characterisation of bacterial communities associated with ......triplicate cell counts using a...
TRANSCRIPT
Characterisation of bacterial communities associated with toxic andnon-toxic dino£agellates:
Alexandrium spp. and Scrippsiella trochoidea
Georgina L. Hold a;*, Elizabeth A. Smith a, Michael S. Rappe b;1,Elizabeth W. Maas c;2, Edward R.B. Moore d, Carsten Stroempl d, John R. Stephen e;3,
James I. Prosser e, T. Harry Birkbeck f , Susan Gallacher a
a Fisheries Research Services, Marine Laboratory, Victoria Road, Aberdeen AB11 9DB, UKb Station Biologique, Centre National de la Research Scienti¢que et Universite Pierre et Marie Curie, BP 74, 29682 Rosco¡ Cedex, France
c Department of Microbiology, University of Otago, P.O. Box 56, Dunedin, New Zealandd Division of Microbiology, National Research Centre for Biotechnology (GBF), 38124 Braunschweig, Germany
e Department of Molecular and Cell Biology, Institute of Medical Sciences, University of Aberdeen, Aberdeen AB25 2ZD, UKf Division of Infection and Immunity, Institute of Biomedical and Life Sciences, University of Glasgow, Glasgow G12 8QQ, UK
Received 27 February 2001; received in revised form 12 July 2001; accepted 13 July 2001
First published online 8 August 2001
Abstract
Several dinoflagellate species have been shown to produce potent neurotoxins known as paralytic shellfish toxins. Evidence is alsoaccumulating that marine bacteria associated with dinoflagellates play a role in the accumulation of paralytic shellfish toxins. In this study,the diversity of bacteria in cultures of both toxic and non-toxic dinoflagellates, Alexandrium spp. and Scrippsiella trochoidea, were comparedusing colony morphology, restriction fragment length polymorphisms, denaturing gradient gel electrophoresis of PCR-amplified 16S rRNAgenes and, ultimately, sequence determination of the 16S rRNA genes. The results suggest that a number of different bacterial species areassociated with dinoflagellates, some of which are common to each of the dinoflagellate cultures examined, whereas others appear to beunique to a particular dinoflagellate. The phylogenetic diversity of the bacteria observed was limited to two bacterial phyla, theProteobacteria and the Cytophaga-Flavobacter-Bacteroides (CFB). Although phylum level diversity was limited, many distinctphylogenetic clades were recovered, including members of both the K- and Q-subclasses of the Proteobacteria. Additionally, several ofthe bacterial phylotypes isolated were not closely related to any published bacterial species but, rather, were identical to isolatescharacterised from Alexandrium cultures 4 years earlier. Finally, many of the bacteria isolated from the dinoflagellate cultures were relatedto microorganisms with known surface-associated life histories (e.g. the CFB phylum, Hyphomonas, Caulobacter and some members of theRoseobacter clade including Ruegeria algicola). ß 2001 Federation of European Microbiological Societies. Published by Elsevier ScienceB.V. All rights reserved.
Keywords: Dino£agellate ; Marine bacterium; Paralytic shell¢sh toxin; Polymerase chain reaction-restriction fragment length polymorphism; Polymerasechain reaction-denaturing gradient gel electrophoresis
1. Introduction
Interactions between algae and bacteria are commonlyobserved in both freshwater and marine ecosystems withbacteria increasingly cited as potentially important regula-tors in processes of algal bloom initiation, maintenanceand decline [1]. Bacteria are an inherent part of the phys-ical environment of dino£agellates, both in vitro and invivo and, as such, may be considered dino£agellate sym-bionts [2,3]. However, the ecological signi¢cance of mostnaturally occurring bacterial^algal associations is unclear
0168-6496 / 01 / $20.00 ß 2001 Federation of European Microbiological Societies. Published by Elsevier Science B.V. All rights reserved.PII: S 0 1 6 8 - 6 4 9 6 ( 0 1 ) 0 0 1 5 7 - X
* Corresponding author. Present address: Rowett Research Institute,Greenburn Road, Bucksburn, Aberdeen AB21 9SB, UK. Tel. : +44 (1224)71 27 51; Fax: +44 (1224) 71 66 87.
E-mail address: [email protected] (G.L. Hold).
1 Present address: Department of Microbiology, 222 Nash Hall,Oregon State University, Corvallis, OR 97331, USA.
2 Present address: Institute of Environmental Science and ResearchLimited, P.O. Box 50-348, Porirua, New Zealand.
3 Present address: Crop and Weed Science, Horticulture ResearchInternational, Wellesbourne, Warwick CV35 9EF, UK.
FEMSEC 1273 9-10-01
FEMS Microbiology Ecology 37 (2001) 161^173
www.fems-microbiology.org
and, in most cases, the bacterial species involved have notbeen identi¢ed. This is particularly the case with algaeassociated with the occurrence of toxins in shell¢sh, suchas the paralytic shell¢sh toxin (PST) producing Alexan-drium spp.
Although the value of 16S rRNA gene sequence datafor identifying bacteria is rarely disputed, this informationis severely limited for bacteria associated with toxic marinedino£agellates. A recent study by Prokic et al. [4] inves-tigated the diversity of bacteria associated with Prorocen-trum lima, a known diarrhoetic shell¢sh poison producingdino£agellate, via microbial cultivation and 16S rRNAgene cloning and sequencing. The authors found Roseo-bacter-related microorganisms of the K-Proteobacteria todominate the P. lima micro£ora. Babinchak et al. [5] used£uorescently labelled 16S rRNA targeted oligonucleotideprobes to identify bacteria from toxic Alexandrium spp. asmembers of the K- and Q-Proteobacteria. However, resolu-tion in this study was low, and a more detailed character-isation would be required to resolve any di¡erences be-tween the bacterial populations of toxic and non-toxicdino£agellates. Several authors have also reported thatbacteria isolated from Alexandrium spp. cultures are capa-ble of producing PST but, again, limited information ontheir identity in the form of 16S rRNA gene sequence datais available [6^9].
The issue of bacterial PST toxigenesis is currentlynot resolved [3]. However, previous studies in our labora-tory have indicated that the bacteria associated with cul-tures of the dino£agellates Alexandrium lusitanicumNEPCC 253 and Alexandrium tamarense NEPCC 407in£uence the toxicity of these dino£agellates [10]. In orderto elucidate the contribution of particular bacterial speciesto dino£agellate/shell¢sh toxicity and bloom dynamics,it is essential that the ecology of bacteria associated withmarine dino£agellates such as Alexandrium spp. bestudied. As a prelude to investigating algal^bacterialdynamics in the environment, it is necessary to establishwhich bacterial taxa are involved, initially requiringhigh resolution characterisation of the bacteria associatedwith dino£agellate cultures. To this aim, using a com-bination of phenotypic (colony morphology) andmolecular-based techniques (16S rDNA restrictionfragment length polymorphisms (RFLP), denaturinggradient gel electrophoresis (DGGE) and, ultimately, se-quence determination of PCR-ampli¢ed 16S rRNA genes),comparisons of the diversity of bacteria associated withPST producing dino£agellate strains, A. lusitanicumNEPCC 253 and A. tamarense NEPCC 407 [11^13] andthose of A. tamarense PCC 173a and Scrippsiella trochoi-dea NEPCC 15 which have no previously reportedlink with toxicity [14], were made. Furthermore compar-isons were made with bacterial isolates obtained fromAlexandrium spp. cultures 4 years prior to this investiga-tion [6].
2. Materials and methods
2.1. Dino£agellate cultures
The dino£agellates A. lusitanicum NEPCC 253,A. tamarense NEPCC 407, A. tamarense PCC 173a andS. trochoidea NEPCC 15 were cultured in Guillards f/2without silicate (Sigma, Dorset, UK) enriched seawater[15]. Cultures were maintained at 15³C with a 14:10 h,light:dark cycle (irradiance level 0.5^1.5U1016 quantas31 cm32). Culture development was monitored daily bytriplicate cell counts using a Sedgwick^Rafter countingchamber and visualised by the addition of Lugol's iodine[16]. NEPCC strains were supplied by the North EastPaci¢c Culture Collection, British Columbia, Canadaand A. tamarense PCC 173a was from the Plymouth Cul-ture Collection, Plymouth, UK.
