This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright
12
Embed
Antagonistic interactions between psychrotrophic cultivable bacteria isolated from Antarctic sponges: a preliminary analysis
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies areencouraged to visit:
Antagonistic interactions between psychrotrophic cultivable bacteriaisolated from Antarctic sponges: a preliminary analysis
Santina Mangano a, Luigi Michaud a, Consolazione Caruso a, Matteo Brilli b, Vivia Bruni a,Renato Fani b, Angelina Lo Giudice a,*
a Department of Animal Biology and Marine Ecology, University of Messina, Salita Sperone 31, 98166 Messina, Italyb Department of Evolutionary Biology, University of Florence, via Romana 17-19, 50125 Florence, Italy
Received 18 June 2008; accepted 29 September 2008
Available online 25 October 2008
Abstract
The present work was aimed at studying antagonistic interactions existing among cultivable bacteria associated with the Antarctic spongesAnoxycalyx joubini and Lissodendoryx nobilis. Overall, bacterial isolates were affiliated with the a- and g-Proteobacteria (17.3 and 65.3%,respectively), the CFB group of Bacteroidetes (10.7%) and the Actinobacteria (6.7%) by 16S rDNA sequencing. The two sponges harboredmicroorganisms belonging to different species/genera and previously retrieved from polar marine environments. Antagonistic interactions,assayed by the cross-streak method and statistically analyzed using the ‘‘network theory’’ approach, were checked among isolates associatedwith the same sponge as well as between isolates retrieved from the two sponge species (‘‘cross-niche inhibition’’). Results suggest thatantagonism could play a significant role in shaping bacterial communities within sponge tissues. Data from this study confirm previousobservations on the antibacterial activity of Antarctic microorganisms and represent a baseline for further investigation of both the ecologicalrole and biotechnological exploitation of Antarctic sponge-associated bacteria.� 2008 Elsevier Masson SAS. All rights reserved.
Sponges harbor numerous prokaryotic and eukaryoticorganisms playing relevant ecological and/or biological roles,including a contribution to sponge mechanical structure,chemical defense, elimination of waste products, autotrophy,bioluminescence, nitrogen metabolism and/or protection fromUV radiation [9,11,29]. Although bacteria might represent themajor source of nutrients for sponges, many of them remainassociated with the host mesohyl, constituting over 40% of itsbiomass [10,30]. Studies carried out on sponge-associatedmicrobial communities revealed that they are generallyphylogenetically diverse, sponge-specific and very differentfrom those of the surrounding seawater, suggesting highly
selective conditions within the sponge mesohyl [4,7,9,10,24,26,27,29,33].
Associations between sponges and bacteria have previouslybeen investigated, especially in temperate and tropical envi-ronments, whereas very little is known about bacteria associ-ated with Antarctic sponges [34]. Studies on the antibacterialactivity of sponge-associated bacteria are of particular interest,as they have primarily elucidated the search for novel anti-microbial compounds against medically relevant microorgan-isms [4,7,23,29,36]. However, from an ecological viewpoint,inhibitory interactions among bacteria inhabiting the sameniche represent an interesting evolutionary strategy, conferringa selective advantage in competition for nutrients and space inthe environment and acting as an effective control of microbialpopulations [7]. Studies concerning this issue suggest that anantagonistic effect exhibited by phylogenetically differentbacterial groups is a widespread trait in marine habitats[3,6,12,17,20].
The aim of this work was to investigate antagonisticinteractions among bacteria isolated from two Antarcticmarine sponge species, Anoxycalyx (Scolymastra) joubini(class Hexactinellida) and Lissodendoryx (Ectyodoryx) nobilis(class Demospongiae), collected from Tethys Bay (Terra NovaBay, Ross Sea).
2. Materials and methods
2.1. Sponge collection
During the XX Italian Expedition to Antarctica (AustralSummer 2004e2005), sampling for analyzing sponge-associ-ated bacterial communities was carried out at Tethys Bay(Terra Nova Bay, Ross Sea; coordinates: 74� 410 S - 164� 040
E). At sampling time, the bay was exceptionally free of icecover; seawater temperatures ranged from �1.6 to þ2.3 �C,while salinity ranged from 33.7 to 37.6&. The sponges A.(Scolymastra) joubini (Topsent 1910) and L. (Ectyodoryx)nobilis (Ridley and Dendy 1886) were chosen to perform thepresent study as they provided a similar number of bacterialisolates (see Section 3). The study was carried out on onespecimen from each sponge species, as collection of a limitednumber of samples was authorized.
