Competitive strategies differentiate closely related species of marine actinobacteria

13 Pages·2015·1.19 MB·English

by  Nastassia V Patin

TheISMEJournal(2016)10,478–490
©2016InternationalSocietyforMicrobialEcology Allrightsreserved 1751-7362/16
www.nature.com/ismej
ORIGINAL ARTICLE
Competitive strategies differentiate closely related
species of marine actinobacteria
Nastassia V Patin1, Katherine R Duncan1, Pieter C Dorrestein2 and Paul R Jensen1
1CenterforMarineBiotechnologyandBiomedicine,ScrippsInstitutionofOceanography,UniversityofCalifornia
San Diego, La Jolla, CA, USA and 2Skaggs School of Pharmacy and Pharmaceutical Sciences, Departments of
Pharmacology, Chemistry and Biochemistry, University of California San Diego, La Jolla, CA, USA
Although competition, niche partitioning, and spatial isolation have been used to describe the
ecologyandevolutionofmacro-organisms,itislesscleartowhatextenttheseprinciplesaccountfor
the extraordinary levels of bacterial diversity observed in nature. Ecological interactions among
bacteria areparticularlychallenging to addressdueto methodological limitations anduncertainties
over how to recognize fundamental units of diversity and link them to the functional traits and
evolutionary processes that led to their divergence. Here we show that two closely related marine
actinomycetespeciescanbedifferentiatedbasedoncompetitivestrategies.Usingadirectchallenge
assaytoinvestigateinhibitoryinteractionswithmembersofthebacterialcommunity,weobserveda
temporal difference in the onset of inhibition. The majority of inhibitory activity exhibited by
Salinispora arenicola occurred early in its growth cycle and was linked to antibiotic production. In
contrast, most inhibition by Salinispora tropica occurred later in the growth cycle and was more
commonly linked to nutrient depletion or other sources. Comparative genomics support these
differences, with S. arenicola containing nearly twice the number of secondary metabolite
biosynthetic gene clusters as S. tropica, indicating a greater potential for secondary metabolite
production. In contrast, S. tropica is enriched in gene clusters associated with the acquisition of
growth-limiting nutrients such as iron. Coupled with differences in growth rates, the results reveal
that S. arenicola uses interference competition at the expense of growth, whereas S. tropica
preferentially employs a strategy of exploitation competition. The results support the ecological
divergenceoftwoco-occurringandcloselyrelatedspeciesofmarinebacteriabyprovidingevidence
theyhaveevolvedfundamentallydifferent strategiestocompete inmarine sediments.
TheISME Journal (2016) 10,478–490;doi:10.1038/ismej.2015.128; published online4August 2015
Introduction identify functional traits that distinguish related
groups of bacteria (Ferris et al., 2003; Sikorski and
Molecular analyses reveal extraordinary levels of
Nevo, 2005; Johnson et al., 2006) including links
bacterial diversity in ocean environments (Sogin
betweensympatricspeciation,competition-dispersal
etal.,2006;Huberetal.,2007).Thisdiversitycreates
tradeoffs(Yawataetal.,2014)andresourcepartition-
a paradox in terms of how a limited range of
ing(Huntetal.,2008;Oakleyetal.,2010).However,
resources can support unexpectedly large numbers
thedelineationoffundamentalunitsofdiversitythat
of species, as has long been observed among
maintains species-like properties and the ecological
phytoplankton (Hutchinson, 1961). The majority of
traits that make them distinct remains a major
bacterial diversity observed in nature falls into
challenge in the field of microbial ecology (Fraser
closely related groups of sequences that have been
et al., 2009).
described as microdiverse sequence clusters (Acinas
Resource competition is thought to be a major
et al., 2004). The ecological implications of this
driver of evolutionary diversification (Svanbäck and
microdiversity remain largely unknown, yet are
Bolnick,2007).Twocompetitivestrategiesbywhich
crucial for understanding community structure and
organisms compete for resources are exploitation
function (Koeppel et al., 2008). Among examples
competition, which is characterized by rapid nutri-
where phylogenetic diversity has been linked to
ent utilization, and interference competition, which
ecological differentiation, it has been possible to
occurs when one organism directly harms another.
Eachstrategyinvolvestradeoffsinenergyinvestment
thataremanifestedasdifferencesingrowthratesand
Correspondence:PRJensen,CenterforMarineBiotechnologyand
Biomedicine, Scripps Institution of Oceanography, University of reproduction (Nicholson, 1954; Case and Gilpin,
CaliforniaSanDiego,9500GilmanDrive,92093LaJolla,CA,USA. 1974; Little et al., 2008). Among bacteria, there is
E-mail:[email protected] evidence that exploitation competition drives diver-
Received13March2015;revised26May2015;accepted18June
sification in lab cultures (Hibbing et al., 2010),
2015;publishedonline4August2015
Competitivestrategiesdifferentiatebacterialspecies
NVPatinetal
479
whereas interference competition is possibly best areas(Jensenetal.,2005).Existingevidencesuggests
known in regards to suppressive soils (Weller et al., that S. tropica and S. pacifica are geographically
2002) and has been proposed as a mechanism to isolated (Mincer et al., 2005; Freel et al., 2012),
explain the high levels of diversity observed in whereas the co-occurrence of both species with the
temporally constant and spatially homogenous more cosmopolitan and abundant species S. areni-
environments (Czárán et al., 2002). Chemically cola has been used as evidence for ecological
mediated interference competition has been linked divergence (Jensen and Mafnas, 2006). Although
to improved fitness in biofilm-forming bacteria patterns of secondary metabolite production
(Rao et al., 2005; Shank et al., 2011) and shown to (Jensen et al., 2007) and biosynthetic gene cluster
affect community structure in hypersaline mats distribution (Penn et al., 2009; Ziemert et al., 2014)
(Long et al., 2013) and freshwater sediments have been linked to Salinispora species-level diver-
(Pérez-Gutiérrez et al., 2013), with theoretical gence,functionalsupportforsympatryhasremained
models supporting the hypothesis that antibiotic elusive.
