Origin of saxitoxin biosynthetic genes in cyanobacteria.

Origin of Saxitoxin Biosynthetic Genes in Cyanobacteria
Ahmed Moustafa1., Jeannette E. Loram2., Jeremiah D. Hackett3, Donald M. Anderson4, F. Gerald
Plumley2, Debashish Bhattacharya1,5*
1 Interdisciplinary Program in Genetics, University of Iowa, Iowa City, Iowa, United States of America, 2 Bermuda Institute of Ocean Sciences, St. George’s, Bermuda,
3 Ecology and Evolutionary Biology Department, University of Arizona, Tucson, Arizona, United States of America, 4Woods Hole Oceanographic Institution, Woods Hole,
Massachusetts, United States of America, 5Department of Biology, University of Iowa, Iowa City, Iowa, United States of America
Abstract
Background: Paralytic shellfish poisoning (PSP) is a potentially fatal syndrome associated with the consumption of shellfish
that have accumulated saxitoxin (STX). STX is produced by microscopic marine dinoflagellate algae. Little is known about
the origin and spread of saxitoxin genes in these under-studied eukaryotes. Fortuitously, some freshwater cyanobacteria
also produce STX, providing an ideal model for studying its biosynthesis. Here we focus on saxitoxin-producing
cyanobacteria and their non-toxic sisters to elucidate the origin of genes involved in the putative STX biosynthetic pathway.
Methodology/Principal Findings: We generated a draft genome assembly of the saxitoxin-producing (STX+)
cyanobacterium Anabaena circinalis ACBU02 and searched for 26 candidate saxitoxin-genes (named sxtA to sxtZ) that
were recently identified in the toxic strain Cylindrospermopsis raciborskii T3. We also generated a draft assembly of the non-
toxic (STX2) sister Anabaena circinalis ACFR02 to aid the identification of saxitoxin-specific genes. Comparative
phylogenomic analyses revealed that nine putative STX genes were horizontally transferred from non-cyanobacterial
sources, whereas one key gene (sxtA) originated in STX+ cyanobacteria via two independent horizontal transfers followed
by fusion. In total, of the 26 candidate saxitoxin-genes, 13 are of cyanobacterial provenance and are monophyletic among
the STX+ taxa, four are shared amongst STX+ and STX-cyanobacteria, and the remaining nine genes are specific to STX+
cyanobacteria.
Conclusions/Significance: Our results provide evidence that the assembly of STX genes in ACBU02 involved multiple HGT
events from different sources followed presumably by coordination of the expression of foreign and native genes in the
common ancestor of STX+ cyanobacteria. The ability to produce saxitoxin was subsequently lost multiple independent
times resulting in a nested relationship of STX+ and STX2 strains among Anabaena circinalis strains.
Citation: Moustafa A, Loram JE, Hackett JD, Anderson DM, Plumley FG, et al. (2009) Origin of Saxitoxin Biosynthetic Genes in Cyanobacteria. PLoS ONE 4(6):
e5758. doi:10.1371/journal.pone.0005758
Editor: Jason E. Stajich, University of California, Berkeley, United States of America
Received February 4, 2009; Accepted April 20, 2009; Published June 1, 2009
Copyright: � 2009 Moustafa et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was funded by a collaborative grant from the National Science Foundation (EF-0732440) awarded to FGP, DB, JDH, and DMA. AM was
supported by an Institutional NRSA (T32 GM98629) from the National Institutes of Health and a NIEHS grant (R01 ES013679-01A2) awarded to DB and DMA.
Funding support was also provided through grants from the NSF/NIEHS Centers for Oceans and Human Health, NIEHS P50 ES 012742 and NSF OCE-043072 to
DMA. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: debashi-bhattacharya@uiowa.edu
. These authors contributed equally to this work.
Introduction
Paralytic shellfish poisoning (PSP) is a potentially fatal syndrome
associated with the consumption of shellfish that have accumulated
toxins produced by microscopic marine algae. This phenomenon
is the most widespread of the poisoning syndromes caused by
blooms of toxic algae, commonly referred to as ‘‘red tides’’ or
‘‘harmful algal blooms’’, (HABs). The impacts of HABs on marine
ecosystems and the seafood industry are substantial. STX, the
etiological agent of PSP, is produced by a small number of marine
dinoflagellates and freshwater filamentous cyanobacteria [1].