Sponge specimens were immediately washed at least threetimes with filter-sterilized natural seawater to remove transientand loosely attached bacteria and/or debris. Specimens werethen placed in individual sterile plastic bags containing filter-sterilized natural seawater and transported directly to thelaboratory at 4 �C for microbiological processing (within 2 hafter sampling). A fragment of each specimen was alsopreserved in 70% ethanol for taxonomic identification.
2.2. Estimation of total and culturable bacteria
A central core of the sponge tissue was cut using an EtOH-sterilized corkborer or a sterile scalpel. The sponge tissue wasthen aseptically weighted and manually homogenized in0.22 mm filtered seawater in a sterile mortar. For viable counts,tissue extracts were serially diluted using filter-sterilizedseawater. Aliquots (100 ml) of each dilution were plated in trip-licate on Marine Agar 2216 (MA, Difco). Plates were incubatedin the dark at 4 �C for 1 month. Data obtained are expressed asColony Forming Units (CFU) per gram of sponge tissue.
Samples for estimation of total bacterial number were treatedas reported by Friedrich et al. [5], after fixing aliquots of the tissueextracts with an equal volume of 8% paraformaldehyde/phos-phate buffered saline (PBS), and were then stored at 4 �C until use.Briefly, tissue extracts were centrifuged for 5 min at 13,000 rpm,washed twice with PBS and fixed in ethanol series (50, 70, 100%).The fluorochrome DAPI (4’,6-diamidino-2-phenylindole) wasadded to tissue extracts (final concentration, 5 mg/ml) and, afterdilution with sterile PBS, 1 ml aliquots of DAPI-stained dilutedextracts were filtered under a vacuum onto polycarbonatemembranes (25 mm, diameter; 0.22 mm, pore size). Membraneswere then mounted on microscope slides and examined byepifluorescence microscope (Axioplan, Zeiss). All cells in
a minimum of 20 eyefields per sample were counted. Bacterialabundance is expressed as cells per gram of sponge tissue.
2.3. Bacterial isolation
Bacterial colonies grown on MA were isolated at randomand streaked at least three times before being considered pure.Cultures were routinely incubated in the dark at 4 �C underaerobic conditions. All bacterial isolates were included in theItalian Collection of Antarctic Bacteria (CIBAN) of theNational Antarctic Museum (MNA) ‘‘Felice Ippolito’’ kept atthe Department of Animal Biology and Marine Ecology of theUniversity of Messina (Italy).
2.4. DNA manipulation
PCR amplification and restriction analysis (AmplifiedrDNA Restriction Analysis, or ARDRA) [31] of 16S rRNAgenes from bacterial isolates were carried out as previouslydescribed [18]. On the basis of restriction patterns obtained,Antarctic isolates were clustered into Operational TaxonomicUnits (OTUs), assuming that each OTU (assigned to a number)included strains belonging to the same species. Whenrequired, a letter was added after the OTU number in order todifferentiate similar restriction patterns.
For each OTU, one to three representative strains showingthe identical ARDRA pattern were randomly selected for 16SrDNA sequencing. All singletons were sequenced.
Each sequence was then used as a query in a BLASTn [1]search and further aligned to the most similar sequencesretrieved from the database using the program ClustalW [28].
Sequences were submitted to GenBank and assigned thefollowing accession numbers: EU237120 to EU237146.
2.5. Screening for antagonistic interactions amongbacterial isolates
Bacteria isolated from the same sponge were screened forantagonistic interactions by the cross-streak method [15] usinga 38 � 38 (1444 tests) and 37 � 37 (1369 tests) array of testsfor A. joubini and L. nobilis bacterial strains, respectively. Inaddition, cross-activity between bacteria retrieved from thetwo sponges was investigated by testing the inhibitory inter-actions of bacteria from the first sponge against those derivedfrom the second and vice-versa (‘‘cross-niche inhibition’’).This means that each strain was tested against one another.
In a previous investigation of antibacterial activity ofAntarctic bacteria [16], neither media composition nor incuba-tion temperature below 15 �C affected the inhibition process;hence, in the present study, a single medium (MA) and an incu-bation temperature of 15 �C (due to the psychrotrophic nature ofthe isolates, data not shown) were chosen for the screening.