production can improve fitness and stimulate bio- In this study, we assessed the effects of
diversity (Little et al., 2008; Hibbing et al., 2010). S. arenicola and S. tropica on the growth of a
In the marine environment, it has been shown that diverse collection of co-occurring environmental
interference competition is greater among Vibrio bacteria in an effort to determine the extent to
populations than within them (Cordero et al., 2012) which secondary metabolites mediate competitive
and more common among particle-associated bac- interactions. Direct challenge assays revealed dis-
teria than those that are free living (Long and Azam, tinct temporal patterns in the onset of growth
2001).However,theextenttowhichbacteriaemploy inhibition,withS.arenicolaemployinginterference
interference vs exploitation competition in nature competition mediated by antibiotic production and
remainsunknownduetoapoorunderstandingofthe the relatively fast growing S. tropica preferentially
spatiotemporal dynamics of microbial interactions employing exploitation competition. Although
and a lack of suitable methodologies for characteriz- these two mechanisms of competition are well
ing these processes. Furthermore, microbes experi- known to occur among plants, the results provide
enceothermetabolictradeoffs(Litchmanetal.,2007; evidence that competitive strategies represent func-
Flamholz et al., 2013) that complicate efforts to link tionaltraitsthatcanbeusedto distinguishbetween
phylogeny with specific functional traits. Although closely related yet ecologically distinct populations
new genomic and spectral-imaging techniques have of bacteria.
provided improved methods to address the ecology
of both cultured and uncultured microbes (Watrous
et al., 2012; Hugoni et al., 2013; Ottesen et al., 2013; Materials and methods
Rinke et al., 2013), competitive interactions among
groups of closely related bacteria remain largely Sediment collection and processing
unknown. Sediment samples were collected via SCUBA at
As a major source of biologically active natural depthsfrom3to16minJuly2012duringaresearch
products (Berdy, 2005), actinomycetes represent a cruise aboard the R/V Walton Smith (U Miami).
particularly interesting subject for chemical ecology Individual sediment samples (5–10g per sample)
studies. They are common inhabitants of complex were collected from the sediment surface to depths
environments such as terrestrial soils and marine of ca. −3cm using sterile Whirl-Pak bags (Nasco, Ft.
sediments,whereithas longbeenhypothesizedthat Atkinson, WI, USA). Locations included sites off
the secondary metabolites they produce mediate MiamiandtheDryTortugasintheUnitedStatesand
interactions with competing microbes (Williams Cancún, Cozumel, Akumal, and Banco Chinchorro
et al., 1989; Jarvis, 1995). The actinomycete genus in the Mexican Caribbean. All samples were
Salinispora is readily cultured from marine sedi- processed immediately aboard ship using two
ments (Jensen et al., 2005; Mincer et al., 2005) and methods: drying and stamping for selective actino-
has proven to be a useful model to address bacterial mycete cultivation (Mincer et al., 2002) and serial
biogeography (Jensen and Mafnas, 2006), species dilution and plating for the general cultivation of
concepts (Jensen, 2010), and the evolution of marine bacteria. For the latter, ca. 1g of sediment
secondary metabolism (Freel et al., 2011; Ziemert was serially diluted 1:1, 1:10 and 1:100 in sterile
et al., 2014). It is comprised of three closely related seawater, vortex mixed, and 50μl ofthe supernatant
species, Salinispora arenicola, Salinispora tropica inoculatedontoagarmediaandspreadwithasterile
and Salinispora pacifica (Maldonado et al., 2005; glass rod. Three types of media were used for both
Ahmed et al., 2013), which share 99% 16S riboso- methods: (1) 25% marine agar (9.3g Marine Broth
mal RNA gene sequence identity (Jensen and Difco 2216, 16-g agar, 750-ml 0.2-μm filtered sea-
Mafnas, 2006), thus placing them within a micro- water, 250-ml deionized water), (2) seawater-agar
diverse sequence cluster (Acinas et al., 2004). The (16-g agar, 1-l 0.2-μm filtered seawater) and (3) 25%
cells form branching filaments that develop into a A1 (2.5-g starch, 1-g yeast extract, 0.5-g peptone,
mycelium and produce dormant, non-motile spores 16-g agar, 750-ml 0.2-μm filtered seawater, 250-ml
that are broadly distributed over large geographic deionized water).
TheISMEJournal
Competitivestrategiesdifferentiatebacterialspecies
NVPatinetal
480
Strain isolation (Supplementary Figure S1). Environmental isolates
Salinisporastrainswererecognizedbasedoncolony were then inoculated in triplicate from established
morphology (Mincer et al., 2002) and repeatedly plate colonies to within 1–2mm of the Salinispora
transferred onto new agar media until pure cultures lawn (cross-streaking) with up to 50 perpendicular
were obtained as evidenced by uniform colony inoculations made per plate using sterile toothpicks
morphology. The collection of sediment-derived (replicates inoculated onto different plates). Growth
bacteria used in the direct challenge assays were of the test strains was considered inhibited if a
purified in a similar manner and selected to clearing zone ⩾5mm was observed in the area
represent a diverse range of colony morphologies adjacent to the Salinispora lawn in at least two of
and pigmentation. All strains were maintained on three replicate assays. The pH within the zones of
medium A1 prepared with 75% artificial seawater inhibition was tested in comparison with a medium
(22g/l Instant Ocean, United Pet Group, Cincinnati, control using pH test strips (Micro Essential Labora-
OH, USA), grown with shaking in A1 without agar tories Inc., New York, NY). Given that antibiotic
(hereafter ‘A1’) and cryopreserved at −80°C with production can be time-dependent (Bibb, 1996), a
10% glycerol. second series of assays was performed in which
strains from both Salinispora spp. were allowed to
grow on agar plates for 23 days prior to adding the
DNA extraction, PCR and 16S ribosomal RNA gene bacterial strains. Only strains that showed no
sequencing evidence of inhibition in the initial assays were
All strains presumed to be actinomycetes based on tested at this second time point.