These latter taxa are a promising model for identifying putative
saxitoxin genes and to elucidating their evolutionary history
because of the small genome sizes of cyanobacteria and the wealth
of available prokaryotic genomic data. Knowledge about STX
genes and their regulation in cyanobacteria could potentially help
in the future ameliorate the most significant impacts from STX
toxicity that derive from dinoflagellate red tides in marine systems.
Recent analyses of STX-producing cyanobacteria have signif-
icantly advanced our understanding of saxitoxin biosynthesis via
isolation and characterization of enzymes putatively involved in
the STX pathway; e.g., S-adenosylhomocysteine hydrolase,
methionine aminopeptidase [2], sulfotransferase [3], Na(+)-depen-
dent transporter [4], aminotransferase, and O-carbamoyltransfer-
ase [5]. A recent study provided the complete sequence (,35 kb)
of a putative STX biosynthesis gene cluster encoding 26 proteins
in the toxic (STX+) strain Cylindrospermopsis raciborskii T3. The
cluster (with genes named sxtA to sxtZ) was identified using
genome-walking upstream and downstream of the gene encoding
O-carbamoyltransferase, which was initially isolated using degen-
erate PCR [6]. Preliminary sequence similarity analyses predicted
the putative functions and origins for each of the 26 STX genes.
PLoS ONE | www.plosone.org 1 June 2009 | Volume 4 | Issue 6 | e5758

Although not yet substantiated by biochemical or molecular
genetic analyses these promising data imply an important role for
horizontal gene transfer (HGT) in the assembly of the STX gene
cluster in Cylindrospermopsis raciborskii T3.
Given this information, here we utilized a combined genomic
and phylogenetic approach to investigate in detail the evolutionary
history and genomic characteristics of putative STX genes in
saxitoxin-producing cyanobacteria. We used the predicted STX
gene cluster in Cylindrospermopsis raciborskii T3 as ‘‘bait’’ to identify
homologs in other toxic and non-toxic strains. To this end, we
generated novel genome data from the STX+ cyanobacterium
Anabaena circinalis ACBU02 and posed the following questions: 1)
Are all of the Cylindrospermopsis raciborskii T3 candidate STX genes
present in ACBU02? 2) If not, which genes and functions form the
core set that is conserved across different taxa? and, 3) what are
the phylogenetic origins of these genes?
Results and Discussion
16S rRNA phylogeny
We examined the evolutionary relationships between STX+
and STX2 cyanobacterial strains by constructing a 16S rRNA
phylogenetic tree that focused on this branch of the tree of life (see
Figure 1A). The phylogeny indicates that STX+ and STX2
strains of the cyanobacterial species, Anabaena circinalis, Aphanizo-
menon flos-aquae, and Cylindrospermopsis raciborskii are evolutionary
closely related and each form monophyletic clades. This suggests
that toxicity as a character was either gained in the toxic strains
through independent HGTs or alternatively, is an ancestral trait
for these cyanobacteria and was lost from the non-toxic strains. An
argument that favors the latter scenario is the known complexity of
the STX biosynthetic pathway and the low likelihood of
convergent evolution of this rare trait among these closely related
species. Therefore, we propose that toxicity is an ancestral state
that was assembled in the common ancestor of the toxic and non-
toxic strains we have studied and most likely some core genes
involved in the pathway were lost from STX- taxa. A similar
scenario was recently proposed to explain the distribution of toxic
and non-toxic strains of the cyanobacterium Planktothrix regarding
the biosynthesis of microcystin, a cyclic heptapeptide cyanotoxin
[7]. In an earlier study, Beltran and Neilan [8] suggested that toxic
and non-toxic strains of Anabaena circinalis would segregate into two
distinct clades. Our Anabaena circinalis-specific 16S rRNA tree
(Figure 1B) is inconsistent with this idea, clearly showing that
STX2 taxa are nested within STX+ clades, supporting loss of
toxicity in the former group. This pattern results in a polyphyletic
collection of STX2 strains.