Hereafter, bacterial isolates tested for inhibitory activity willbe termed ‘‘tester’’ strains, whereas those used as targets will becalled ‘‘target’’ strains. Briefly, tester strains were streakedacross one-third of an MA plate and incubated at 15 �C untilsatisfactory growth was obtained. Target strains were streaked
28 S. Mangano et al. / Research in Microbiology 160 (2009) 27e37
perpendicular to the initial streak and plates were further incu-bated at 15 �C. The antagonistic effect was indicated by failureof the target strain to grow in the confluence area. To obtainreliable results, isolates producing too scarce a biomass on agarplates when streaked as testers were used exclusively as targets.
Bacterial isolates were then operationally distinguished intothree different interactivity clusters, termed: (1) active, if theywere able to inhibit growth of at least one bacterial target; (2)sensitive, if their growth was inhibited by at least one isolateused as a tester; and (3) resistant if their growth was neverinhibited by tester strains. It must be noted that an individualstrain could be included in one or two interactivity clusters.Thus, in Section 3: (1) the ‘‘active cluster’’ includes strainswhich proved to be active plus sensitive as well as active plusresistant; (2) the ‘‘sensitive cluster’’ includes strains shown tobe exclusively sensitive, as well as sensitive plus active; and(3) the ‘‘resistant cluster’’ includes strains shown to be resis-tant plus active, as well as exclusively resistant.
2.6. Inhibitory network analysis
The antagonistic relationship was encoded in the form ofa squared asymmetric binary matrix (L) where Lij ¼ 1 if straini inhibits strain growth j, and zero otherwise; similarly, Lji ¼ 1if strain j inhibits strain i. This matrix can be studied asa network of inhibitory relationships where each node corre-sponds to a bacterial strain and each connection (or link)between two strains indicates an antagonistic relationshipbetween them. Connections are directed, since inhibition maybe non-symmetrical. We studied two main properties of thenetwork: assortativity and degree of distribution. The ‘degree’of a given node A is the number of links with other nodes ofthe network. In a directed graph, the degree can be subdividedinto an in- and an out-degree, that is the number of strainsinhibiting A or inhibited by strain A, respectively. Theassortativity coefficient (r) [21] is essentially the Pearsoncorrelation coefficient between the degrees of all pairs ofconnected nodes and quantifies the extent of degreeedegreeassociation in a network, i.e. how many nodes with a givendegree tend to inhibit nodes with similar/dissimilar degrees.As a correlation coefficient, assortativity ranges between �1and þ1, and can be calculated on the entire network or part(s)thereof (see Supplementary file 1 for formulas). Positive ornegative values of r indicate a correlation between nodes ofsimilar or different degrees, respectively. In particular, whenr ¼ 1 the network has perfect assortative mixing patterns(nodes with degree X inhibit nodes with degree X ), while,when r ¼ �1 the network is not assortative (high-degree nodesinhibit very low-degree nodes). The statistical significance ofeach assortativity coefficient was evaluated by repeating thesame calculation for 200,000 random matrices obtained byrandomizing existing connections and counting the fraction oftimes the random matrix has an assortativity coefficientgreater than the observed one. This gives a p-value accountingfor the statistical significance of the assortativity coefficient ofthe original matrix. To calculate both in- and out-assortativitycoefficients, we used the formula of Newman et al. [21];
moreover, we also used statistics to measure so-called discreteassortative mixing, which quantifies biases in antagonisticrelationships when we take into account additional features ofthe data, such as the taxonomic affiliation or the host sponge,i.e. to test whether strains of a taxonomic group tend to inhibitstrains of the same (different) taxonomic group(s). Thestatistical significance of our measurement was assessed usingpermutations of the original matrix (see Supplementary file 1for formula and statistical significance assessment).
Analyses were performed using the program ecoNetwork([15], and Brilli et al. in preparation) and the networks werevisualized using Visone software (http://visone.info/).
3. Results and discussion
3.1. Bacterial abundance and phylogenetic affiliation ofisolates
The total number of bacteria from sponge extracts rangedbetween 3.2 � 0.9 � 107 and 1.9 � 1.2 � 108 cells per gram ofL. nobilis and A. joubini tissues, respectively. The percentage ofcultivability was in the range of 0.3e0.6% (1.8 � 0.7 � 105
and 5.2 � 0.5 � 105 CFU per gram of L. nobilis and A. joubinitissues, respectively). Viable counts were comparable to datareported by Muscholl-Silberhorn et al. [19] on different culturemedia for Mediterranean sponges.