morphology were grown in 7-ml A1 for 3–10 days
while shaking at 160rpm. DNA was extracted from
1mloftheresultingcultureaccordingtotheDNeasy Interference vs exploitation competition assays
protocol (Qiagen Inc., Valencia, CA, USA) with Follow-up assays were performed to distinguish
previously described changes (Gontang et al., between interference competition (the production
2007). For all other strains, colony PCR was of diffusible growth inhibitors) and exploitation
performed by suspending a single colony in 5-μl competition (nutrient depletion) as the source
dimethyl sulfoxide and using 1μl as the PCR of the growth inhibition detected in the direct
template. 16S ribosomal RNA PCR primers are challenge assays. Agar diffusion assays were
described in Supplementary Table S4. Each PCR performed in triplicate using two S. arenicola
consisted of a 25μl mixture containing 10× PCR (CNY-679 and CNY-685) and two S. tropica
Buffer (Applied Biosciences, Foster City, CA, USA), (CNY-678 and CNY-681) strains. These strains were
2.5mM MgCl2 (Applied Biosciences), 0.7% dimethyl grown in the same manner used in the direct
sulphoxide, 10mM dNTPs, 1.5 U AmpliTaq Gold challenge assays, after which1cm2agarblockswere
DNAPolymerase(AppliedBiosciences)and10μmol cutfromtheareaimmediatelyadjacenttotheculture
of each primer. PCR thermocycling conditions were and placed over a freshly inoculated lawn of an
as follows: 5min of initial denaturation at 95°C environmentalisolatethatpreviouslytested positive
followed by 30 cycles of denaturation at 94°C for for growth inhibition. The lawns were periodically
30s,annealingat55°Cfor30sandextensionat72°C checked (1–10 days) for zones of inhibition
for 1min. Sequencing was performed by SeqXcel, surrounding the agar block and scored as positive
Inc. (San Diego, CA, USA). Sequences were sub- whenaclearzone(⩾5mm)wasobserved.Theblocks
mitted to the National Center for Biotechnology werevisuallyassessedthroughouttheteststobefree
Information Basic Local Alignment Search Tool of Salinispora colonies, which are easily recognized
(BLASTN) and identified based on the taxonomic based on morphology, so that any observed inhibi-
assignment of the closest Basic Local Alignment tion could be linked to the diffusion of compounds
Search Tool match. thatweretransferredwiththeagarblocks.Allstrains
that were not inhibited in the agar diffusion assay
were further analyzed to determine whether the
Direct challenge assays causeoftheactivityobservedinthedirectchallenge
Salinispora cultures were inoculated from frozen assay was due to iron depletion. In this case, the
stocks into 25ml A1. Cultures were shaken at directchallengeassayswererepeatedusingstandard
160rpm for 6 days (S. tropica) or 11 days A1 and A1 supplemented with FeSO (10μgml−1
4
(S. arenicola) after which 60μl was transferred by final concentration). If growth was inhibited on A1
pipet to square 150×150mm petri plates containing but not on iron-replete A1, the inhibition was
100ml of A1 agar media. The cultures were spread attributed to iron depletion.
with a sterile loop down the center of the plate to
create a narrow (ca. 3–5mm) lawn of growth and
allowed to grow for 7 days (S. tropica) or 10 days Growth rates
(S. arenicola), based on the time required to reach Salinispora growth rates were determined by
confluence. These incubation periods correspond to changes in liquid culture dry weight biomass over
the entry of liquid cultures into stationary phase time. Salinispora cultures were inoculated from
TheISMEJournal
Competitivestrategiesdifferentiatebacterialspecies
NVPatinetal
481
frozen bacterial stocks (1.8ml) into 50ml A1. was chosen based on the consistently large zones of
Following 7 days of growth (25°C, 160rpm), 1-ml inhibition observed in the direct challenge assays
of this culture was inoculated into each of 24 glass with all four S. arenicola strains. The interaction
tubes containing 10ml A1 and shaken at 160rpm zone, along with monocultures of both strains and a
(25°C). On days 0, 3, 6, 9, 12, 15 and 18, triplicate medium control, were processed for MALDI-based
tubes were filtered onto pre-weighed 47-mm glass imagingmassspectrometryinpositive modeusinga
fiber filters (Pall Corporation, Ann Arbor, MI, USA), Microflex Bruker Daltonics mass spectrometer as
dried overnight (32°C) and weighed. Cell mass was described in the Supporting Information.
calculated as mg dry weight per ml and growth
curves generated by plotting the log of cell mass vs
time. Growth rates were calculated as the change in Statistical analyses
biomass over time during the exponential phase of A non-parametric PERMANOVA analysis was used
growth. to test for correlations between growth inhibition
and the taxonomy of the strains used in the direct
challenge assays (Anderson, 2001). This test was
Chemical extractions and disc-diffusion assays performed using the ‘adonis’ function provided by
S. arenicola CNY-679 and S. tropica CNY-678 were the vegan package (Oksanen et al., 2014) and run in
grown on A1 agar plates for 10 and 23 days, the statistical program R (R Core Team). Welch’s
respectively. Cell-free agar adjacent to Salinispora two-sample t-tests (Welch, 1947) were performed in
growthwasremovedandcutintosmallpiecesusing R to test for significant differences between growth
a sterile scalpel and extracted using methanol rates (n=4 for each species), average zones of
(500ml, 160rpm, 2h). The volume of agar extracted inhibition (n=4 for each species) and average
was measured by solvent displacement. The extract percentage of strains inhibited (n=12 for S. areni-
was filtered (0.45μm Whatman), dried in vacuo, cola, n=13 for S. tropica).