Genome annotation of saxitoxin-producing Anabaena
circinalis strain ACBU02
A total of 181 contigs were assembled from the ACBU02 ‘‘454’’
genome data using the Newbler and SeqMan Pro assemblers.
These data total 4.5 Mb of unique sequence that agrees well with
the pulse-field gel electrophoresis estimate of genome size (4.0–4.5
Mb) in this taxon [9]. We believe therefore that the draft assembly
of ACBU02 represents a near-complete genome for downstream
analyses. A total of 5,188 putative protein-coding genes were
identified in the assembled genome of ACBU02 using RAST [10]
and GeneMarkS [11]. These data are publicly available at the
RAST repository ([10]; genome ID: 109265.7). Using BLAST
searches against the NCBI non-redundant database ‘‘nr’’, 4,540
proteins (87.5%) had significant matches (e-value,1E-9), 231
proteins (4.5%) had matches with an e-value between 1E-3 and
1E-9, and 417 proteins (8.0%) did not have matches (i.e., with an
e-value,1E-3). Of the 4,771 predicted proteins that had significant
hits, 4,474 (93.8%) had the closest matches to cyanobacterial taxa,
of which the most frequent matching taxa were Nostoc punctiforme
(1,450 hits), Nodularia spumigena (882 hits), Anabaena variabilis (752
hits), and Nostoc sp. (594 hits). For the remaining 6.2% of the
predicted proteins, the best hits were distributed among other
bacterial phyla, of which the most frequent were proteobacteria,
firmicutes, and planctomycetes, in descending order.
Gene families
We identified 3,483 gene families, of which 3,024 (86.8%) are
single-copy genes and 30 families (0.86%) contain greater than 10
genes per family. The largest gene family has 212 members (4% of
the predicted genes) encoding proteins that contain serine/
threonine-specific protein kinases and tetratricopeptide repeat
(TPR) domains. The second largest family is of size 84 genes (1.6%
of the predicted set) encoding ATP-binding cassette (ABC)
transporters.
Broad patterns of STX gene origins
The 26 putative STX genes identified by Kellmann et al. [6] in
the toxic cyanobacterium Cylindrospermopsis raciborskii T3 were used
as a query to retrieve homologs in GenBank. Taxon sampling was
maximized with the use of the NCBI nr protein database with the
addition of the predicted proteomes from the STX+ and STX2
strains of Anabaena circinalis ACBU02 and ACFR02, respectively.
Our preliminary assembly of the ACFR02 data identified 583
contigs that sum to 4.4 Mb of unique genome sequence, encoding
5,203 predicted proteins. We have used these data as a rough
guide to strengthen our inferences about ACBU02 genome
evolution. The ACFR02 genome is currently being completed
and will be published in the near future.
Each STX gene query resulted in a comprehensive alignment
that was submitted to a bootstrapped maximum-likelihood (ML)
phylogenetic analysis [12,13]. Examination of these 26 phyloge-
nies (see Figure S1) led us to identify three broad evolutionary
patterns that underlie and explain the evolutionary history of these
putative STX genes. The first pattern consists of genes that are
common to STX+ and STX2 cyanobacteria (Table 1, group I).
The second pattern defines genes that have a cyanobacterial origin
in the putative common ancestor of only STX+ cyanobacteria
(Table 1, group II). The third pattern (Table 1, groups III, IV, V,
and VI) includes genes that have originated via HGT in the
genome of the putative ancestor of STX+ cyanobacteria from a
non-cyanobacterial source. The genes in the second (putatively
vertically inherited) and third (putatively horizontally transferred)
patterns could have been established in the common ancestor of
STX+ and STX2 cyanobacteria; however, they were subse-
quently lost differentially from STX2 cyanobacteria. The
following sections provide a detailed account and our interpreta-
tion of these phylogenetic patterns of STX gene origin.
Genes of cyanobacterial origin
We find that 17 of the predicted 26 STX genes in
Cylindrospermopsis raciborskii T3 are of cyanobacterial origin. These
genes are unambiguously categorized into two distinct evolution-
ary groups. One group includes four genes (Table 1, group I) that
are broadly distributed across cyanobacteria with their phylogeny
agreeing with the typical 16S rRNA tree for these taxa (see
Figure 1A). In these cases, there is no phylogenetic distinction
between STX+ and STX2 species. The phylogeny of sxtY
(phosphate uptake regulator) and sxtZ (histidine kinase) exemplify
this general topological pattern (Figures 2A and 2B, respectively).