From a total of 75 bacterial isolates (38 from A. joubini and37 from L. nobilis), 27 different AluI ARDRA patterns wereobtained (Table 1), most of which (22 out of 27) were repre-sented by one to three isolates. In agreement with Websteret al. [34], we observed wide phylogenetic affiliation withinthe bacterial cultivable fraction obtained from the two spongesanalyzed. Isolates were placed within the g-Proteobacteria(65.3%), a-Proteobacteria (17.3%), CFB group of Bacter-oidetes (10.7%) and Actinobacteria (6.7%), with most of them(56) having as closest relatives Arctic or Antarctic bacteria.We also found Actinobacteria that had been previously iso-lated from sponges originating from marine locations differentfrom Antarctica [7,8,13,14,19,23,24,26,33,35].
Strains belonging to the a-Proteobacteria and the Bacter-oidetes were exclusively isolated from A. joubini, whereasmembers of the g-Proteobacteria predominated in both sponges.The two sponges exclusively shared members of OTU 3 (Table 1).The differences recorded in the compositions of the two bacterialcultivable communities analyzed in this study suggest thatdifferent sponge species harbor bacteria belonging to differentspecies/genera, confirming previous observations [14,24,34].
3.2. Antagonistic interactions among bacterial isolates
When testing inhibitory interactions among bacteriaobtained from the same sponge, active isolates accountedfor 62.2 and 90% of the total number of bacteria isolated fromL. nobilis and A. joubini, respectively (Tables 2 and 3).
As previously observed [6,15,17], inhibition patterns variedfor different bacterial isolates, even if they were grouped in thesame OTU, thus presumably belonging to the same species.
29S. Mangano et al. / Research in Microbiology 160 (2009) 27e37
Author's personal copy
This finding suggests that antagonism might be due todifferent inhibitory mechanisms within the same species. Inparticular, inhibitory interactions were rarely detected amongisolates belonging to the same OTU.
Autoinhibition was observed for two (TB 26 and TB 3) andfour (TB 80, TB 79, TB 47 and TB 81) bacterial isolatesassociated with L. nobilis and A. joubini, respectively. Thisphenomenon, acting as a controlling factor in the maintenanceof species diversity, is frequent in environments inhabited bytaxonomically different bacteria. It is generally related to thesynthesis of bacteriocins, with polypeptide killing belongingto strongly related species, providing the producer bacteriaa selective advantage, i.e. partially limiting itself and coex-isting with competitors [7,20].
As antagonistic activity was investigated by testing indi-vidual bacterial isolates against one another, several activeisolates also proved to be sensitive or resistant to the effects ofthe other tester strains. Results revealed that the 54 and 76.3%of bacteria isolated from L. nobilis and A. joubini, respectively,were sensitive. Among resistant isolates (nearly 46% fromL. nobilis and 23.7% from A. joubini), only eight showed noantagonistic interactions, being neither active nor sensitive.
Data obtained from cross-niche inhibitory interaction assaysare reported in Tables 4 and 5. The antagonistic activity of the 33(out of 37) L. nobilis-associated bacterial strains was testedagainst 25 (out of 38) A. joubini-associated isolates (Table 4).Each of the 33 bacterial isolates inhibited growth of one to seven
targets. In particular, 11 strains (underlined in Table 4) previouslyunable to inhibit growth of bacteria retrieved from the samesponge (i.e. L. nobilis) inhibited A. joubini-associated bacterialisolates. The assays also revealed that 10 (out of 25) A. joubinibacterial isolates, mainly belonging to the a-Proteobacteria, weresensitive, whereas strains assigned to the g-Proteobacteriagenerally were resistant. L. nobilis-associated bacterial isolatesgrouped in the same OTU showed slightly different inhibitionpatterns. In particular, strains assigned to Pseudoalteromonas sp.were not active against isolates clustering in the same OTU(obtained from both sponges).
The antagonistic activity of 21 (of 38) strains isolated fromA. joubini was tested against all 37 L. nobilis-associatedisolates (Table 5). All but one (TB 57) isolate was active, asthey inhibited growth of one to eight targets. A number of L.nobilis targets (21 bacterial isolates) appeared to be sensitiveto inhibition by A. joubini-associated bacteria. Eight L. nobilisbacterial isolates (TB 30, TB 33, TB 34, TB 28, TB 2, TB 15,TB 16 and TB 23), previously shown to be resistant to theinhibitory activity of bacteria retrieved from the same sponge,were sensitive to A. joubini bacterial strains. As observed forL. nobilis, bacterial isolates had different inhibition patternseven if they clustered in the same OTU.