2.2. Isolation of bacteria from dino£agellate cultures
Samples (1 ml) of dino£agellate cultures from early, lateexponential and stationary growth phases were seriallydiluted (10-fold dilutions; neat ^ 1039) in sterile seawater.Aliquots (100 Wl) of each dilution were spread, in tripli-cate, onto marine agar medium (Difco 2216, Michigan,USA) and subsequently incubated for 14 days at 20³C.Bacteria from the dilution containing between 50 and100 colonies were isolated from one of the replicate platesand replated individually onto marine agar, to obtain purecultures. Resultant bacterial isolates were categorised bycolony morphology and examined microscopically fortheir reaction to Gram stain. Isolates were stored in ma-rine broth plus 10% (v/v) glycerol at 370³C, prior toRFLP investigation.
2.3. RFLP analysis
Bacteria isolated from dino£agellate cultures were resus-pended in ATL bu¡er (Qiagen, Dorking, UK) and ge-nomic DNA from each isolate was extracted followingthe QIAmp tissue protocol (Qiagen, Dorking, UK).DNA was stored at 320³C prior to PCR ampli¢cation.Small subunit (SSU) 16S rDNA was PCR-ampli¢ed fromtotal genomic DNA, using the universal eubacterial prim-ers 27F and 1522R (Bioline, London, UK) [17]. In a ¢nalvolume of 50 Wl, the reaction mixture contained 100 pmolof each ampli¢cation primer, 100 WM each deoxynucleo-side triphosphate, 1.5 mM MgCl2, 10 ng template and 5 Uof Taq DNA polymerase (Bioline, London, UK), whichwas added after a pre-cycling step in which the reactionmixture was heated to 95³C for 5 min. Ampli¢cation con-ditions were 95³C for 1 min, 55³C for 1 min, and 72³C for1 min, for a total of 35 cycles followed by 72³C for 10 min[17]. Ampli¢cation products were visualised by electropho-resis using a 2% (w/v) agarose gel (Sigma, Dorset, UK)
FEMSEC 1273 9-10-01
G.L. Hold et al. / FEMS Microbiology Ecology 37 (2001) 161^173162
containing ethidium bromide (0.5 mg ml31), with a 1-kbmarker (Gibco, Glasgow, UK) included on the gel forreference.
To generate RFLP patterns, approx. 700 ng of eachPCR product was digested with 5 units of the restrictionendonuclease HaeIII (Promega, Southampton, UK), fol-lowing the manufacturer's speci¢cations. Reactions werestopped by the addition of EDTA, and fragments wereresolved by gel electrophoresis in 2.5% low melting pointagarose gel (Gibco, Glasgow, UK), containing ethidiumbromide (0.5 g ml31). DNA fragments were visualisedunder UV illumination with RFLP patterns grouped usingPhoretix ID v 4 computer software (Phoretix Internation-al, Newcastle upon Tyne, UK).
2.4. DGGE analysis of 16S rDNA fromdino£agellate cultures
Cell pellets from 1-l cultures, at early, late exponentialand stationary growth phases, of A. lusitanicum NEPCC253, A. tamarense NEPCC 407 and PCC 173a andS. trochoidea NEPCC 15 were obtained by centrifugationat 10 000Ug for 10 min. The cell pellet was resuspended inTE bu¡er (1 ml) and stored at 320³C until extraction ofDNA. Following thawing, glass beads (0.5 g) of 0.16^1.17mM diameter (Sigma, Dorset, UK) were added to thesamples, which were alternately vortexed and placed onice for 60 s, until visual signs of dino£agellate cell lysiswere observed. Following addition of 50 Wl lysozyme(50 mg ml31 in distilled water) (Sigma, Dorset, UK),samples were incubated for 30 min at room temperature.Sodium dodecyl sulfate (SDS) (100 Wl, 10% (w/v) in dis-tilled water) (Sigma, Dorset, UK) was mixed with eachsample, followed by addition of 50 Wl proteinase K(20 mg ml31 in distilled water) (Sigma, Dorset, UK) andfurther incubation at room temperature for 30^60 minbefore phenol/chloroform extraction [18].
Bacterial 16S rDNA was ampli¢ed using primers de-scribed by Muyzer et al. [19] which generated ampli¢-cation products corresponding to nucleotide positions341^534 in Escherichia coli. Ampli¢cation conditionswere 94³C for 30 s, 55³C for 45 s, and 72³C for 30 s,for 30 cycles followed by 72³C for 10 min. Ampli¢ca-tion products were visualised by electrophoresis usinga 2% (w/v) agarose gel (Sigma, Dorset, UK) con-taining ethidium bromide (0.5 mg ml31), with a 100-bpmarker (Gibco, Glasgow, UK) included on the gel forreference.
DGGE was performed with a D-code 16/16 cm gel sys-tem using a 1.0 mm gel width (Bio-Rad, Herts, UK) main-tained at a constant temperature of 65³C in 6 l of 1UTAEbu¡er (20 mM Tris acetate, 0.5 mM EDTA, pH 8.0).Gradients were formed between 20 and 60% denaturant(with 100% denaturant de¢ned as 7 M urea and 40% (v/v)formamide). Gels were run at 130 V for 4^6 h and stained
in 1UTAE containing ethidium bromide (0.5 mg l31) be-fore destaining in distilled water for 15 min. Pro¢les wereinspected under UV illumination.
DGGE bands were excised using a sterile razor bladeand DNA eluted overnight in 100 Wl of TE bu¡er. DNAwas recovered using the Wizard PCR prep DNA puri¢ca-tion system (Promega, Southampton, UK), with 5-Wl ali-quots of DNA from each excised band used as template inPCR ampli¢cations as described above. PCR productswere puri¢ed using Qiagen PCR spin columns (Qiagen,Dorking, UK).
2.5. 16S rDNA sequence determination andphylogenetic analysis
Representative bacterial isolates, from dino£agellatecultures, possessing unique colony morphologies orRFLP patterns and bacteria previously isolated from di-no£agellate strains A. tamarense UW2c, A. tamarenseUW4, A. tamarense NEPCC 407, A. lusitanicum NEPCC253 and Alexandrium a¤ne NEPCC 667 [6] were selectedfor DNA sequence analysis. Several bacterial isolates fromeach group were partially sequenced in order to con¢rmcolony morphology and RFLP analyses were e¡ective atclassifying the cultured isolates. However, only one repre-sentative sequence from each group, spanning nearly theentire 16S rRNA gene, was submitted to the database. 16SrRNA gene PCR amplicons were puri¢ed using Qiaquick-spin columns (Qiagen, Dorking, UK) and subsequentlysequenced bidirectionally, using universal 16S rRNAprimers [20]. PCR amplicons obtained from DGGE pro-¢les were also sequenced bidirectionally, using primers de-scribed by Muyzer et al. [19], on a 373A automated DNAsequencer using Taq polymerase initiated cycle sequencingwith £uorescent dye-labelled dideoxynucleotides with pro-tocols recommended by the manufacturer (Perkin Elmer,Bucks, UK). Nucleotide sequence fragments generatedwere assembled into consensus sequences using the GCGpackage v 7.0-UNIX [21].
The automatic alignment function of the ARB sequenceanalysis software package [22] was used initially to alignthe 16S rRNA gene sequences with reference sequencesobtained from GenBank [23], the Ribosomal DatabaseProject (RDP) [24], and the ARB database (Departmentof Microbiology, Technical University of Munich, Mu-nich, Germany). The Genetic Data Environment (GDE)v 2.2 [25] sequence analysis software package was usedsubsequently to re¢ne the alignment generated by ARBmanually. Pairwise sequence similarities were generated,using GDE, and individual sequence masks employedfor each pair of sequences analysed.
2.6. Nucleotide sequence accession numbers
Nucleotide sequences from the cultured bacteria were
FEMSEC 1273 9-10-01
G.L. Hold et al. / FEMS Microbiology Ecology 37 (2001) 161^173 163
¢led in GenBank under the following accession numbers:253-13 (AJ294351), 253-16 (AJ294352), 667-2 (AJ294353),667-4 (AJ294354), 667-19 (AJ294355), 667-12 (AJ294356),4KVS17 (AJ294358), 407-13 (AJ294359), 407-2(AJ294360), 2C3 (AJ294361), 253-19 (AJ294362), 667-16(AJ296288), ALUS253_28 (AF359524), ATAM407_61(AF359525), ALUS253_43 (AF359526), ALUS253_25(AF359527), ATAM407_68 (AF359528), ATAM407_18(AF359529), ATAM407_62 (AF35930), ATAM173a_16(AF359531), ATAM173a_17 (AF359532), ATAM407_58(AF359533), SCRIPPS_732 (AF359534), ATAM407_56(AF359535), SCRIPPS_101 (AF359537), ATAM173a_51(AF359538), ATAM173a_2 (AF359539), ATAM173a_3(AF359540), ATAM173a_6 (AF359541), ATAM173a_5(AF359542), ATAM173a_36 (AF359543), ATAM173a_9(AF359544), SCRIPPS_94 (AF359545), SCRIPPS_739(AF359546), SCRIPPS_423 (AF359547), SCRIPPS_413(AF359548), SCRIPPS_426 (AF359549), SCRIPPS_740(AF359550), ALUS253_6 (AF359551), ATAM407_2035(AF359552).
