Identification of a Polyketide Synthase Coding Sequence Specific for Anatoxin-a-Producing Oscillatoria Cyanobacteria

Abstract

We report the identification of a sequence from the genome of Oscillatoria sp. strain PCC 6506 coding for a polyketide synthase. Using 50 axenic cyanobacteria, we found this sequence only in the genomes of Oscillatoria strains producing anatoxin-a or homoanatoxin-a, indicating its likely involvement in the biosynthesis of these toxins.

Anatoxin-a and homoanatoxin-a (Fig. 1) are potent neurotoxins produced by cyanobacteria (3, 20, 21). Cases of animal death, after the animals ingested water contaminated by cyanobacteria producing these alkaloids, are regularly reported in different places around the world (2, 6-8, 20, 22). Therefore, the release of such cyanobacterial toxins into water bodies and water supplies is a major public health and environmental issue (4).

FIG. 1.

Structures of anatoxin-a and homoanatoxin-a.

To find a reliable genetic marker for the detection of anatoxin-a- and homoanatoxin-a-producing cyanobacteria, we searched for a gene involved in the biosynthesis of these toxins. Based on the biosynthetic route proposed by Hemscheidt et al. (9), we suspected that a polyketide synthase (PKS) was involved in the elongation of the glutamic semialdehyde starter. We chose Oscillatoria sp. strain PCC 6506 for this work because it produces anatoxin-a and homoanatoxin-a, it is a fast-growing cyanobacterium under standard conditions, and it has lost its benthic character after multiple subcultures since its isolation (R. Rippka and A. Méjean, unpublished observations). To amplify PKS genes from the genome of PCC 6506, we used the degenerate primers DKF and DKR, designed by Moffitt and Neilan (13). The PCC 6506 strain was cultivated (1, 10, 16) and its genomic DNA was isolated (12) by using previously described protocols. The PCR amplification, achieved under conditions described previously (2, 13), afforded a diffuse band corresponding to about 650 bp, as judged by gel electrophoresis (17), and the product was purified and cloned into the pGEM-T vector (Promega). After the transformation of Escherichia coli JM109, the selection of positive clones, the purification of the plasmids, and sequencing by the GATC Company (Konstanz, Germany), we obtained three distinct DNA sequences, designated ks1-630, ks2-682, and ks3-716 (corresponding to fragments ks1, ks2, and ks3, of 630, 682, and 716 bp, respectively). Using Southern blotting experiments with digoxigenin-labeled probes (17), we obtained a larger ks1 fragment: ks1-3711. Using the adaptor-mediated gene walking technology (18) with the APAgene BT 501 kit (BIO S&T, Canada), we obtained a longer ks2 fragment: ks2-1680. A BLAST analysis of the ks2 sequence against the GenBank database revealed three sequences (accession no. AY768507, AY768508, and AY210784) almost identical to ks2-1680. Interestingly, the AY210784 sequence was from Anabaena flos-aquae NRC 44-1, a cyanobacterium producing anatoxin-a (3, 14), while the AY768507 and AY768508 sequences, which were identical, were from PCC 6506.

The putative functions of the products translated from partial coding DNA sequences (CDS) in the fragments ks1-3711, ks2-1680, and ks3-676 (Table 1) were inferred from bioinformatic analyses (using NCBI BLAST and the websites http://www.tigr.org/jravel/nrps and http://www.nii.res.in/nrps-pks.html for predicting PKS domains). As expected, each CDS coded for a ketosynthase (KS) domain of PKS. Furthermore, an acyltransferase (AT) domain, a classical feature in PKS (5), was found downstream of the KS domain in the product of the ks2-1680 CDS. Comparison of the sequence translated from ks2-1680 with PKS sequences involved in the biosynthesis of cyanobacterial secondary metabolites, such as curacin, jamaicamid, and microcystin, showed a degree of sequence identity higher than 50% (see Fig. S1 in the supplemental material), suggesting that the ks2 gene product is likely a PKS (Table 1). Fragment ks3-676 coded for a KS domain homologous (89% identical) to CyrB, a PKS involved in the biosynthesis of cylindrospermopsin (11). The fragment ks1-3711 encoded, beside the KS domain, a partial pyridoxal phosphate-dependent decarboxylase and a methyltransferase. It is very likely that other PKS genes are present in the genome of PCC 6506. However, the degenerate primers did not anneal to these sequences, preventing their isolation by our strategy.

TABLE 1.

Putative functions of the CDS obtained from Oscillatoria sp. strain PCC 6506

DNA sequence designation	Size (bp)	Position (bp) of partial CDS	Accession no. and corresponding protein (organism) for Psi-BLAST similarity match	% Identity	Description of putative product
ks1-3711	3,711	1-1392	EAV71836.1 (Clostridium cellulolyticum H10)	58	Pyridoxal phosphate-dependent decarboxylase
		1398-2204	ACC81856.1 (Nostoc punctiforme PCC 73102)	42	Methyltransferase
		2254-3711	ACC81837.1 (Nostoc punctiforme PCC 73102)	73	KS domain of PKS
ks2-1680	1,680	1-1680	AAT70104.1; CurI (Lyngbya majuscula)	60	KS and AT domains of PKS
ks3-676	676	1-676	ABX60161.1; CyrB (Cylindrospermopsis raciborskii AWT205)	89	KS domain of PKS

The genomes of 49 other axenic cyanobacteria belonging to several genera, obtained from the Pasteur Culture Collection of Cyanobacteria (PCC) (10), were searched for the presence of the PKS-encoding DNA sequences identified in the genome of strain PCC 6506 (Table 2; see also Table S1 in the supplemental material). We designed specific primers (see Table S2 in the supplemental material) to amplify a 400-bp sequence within fragment ks1, ks2, or ks3. The PCR was run using purified genomic DNA (19) or cell lysate (2) under standard conditions (17). When a PCR product was obtained, it appeared, in each case, as a clean single band of the predicted size upon agarose gel electrophoresis. Therefore, it is reasonable to postulate that the presence of a PCR product indicated the presence of the corresponding gene in the genome of the strain of interest and that the absence of a PCR product indicated the absence of the corresponding gene.