Overall, the percentage of active bacteria (81.3%) wasmuch higher than those (5e50%) reported for particle-attached and epibiotic bacteria and (15%) for their Antarcticseawater counterparts [15], suggesting that microorganisms
Table 1
16S rRNA gene sequence affiliation, with their closest phylogenetic neighbors, of Antarctic bacterial isolates representing each OTU obtained by ARDRA
30 S. Mangano et al. / Research in Microbiology 160 (2009) 27e37
Author's personal copy
derived from Antarctic sponges could be strongly competitive,also confirming the assumption that bacteria benefit more fromantagonistic interactions when colonizing particles (bothliving and dead) for which high bacterial densities occur ratherthan free-living in seawater, with this being a key factor inregulating bacterial populations [12,17,20]. Antagonism wasapparently not linked to the presence of plasmid molecules, asplasmids (size range 1.5e3.5 kb) were detected only in threePseudoalteromonas isolates (TB 14, TB 19 and TB 43), allbelonging to OTU 3 (data not shown).
3.3. Network analysis
In order to visualize and statistically study the antagonisticinteractions occurring among bacterial isolates, the dataobtained were analyzed using the network theory [2,21]. Thenetworks shown in Fig. 1 visualize inhibitory relationshipsexisting between strains isolated from either the same sponge or
the two sponges. Strains were stratified on the basis of thenumber of sent (out-degree Fig. 1a) or received (in-degreeFig. 1b) connections. The greater the size of each node, thegreater the number of in- or out-connections a strain establisheswith the others.
The degree of distribution of the network, which providesinformation on several of its properties [2], such as resistanceto random node removal (the disappearance of strain(s)), wasalso evaluated. The analysis performed (see SupplementaryFigs. 1 and 2) revealed that in- (out-)degree distribution can bewell-fitted by a power-law of the in- (out-)degree, i.e. moststrains have just a few connections, whereas the remainingsmall fraction of isolates shows many more connections.Power-law degree distributions have major consequences interms of robustness of the network when confronted withperturbations such as random node removal.
Lastly, in- and out-degree assortativity coefficients werecalculated (Table 6). Both coefficients were statistically
Table 2
Inhibitory activity and susceptibility of Antarctic bacterial isolates from L. nobilis, in relation to their phylogenetic affiliation
Phylum or class Genus Isolate OTU Inhibition against
(no. of sensitive isolates
for phylum or class)
Inhibited by
(no. of active isolates
for phylum or class)
GAM ACT GAM ACT
g-Proteobacteria Shewanella TB 1 1 1 1
TB 3 1 3 5
TB 4 1 2 5
TB 21 1 3
TB 28 1 1
TB 37 1 1
TB 29 8 6
TB 31 8 1 1 7
TB 7 1a 9
TB 36 10 2 2
Pseudoalteromonas TB 8 3 1
TB 9 3 1 8
TB 10 3 2
TB 11 3 1 1 2
TB 5 3 2
TB 6 3 3 1
TB 12 3 2 1
TB 14 3 4
TB 17 3 7
TB 19 3 5 1 1
TB 22 3 1
TB 24 3 3
TB 25 3
TB 27 3 4 1 1
TB 30 3 4
TB 32 3 1
TB 33 3 6
TB 34 3 3 1
TB 13 4
Psychrobacter TB 15 2
TB 2 2
TB 20 2 1
nd TB 35 61 3 1
Actinobacteria Arthrobacter TB 16 5
TB 23 5
TB 18 55 1
TB 26 53 1 1 5 1
nd, not determined. GAM, g-Proteobacteria; ACT, Actinobacteria.
31S. Mangano et al. / Research in Microbiology 160 (2009) 27e37
Author's personal copy
significant ( p-value < 0.0001). The negative values of thecoefficients suggested that: (i) in the out-degree network,connections mainly go from high degree nodes to low-degreenodes, and that; (ii) in the in-degree graph, nodes with a highdegree are mainly inhibited by nodes with a low degree. Thissuggests a sort of hierarchical organization of antagonisticrelationships and a preference concerning antagonisticactivity, i.e. the ability of a strain to inhibit other strains seemsto be non-random.