3. Results
3.1. Diversity of culturable bacteria
Table 1 lists the 16S rDNA sequences most closely re-lated to those of the 12 bacterial strains previously isolatedfrom stationary phase batch cultures of Alexandrium spp.by Gallacher et al. [6]. All strains examined were clearlyplaced in the K- and Q-Proteobacteria subphyla, with theexception of strain 667-16, which was Gram-positive andmost closely related to Micrococcus luteus. Other bacteriaisolated from the weakly toxic A. a¤ne NEPCC 667 be-longed to the Roseobacter clade. Some bacteria isolatedfrom A. lusitanicum NEPCC 253 were also associatedwith the Roseobacter clade and showed high sequence sim-ilarities with bacterial 16S rRNA gene clones recoveredfrom a culture of the toxic dino£agellate P. lima [4]. How-ever, these strains were found to be related only distantlyto the original Roseobacter algicola (now Ruegeria algico-la) isolated from the phycosphere of P. lima (data not
Fig. 1. Distribution of bacteria associated with the three growth phases of the dino£agellates A. lusitanicum NEPCC 253, A. tamarense NEPCC 407, A.tamarense PCC 173a and S. trochoidea NEPCC15. Results are expressed as mean values from triplicate analyses þ S.D.
FEMSEC 1273 9-10-01
G.L. Hold et al. / FEMS Microbiology Ecology 37 (2001) 161^173164
shown). The 16S rDNA sequences of other bacterialstrains isolated from A. lusitanicum, A. tamarense NEPCC407 and UW2c, possessed high sequence similarities withthose of the Q-Proteobacterium Alteromonas macleodii, acommon inhabitant of coastal water and open oceans [26],and Marinobacter aquaeolei, originally isolated from theNorth Sea [27].
All bacteria isolated from Alexandrium spp. and S. tro-choidea cultures in this investigation were identi¢ed asGram-negative rods, with the majority of isolates (90%)
forming distinct non-pigmented colonies ; other bacteriaexhibited pink, brown and yellow pigmentation followinggrowth on marine agar (Table 2). Using colony morphol-ogy and RFLP pro¢les, the cultured bacteria were cate-gorised into 26 groups (Fig. 1, Table 2). Based on 16SrDNA sequence analysis of representative isolates fromeach RFLP group, bacteria were identi¢ed as belongingto the K-Proteobacteria (with many sequences falling with-in the Roseobacter clade), Q-Proteobacteria and the Cyto-phaga-Flavobacter-Bacteroides (CFB) phylum (Table 2).
Table 116S rDNA sequence comparison of strains previously isolated from toxic Alexandrium spp. cultures, with 16S rDNA database sequence entries
Bacterialstrain
Phylogenetic group Most closely related sequencefrom a validly describedbacterial species (GenBankaccession No.)a
Sequencesimilarity(%)
Most closely related databasesequence (GenBank accessionNo.)b
Sequencesimilarity(%)
A. a¤ne NEPCC 667667-2 K-Proteobacteria (Roseobacter
clade)Roseobacter gallaeciensis (Y13244) 96.9 ATAM407_56 100
667-4 K-Proteobacteria (Roseobacterclade)
Staleya guttiformis (Y16427) 99.2 ATAM173a_16 100
667-12 K-Proteobacteria (Roseobacterclade)
Antarctobacter heliothermus(Y115522)
99.4 ATAM173a_51 100
667-19 K-Proteobacteria (Roseobacterclade)
Sul¢tobacter pontiacus (Y13155) 95.7 env.PRLIST06 (Y15343) geneclone recovered fromProrocentrum lima
100
667-16 Gram-positive Micrococcus luteus (AJ296288) 98.6 ^* ^A. tamarense NEPCC 407407-2 Q-Proteobacteria Alteromonas macleodii (L10938) 98.7 ATAM407_18 100407-13 Q-Proteobacteria Marinobacter aquaeolei
(AF173969)99.2 ATAM407_2035 99.8
A. lusitanicum NEPCC 253253-13 K-Proteobacteria (Roseobacter
clade)Rosevarius tolerans (Y11551) 96.4 ATAM407_61 100
env. kpc 34 (AF194397) geneclone recovered from Aureococcusanophage¡erens
99.9
env.PRLISY03 (Y15348) geneclone recovered fromProrocentrum lima
99.7
253-16 K-Proteobacteria (Roseobacterclade)
Ruegeria gelatinovora (D88523) 96.8 env.PRLISY01 (Y15346) geneclone recovered fromProrocentrum lima
99.9
env.PRLIST02 (Y15339) geneclone recovered fromProrocentrum lima
99.9
env.kpc101f (AF194394) geneclone recovered from Aureococcusanophage¡erens
99.9
253-19 Q-Proteobacteria Alteromonas macleodii (L10938) 98.3 ^ ^A. tamarense UW2c2c3 Q-Proteobacteria Alteromonas macleodii (L10938) 98.3 Alteromonas macleodii related
isolate CH-516 (Y18234)99.6
endocytic bacterium Noc15(AF262741)
99.4
A. tamarense UW4c4Kvs17 K-Proteobacteria Aquaspirillum itersonii (Z29620) 89.5 endocytic bacterium Noc17
(AF262750)99.5
^* indicates validly published sequence showed highest sequence similarity.aMost closely related validly published sequence indicates the de¢ned type strain with highest 16S rDNA sequence similarity to each particular isolate.bMost closely related database sequence indicates the 16S rDNA sequence within the database which has highest 16S rDNA sequence similarity to eachparticular isolate.
FEMSEC 1273 9-10-01
G.L. Hold et al. / FEMS Microbiology Ecology 37 (2001) 161^173 165
Tab
le2
Cha
ract
eris
atio
nof
bact
eria
isol
ated
onm
arin
eag
ar,
from
the
four
dino
£age
llate
cult
ures
A.
lusi
tani
cum
NE
PC
C25
3,A
.ta
mar
ense
NE
PC
C40
7,A
.ta
mar
ense
PC
C17
3aan
dS
.tr
ocho
idea
NE
PC
C15
RF
LP
patt
ern
Col
ony
mor
phol
ogy
Rep
rese
ntat
ive
isol
atea
Num
ber
ofsi
mila
rba
cter
iase
quen
ced
Mos
tcl
osel
yre
late
dse
quen
cefr
oma
valid
lyde
scri
bed
bact
eria
lsp
ecie
s(G
enB
ank
acce
ssio
nN
o.)b
Sequ
ence
sim
ilari
ty(%
)
Mos
tcl
osel
yre
late
dda
taba
sese
quen
ce(G
enB
ank
acce
ssio
nN
o.)c
Sequ
ence
sim
ilari
ty(%
)
K-P
rote
obac
teri
a1
circ
ular
,sm
ooth
surf
ace,
umbo
nate
,cr
eam
AL
US2
53_2
87
Ahr
ensi
aki
elen
se(D
8852
4)98
.5^*
^
2ir
regu
lar,
wit
hlo
bate
mar
gin,
crea
mA
LU
S253
_25
1S
tapp
iast
ellu
lata
(D88
525)
98.6
PT
B1
(Y10
913)
99.7
PT
B5
(AJ0
0064
6)99
.43
circ
ular
,sm
ooth
,ra
ised
,cr
eam
AT
AM
407_
681
Phy
lloba
cter
ium
myr
sina
cear
um(D
1278
9)96
.8en
v.R
SHD
3S10
(AF
1902
14)
gene
clon
efr
omP
¢est
eria
dino
£age
llate
cult
ure
CC
MP
1829
97.9
env.