Within this group of genes, STX+ and STX2 strains of Anabaena
Origin of Saxitoxin Genes
PLoS ONE | www.plosone.org 2 June 2009 | Volume 4 | Issue 6 | e5758

circinalis are monophyletic (i.e., they share a common ancestor) as
expected (i.e., based on the 16S rRNA tree). This clade is
positioned within a larger group of cyanobacteria in which the
most closely related sequences are members of the Nostocaceae
such as the STX2 species Anabaena variabilis, Nostoc sp., and
Nodularia spumigena. This phylogenetic pattern suggests these genes
are not specific to saxitoxin biosynthesis in the STX+ cyanobac-
teria, but rather have putatively been co-opted for this function. It
is worth noting that one gene, sxtW (ferredoxin), did not have a
significant hit in the genome of ACBU02, suggesting that it may
Figure 1. Phylogenetic tree of 16S rRNA from cyanobacteria. Unrooted ML phylogenetic trees inferred from small subunit (16S) rRNA that
include (a) different cyanobacterial orders and (b) only saxitoxin-producing (STX+) and STX2 strains of Anabaena circinalis. ML bootstrap values
(when $50%) are indicated at the nodes. The branch lengths are proportional to the number of substitutions per site (see scales in the figure). The
toxic and non-toxic strains are indicated by the plus sign (+) and minus sign (2), respectively.
doi:10.1371/journal.pone.0005758.g001
Origin of Saxitoxin Genes
PLoS ONE | www.plosone.org 3 June 2009 | Volume 4 | Issue 6 | e5758

not provide a core function involved in STX synthesis.
Alternatively, the function of sxtW might have been substituted
in Anabaena circinalis ACBU02 by another gene product.
The other STX genes of cyanobacterial origin (13 sequences in
11 gene families) in which the STX-producing taxa form a well-
supported monophyletic clade that is positioned within cyanobac-
teria (Table 1, group II). However, the clade containing STX+
taxa is relatively evolutionarily distant from its closest STX2
neighbors (Figure 3). These 12 genes can be further subdivided
into four classes based on their closest STX2 cyanobacterial
neighbor and their distribution among different cyanobacteria.
The first comprises three genes that are shared with other
Nostocaceae cyanobacteria similar to that of the non-saxitoxin-
specific genes identified above, except that STX+ taxa form a
clade that excludes STX2 taxa. The three genes in this group
(Figure S1) are saxitoxin-binding protein (sxtP), short-chain alcohol
dehydrogenase (sxtU), and succinate dehydrogenase/fumarate
reductase (sxtV). The second class is a set of three genes that are
shared with Chroococcales cyanobacteria (e.g., Synechococcus sp).
This group includes two genes that encode toxic compound
extrusion proteins (sxtM and sxtF) and acyl-CoA N-acyltransferase
(sxtR). The third class includes genes that are shared with
Oscillatoriales cyanobacteria, specifically, Lyngbya sp., that belongs
to the same genus of the STX+ Lyngbya wollei [14] and/or the
bloom-forming and neurotoxin-producing Trichodesmium erythraeum
[15]. This group is comprised of seven genes: sterole desaturase
(sxtD), GDSL-lipase (sxtL), O-carbamoyltransferase (sxtI), sulfo-
transferase (sxtN), cephalosporin hydroxylase (sxtX), and two
hypothetical genes (sxtJ and sxtK). The phylogeny of sxtN
(Figure 3) provides an example of a STX+ gene of a non-
Nostocaceae cyanobacterial provenance. SxtN is postulated to
encode a sulfotransferase that carries out the transfer of the sulfate
group from 39-phosphoadenosine 59-phosphosulfate (PAPS) to the
carbamoyl group of saxitoxin compounds, i.e., based on work
done with the STX+ dinoflagellate Gymnodinium catenatum [3].