In appearance, analysis of the networks reported in Fig. 1suggests inhibitory interactions occurring within the Proteo-bacteria which are more numerous than those existing withother strains; however, the degree of assortativity was notsignificant (Table 6) suggesting that the high value for thecoefficient of assortativity might depend on the high number ofProteobacteria in the dataset. In contrast, members of theActinobacteria appear to be less assortative than Proteobacteria,
and tend to inhibit species of the same taxonomic group. Wealso found that the L. nobilis bacterial community was assor-tative, that is most antagonistic relationships occurred mainlybetween strains isolated from the same sponge species ratherthan between isolates from different sponge species. The sameis true for A. joubini; however, in the latter case there is nostatistical support.
A. joubini-derived bacterial strains interacted much morewith each other than those recovered from L. nobilis: theirinhibition patterns were more diversified and the number ofsensitive strains for every active isolate was generally higherthan that recorded among bacteria from L. nobilis. In contrast,cross-niche inhibition analysis revealed that, although the totalnumber of sensitive A. joubini-retrieved isolates was lower,inhibition patterns of L. nobilis-derived active bacteria weregenerally wider. As previously suggested by Long and Azam[17] for particle-attached and free-living bacteria, antagonistic
Table 3
Inhibitory activity and susceptibility of Antarctic bacterial isolates from A. joubini, in relation to their phylogenetic affiliation (unidentified isolates are not
included)
Phylum or class Genus Isolate OTU Inhibition against (no. of sensitive
isolates for phylum or class)
Inhibited by (no. of active
isolates for phylum or class)
ALF GAM BAC ACT ALF GAM BAC ACT
ALF Roseobacter TB 44 7 1 1 4 1
TB 73 3a nt nt nt nt 2 4
TB 60 12 3 3 7 1
Octadecabacter TB 80 6 8 1 5 8 1
TB 71 6 nt nt nt nt 5 6 1 1
TB 72 6 nt nt nt nt 5 9 1 1
TB 74 6 1 1 3 8 1
TB 77 6 nt nt nt nt 1 5 1
TB 66 6 nt nt nt nt 5 8 1
Sphingopyxis TB 79 56 10 2 1 4 4 1
TB 82 57 10 4 7
GAM Psychrobacter TB 40 9a 3
TB 47 9a 6 1 2
TB 67 9a 9 5 1 1
TB 54 9 6 1 1
TB 61 9 1 1
TB 55 9 2 3 1
TB 56 9 5 1
TB 57 9 1
TB 58 9 1
Pseudoalteromonas TB 41 3 10 1 1
TB 42 3 3 1
TB 43 3 7 1 2
TB 51 3 11 1 1 2
TB 64 58 5 1
TB 49 64 5 1 1 1
nd TB 59 29 6 1 1 2 3
TB 81 6a 8 1 1 5 11
TB 76 60 6 1 2 6
BAC Psychroserpens TB 39 11 7 2
Formosa TB 65 17 nt nt nt nt 3 6 1
TB 53 17 4 4 1
nd TB 38 2a
TB 45 2a nt nt nt nt 3 4
TB 48 2a 4 3 5 1
TB 50 2a nt nt nt nt 1 1
TB 52 2a 1 1
ACT Arthrobacter TB 69 54 4 1 3
ALF, a-Proteobacteria; GAM, g-Proteobacteria; BAC, Bacteroidetes; ACT, Actinobacteria; nt, not tested. nd, not determined.
32 S. Mangano et al. / Research in Microbiology 160 (2009) 27e37
Author's personal copy
interactions occurring among bacteria isolated from thesponges A. joubini and L. nobilis might contribute to shapingbacterial communities. Based on data obtained in the presentwork dealing only with antagonism, even if it occurs amongcultivable bacteria, it can be speculated that this phenomenonwould act differently within bacterial-associated communitiesanalyzed, since the A. joubini bacterial community appears toautoregulate itself, whereas that of L. nobilis tends to defendits structure by acting against ‘‘alien’’ microorganisms. Thishypothesis seems to be supported by the analysis of antago-nistic interactions in relation to phylogenetic affiliation ofbacterial strains. Among L. nobilis bacterial isolates, antago-nistic interactions occurred between closely related isolatesalso belonging to the same OTU, suggesting (as presumed forautoinhibition) potential production of bacteriocin-likecompounds. In addition, as reported by Hentschel et al. (2001)for bacteria isolated from a Mediterranean sponge, Gram-positive strains were generally active against Gram-positivebacteria, and Gram-negative strains were generally active
against Gram-negative bacteria. Data obtained are indisagreement with those reported by Grossart et al. [6], whonever observed reciprocal inhibition between closely relatedisolates.