HN
SS21
from
mar
ine
spon
ge(Z
8857
2)97
.0
4ir
regu
lar
wit
hun
dula
tem
argi
n,co
nvex
,br
own
cent
rew
ith
crea
mm
argi
n
SCR
IPP
S_94
3A
quas
piri
llum
iter
soni
i(Z
2962
0)89
.5en
docy
tic
bact
eriu
mN
oc17
(AF
2627
50)
99.5
4Kvs
17(s
eeT
able
1)99
.15
circ
ular
,en
tire
,co
nvex
,sm
ooth
,br
own
cent
rew
ith
crea
mm
argi
nSC
RIP
PS_
739
2H
ypho
mon
asjo
hnso
nii
(AF
0827
91)
98.7
^^
6ci
rcul
ar,
pulv
inat
e,en
tire
,sm
ooth
,w
hite
cent
rew
ith
crea
mm
argi
nSC
RIP
PS_
423
2H
ypho
mon
asoc
eani
tis
(AF
0827
97)
90.9
uncu
ltur
edba
cter
ium
BU
RT
ON
-39
(AF
1428
57)
96.3
7ci
rcul
ar,
pulv
inat
e,sm
ooth
,cr
eam
SCR
IPP
S_42
61
Cau
loba
cter
vibr
ioid
es(A
J227
754)
93.7
uncu
ltur
edC
aulo
bact
er5C
2(A
F24
5033
)94
.8
K-P
rote
obac
teri
a(R
oseo
bact
ercl
ade)
8ci
rcul
ar,
conv
ex,
crea
mw
ith
light
pink
cent
reA
LU
S253
_13
Rue
geri
aal
gico
la(X
7831
3)94
.525
3-13
(see
Tab
le1)
99.9
env.
PR
LIS
Y03
(Y15
348)
gene
clon
ere
cove
red
from
Pro
roce
ntru
mlim
a99
.7
AT
AM
407_
612
Ros
eova
rius
tole
rans
(Y11
551)
96.5
env.
kpc
34(A
F19
4397
)ge
necl
one
reco
vere
dfr
omA
ureo
cocc
usan
opha
ge¡
eren
s
99.9
9ci
rcul
ar,
conv
ex,
crea
mw
ith
light
pink
cent
reA
LU
S253
_43
10R
oseo
vari
usto
lera
ns(Y
1155
1)99
.6^
^
AT
AM
407_
254
asab
ove
10ir
regu
lar
wit
han
undu
late
mar
gin,
smoo
th,
conv
ex,
dark
brow
nA
TA
M40
7_62
2O
ctad
ecab
acte
ran
tarc
ticu
s(U
1458
3)94
.3en
v.D
049
(AF
1775
68)
gene
clon
ere
cove
red
from
seaw
ater
95.5
env.
EK
H01
8(A
F14
2900
)ge
necl
one
reco
vere
dfr
omse
awat
er95
.4
env.
MC
-1(A
F18
8164
)ge
necl
one
from
P¢e
ster
iasp
.95
.4
11ir
regu
lar
wit
hlo
bate
mar
gin,
conv
ex,
smoo
th,
brow
nce
ntre
AT
AM
407_
581
Sul
¢tob
acte
rm
edit
erra
neus
(Y17
387)
98.7
^10
0
AT
AM
173a
_16
1S
tale
yagu
ttif
orm
is(Y
1642
7)99
.266
7-4
(see
Tab
le1)
^A
TA
M17
3a_1
71
Sul
¢tob
acte
rm
edit
erra
neus
(Y17
387)
99.7
^99
.9A
TA
M17
3a_4
93
Sul
¢tob
acte
rpo
ntia
cus
(Y13
155)
96S
ul¢t
obac
ter
sp.
DSS
-2(A
F09
8490
)99
.9SC
RIP
PS_
732
2S
ul¢t
obac
ter
pont
iacu
s(Y
1315
5)99
.8S
ul¢t
obac
ter
sp.
DSS
-2(A
F09
8490
)10
0
FEMSEC 1273 9-10-01
G.L. Hold et al. / FEMS Microbiology Ecology 37 (2001) 161^173166
Tab
le2
(con
tinu
ed)
RF
LP
patt
ern
Col
ony
mor
phol
ogy
Rep
rese
ntat
ive
isol
atea
Num
ber
ofsi
mila
rba
cter
iase
quen
ced
Mos
tcl
osel
yre
late
dse
quen
cefr
oma
valid
lyde
scri
bed
bact
eria
lsp
ecie
s(G
enB
ank
acce
ssio
nN
o.)b
Sequ
ence
sim
ilari
ty(%
)
Mos
tcl
osel
yre
late
dda
taba
sese
quen
ce(G
enB
ank
acce
ssio
nN
o.)c
Sequ
ence
sim
ilari
ty(%
)
12ci
rcul
arw
ith
enti
rem
argi
n,sm
ooth
,cr
eam
AT
AM
407_
567
Ros
eoba
cter
galla
ecie
nsis
(Y13
244)
96.9
667-
2(s
eeT
able
1)10
0
circ
ular
wit
her
ose
mar
gin,
gran
ular
,um
bona
tew
ith
abr
own
cent
rean
dpa
lem
argi
n
SCR
IPP
S_10
12
Sul
¢tob
acte
rbr
evis
(Y16
425)
97.2
Sul
¢tob
acte
rsp
.D
SS-2
(AF
0984
90)
97.3
13ci
rcul
ar,
smoo
th,
gela
tino
us,
rais
ed,
crea
mA
TA
M17
3a_5
12
Ant
arct
obac
ter
helio
ther
mus
(Y11
552)
99.6
667-
12(s
eeT
able
1)10
0
14pu
ncti
form
,ge
lati
nous
,pi
nkA
TA
M17
3a_9
2R
oseo
vari
usto
lera
ns(Y
1155
1)96
.3en
v.P
RL
ISY
03(Y
1534
8)ge
necl
one
reco
vere
dfr
omP
roro
cent
rum
lima
98.3
Q-P
rote
obac
teri
a15
circ
ular
,sm
ooth
,ge
lati
nous
,£a
t,cr
eam
AT
AM
407_
183
Alt
erom
onas
mac
leod
ii(X
8214
5)98
.940
7-2
(see
Tab
le1)
99.9
2c3
(see
Tab
le1)
97.9
253-
19(s
eeT
able
1)97
.716
circ
ular
,co
nvex
,ge
lati
nous
,ye
llow
AT
AM
173a
_51
Alt
erom
onas
mac
leod
ii(A
F17
3965
)93
.7po
lar
gas
vacu
olat
est
rain
214.
6(U
7372
4)97
.1
17ci
rcul
arm
ucoi
d,ge
lati
nous
,la
rge
crea
mA
TA
M40
7_20
352
Mar
inob
acte
raq
uaeo
lei
(AF
1739
69)
99.3
Mar
inob
acte
rsp
.D
S40M
8(A
F19
9440
)99
.8
407-
13(s
eeT
able
1)99
.8A
TA
M17
3a_3
61
Pse
udoa
lter
omon
asha
lopl
ankt
issu
bsp.
tetr
aodo
nis
(AF
2147
30)
99.8
^^
18ir
regu
lar,
loba
te,
£at,
yello
wSC
RIP
PS_
740
1P
seud
omon
asst
utze
ri(U
2641
5)99
.2un
cult
ured
gam
ma
prot
eoba
cter
ium
(AB
0108
54)
99.4
Cyt
opha
ga-F
lavo
bact
er-B
acte
roid
es(C
FB
)19
circ
ular
,ge
lati
nous
,ra
ised
,br
ight
yello
wA
TA
M17
3a_2
2P
olar
ibac
ter
irge
nsii
(M61
002)
96.6
EB
AC
34(A
F26
8227
)ge
necl
one
isol
ated
from
seaw
ater
98.0
AT
AM
173a
_33
Gel
idib
acte
ral
gens
(U62
914)
)92
.3m
arin
eps
ychr
ophi
leSW
17(A
F00
1368
)94
.6
20ci
rcul
ar,
£at,
yello
wA
TA
M17
3a_6
1C
ellu
loph
aga
ulig
inos
a(M
6279
9)93
.7C
ytop
haga
sp.
KT
02ds
18(A
F23
5126
)98
.1
21ci
rcul
ar,
conv
ex,
smoo
th,
crea
mSC
RIP
PS_
413
1P
sych
rose
rpen
sbu
rton
ensi
s(U
6291
3)93
.1F
lavo
bact
eriu
msp
.5N
-3(A
B03
2514
)94
.822
circ
ular
,en
tire
,ra
ised
,ge
lati
nous
,ye
llow
AL
US2
53_6
2G
elid
ibac
ter
alge
ns(A
F00
1367
)92
.0C
ytop
haga
sp.