However, unlike other STX genes, we could not identify a
homolog of sxtX in ACBU02 (although it is encoded by three other
STX+ cyanobacteria, Cylindrospermopsis raciborskii T3, Lyngbya wollei
Carmichael/Alabama, and Aphanizomenon flos-aquae NH-5). This
result is consistent with the prediction of Kellmann et al. [6], that
the loss of sxtX from Anabaena circinalis (e.g., ACBU02) explains the
inability of the toxic strains of this species to produce N-1-
hydroxylated analogs of STX.
It is important to note that the different phylogenetic patterns
observed in this group of genes could be explained by inter-phylum
HGT, whereby the STX genes were horizontally transferred among
specific cyanobacterial phyla, including the common ancestor of
STX+ cyanobacterium. Alternatively, there may have been inde-
Table 1. Genes involved in the biosynthesis of saxitoxin in cyanobacteria, their putative functional annotation, and origin
STX Gene Accession Annotation Origin Monophyly Group
sxtO ABI75115 Adenylylsulfate kinase Cyanobacteria N I
sxtW ABI75106 Ferredoxin Cyanobacteria N I
sxtY ABI75117 Phosphate uptake regulator Cyanobacteria N I
sxtZ ABI75118 Two-component sensor histidine kinase Cyanobacteria N I
sxtD ABI75089 Sterole desaturase Cyanobacteria Y II
sxtF ABI75096 Toxic compound extrusion protein Cyanobacteria Y II
sxtI ABI75099 O-carbamoyltransferase Cyanobacteria Y II
sxtJ ABI75100 Hypothetical protein Cyanobacteria Y II
sxtK ABI75101 Hypothetical protein Cyanobacteria Y II
sxtL ABI75102 GDSL-lipase Cyanobacteria Y II
sxtM ABI75103 Toxic compound extrusion protein Cyanobacteria Y II
sxtN ABI75104 Sulfotransferase Cyanobacteria Y II
sxtP ABI75114 Saxitoxin-binding protein Cyanobacteria Y II
sxtR ABI75112 Acyl-CoA N-acyltransferase Cyanobacteria Y II
sxtU ABI75108 Short-chain alcohol dehydrogenase Cyanobacteria Y II
sxtV ABI75107 Fumarate reductase Cyanobacteria Y II
sxtX ABI75105 Cephalosporin hydroxylase Cyanobacteria Y II
sxtB ABI75093 Cytidine deaminase Proteobacteria Y III
sxtE ABI75095 Chaperone-like protein Proteobacteria Y III
sxtG ABI75097 Amidinotransferase Proteobacteria Y III
sxtQ ABI75113 Unknown Proteobacteria Y III
sxtS ABI75110 Phytanoyl-CoA dioxygenase Proteobacteria Y III
sxtC ABI75092 Amidohydrolase Firmicutes Y IV
sxtH ABI75098 Phenylpropionate dioxygenase Unknown Y V
sxtT ABI75109 Phenylpropionate dioxygenase Unknown Y V
sxtA ABI75094 Polyketide synthase Chimeric Y VI
The rows are grouped based on the origin of the genes and the monophyly of the STX+ taxa for the respective gene as following: I = Cyanobacteria and not
monophyletic, II = Cyanobacteria and monophyletic, III = Proteobacteria, IV = Firmicutes, V =Unknown, VI = Chimeric.
doi:10.1371/journal.pone.0005758.t001
Origin of Saxitoxin Genes
PLoS ONE | www.plosone.org 4 June 2009 | Volume 4 | Issue 6 | e5758

Figure 2. Phylogeny of sxtY and sxtZ. ML phylogenetic trees of (a) sxtY (phosphate uptake regulator) and (b) sxtZ (histidine kinase). These
represent the class of saxitoxin-related genes of cyanobacterial origin in which taxa most closely related to the STX+ strains are members of the
Nostocaceae. ML bootstrap values (when $50%) are indicated at the nodes. The branch lengths are proportional to the number of substitutions per
site (see scales in the figure). Cyanobacterial taxa are shown in blue text, non-cyanobacterial prokaryotes in black text, and eukaryotes (Plantae) in
green except for the photosynthetic filose amoeba Paulinella chromatophora in red. The trees have been rooted arbitrarily.
doi:10.1371/journal.pone.0005758.g002
Origin of Saxitoxin Genes
PLoS ONE | www.plosone.org 5 June 2009 | Volume 4 | Issue 6 | e5758