Similarly to results reported by Long and Azam [17] andLo Giudice et al. [15], A. joubini-derived bacteria generallyinhibited growth of both closely related and taxonomicallydistant bacteria. In particular, contrary to the L. nobilisbacterial community, A. joubini-derived g-Proteobacteriashowed higher inhibitory activities against a-Proteobacteriarather than members of the same class.
Cross-niche inhibitory analyses revealed that L. nobilis-derived isolates acted predominantly against A. joubinia-Proteobacteria and Bacteroidetes, suggesting that antago-nistic features expressed by these bacteria may inhibit othersponge-associated microorganisms, thus resulting in theobserved dominance of the cultured bacterial community. Thesame did not hold true for A. joubini-derived isolates, mainlyactive against g-Proteobacteria.
Table 4
Antagonistic interactions of L. nobilis-derived bacterial isolates against A. joubini-derived bacterial isolates
Text in bold indicates isolates lacking inhibitory activity against bacteria derived from the same sponge (L. nobilis). ALF, a-Proteobacteria; GAM, g-Proteo-
bacteria; BAC, Bacteroidetes. List of the other 14 A. joubini isolates used as targets, but revealed as resistant: Psychrobacter spp. TB 40, TB 47, TB 55, TB 56, TB
57, TB 58, TB 61 and TB 67; Pseudoalteromonas spp. TB 41, TB 42, TB 43, TB 49, TB 51 and TB 64; Arthrobacter sp. TB 69. See Table 5 for identification of
sensitive A. joubini isolates and text for more details. þ, positive result.
33S. Mangano et al. / Research in Microbiology 160 (2009) 27e37
Author's personal copy
Table 5
Antagonistic interactions of A. joubini-derived bacterial isolates against L. nobilis-derived bacterial isolates
nd 60 TB 76 þ þ þ þ þ þ þ þBacteroidetes nd 2a TB 48 þActinobacteria Arthrobacter 54 TB 69 þ þ þ þ þnd, not determined. ALF, a-Proteobacteria; GAM, g-Proteobacteria; BAC, Bacteroidetes. List of the other 16 L. nobilis isolates used as targets, but shown to be resistant: Shewanella spp. TB 1 and TB 37;
Pseudoalteromonas spp. TB 5, TB 6, TB 8, TB 9, TB 10, TB 12, TB 13, TB 14, TB 17, TB 19, TB 22, TB 25, TB 27 and TB 32. See Table 3 for identification of sensitive L. nobilis isolates and text for more
details. þ, positive result.
34
S.M
angano
etal.
/R
esearchin
Microbiology
160
(2009)
27e37
Author's personal copy
Fig. 1. Network analysis of antagonistic relationships between sponge-derived bacteria. Each node represents a bacterial strain and has been assigned a score based
on the number of connections with other isolates, which, in turn, is used both to assign a proportional size to each node and to stratify nodes. In (a) nodes are
stratified on the basis of their out-degree, i.e. the number of links they send, while in (b) node size and stratification are based on the in-degree, i.e. how many
connections a link receives. See details in the text.
35S. Mangano et al. / Research in Microbiology 160 (2009) 27e37
Author's personal copy
Overall, active isolates were mainly affiliated with theg-Proteobacteria, followed to a lesser extent by thea-Proteobacteria, Actinobacteria and finally, Bacteroidetes.Data obtained in this work are consistent with those reportedfor pelagic microbes by Long and Azam [17] who reportedg-Proteobacteria to be dominant producers, followed bya-Proteobacteria and, to a lesser extent, Bacteroidetes. Incontrast, Grossart et al. [6] and Lo Giudice et al. [15] reportedActinobacteria isolated from pelagic marine environments tobe the dominant inhibitors.
Altogether, active g-Proteobacteria belonged to the generaPseudoalteromonas (for both sponges), Shewanella (only forL. nobilis) and Psychrobacter (for both sponges). Pseudoalter-omonas members, frequently associated with eukaryotic hostsas well as algae, are well-known producers of antimicrobialcompounds, while the antibacterial activity of Shewanella sp.was reported for the first time by Bhattarai et al. [3]. To ourknowledge, the genus Psychrobacter could represent a novelsource of antimicrobial compounds. The antibacterial activity ofthe genus Roseobacter has been previously reported [3,6,7,17],whereas, to date, no reports exist on members of Octadecabacterand Sphingopyxis, some of which proved to be particularlyactive against bacteria isolated from the same sponge. Actino-bacteria have considerable value as prolific producers of anti-biotics and other therapeutic compounds [35]. Nevertheless, theantimicrobial activity of Arthrobacter spp., as observed in thepresent study, has rarely been reported [7,22].