dissolvedinca.10mlofwaterandextractedwithan
equalvolumeofethylacetate.Theethylacetatelayer
was separated, filtered (0.45μm Whatman), dried Genome sequencing and analysis
All genome sequences were generated as previously
under N , and weighed. A1 agar media control
2 described (Ziemert et al., 2014) according to the
extracts were similarly prepared. Extracts were
guidelinesoftheDepartmentofEnergyJointGenome
dissolved in methanol at 1×, 10× and 100×
Institute. Twenty-four genomes (Supplementary
volumetrically equivalent concentrations with 1×
Table S3) were downloaded from the Joint Genome
equal to the extract being dissolved in a volume of
Institute website and submitted to antiSMASH for
solvent equivalent to the volume of agar extracted.
identification of genes associated with secondary
Extracts were tested for antibiotic activity against
metabolism (Medema et al., 2011). Gene clusters
environmental isolates using standard disc-diffusion
assays.Fortheseassays,15-μlofSalinisporaextract, linked to rifamycin biosynthesis were submitted to
NaPDoS (Ziemert et al., 2012) to confirm their
media extract or solvent controls (MeOH) were
identity based on a phylogenetic analysis of the
added to paper discs, allowed to dry and placed
associated ketosynthase domains. To assess the
onto A1 agar plates, along with an antibiotic control
disc(5μgciprofloxacin;BD,Sparks,MD,USA),onto potential for siderophore biosynthesis, previously
identifiedsiderophoregeneclusters(Supplementary
which a bacterial strain had been inoculated as per
TableS5)wereextractedfromtheclosedgenomesof
the agar diffusion assays. Zones of inhibition were
S. tropica CNB-440 and S. arenicola CNS-205 (Penn
recorded as the diameter of clear halos surrounding
et al., 2009) using Geneious Pro 5.5.9 (created by
the discs. Thirteen of the strains showing sensitivity
Biomatters, available at http://www.geneious.com).
to S. arenicola CNY-679 culture extracts were
These gene clusters were used as queries against a
similarly tested for sensitivity to commercially
database created from 24 Salinispora genomes
available rifamycin SV (Sigma-Aldrich, St Louis,
(12 S. arenicola and 12 S. tropica) using Multi-
MI, USA) at concentrations of 0.01, 0.1, and
1mgml−1. GeneBlast 1.1.13 (Medema et al., 2013). Genome
sequences were considered to contain a gene cluster
if the sequence coverage and identity values were
450% to the query sequences.
Liquidchromatography-tandemmassspectrometryand
MALDI-TOF imaging mass spectrometry
High-resolution liquid chromatography-tandem Results
mass spectrometry was performed using an Agilent
6530 Accurate Mass Q-TOF coupled to an Agilent Strain isolation and identification
1260LCsystem(SantaClara,CA,USA)asdescribed Atotalof289sedimentsampleswerecollectedfrom
in the Supporting Information. Imaging mass spec- 22 locations and processed for the selective cultiva-
trometry was performed on agar plates prepared as tion of actinomycetes. These efforts yielded 22
per the direct challenge assays using S. arenicola S. arenicola and four S. tropica strains from seven
CNY-679 and Kytococcus sp. CUA-766. The latter locations ranging from the Dry Tortugas to the
TheISMEJournal
Competitivestrategiesdifferentiatebacterialspecies
NVPatinetal
482
Yucatán peninsula. These results provide further least one Salinispora strain indicating the genus
support for the co-occurrence of the two Salinispora has the capacity to inhibit a broad spectrum of
species (Jensen and Mafnas, 2006) and the higher bacteria. The overall levels of inhibition were
relative abundance of S. arenicola (Mincer et al., similar for the two species, with S. arenicola
2005). Based on 16S ribosomal RNA gene sequen- inhibiting on average slightly fewer strains (82)
cing, the Salinispora strains all belong to previously than S. tropica (87) when both time points were
identified sequence types (Freel et al., 2013). considered (Table 1). However, the potency of the
In addition, 127 taxonomically diverse marine inhibition, as measured by the average size of the
bacteria were isolated from the same habitats that zones of inhibition produced by each Salinispora
yielded the Salinispora strains (Supplementary strain (Supplementary Figure S1), was significantly
Table S1). These bacteria were assigned to 23 greater for S. arenicola (P=0.008, Welch’s two-
families in four phyla (Supplementary Table S2) sample t-test).
and used in a direct challenge assay designed to The patterns of inhibition varied both within and
detect the ability of established Salinispora cultures between Salinispora species (Figure 2). A PERMA-
to inhibit the growth of potential competitors. NOVA analysis showed no correlation between the
The genus Bacillus comprised the largest taxonomy of the potential competitors and the
number of strains (32) followed by Streptomyces likelihood they would be inhibited (P40.05 for all
(phylum Actinobacteria; 12 strains), Erythrobacter taxonomiclevels).Thepercentagesofenvironmental
(subphylum alphaproteobacteria; 11 strains) and the isolates inhibited by all four Salinispora strains
genera Pseudoalteromonas and Microbulbifer of either species were roughly the same
(both subphylum gammaproteobacteria; six strains). (SupplementaryFigureS2).However,incaseswhere
Forty-five strains had no matches at 100% identity bacteria were inhibited by at least one Salinispora
to sequences in GenBank suggesting they have strain,theywereinhibited byallfourstrainsin42%
not previously been cultured (Supplementary of the cases for S. tropica relative to 28% for S.
Table S1). arenicola.Thisdifferencemaylargelybeattributable
to S. arenicola strain CNY-694, which grew poorly
relativetotheotherstrains(seebelow)andinhibited
Direct challenge assays only 34 of the strains against which it was tested
Four S. arenicola and four S. tropica strains were relative to an average of 98 for the other three
testedat two time pointsinadirect challenge assay S. arenicola strains (Table 1). An antiSMASH
designedtodetecttheirabilitytoinhibitthegrowth analysis of the secondary metabolite gene clusters
ofpotentialbacterialcompetitors(Figure1).Intotal, in CNY-694 revealed a biosynthetic potential that is
growth inhibition was detected in 671 (38%) of the equivalent to other S. arenicola strains for which
1769 interactions tested. The vast majority of the genome sequences were analyzed (Supplementary
127 strains tested (119 or 93%) were sensitive to at Table S3).
