	−	−
ANA354A	River Murray, SA	Y	−	NA
Anabaena spiroides f. spiroides
ANA139A	Torrumbarry Weir, VIC	Y	+	NA	II
ANA281A	Windamere Dam, NSW	ND	−	NA
ANA292C	Lake Ballyrogan, NSW	Y	−	NA
Anabaena spiroides f. minima
ANA044A	Lake Albert, SA	ND	−	NA
ANA084E	Lake Alexandrina, SA	ND	−	NA
ANA234B	Menindee Lake, NSW	ND	−	NA
Anabaena sp.
ANA062C	Moolooroo, SA	ND	−	NA
ANA098C	Lagoon, NSW	ND	−	NA
ANA193E	Binghi, NSW	ND	−	NA
ANA238A	Darling Anabranch, NSW	ND	−	NA
ANA238C	Darling Anabranch, NSW	Y	−	NA
ANA255C	Darling Anabranch, NSW	Y	−	NA
Anabaena aphanizomenoides
ANA023C	Murtho Park, SA	ND	−	NA
ANA214B	Nildottie Lagoon, SA	ND	−	NA
ANA217A	Nildottie Lagoon, SA	ND	−	NA
ANA259C	Darling Anabranch, NSW	ND	−	NA
ANA280A	Chaffey Dam, NSW	Y	−	−
ANA303C	Riverslea, QLD	ND	−	NA
Anabaena pertubarta f. tumida
ANA112D	Cobram, VIC	ND	−	NA
ANA146C	Fish Creek Farm Dam, VIC	ND	−	NA
ANA187A	Mungindi, NSW	ND	−	NA
ANA221E	Yarramundi, SA	Y	−	NA
ANA297B	Deniliquin, NSW	Y	−	NA
ANA313C	Bundaleer Reservoir, SA	ND	−	NA
ANA313D	Bundaleer Reservoir, SA	ND	−	NA
Anabaena solitaria
ANA060B	Renmark, SA	ND	−	NA
ANA075C	Iraak Creek, VIC	ND	−	NA
ANA207A	Bridgewater Farm Dam, SA	Y	−	NA
ANA282B	Windamere Dam, NSW	Y	−	NA
ANA337A	Goulbourn, NSW	Y	−	NA
Aphanizomenon issatschenkoi
APH016B	Torrumbarry Weir, VIC	ND	−	NA
APH025A	Wongulla Lagoon, SA	ND	−	−
APH027A	Rockhampton, QLD	ND	−	−
Aphanizomenon gracile
APH015B	Bundaleer Reservoir, SA	ND	−	NA
APH026E	Joan Powling, VIC	Y	−	NA
Aphanizomenon ovalisporum
APH028A	Palm Lakes, QLD	Y	−	NA

PCR amplification using cyanobacterium-specific primers targeted to the rpoC1 gene (rpoC1-1 [5′-GAGCTCYAWNACCATCCAYTCNGG] and rpoC1-T [5′-GGTACCNAAYGGNSARRTNGTTGG]) has been previously described (9). Primers used for the A. circinalis-specific PCR assay detailed in this study were Ana2 (5′-GATAGCATCCTCAATTTCTAGCCATTGG), Ana4 (5′-CTCTGAAGCCAGAAATGGACGGC), and Ana-ICF (5′-TAGCCATTGGCATATCCAAGAGAATAGC) and were constructed from the sequence determined in this study. Each 50-μl PCR mixture contained 1 to 10 ng of genomic DNA, 20 pmol of Ana2, 20 pmol of Ana4, 200 μM deoxynucleoside triphosphates, 2.5 mM magnesium chloride, 1× PCR buffer II, 2.5 U of Ampli Taq Gold, and 2 pg of an internal control fragment (ICF). The following protocol was used: 94°C for 10 min for 1 cycle; 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s for 35 cycles; 72°C for 15 min for 1 cycle; and holding at 4°C. Nucleotide data was analyzed with GeneJockeyII (Biosoft, Cambridge, United Kingdom), sequence alignments were done with ClustalX (13), and phylogenetic trees were constructed with the MEGA analysis platform (6). The Jukes-Cantor method was used to calculate pairwise distances, and a tree was constructed with the neighbor-joining algorithm. Bootstrap analyses were performed with 500 replicates.

rpoC1 sequence and phylogeny. A 612-bp fragment of the rpoC1 gene from 24 Anabaena strains and 2 Aphanizomenon strains was amplified and sequenced (Table 1). An alignment of the sequences showed that in 4 out of 12 A. circinalis strains examined, identical nucleotide changes were observed at 18 of 540 positions. This suggests that genotypic groups of A. circinalis exist (termed types I and II). It is interesting that at position 57, three out of eight type I strains (ANA019A, ANA209F, and ANA301A) had the same nucleotide change, which coincided with the same change in type II isolates. Subsequent analysis failed to associate the observed differences in rpoC1 sequence to strain origin or toxin production. Importantly, the rpoC1 gene provided sufficient sequence variation to differentiate the genus Anabaena to the species level and A. circinalis to the strain level.

A phylogenetic analysis of a range of cyanobacteria based on partial rpoC1 sequences has previously been presented (15) and is combined here with the additional rpoC1 gene sequences obtained for a range of Anabaena and Aphanizomenon species (Fig. 1). The two types of A. circinalis are evident, with the Anabaena genus fitting into the previously identified heterocyst-forming cluster (15). Strain ANA139A was originally identified as A. spiroides on the basis of morphological criteria (Table 1). However, analysis of its rpoC1 sequence showed 100% identity to A. circinalis type II and as shown in Fig. 1, it is within the A. circinalis grouping. This result prompted its identification to be changed and highlighted the importance of rpoC1 typing for Anabaena, as clearly species assignment based on morphological criteria alone can lead to misidentification. Due to the association of A. circinalis with toxicity, misidentification of this species is of considerable concern.