(AB
0170
46)
96.5
^*in
dica
tes
valid
lyde
scri
bed
sequ
ence
show
edhi
ghes
tse
quen
cesi
mila
rity
.aR
epre
sent
ativ
eis
olat
esw
ere
obta
ined
from
:A
LU
S_25
3,A
.lu
sita
nicu
mN
EP
CC
253
;A
TA
M_4
07,
A.
tam
aren
seN
EP
CC
407
;A
TA
M_1
73a,
A.
tam
aren
seP
CC
173a
;SC
RIP
PS,
S.
troc
hoid
eaN
EP
CC
15.
bM
ost
clos
ely
rela
ted
sequ
ence
from
ava
lidly
desc
ribe
dba
cter
ial
spec
ies
indi
cate
sth
ede
¢ned
type
stra
inw
ith
high
est
16S
rDN
Ase
quen
cesi
mila
rity
toea
chpa
rtic
ular
isol
ate.
c Mos
tcl
osel
yre
late
dda
taba
sese
quen
cein
dica
tes
the
16S
rDN
Ase
quen
cew
ithi
nth
eda
taba
sew
hich
has
high
est
16S
rDN
Ase
quen
cesi
mila
rity
toea
chpa
rtic
ular
isol
ate.
FEMSEC 1273 9-10-01
G.L. Hold et al. / FEMS Microbiology Ecology 37 (2001) 161^173 167
In general, each colony morphotype was associated witha unique HaeIII RFLP pattern. Exceptions to this obser-vation were bacterial isolates from Alexandrium culturesdescribed as circular, convex and cream-coloured with arose centre, which gave two distinct RFLP patterns (pat-terns 8 and 9) (Table 2). In contrast, examination of 16SrDNA sequence data from representative isolates fromA. tamarense NEPCC 407 and S. trochoidea cultures, cat-egorised as possessing RFLP pattern 12 but showing dif-ferent colony morphologies, indicated diverse phylogeneticneighbours (Table 2). Representative sequences fromRFLP patterns 11, 17 and 19 sharing the same colonycharacteristics also had distant phylogenetic neighbours.
3.2. Bacterial diversity associated with A. lusitanicumNEPCC 253
In this study, A. lusitanicum NEPCC 253 was observedto support the most limited culturable bacterial popula-tion, with bacterial isolates, characterised into ¢ve RFLPpatterns (patterns 1, 2, 8, 9 and 22), being prominent at allstages of sampling (Fig. 1). Sequence analysis of 16SrDNA from representative isolates indicated that the ma-jority of the bacteria associated with this dino£agellatewere K-Proteobacteria. RFLP pattern 8 and 9 isolateswhich grouped within the Roseobacter clade were closelyrelated with only a 2.5% di¡erence in their 16S rDNAsequence over the 1250 base pairs analysed. Bacteria ex-hibiting RFLP pattern 22 with strongest sequence similar-ity to Gelidibacter algens were the only isolates from thisdino£agellate to be located outside the K-Proteobacteriasubphylum (Table 2).
3.3. Bacteria associated with A. tamarense NEPCC 407
As observed in A. lusitanicum, bacteria generatingRFLP patterns 8 and 9 were present at all stages ofgrowth in cultures of A. tamarense NEPCC 407 but atmuch lower numbers (Fig. 1). However, in this culture,bacteria possessing pattern 15, with strong sequence sim-ilarity to the previously isolated 407-2, 2c3 and 253-19(Table 1), and RFLP pattern 12 isolates were also consis-tently present. Pattern 11 isolates were detected only atlate exponential and stationary growth phases of this cul-ture but at low levels (Fig. 1). Other K-Proteobacterialisolates possessing 16S rDNA sequence types formingRFLP patterns 3 and 10 only reached detectable levelsin the stationary growth phase of A. tamarense NEPCC407. Q-Proteobacterial isolates possessing 16S rDNA se-quence types showing banding pattern 17 reached onlylow levels in late exponential samples (Fig. 1), with 16SrRNA gene sequences from these bacteria(ATAM407_2035) almost identical to that of isolate 407-13 (Table 1) and also showing high similarities to 16SrDNA sequences obtained from environmental clone li-braries of S. trochoidea [28].
3.4. Bacteria associated with A. tamarense PCC 173a
Bacteria exhibiting RFLP patterns 8, 11, 13, 14, 16, 19and 20 were consistently present at all growth points ofA. tamarense PCC 173a, while pattern 17 isolates weredetected only during the exponential phase (Fig. 1, Table2). RFLP patterns 13, 16, 19 and 20 were characteristic ofisolates from A. tamarense PCC 173a in the current studywith the representative sequence from pattern 13 identi¢edas an K-Proteobacterium with 100% sequence similarity tothe previously isolated strain 667-12 (Table 1). The pattern16 representative belonged to the Q-Proteobacteria subphy-la and patterns 19 and 20 clustered within the CFB phy-lum.
3.5. Bacteria associated with S. trochoidea NEPCC 15
Of the dino£agellates studied, the culturable bacteriaassociated with S. trochoidea showed the highest degreeof phylogenetic variation over the course of the dino£a-gellate growth cycle (Fig. 1), with 12 di¡erent RFLP pat-terns identi¢ed. Following analysis of the sequence datafrom RFLP pattern representatives, isolates identi¢ed aspattern 12 were clearly distinguished from those bacteriacategorised as pattern 12 from A. tamarense NEPCC 407.Some of the bacteria had 16S rDNA sequences similar oridentical to those obtained from environmental clones ofthis dino£agellate (patterns 4 and 21) [28] or isolate4Kvs17, again previously obtained from Alexandriumspp. (pattern 4) (Tables 1 and 2).
On continued subculture, isolates showing RFLP pat-terns 23^26 could not be maintained and subsequently 16SrDNA phylogenetic analysis was not possible.
3.6. DGGE analysis of dino£agellate cultures
Examination of PCR-ampli¢ed 16S rDNA DGGE pro-¢les demonstrated that di¡erent pro¢les were detected foreach dino£agellate culture (Fig. 2). Each pro¢le remainedconstant throughout the di¡erent growth phases althoughindividual band intensities varied during growth. In allcases, more DGGE bands were identi¢ed than culturedisolates, as distinguished by RFLP patterns and colonymorphologies (Figs. 1 and 2, Table 2). The PCR productsderived from total genomic DNA from A. lusitanicumNEPCC 253 and A. tamarense NEPCC 407 each generated10 distinct DGGE bands, while those of A. tamarensePCC 173a and S. trochoidea NEPCC 15 generated 16bands (Fig. 2). 16S rDNA DGGE bands were excisedfor sequence analysis although some bands failed to re-amplify or contained mixed sequences, so further analysiswas not possible. Thus, overall, DGGE pro¢ling did notdetect the same diversity of bacteria which could be cul-tured on marine agar. In A. lusitanicum NEPCC 253,A. tamarense NEPCC 407 and PCC 173a, identically posi-tioned bands from the di¡erent growth points were excised
FEMSEC 1273 9-10-01
G.L. Hold et al. / FEMS Microbiology Ecology 37 (2001) 161^173168
to assess reproducibility in the DGGE pro¢les from di¡er-ent growth samples. The sequences recovered from theseco-migrating bands from the di¡erent sampling points(early and late exponential and stationary), were virtuallyidentical with di¡erences observed assumed to be artifactsdue to single sequencing reactions.
3.7. 16S rDNA DGGE analysis of A. lusitanicumNEPCC 253
A representative cultured isolate from each of the ¢veRFLP patterns obtained from A. lusitanicum NEPCC 253was successfully ampli¢ed using the PCR conditions forDGGE analysis and demonstrated 100% sequence recog-nition with the primers. However, DGGE analysis of thisdino£agellate culture failed to identify the presence of allRFLP groups. 16S rDNA sequence analysis of excisedDGGE bands in A. lusitanicum cultures demonstratedthe presence of members of the Roseobacter clade (Fig.2, bands 253/6, 7, 10) with strong sequence similarity toRFLP pattern 3 and other K-Proteobacteria most closelyrelated to SCRIPPS_739 ^ a Hyphomonas-related isolate(Fig. 2, 253/2, 4). A further band (253/3) was related to theCytophagales and possessed an identical 16S rDNA se-quence to that of ALUS253_6 (RFLP pattern 22) andG. algens (AF001367). Proteobacteria from the Q-subphy-lum were not cultured from this dino£agellate at anygrowth phase examined and, subsequently, 16S rDNA se-quences belonging to this group were also not recoveredfrom DGGE gels (Table 3).