Work Flow
+
-
Direct Challenge Assay
8 Salinispora strains
n=3
+
-
Agar Diffusion Assay
4 Salinispora strains
n=3
+ - +
- Iron Supplement Assay Extract (Disc) Diffusion Assay
4 Salinispora strains 2 Salinispora strains
n=3 D n=1 A
B
E
C
Figure1 Workflow.AdirectchallengeassaywasusedtodetecttheabilityofestablishedSalinisporaculturestoinhibitthegrowthofco-
occurringbacterialstrains.Allstrainsthatwereinhibitedinthedirectchallengeassay(+)weretestedfurtherinanagardiffusionassayto
determinewhethertheactivitywasduetoadiffusiblemolecule.Apositiveresult(+)wasrecordedwhengrowthoftheteststrainwas
inhibitedaroundtheagarblockbutnotaroundamediumcontrol.Organicextractsgeneratedfromsimilaragarblockswerethentested
againstthesensitivestrainsinadisk-diffusionassaytodeterminewhethertheactivitywasorganicsoluble.Activeorganicextracts(a–c)
wereidentifiedbasedonthedetectionofzonesofinhibitionaroundthediscs.Strainsthatwereinhibitedinthedirectchallengeassaybut
not in the agar diffusion assay were tested further to determine whether the inhibition was due to iron depletion. Iron depletion was
identifiedasthesourceoftheinhibitionwhengrowthwasrestoredoniron-supplementedmedia(+).
TheISMEJournal
Competitivestrategiesdifferentiatebacterialspecies
NVPatinetal
483
Table1 Resultsfromthedirectchallengeassayforall8Salinisporastrainsateachtimepointandforbothtimepointscombined
Species Strain 10-daytimept 23-daytimept Combinedtimepts
#Tested #Inhibited(%) #Tested #Inhibited(%) #Inhibited %Inhibited
S.arenicola CNY-679 127 59(47) 68 39(31) 98 77
S.arenicola CNY-680 127 83(65) 44 23(18) 106 84
S.arenicola CNY-685 127 58(46) 69 31(24) 89 79
S.arenicola CNY-694 127 21(17) 106 13(10) 34 27
Avg.±SD 55±26(44±20) 27±11(21±9) 82±33 65±26
Species Strain 7-daytimept 23-daytimept Combinedtimepts
#Tested #Inhibited(%) #Tested #Inhibited(%) #Inhibited %Inhibited
S.tropica CNY-678 127 14(11) 113 59(47) 73 57
S.tropica CNY-681 127 20(16) 107 80(63) 100 79
S.tropica CNY-682 127 26(20) 101 71(56) 97 76
S.tropica CNY-684 127 21(17) 106 58(46) 79 62
Avg.±SD 20±5(16±4) 67±10(53±8) 87±13 69±10
Onlystrainsthatwerenotinhibitedatthefirsttimepointwererepeatedatthesecondtimepoint.
Color Key
0 1 2 3 4 5
Average ZOI Size (cm)
Actinobacteria
s
e
talo Proteobacteria (α)
s
I la
tn Proteobacteria (γ)
e
m
no Bacteroidetes
riv
n
E
Firmicutes
CNY-679 CNY-680 CNY-685 CNY-694 CNY-681 CNY-682 CNY-684 CNY-678
S. arenicola S. tropica
Strain
Figure2 Growthinhibitionobservedinthedirectchallengeassaysoverbothtimepoints.Eachrowcorrespondstoonebacterialstrain
testedagainsteightSalinisporastrains.Thebacteriaaregroupedbyphyla,withthephylumBacteriodetesrepresentedbyonlyonestrain.
Thecolorintensityrepresentstheaveragesize(cm)ofthezoneofinhibition(ZOI)fortriplicateassays.
Temporal variability were tested against a subset of 30 bacteria chosen to
Although the total number of inhibitory interactions represent a range of the taxa encompassed by the
recorded for each Salinispora species was similar, original127strains(SupplementaryTableS1).When
therewasamajordifferenceinthetemporalonsetof thedatafromallassayswerecombined,S.arenicola
the activities (Figure 3, Supplementary Figure S3). remained significantly more active at the early time
Most notably, S. arenicola exhibited on average 55 point (t-test; P=0.02), whereas there was no differ-
inhibitory interactions at the first time point relative ence between the two species when the averages of
to 20 for S. tropica (Table 1). The trend of greater the combined time points were compared (Figure 3)
activity at the first time point was also observed for (t-test; P=0.49). Given that growth inhibition in
S. arenicola strain CNY-694 despite the relatively directchallengeassayscanbecausedbyfactorssuch
low number of bacteria inhibited. To determine asnutrientdepletioninadditiontotheproductionof
whether these patterns were consistent features of inhibitorycompounds,wenextsoughttodistinguish
the two Salinispora species, eight additional between these two possibilities in an effort to better
S. arenicola and nine additional S. tropica strains categorize the observed activities.
TheISMEJournal
Competitivestrategiesdifferentiatebacterialspecies
NVPatinetal
484
80
*
ntage of bited 60
ehi
erage PercStrains In 2400
v
A
0
S. arenicola S. tropica
Species
Figure 3 Average percentage of strains inhibited by 12
S. arenicola and 13 S. tropica strains. White bars represent the
results for the first time point (7 and 10 days for S. tropica and
Figure4 Sourcesofgrowthinhibition.Thetoppanelpresentsthe
S. arenicola, respectively). Gray bars represent the combined
resultsfromthefirsttimepoint(7daysforS.tropica,10daysfor
percentageofstrainsinhibitedatthefirstandsecond(23daysfor
S. arenicola), whereas the bottom panel presents cumulative
both Salinispora species) time points. There was a significant
results from the first and second time points (23 days for both
difference betweenthe averagepercentageof strains inhibited at
species).Sensitiveisolatesweretestedinanagardiffusionassayto
thefirsttimepointbyeachspecies(*).