FIG. 1 Phylogenetic tree constructed from rpoC1 nucleotide sequence alignments. Strains sequenced in this study are in bold type (GenBank accession no. AF199423 to AF199433). The A. spiroides ANA139A isolate reclassified as A. circinalis ANA139A type II is underlined. (more ...)

Interestingly, Anabaena bergii ANA283A clusters with Aphanizomenon ovalisporum APH028A (11), separately from the Anabaena genus. In addition, this cluster is distant from the position of Aphanizomenon gracile APH026E (Fig. 1). Both Aphanizomenon ovalisporum and A. bergii are also known to produce the cyanobacterial toxin cylindrospermopsin. rpoC1 sequence analysis revealed 100% identity between the two isolates, and this is supported by 99% 16S rRNA gene sequence identity (M. A. Schembri, unpublished data). Despite morphological data initially indicating that strain ANA283A belongs to the Anabaena genus, sequence analysis demonstrated that this is not the case and that this isolate and Aphanizomenon ovalisporum are morphological variants of the same cyanobacterium. A more detailed genetic and morphological analysis of the taxonomy of these two organisms must now be performed.

Morphological differences observed between strains of Anabaena flos-aquae II led Cronberg and Komárek (2) to define a new cyanobacterial species, Anabaena pertubarta f. tumida. A number of strains from the culture collection of the Australian Water Quality Centre were reclassified accordingly. While previously based on morphological observations, definition of the species A. pertubarta f. tumida is now also supported by genetic data, with the rpoC1 sequence analysis presented here clearly indicating distinct phylogenetic separation between A. flos-aquae and A. pertubarta strains (Fig. 1).

A. circinalis PCR assay. A PCR assay to specifically detect A. circinalis in reference and environmental samples was developed using primers designed on the basis of conserved regions of the A. circinalis rpoC1 gene. An ICF was constructed using a previously described method (15) employing primers Ana2, Ana4, and Ana-ICF. Ana-ICF recognized a sequence internal to the region bounded by Ana2 and Ana4. A PCR with Ana2 and Ana-ICF yielded a fragment which was used in a subsequent PCR with Ana2 and Ana4, which recognizes a terminal decameric sequence incorporated using Ana-ICF. The final product was 276 bp in length and contained 3′ and 5′ sequences which exactly matched Ana2 and Ana4. Two picograms of this internal control was spiked into each PCR mixture. Amplification of the ICF yielded a 276-bp fragment, with the presence of A. circinalis chromosomal DNA resulting in an additional 377-bp product. The ICF indicates if failure to amplify the diagnostic product is due to genuine absence of the target sequence or because the PCR failed due to sample inhibition. In this case, the detection limit is approximately 2,000 cells but lower levels should be detectable with careful optimization of PCR conditions.

Reference cultures of Anabaena and Aphanizomenon (Table 1) and four environmental samples were screened with the assay, including an environmental sample from Bahia Blanca, Argentina (Fig. 2). This sample was known to contain A. circinalis and Microcystis and Ceratium spp. The 377-bp diagnostic band was amplified in all A. circinalis reference cultures (lanes 1 to 15) and in three of the environmental samples known to contain A. circinalis following microscopic analysis (lanes 56 to 58). Sequence analysis of the three diagnostic PCR products confirmed the identification of A. circinalis in the environmental samples. In addition, the specific PCR also supported the definition of A. spiroides f. spiroides ANA139A as A. circinalis (lane 20). This is the first report of a PCR assay that is specific for A. circinalis and able to identify this species in an environmental water sample. A PCR test which targets toxin-encoding genes in A. circinalis would be ideal. However, in the absence of DNA sequence information regarding these genes, but the strong association of this species with toxicity, the rapid and specific identification of A. circinalis is a useful tool in the management of toxic algal blooms.

FIG. 2 A. circinalis-specific PCR assay. The diagnostic product (377 bp) and the ICF (276 bp) are indicated. Lanes 1 to 15 are A. circinalis isolates (019A, 059C, 118A, 118C, 125A, 150A, 175A, 209F, 301A, 306A, 311E, 323B, 332H, 349H, and 350C), lanes 16 to (more ...)

Nucleotide sequence accession numbers. The nucleotide sequences obtained in this work have been deposited with the GenBank database under the following accession numbers: A. circinalis type I rpoC1, AF199423; A. circinalis type IA (representing strains ANA019A, ANA209F, and ANA301A), AF199424; A. circinalis type II rpoC1, AF199425; Anabaena solitaria rpoC1, AF199426; A. spiroides ANA292C rpoC1, AF199427; A. pertubarta rpoC1, AF199428; A. flos-aquae rpoC1, AF199429; Anabaena aphanizomenoides rpoC1, AF199430; Aphanizomenon gracile rpoC1, AF199431; Anabaena sp. strain ANA238C rpoC1, AF199432; Anabaena sp. strain ANA255C rpoC1, AF199433.

This work was supported by the Cooperative Research Centre for Water Quality and Treatment.

We thank Peter Baker for expert advice and provision of strains, Mark Schembri for unpublished data on A. bergii, Glen Shaw for provision of Aphanizomenon ovalisporum, Dennis Steffensen for the sample from Argentina, and Paul Monis for assistance with phylogenetic tree analysis.