3.8. 16S rDNA DGGE analysis of A. tamarenseNEPCC 407
As with A. lusitanicum NEPCC 253, all cultured isolatesshowed 100% sequence similarity to DGGE primers, with
representatives from each RFLP pattern identi¢ed fromthis dino£agellate amplifying under the PCR conditionsused for DGGE analysis, although again not all RFLPpatterns were represented in the community pro¢les. 16SrDNA DGGE analysis of A. tamarense NEPCC 407 in-dicated the presence of many Roseobacter-related sequen-ces with six bands excised from di¡erent positions in theDGGE pro¢le showing greater than 98% sequence simi-larity to Roseobacter-related bacterial 16S rDNA sequen-ces over the approx. 200 nucleotides analysed (Fig. 2).Roseobacter-related cultured isolates classi¢ed as belong-ing to RFLP patterns 9, 11 and 12 were all identi¢edwithin the DGGE pro¢le with s 99% sequence similarity.Other bacterial 16S rDNA sequences identi¢ed from theDGGE pro¢les included a Hyphomonas-related sequence(isolate SCRIPPS_739 (97.7%)), within the K-Proteobacte-ria (band 407/3). However, K-Proteobacteria, excludingRoseobacter strains, were not isolated by culturing at earlygrowth phases of this dino£agellate. Other DGGE 16SrDNA bands possessed low similarities (6 90%) to Q-Pro-teobacteria (407/4) and an unknown marine psychrophilewithin the CFB phylum (407/2), although CFB phylumbacteria were not detected from this dino£agellate cultureusing enrichment techniques.
3.9. 16S rDNA DGGE analysis of A. tamarense PCC 173a
The majority of 16S rDNA DGGE bands excised werecon¢rmed as Roseobacter-related sequences. This resultcorresponds with the observation of the high percentageof these bacteria cultured from this dino£agellate on ma-rine agar medium (Figs. 1 and 2). However, other K-pro-teobacterial sequences detected by DGGE were not ob-served in this culture by cultivation on marine agar, forexample Sphingomonas sp., corresponding to band 173a/2(Fig. 2, Table 3). Also, con¢rmation of the presence of
Table 3Comparison of the percentage of cultured isolates classi¢ed as being Roseobacter related, K-Proteobacteria (not including Roseobacter-related isolates),Q-Proteobacteria and CFB and non-cultured (DGGE) pro¢les
Percent of isolates
A. lusitanicum NEPCC 253 A. tamarense PCC 407 A. tamarense PCC 173a S. trochoidea NEPCC 15
Earlyexp.
Lateexp.
Station-ary
Earlyexp.
Lateexp.
Station-ary
Earlyexp.
Lateexp.
Station-ary
Earlyexp.
Lateexp.
Station-ary
K-Proteo-bacteria(Roseobacter)
Cultured 66.7 52.3 30.8 55.7 76 70.9 43.5 66.7 64.2 33.3 31.3 25
DGGE 6 6 6 6 6 6 6 6 6 6 6 6
K-Proteo-bacteria
Cultured 16.7 4.1 30.8 ND ND 11.2 ND ND ND 66.7 62.4 50
DGGE 6 6 6 6 6 6 6 6 6 6 6 6
Q-Proteo-bacteria
Cultured ND ND ND 44.3 24 17.9 21.8 6.6 10.1 ND ND 25
DGGE ND ND ND 6 6 6 6 6 6 6 6 6
CFB Cultured 16.6 47.3 38.4 ND ND ND 34.7 26.7 25.7 ND 6.3 NDDGGE 6 6 6 6 6 6 6 6 6 6 6 6
6 denotes the presence of bacterial 16S rDNA detected sequences by DGGE analysis ; ND, not detected.
FEMSEC 1273 9-10-01
G.L. Hold et al. / FEMS Microbiology Ecology 37 (2001) 161^173 169
bacteria belonging to the CFB phylum was obtained fromDGGE (Fig. 2, bands 173a/12 and 16). Q-Proteobacteria-related sequences were also detected (band 173a/13), withother bands possessing low sequence similarity values todatabase 16S rDNA sequences (Fig. 2, ATAM173a/7).Examination of the 16S rDNA sequences obtained fromcultured isolates from A. tamarense PCC 173a indicatedthat bacterial isolates grouped as RFLP patterns 14, 16, 17and 19 did not have 100% sequence similarity with theDGGE primers and would therefore not be representedin the DGGE pro¢les. However, all other RFLP patternsfrom this algal culture were con¢rmed by DGGE analysis
3.10. 16S rDNA DGGE analysis of S. trochoideaNEPCC 15
Sequence determination and analysis of DGGE bandsexcised from pro¢les from the PCR-ampli¢ed 16S rDNAof S. trochoidea cultures was less successful, with onlylimited sequence information being generated. 16SrDNA DGGE pro¢les from Scrippsiella cultures, however,along with A. tamarense PCC 173a contained the highestnumber of bands (Fig. 2). As with A. lusitanicum andA. tamarense NEPCC 407, all cultured isolates showed100% sequence similarity with the DGGE primers,although only limited RFLP patterns were con¢rmed inthe DGGE analysis. The DGGE bands which could besequenced were a¤liated to all phyla in this alga detectedby culturing (Table 3). However, culturing on marine agarfailed to detect members of the Q-Proteobacteria in theexponential growth phase of the dino£agellate culture,although two adjacent bands from within the DGGE pro-¢le possessed between 94 and 96% sequence similarity toQ-Proteobacteria belonging to the Pseudomonas genus (Fig.1, RFLP pattern 25; Fig. 2, SCRIPP/6 and 7).
In contrast, DGGE band SCRIPP/13 possessed highsimilarity with the 16S rDNA sequences of isolates form-ing RFLP pattern 12 (ATAM407_56 and 667-2), culturedfrom this dino£agellate only in earlier growth phases.
4. Discussion
In this study, bacterial strains isolated 4 years previ-ously from a range of dino£agellate cultures were charac-terised by analysis of their 16S rRNA gene sequences (Ta-ble 1). Previous HPLC and CE-MS evidence suggestedthat bacterial isolate 667-2 belonging to the Roseobacterclade, and isolates 407-2 and 253-19 both closely related to
Fig. 2. DGGE pro¢les of stationary phase samples from the 16S rRNAgene V3 region of dino£agellate cultures A. lusitanicum NEPCC 253, A.tamarense NEPCC 407, A.tamarense PCC 173a and S. trochoideaNEPCC 15. Closest relative sequences are indicated for each band, withpercentage sequence identity noted. Bands classi¢ed as `poor sequence'indicate no closest relative sequence could be de¢ned.6
FEMSEC 1273 9-10-01
G.L. Hold et al. / FEMS Microbiology Ecology 37 (2001) 161^173170
Am. macleodii, may be capable of synthesising multiplePSP derivatives [6]. Other putatively toxigenic bacteria in-clude isolate 253-13, also belonging to the Roseobacterclade, and isolate 2c3, again showing high similarity withAm. macleodii (Gallacher et al., unpublished). Thesestrains were associated with stationary phase cultures ofthe dino£agellate and it was considered to be of interest todetermine the predominance of bacterial strains, particu-larly those which are considered toxic, at di¡erent stagesof the dino£agellate life cycle. Hence we have further ex-tensively characterised and compared the bacterial consor-tia living with both toxic and non-toxic Alexandrium spe-cies and a dino£agellate species not associated withtoxicity (S. trochoidea). Toxin pro¢les of the dino£agel-lates strains A. lusitanicum NEPCC 253 and A. tamarenseNEPCC 407 used in this study have previously been in-vestigated on several occasions [10^13] and prior to thisstudy their toxicity was con¢rmed by HPLC (data notshown).
A combination of colony morphology and RFLP anal-ysis of PCR-ampli¢ed 16S rRNA genes was used to sortthe large number of cultivated isolates obtained from thedino£agellates into groups, thereby reducing the extent of16S rDNA sequence analysis. Previous work by Lebaronet al. [29] on marine bacteria demonstrated that colonymorphology underestimated taxonomic diversity. It wassuggested that, unless used in conjunction with other se-lection criteria, such as RFLP analysis of 16S rRNAgenes, e¡ective di¡erentiation between isolates could notbe achieved. In general, the majority of bacterial strainsisolated in the current study were distinguishable fromeach other using detailed colony morphology and RFLPindependently (Table 2) as bacterial strains sharing thesame RFLP pattern gave identical sequences (Table 2).Combining both techniques provided a powerful toolwhich proved useful in identifying strains for further phy-logenetic analysis.