determine whether activity could be linked to a diffusible
molecule. If negative, further tests were performed to determine
whethergrowthwasrestoredwhenexcessironwasaddedtothe
Interference vs exploitation competition medium, in which case the activity was attributed to iron
Growth inhibition due to the production of anti- depletion. If both assays were negative, growth inhibition was
biotics or other inhibitory substances can bedefined attributed to ‘other’ sources. Each Salinispora strain was tested
as interference competition, whereas inhibition against the environmental isolates they inhibited in the direct
challengeassays.
owing to nutrient depletion can be considered
exploitation competition (Hibbing et al., 2010). Agar
diffusion assays were performed to determine
whether the activities detected in the direct chal- Table2 Numberofstrainsinhibitedbyadiffusiblemoleculeand
lenge assays could be linked to the presence of organicextract
diffusiblemolecules(Figure1)andthusindicativeof
Inhibitoryfactor Strainextracted
antibioticproduction.Agartakenatbothtimepoints
from areas adjacent to two S. arenicola and two
S.arenicola S.tropica
S. tropica strains was tested against all environmen-
CNY-679 CNY-678
talisolatesthatshowedsensitivitytothesestrainsin
the direct challenge assays. Together, these isolates
Diffusiblemolecule 44(35) 12(9)
represent103(95%)and94(87%)ofthebacteriathat MeOHextract(⩾1×) 11 0
displayed sensitivity to any of the S. arenicola or MeOHExtract(⩾10×) 32 0
S. tropica strains, respectively. At the first time MeOHExtract(100×) 42 0
point,theinhibitiongeneratedbyS.arenicolastrains
Forthediffusiblemoleculetest,percentageoftotaltestpanelisalso
CNY-679 and CNY-685 was diffusible in 72% and
includedinparentheses.Cell-freeareasadjacenttoSalinispora
40%ofthetestsperformed,respectively. Incompar- culturesgrowingonagarmediawereextractedfromS.arenicola
ison, the inhibition generated by S. tropica strains andS.tropicastrainsandtestedatthreevolumetricconcentrations
CNY-678 and CNY-681 was diffusible in 9% and (1×,10× and100×)againstallenvironmentalisolatesthatwere
inhibitedintheagardiffusionassays.Extractionswereperformedat
14% of theassays, respectively (Figure4). When the
thetimepointthatgeneratedthemostactivityinthedirectchallenge
combined time points were considered, the inhibi- assays(10daysforS.arenicola,23daysforS.tropica).
tiongeneratedby S. arenicolawasdiffusible in 51%
of the assays compared with 20% for S. tropica. We
next asked if the diffusible activities could be supplementation assay was performed to determine
extracted with an organic solvent (Figure 1). whether iron depletion was the cause of the inhibi-
At 10× concentration, an extract generated from tion(Figure1).Onaverageoverbothtimepoints,9%
S. arenicola strain CNY-679 inhibited 32 of the 44 of the inhibition generated by S. arenicola could be
strainsthatwereinhibitedintheagardiffusionassay linked to iron depletion compared with 20% for
(Table 2). In contrast, an extract prepared from S. tropica (Figure 4). The remaining inhibitory
S. tropica strain CNY-678 did not exhibit any activities (onaverage,40%and60%forS.arenicola
inhibition, even when tested at 100× concentration. and S. tropica, respectively) were ascribed to ‘other’
In cases where the inhibitory activities could not causes. There were no measurable differences in pH
be linked to a diffusible molecule, an iron between the zones of inhibition and media controls
TheISMEJournal
Competitivestrategiesdifferentiatebacterialspecies
NVPatinetal
485
(data notshown)suggesting thatachangeinpHwas Inaddition,thegenomesequenceswereinvestigated
not the cause of the ‘other’ inhibition. for gene clusters previously linked to siderophore
biosynthesis in Salinispora spp. (Penn et al., 2009;
Roberts et al., 2012). On average, 31.3 secondary
Growth rates metabolite gene clusters were identified in the
To explore the hypothesis that S. tropica preferen- S. arenicola strains in comparison with 23.1 for
tially employs exploitation competition as a compe- S. tropica (Supplementary Table S3). These results
titive strategy, the growth rates of the two species conformwellwiththe31and19clustersdetectedin
were assessed. Growth curves generated for all eight an earlier analysis of two closed S. arenicola and
strainsrevealedthatS.tropicahasfastergrowthrates S. tropica genomes, respectively (Penn et al., 2009).
as determined by changes in dry weight biomass Thepresenceoftherifamycin(rif)genecluster(Floss
over time, which were significantly different and Yu, 2005) was confirmed in all of the
between the two species (Figure 5; t-test, P=0.02). S. arenicola strains, whereas none of the pathways
In addition to faster growth rates, cellular biomass detected in S. tropica could be linked to the
wasgreateronaverageinstationaryphase S.tropica productionofaknownantibiotic.Althoughthetotal
cultures (mean=4.13±SD 1.11mgml−1) than in numberofgeneclustersisgreaterinS.arenicola,the
stationary phase S. arenicola cultures (mean=2.98 trend is reversed when it comes to siderophore
±SD 1.19mgml−1) (Supplementary Figure S4). biosynthesis. In an earlier report, both S. arenicola
S. arenicola CNY-694, which produced relatively CNS-205 and S. tropica CNB-440 were identified as
few inhibitory interactions, also produced the least possessing two siderophore-related gene clusters,
biomass per volume of any strain (Supplementary one predicted to encode desferrioxamines (des) and
Figure S4). a second (sid2) related to the gene cluster for
yersiniabactin biosynthesis. However, two addi-
tionalsiderophoregeneclusters(sid3andsid4)were
also identified in S. tropica CNB-440 (Penn et al.,
Bioinformatic analyses
A prior bioinformatic analysis revealed that 2009). The production of desferrioxamines B and E
S.arenicolamaintainsalargernumberofpolyketide was later confirmed for both S. tropica and
synthase and non-ribosomal peptide synthetase S. arenicola and linked to the des locus (Roberts
biosynthetic gene clusters than S. tropica (Ziemert etal.,2012).AMultiGeneBlastanalysisrevealedthat
et al., 2014). To more broadly assess the secondary all12S.tropicastrainsforwhichgenomesequences
metabolite potential of the two species, genome areavailablepossessallfourofthesiderophoregene
sequencesfrom 12strainsof eachspecies(including clusters, whereas the 12 S. arenicola genomes
two S. tropica and three S. arenicola strains used in chosen for this study only possess the des and sid2
thedirectchallengeassays)(SupplementaryTableS1) clusters (Supplementary Table S3).