The analysis of the degree of distribution of the networkof antagonistic relationships suggests that network topologyis very resistant to random node (strain) removal [2],because its structure is dependent on a very low number ofhubs (more active strains). However, the elimination of hubstrains might compromise overall network structure, i.e. itmight cause a global change in community composition.Considering the importance of the sponge-associatedmicrobiota, via photosynthesis, nitrogen fixation, or activecompound production [25], it is plausible that a drasticchange in the relative abundances of bacterial species caused
by removal of hub strains might have a strong impact onsponge fitness.
In conclusion, microbiological investigations of theAntarctic marine environment have mainly examined sedi-ment, sea-ice and seawater microbial communities. Onlyrecently, Webster et al. [34] and Webster and Bourne [32],primarily using a culture-independent approach, analyzed themicrobial community associated with Antarctic invertebrates.To the best of our knowledge, this is the first investigation onsponge-associated cultivable Antarctic bacteria. We are awarethat the isolated strains represent only a small part of thesponge-associated bacterial community and cannot depict theentire sponge microbiota. However, some ecological extrapo-lations can be made based on data obtained in this work. Itcould be concluded that the sponge-associated cultivablebacterial community is: (1) mainly composed of bacteriapreviously retrieved from polar marine environments anddifferent from those associated with sponges from otherclimates; and (2) probably sponge-specific, as L. nobilis and A.joubini harbored microorganisms belonging to differentspecies/genera.
Results from inhibitory assays suggest that antagonismcould play a significant role in structuring bacterial commu-nities within sponge tissue. In laboratory-scale experiments itis difficult to exactly reproduce all biotic and abiotic featurescharacterizing an environment, due to well-known biasesarising from isolation and cultivation procedures. In fact, itmust be pointed out that various environmental factors andspecific sponge biological properties (e.g. difference inmorphology, structure, nutritional behavior), as well as theoccurrence of other kinds of microbial interactions (such ascommensalism and symbiosis, which could also involveuncultivable bacteria) affect the true bacterial communitycomposition. Nevertheless, results from inhibitory assaysamong cultivable bacteria in artificial systems could provideprecious preliminary indications on bacterial interactionsoccurring in a natural environment, representing a baseline forfurther investigation of the ecological role of Antarcticsponge-associated bacteria.
Our results confirm the antibacterial features of Antarcticmicroorganisms [15,16,22] and enable inclusion of sponge-associated Antarctic bacteria among potential novel sources ofantibacterial molecules for applications in a pharmaceuticalcontext.
Acknowledgements
L. Michaud and A. Lo Giudice are grateful to Prof. G.Odierna (University of Naples, Italy) and the crew of the M/N‘‘Malippo’’ for assistance with sponge collection, and to theentire staff of ‘‘Mario Zucchelli’’ Station for logistic help andsupport which made the expedition possible. Our thanks go toProf. M. Pansini and Dr M. Bertolino (both from the Universityof Genoa, Italy) for sponge identification. This research wassupported by grants from PNRA (Programma Nazionale diRicerche in Antartide), the Italian Ministry of Education andResearch (PEA 2004, Research Project PNRA 2004/1.6).
Table 6
Several assortativity measures for the network of A. joubini and L. nobilis
Category Coefficient Significance
Out- �0.129 ***
In- �0.264 ***
Out-assortativity by groupProteobacteria 0.612067
Actinobacteria 0.166667 ***
Bacteroidetes 0
Miscellaneous 0.25
A. joubini 0.485863
L. nobilis 0.442177 ***
Out- and in-rows contain assortativity coefficients for out- and in-degrees. The
in-degree proved to be significantly assortative in nature: high in-degree nodes
tended to be inhibited by low in-degree nodes. Assortativity calculated for
taxonomic groups revealed that Actinobacteria tend to inhibit species of their
group. By considering the two different sponges, we obtained statistically
significant assortativity for isolates from L. nobilis. Statistical significance
code: ***p < 0.0001 calculated as explained in Section 2.
36 S. Mangano et al. / Research in Microbiology 160 (2009) 27e37
Author's personal copy
Appendix A. Supplementary material
Supplementary information for this manuscript can bedownloaded at doi: 10.1016/j.resmic.2008.09.013.