Overall, the phylogenetic diversity was limited to twobacterial phyla, the Proteobacteria and CFB, with the for-mer restricted to the K- and Q-Proteobacteria subclasses.This is similar to the ¢ndings of Babinchak et al. [5] who,using oligonucleotide probes, found similar classes of bac-teria in a range of dino£agellates. Interestingly, compara-tive 16S rRNA gene sequence analyses revealed that manyof our isolates showed close identity to clades of predom-inantly marine bacterial isolates that do not contain val-idly described representative taxa. This result is commonin phylogenetic studies of naturally occurring marine bac-terial communities [28], and indicates that novel microbialspecies can be relatively easily cultivated from marine di-no£agellate cultures using standard marine agar media.
In this study, the K-Proteobacteria, particularly those ofthe Roseobacter clade, dominated the micro£ora of alldino£agellate cultures. One of these strains(ALUS253_25), isolated from A. lusitanicum, was foundto be closely related to PTB1 and PTB5 strains previously
isolated from a highly toxic Japanese A. tamarense clone[8]. Further, Roseobacter-related isolates from toxic A. lu-sitanicum and A. tamarense NEPCC 407 classi¢ed asRFLP pattern 8 demonstrated high sequence similaritywith the toxigenic bacterium 253-13, while pattern 12showed complete sequence identity with the sodium chan-nel blocking isolate 667-2 obtained by Gallacher et al. [6].
Many of the bacteria were related to microorganismswith known surface-associated life histories (CFB phylum,Hyphomonas, Caulobacter [30] and some members of theRoseobacter clade, for example Roseobacter algicola (re-classi¢ed as Ruegeria algicola [31] and found in the phyco-sphere of P. lima [32]), Roseobacter litoralis and Roseo-bacter denitri¢cans, obtained from the surface of greenseaweed [33]). This may be important in bacterial/dino£a-gellate interactions in relation to bacterial e¡ects on PSTproduction. One of the few studies to take into accountpossible e¡ects of bacterial adhesion on dino£agellatesdemonstrated a decrease in culture toxicity when toxigenicbacteria were prevented from physical contact with anaxenic A. lusitanicum culture [34].
Microbial cultivation also demonstrated that a dynamicsystem exists within dino£agellate culture, with shifts innumbers and/or composition of the bacterial £ora occur-ring throughout the dino£agellate vegetative cell cycle.However, PCR-DGGE ¢ngerprinting analyses of the mi-crobial community showed consistent pro¢les from vari-ous stages of the growth cycle. One possible explanationfor this discrepancy is that the active portion of the micro-bial community shifts throughout the dino£agellategrowth cycle, which is re£ected in the types of bacteriawe were able to culture at the di¡erent growth stages.Regardless of the reason, it is apparent that our studysu¡ers from the same cultivation biases that have beendocumented for other studies employing both microbialcultivation and cultivation-independent molecular ecologyapproaches. The DGGE analysis identi¢ed some gene se-quences which were not detected using the culture-basedmethod, for example the identi¢cation of K-Proteobacteriaoutwith the Roseobacter clade from A. tamarense PCC173a, thus highlighting the need to use both approaches(Table 3). In this study, the K-Proteobacteria, particularlythose of the Roseobacter clade, dominated the culturablemicro£ora of all dino£agellate cultures. A predominanceof Roseobacter-related sequences (with the exception ofS. trochoidea) was also identi¢ed by DGGE analysis,with more than 50% of sequences from the three Alexan-drium cultures being grouped within the clade.
Previously it was suggested that maximum bacterialnumbers occurred during stationary phase of the dino£a-gellate life cycle in batch culture, where dino£agellate celldeath could lead to increased availability of nutrients [3].This study indicated that the bacterial numbers £uctuatedaccording to cell type (as de¢ned by RFLP) independentlyof the dino£agellate growth phase.
Bacterial isolates from the non-toxin producing dino£a-
FEMSEC 1273 9-10-01
G.L. Hold et al. / FEMS Microbiology Ecology 37 (2001) 161^173 171
gellates S. trochoidea and A. tamarense PCC 173a showedthe most diverse populations with many of the strainsunique to these particular cultures. This suggests that aspecies-speci¢c association may exist between some bacte-ria and certain algal species and that di¡erences exist inthe micro£ora between toxic and non-toxic dino£agellates.However, further work using more dino£agellate culturesis required to determine if this is the case. Future studiesshould take into account the e¡ect on the isolated micro-£ora of using media other than marine agar and also thepotential of bacteria to resist cultivation because of theirinterdependence with other microbes [35].
Further work is also needed to ascertain if these bacte-rial populations are stable amongst di¡erent subcultures ofthe same dino£agellate strain. Spurious contamination ofalgal cultures in the laboratory cannot be discounted, butevidence demonstrating a functional signi¢cance of bacte-rial^algal associations (e.g. modulation of toxicity) wouldargue for a selective component to these relationships.This may indeed be the case as some bacterial strains,although isolated 4 years apart, were 99.9% similar basedon 16S rDNA sequence identity. This includes isolatesfrom the Alteromonas clade which were identical to thosepreviously reported as producing PST [6]. Also many ofthe bacterial gene sequences recovered from S. trochoideausing marine agar showed high phylogenetic similarity toenvironmental clone sequences obtained previously fromthe same culture [28].
In conclusion, this study has identi¢ed bacteria associ-ated with three growth phases of PST-associated dino£a-gellates, some of which are unde¢ned species/genera. Thestudy signi¢cantly adds to the limited information which iscurrently available within the sequence databases on bac-teria of algal origin. The results can also be used to inves-tigate the interaction of bacteria with dino£agellates inculture and in the environment. Speci¢cally, oligonucleo-tide probes for in situ hybridisation studies can now bedesigned to investigate the kinetics and physical associa-tion of these bacteria with dino£agellates in relation toboth toxicity and bloom dynamics.
Acknowledgements
We thank Jennifer Graham, Sharleen Alexander, PaulBurgess, Emma Collins and Faye Grant for their technicalassistance. This work was supported by Fisheries ResearchServices, Marine Laboratory, Aberdeen, and by EU grantFAIR CT96-1558. The work was also in part funded bythe European Community Human Capital and MobilityNetwork grant CHRX CT93-0194, and by the German^New Zealand Science and Technology Cooperation Proj-ect (NEU-014). The authors acknowledge the support ofProf. K.N. Timmis (Head of Division of Microbiology,National Research Centre for Biotechnology, Braun-schweig, Germany).
References
[1] Doucette, G.J. (1995) Assessment of the interaction of prokaryoticcells with harmful algal species. In: Toxic Marine Phytoplankton(Lassus, P., Arzul, G., Erard-Le Denn, E., Gentien P. and Marcail-lou-Le Baut, C., Eds.), pp. 385^394. Lavoisier, Paris.
[2] Doucette, G.J. (1995) Interactions between harmful algae and bacte-ria: a review. Nat. Toxins 3, 65^74.
[3] Gallacher, S. and Smith, E.A. (1999) Bacteria and paralytic shell¢shtoxins. Protist 150, 245^255.
[4] Prokic, I., Brummer, F., Brigge, T., Gortz, H.D., Gerdts, G., Schutt,C., Elbrachter, M. and Muller, W.E.G. (1998) Bacteria of the genusRoseobacter associated with the toxic dino£agellate Prorocentrumlima. Protist 149, 347^357.
[5] Babinchak, J.A., McGovern, E.R. and Doucette, G.J. (1998) Isola-tion and characterization of the bacterial £ora associated with PSP-related dino£agellate species. In: Harmful Algae (Reguera, B., Blan-co, J., Fernandez, M.L. and Wyatt, T., Eds.), pp. 410^413. Xunta deGalicia and Intergovernmental Oceanographic Commission ofUNESCO, Gra¢sant, Spain.
[6] Gallacher, S., Flynn, K.J., Franco, J.M., Brueggemann, E.E. andHines, H.B. (1997) Evidence for the production of paralytic shell¢shtoxins by bacteria associated with Alexandrium spp. (Dinophyta) inculture. Appl. Environ. Microbiol. 63, 239^245.
[7] Kodama, M., Ogata, T. and Sato, S. (1988) Bacterial production ofsaxitoxin. Agric. Biol. Chem. 52, 1075^1077.
[8] Kopp, M., Doucette, G.J., Kodama, M., Gerdts, G., Schut, C. andMedlin, L.K. (1997) Phylogenetic analysis of selected toxic and non-toxic bacterial strains isolated from the toxic dino£agellate Alexan-drium tamarense. FEMS Microbiol. Ecol. 24, 251^257.