were analyzed for the presence of secondary meta-
bolite gene clusters using antiSMASH, which can
identify pathway types other than polyketide Chemical analyses
synthase and non-ribosomal peptide synthetase To determine whether any known antibiotics were
including those associated with lantibiotic and produced by either Salinispora species, organic
terpene biosynthesis (Medema et al., 2011). culture extracts of S. tropica CNY-678 and
S. arenicola CNY-679 were examined by liquid
chromatography-tandem mass spectrometry and the
results compared with the AntiMarin database
(Blunt et al., 2006) and an in-house database of
previouslyidentified Salinispora secondarymetabo-
lites. The S. arenicola extract contained the anti-
bioticrifamycinS(SupplementaryFigureS5),which
belongstoaclassofcompoundspreviouslyreported
from this species (Kim et al., 2006; Jensen et al.,
2007). In addition, a parent ion that matched
rifamycin S was observed directly in a zone of
inhibition between S. arenicola CNY-679 and Kyto-
coccus sp. CUA-766 using MALDI-TOF imaging
mass spectrometry (Figure 6). Thirteen strains
showing sensitivity to S. arenicola culture extracts
Figure 5 Salinispora growth rates. For each box, the dark were tested for sensitivity to commercially available
horizontalbarrepresentsthemedianvalueforthechangeindry rifamycin SV (the quinone analog of rifamycin S)
weight/dayofthefilteredcellbiomasscollectedduringexponen- and found to be sensitive at concentrations between
tial growth phase. The box edges represent the upperand lower 10–100μgml−1 (data not shown), which is
quartilesofthedata,andthewhiskersrepresenttheminimumand
comparable to the MIC-50 values reported against a
maximumvalues.Theplotcontainsdatafromfourstrainsofeach
speciesgrownintriplicate. variety of enteropathogens (Farrell et al., 2011).
TheISMEJournal
Competitivestrategiesdifferentiatebacterialspecies
NVPatinetal
486
m/z = 696
Figure6 MALDI-TOFimagingmassspectrometryofanS.arenicolaCNY-679interactionwithKytococcussp.CUA-766.(a)MALDIplate
setupwithcontrolsamplesinthetoprow,fromlefttoright:mediablank,Kytococcussp.CUA-766monoculture,S.arenicolaCNY-679
monoculture. The sample inthe bottom rowcontains the zoneof inhibitionbetweenthe twostrains. The grayboxes define theareas
chosenforimaging.(b)Spatialdistributionofthem/z696ion,anexactmatchtorifamycinS(M+H),showningreensurroundingthe
S.arenicolacolonyanddiffusingoutwardstowardtheinhibitedbacterialstrain.ThisionwasobservedintheS.arenicolamonoculture
butnotinthemediumblankortheKytococcussp.monoculture.
These results provide evidence that the antibiotic experimental evidence linking these differences to
rifamycin S is, at least in part, associated with the divergentecologicalstrategieshasbeenlacking.Here
inhibitory activities generated by S. arenicola. weprovideevidencethatS.tropicaandS.arenicola
Compounds in this class have been similarly linked employ fundamentally different competitive strate-
to the activities observed between sponge-derived gies that are mediated in part by secondary metabo-
S. arenicola strains and Mycobacterium spp. (Izumi lites, with the former investing in growth or
et al., 2010). No known antibiotics were detected in exploitation competition and the latter in interfer-
the S. tropica extract. ence competition via the production of antibiotics
and at the expense of growth. The results create a
link between the fine-scale phylogenetic relation-
ships maintained by these two lineages and func-
Discussion
tional traits that establish them as distinct ecotypes.
Molecular surveys have revealed that bacteria exhi- The two Salinispora species exhibited similar
bit extraordinary levels of phylogenetic diversity overall levels of growth inhibition against a diverse
(Rappe and Giovannoni, 2003). Although the exis- panel of co-occurring marine bacteria. However,
tence of this diversity is widely appreciated, estab- temporal differences in the onset of inhibition
lishing links between the clades observed in suggested there could be fundamental differences
phylogenetic trees and the ecological and evolution- in the mechanisms by which these activities were
aryprocessesthatcreateandmaintainthemremains generated. Subsequent tests made it possible to
one of the great challenges in microbial ecology distinguishbetweeninhibitionduetotheproduction
(Fuhrman,2009;CorderoandPolz,2014).Ithasbeen of allelopathic molecules and that caused by other
proposed that fine-scale phylogeny can be used to factors, including nutrient depletion. The observa-
delineatebacteriaintoecologically cohesiveunitsor tion that S. arenicola generated more than twice the
‘ecotypes’ (Cohan, 2002) and that in the few number of inhibitory activities linked to the produc-
examplesavailable,ecologicalpopulationsrepresent tion of diffusible substances relative to S. tropica
‘gene flow units’ for which genome wide rates of provided the initial line of evidence that these two
homologous recombination are much greater within species differentially invest in interference and
than between clusters (Cordero and Polz, 2014). exploitation competition, respectively. Additional
A multilocus sequence analysis of Salinispora evidence comes from the identification of the
species supports the description of S. tropica and antibioticrifamycinSinS.arenicolacultureextracts
S. arenicola in accordance with what appear to be and directly in the zones of inhibition, whereas no
natural barriers torecombination(Freel et al., 2013). antibiotic activities were detected in the S. tropica
Although there are many processes that can account culture extracts. S. arenicola also averaged 31 gene
for these barriers (Cohan, 2002; Acinas et al., 2004; clusters related to secondary metabolism relative to
Thompsonetal.,2005;Fraseretal.,2007;Hellweger 23 for S. tropica, indicating greater genetic invest-
et al., 2014), ecological differentiation has been mentinthisfunctionaltrait.Thedistinctionbetween
proposed as the mechanism driving diversification interference and exploitation competition was
betweenS. arenicola and thetwo sister specieswith further supported by the significantly faster growth
which it co-occurs (Jensen et al., 2007; Freel et al., ratesrecordedforS.tropica.Onenotableresultisthe
2013). To date, the major adaptive traits that largenumberofactivitiesthatcouldnotbelinkedto
distinguish the three species relate to secondary diffusible molecules or iron limitation (Figure 4).