TABLE 2.

Correlation between the production of anatoxin-a and homoanatoxin-a and the amplification of DNA sequences coding for PKS in PCC strains

PCC strain	Presence of anatoxin-a and homoanatoxin-aa	Presence of PCR productb for:
		ks1	ks2	ks3
Anabaena flos-aquae strains
PCC 9302	−	+	−	+
PCC 9332	−	+	−	−
PCC 9349	−	+	−	−
Aphanizomenon flos-aquae strain
PCC 7905	−	+	−	+
Geitlerinema strain
PCC 10602	−	−	−	−
Lyngbya strains
PCC 7419c	−d	+	−	−
PCC 8103c	−d	+	−	−
PCC 8937c	−d	+	−	−
PCC 8992c	−d	+	−	−
Microcystis strain
PCC 7806	−	−	−	−
Nostoc strains
PCC 7120	−	−	−	−
PCC 73102	−	−	−	−
Oscillatoria strains
PCC 6304	−	−	−	−
PCC 6407	+	+	+	+
PCC 6412	+	+	+	+
PCC 6506	+	+	+	+
PCC 6602	−	−	−	−
PCC 7112	−	+	−	−
PCC 7515	−	−	−	−
PCC 7816	−	−	−	−
PCC 7817c	−d	−	−	+
PCC 7823c	−d	−	−	−
PCC 7907c	−d	−	−	+
PCC 7926	−	−	−	−
PCC 7928c	−d	−	−	−
PCC 7929	−	−	−	−
PCC 7930	−	+	−	+
PCC 8704	−	−	−	+
PCC 8808	−	−	−	−
PCC 8914	−	−	−	+
PCC 8919	−	+	−	−
PCC 8922	−	−	−	−
PCC 8923	−	−	−	−
PCC 8935	−	+	−	+
PCC 8946	−	−	−	−
PCC 8963	−	+	−	−
PCC 8973	−	−	−	−
PCC 8989	−	−	−	−
PCC 9029	+	+	+	+
PCC 9107	+	+	+	+
PCC 9240	+	−	+	−
PCC 9609	−	+	−	−
PCC 10016	−	+	−	−
PCC 10111	+	+	+	+
PCC 10601	+	−	+	−
PCC 10608	+	−	+	−
PCC 10702	+	+	+	+
Planktothrix strains
PCC 7805c	−d	+	−	−
PCC 7811c	−d	+	−	−
Synechocystis strain
PCC 6803	−	−	−	−

^aPresence (+) or absence (−) of anatoxin-a and homoanatoxin-a in the cellular extract as analyzed by gas chromatography-mass spectrometry as described previously (2).

^bPresence (+) or absence (−) of a single PCR amplicon obtained using primers specific for the ks1, ks2, or ks3 sequence.

^cCell lysate instead of purified genomic DNA was used for the PCR amplifications.

^dData are from Aráoz et al. (1).

In parallel, we tested the abilities of each axenic strain to produce anatoxin-a and homoanatoxin-a (Table 2). The toxin contents of the cellular extracts were determined as described previously (2). Ten strains among the 50 strains studies produced anatoxin-a and homoanatoxin-a, and these 10 strains gave a ks2 PCR product (Table 2), whereas the 40 nonproducers gave no ks2 PCR product. In contrast, no such correlation for ks1 and ks3 PCR products could be observed. The 10 ks2 amplicons were purified using a QIAquick PCR purification kit (Qiagen) and sequenced by the GATC Company to confirm their identities or similarities. The alignment (see Fig. S2 in the supplemental material) of the 10 DNA sequences showed that they were of exactly the same length (412 bp), with high degrees of identity (94 to 100%). The corresponding translated sequences also showed high levels of identity (see Fig. S2 in the supplemental material). Thus, we are confident that the 10 ks2 amplicons correspond to the same partial CDS. Interestingly, the A. flos-aquae NRC 44-1 strain, which harbors a homologous ks2 sequence (the sequence translated from the gene with accession no. AY210784 and that translated from ks2-1680 shared 96% identity), produces anatoxin-a (3, 14). Although these facts are only indirect evidence, we suggest that ks2-1680 is part of the gene cluster responsible for the biosynthesis of anatoxin-a and homoanatoxin-a. Insertional gene inactivation was not attempted at this stage of the work because, so far, it has not been possible to perform such genetic experimentation on PCC 6506.

We have provided the first tools toward a simple and reliable PCR-based methodology (using primers CGCAAATCGATGCTCACTTA and CCACTGGCTCCATCTTGATT) to detect, in field samples, the presence of anatoxin-a- and homoanatoxin-a-producing Oscillatoria cyanobacteria that represent a risk for animal and human health (15).

Nucleotide sequence accession numbers.

The three new DNA sequences determined in this study were deposited in the GenBank database with the following accession numbers: ks1-3711, FJ418584; ks2-1680, FJ418585; and ks3-676 (equivalent to ks3-716 but excluding the sequence of the degenerate primers), FJ418586.