[9] Groben, R., Doucette, G.J., Kopp, M., Kodama, M., Amann, R. andMedlin, L.K. (2000) 16S rRNA targeted probes for the identi¢cationof bacterial strains isolated from cultures of the toxic dino£agellateAlexandrium tamarense. Microb. Ecol. 39, 186^196.
[10] Hold, G.L., Smith, E.A., Birkbeck, T.H. and Gallacher, S. (2001)Comparison of paralytic shell¢sh toxin production by the dino£agel-lates Alexandrium lusitanicum NEPCC 253 and Alexandrium tama-rense NEPCC 407 in the presence and absence of bacteria. FEMSMicrobiol. Ecol. 36, 223^233.
[11] Cembella, A.D., Therriault, J.C. and Berland, P. (1988) Toxicity ofcultured isolates and natural populations of Protogonyaulax tama-rensis. Lebour, Taylor from the St. Lawrence Estuary. J. Shell¢shRes. 7, 611^622.
[12] Hummert, C., Donner, G., Reinhardt, K., Ritscher, M., Elbracher,M. and Luckas, B. (1998) Paralytic shell¢sh poisoning in Alexan-drium. A comparison of PSP patterns produced by twelve cultivatedstrains. In: Harmful Algae (Reguera, B., Blanco, J., Fernandez, M.L.and Wyatt, T., Eds.), pp. 397^401. Xunta de Galicia and Intergov-ernmental Oceanographic Commission of UNESCO, Gra¢sant,Spain.
[13] Mascarenhas, V., Alvito, P., Franca, S., Sousa, I., Martinez, A.G.and Rodriguez-Vazquez, J.A. (1995) The dino£agellate Alexandriumlusitanicum isolated from the coast of Portugal. Observations on tox-icity and ultrastructure during growth phases. In: Toxic Marine Phy-toplankton (Lassus, P., Arzul, G., Erard-Le Denn, E., Gentien, P.and Marcaillou-Le Baut, C., Eds.), pp. 71^76. Lavoisier, Paris.
[14] Schmidt, R.J. and Loeblich, A.R. (1979) A discussion of the system-atics of toxic Gonyaulax species containing paralytic shell¢sh poison-ing. In: Toxic Dino£agellate Blooms (Taylor, D.L. and Seliger, H.H.,Eds.), pp. 83^88. Elsevier/North-Holland, New York.
[15] Guillard, R.R.L. (1975) Culture of phytoplankton for feeding marineinvertebrates. In: Culture of Marine Invertebrate Animals (Smith,W.L. and Chanley, M.H., Eds.), pp. 29^60. Plenum Press, New York.
[16] McAlice, B.J. (1971) Phytoplankton sampling with the Sedgwick-Raf-ter cell. Limnol. Oceanogr. 16, 19^28.
[17] Suzuki, M., Rappe, M.S., Haimberger, Z.W., Win¢eld, H., Adair, N.,
FEMSEC 1273 9-10-01
G.L. Hold et al. / FEMS Microbiology Ecology 37 (2001) 161^173172
Stro«bel, J. and Giovannoni, S.J. (1997) Bacterial diversity amongsmall-subunit rRNA gene clones and cellular isolates from thesame seawater sample. Appl. Environ. Microbiol. 63, 983^989.
[18] Sambrook, J., Fritsch, E.F. and Maniatis, T. (1989) Molecular Clon-ing, a Laboratory Manual, 2nd edn. Cold Spring Harbor LaboratoryPress, Plainview, NY.
[19] Muyzer, G., De Waal, E.C. and Uitterlinden, A.G. (1993) Pro¢ling ofcomplex microbial populations by denaturing gradient gel electro-phoresis analysis of polymerase chain reaction-ampli¢ed genes codingfor 16S rDNA. Appl. Environ. Microbiol. 59, 695^700.
[20] Lane, D.J. (1991) 16S/23S rRNA sequencing. In: Nucleic Acid Tech-niques in Bacterial Systematic (Stackebrandt, E. and Goodfellow, M.,Eds.), pp. 115^175. Wiley, New York.
[21] Devereux, J., Haeberli, P. and Smithies, O. (1984) A comprehensiveset of sequence analysis programs for the vax. Nucleic Acids Res. 12,387^395.
[22] Ludwig, W. and Strunk, O. (1997) ARB ^ a software environment forsequence data. Internet address: http://www.biol.chemie.tu-muenchen.de/pub/ARB/documentation/.
[23] Benson, D.A., Boguski, M.S., Lipman, D.J., Ostell, J., Ouellette,B.F.F., Rapp, B.A. and Wheeler, D.L. (1999) GenBank. NucleicAcids Res. 27, 12^17.
[24] Maidak, B.L., Cole, J.R., Parker, C.T., Garrity, G.M., Larsen, N.,Li, B., Lilburn, T.G., McCaughey, M.J., Olsen, G.J., Overbeek, R.,Pramanik, S., Schmidt, T.M., Tiedje, J.M. and Woese, C.R. (1999) Anew version of the RDP (Ribosomal Database Project). NucleicAcids Res. 27, 171^173.
[25] Smith, S.W., Overbeek, R., Woese, C.R., Gilbert, W. and Gillevet,P.M. (1994) The genetic data environment: an expandable GUI formultiple sequence analysis. CABIOS 10, 671^675.
[26] Baumann, L., Baumann, P., Mandel, M. and Allen, R.D. (1972)Taxonomy of aerobic marine eubacteria. J. Bacteriol. 110, 402^429.
[27] Eilers, H., Pernthaler, J., Glockner, F.O. and Amann, R. (2000)
Culturability and In situ abundance of pelagic bacteria from theNorth Sea. Appl. Environ. Microbiol. 66, 3044^3051.
[28] Rappe, M.S. Personal communication.[29] LeBaron, P., Ghiglione, J.-F., Fajon, C., Batailler, N. and Normand,
P. (1998) Phenotypic and genetic diversity within a colony morpho-type. FEMS Microbiol. Lett. 160, 137^143.
[30] Abraham, W.-R., Strompl, C., Meyer, H., Lindholst, S., Moore,E.R.B., Christ, R., Vancanneyt, M., Tindall, B.J., Bennaser, A.,Smit, J. and Tesar, M. (1999) Phylogeny and polyphasic taxonomyof Caulobacter species. Proposal of Maricaulis gen. nov. with Mari-caulis maris (Poindexter) comb. nov. as the type species, and emendeddescription of the genera Brevundimonas and Caulobacter. Int. J. Syst.Bacteriol. 49, 1053^1073.
[31] Uchino, Y., Hirata, A., Yokota, A. and Sugiyama, J. (1998) Reclas-si¢cation of marine Agrobacterium species: proposals of Stappia stel-lulata gen. nov., comb. nov., Stappia aggregata sp. nov., nom. rev.,Ruegeria atlantica gen. nov., comb. nov., Ruegeria gelatinovora comb.nov., Ruegeria algicola comb. nov., and Ahrensia kieliense gen. nov.,sp. nov., nom. rev. J. Gen. Appl. Microbiol. 44, 201^210.
[32] Lafay, B., Ruimy, R. and Rausch De Traubenberg, C. (1995) Roseo-bacter algicola sp. nov., a new marine bacterium isolated from thephycosphere of the toxin-producing dino£agellate Prorocentrum lima.Int. J. Syst. Bacteriol. 45, 290^296.
[33] Ruiz-Ponte, C., Cilia, V., Lambert, C. and Nicolas, J.L. (1998)Roseobacter gallaeciensis sp. nov., a new marine bacterium isolatedfrom rearings and collectors of the scallop Pecten maximus. Int. J.Syst. Bacteriol. 48, 537^542.
[34] Doucette, G.J. and Powell, C.L. (1998) Algal-bacterial interactions:can they determine the PSP-related toxicity of dino£agellates? In:Harmful Algae (Reguera, B., Blanco, J., Fernandez, M.L. and Wyatt,T., Eds.), pp. 406^409. Xunta de Galicia and Intergovernmental Oce-anographic Commission of UNESCO, Gra¢sant, Spain.
[35] Akagi, Y., Taga, N. and Simidu, U. (1977) Isolation and distributionof oligotrophic marine bacteria. Can. J. Microbiol. 23, 981^987.
FEMSEC 1273 9-10-01
G.L. Hold et al. / FEMS Microbiology Ecology 37 (2001) 161^173 173