metabolite production (Jensen et al., 2007), but These activities require further study and may be
TheISMEJournal
Competitivestrategiesdifferentiatebacterialspecies
NVPatinetal
487
linked to the depletion of nutrients other than iron- whole to minimize the total number of competitors.
including carbon and nitrogen sources. The preva- Intra-specific variability also provides a rational
lence of non-diffusible growth inhibition suggests approach to ensure that at least some individuals
that exploitation competition may be widely over- will remain competitive as new challengers are
looked as the source of growth inhibition in direct encountered either when spores are dispersed or
challenge assays. This is not to imply, however, that when new resources become available at an existing
S. tropica does not also engage in interference site. It also provides an effective strategy to avoid
competition via the production of inhibitory com- resistance,asintheapplicationofcombinatorialdrug
pounds;they were simply not detected in the assays therapy, and the subsequent need to enter into a co-
employed. evolutionary arms race (Kinkel et al., 2013). Regard-
Although S. arenicola maintains a larger number less of the ecological benefits, this variability can
of gene clusters devoted to secondary metabolism, likely be linked to recent horizontal gene transfer
this trend is reversed when it comes to siderophore events and the concept that bacteria frequently
biosynthesis. Siderophores are an important ‘sample’ gene clusters from the local gene pool
mechanism by which bacteria acquire growth- (Ziemert et al., 2014), with conservation observed
essential iron (Neilands, 1995) and their production for only those clusters whose products provide
hasbeenlinkedtoantagonisminVibrio(Pybusetal., sufficient selective advantage (Jensen et al., 2007).
1994) and Pseudomonas spp. (Simões et al., 2008). The bacteria used in the direct challenge assays
Although siderophores aresecretedsecondary meta- were all heterotrophic and originated from similar, if
bolites, it can be argued they have a functional role not the same, sediment samples as the Salinispora
in exploitation competition as opposed to more isolates, thus making them potential competitors. As
traditional allelopathic agents such as antibiotics, with any culture-based study, these strains represent
which function in interference competition. only a small fraction of the total bacterial community
S. tropica genome sequences maintain two gene and do not include classes such as the Deltaproteo-
clusters predicted to encode siderophore biosynth- bacteria and the Planctomycetacia, which can repre-
esis(sid3andsid4)thatarenotfoundinS.arenicola sent major components of sediment communities
(Supplementary Table S3 and S5), suggesting an (Schauer et al., 2010). Nonetheless, these strains
additionalinvestmentinironuptakebythisspecies. encompass a wide range of taxonomic diversity,
Although the inactivation of key genes in the S. including common marine families such as the
tropicasid3andsid4clustersdidnotaffectgrowthin Rhodobacteraceae and Pseudoalteromonadaceae
iron-limited media (Roberts et al., 2012), these (Acinas et al., 2004; Gilbert et al., 2012), and share
pathwaysmayhavearoleintheacquisitionofother similar growth requirements with Salinispora spp.,
limiting metals (Bellenger et al., 2008) or be further supporting their ecological relevance.
regulated by factors other than iron limitation, and Although colony growth on agar plates is not the
thus enhanced siderophore production could help naturalstateofsedimentbacteria,theencroachmentof
support higher growth rates in this species. The an established colony by competing bacteria repre-
conservationofsid3andsid4amongall12S.tropica sents a scenario that may be important in dictating
strains for which genome sequences were available defensive strategies in marine sediments. Structured
suggests that there are strong selective pressures to habitats such as sediments are prime locations for
maintain the functions provided by the products of interference competition as has been suggested for
these pathways. The observation that secondary colicinogenic bacteria grown in soft agar matrices
metabolites can have a role in exploitation competi- (Chao and Levin, 1981). The mycelium growth form
tion by facilitating nutrient acquisition as well as exhibited by Salinispora spp. may facilitate the
interference competition via antibiotic production accumulation of antibiotics at concentrations that
emphasizes the importance of distinguishing achieve functional levels in the surrounding micro-
between these two competitive mechanisms when environment. However, translating the results
addressing the ecological functions of secondary obtained here to what occurs in nature will require
metabolites. additional studies aimed toward gaining a spatial
Considerable intra-specific variation was observed contextforcompetitiveinteractionsandmoreaccurate
in the inhibitory activities generated by both species. estimates for in situ compound concentrations.
This supports a growing body of evidence in which Salinispora strains largely occur as dormant
inhibitoryactivitiesarestrain-specific(Grossartetal., spores and represent relatively rare members of the
2004;Rypienetal.,2010;Longetal.,2013)oroccurat sedimentmicrobialcommunity(Minceretal.,2005).
low frequencies within a population (Vetsigian etal., Thus, it is possible for actively growing, localized
2011). It also supports the concept that secondary populationstooriginatefromindividualspores.This
metabolite gene cluster evolution is a dynamic may help explain the apparent lack of ‘social
process (Medema et al., 2014) with high levels of cheaters’ benefiting from the production of antibio-
plasticity within a single species (Ziemert et al., tics by con-specifics, as has been reported in marine
2014).Onerationalizationforthisplasticityisthatthe Vibrionaceae (Cordero et al., 2012). The consistent
targets of the antibiotics produced by any one strain production of antibiotics in the rifamycin class by
maymatterlessthantheabilityofthepopulationasa S. arenicola (Jensen et al., 2007) suggests there are
TheISMEJournal