Biotechnological potential of Actinobacteria from Canadian and Azorean volcanic caves

15 Pages·2016·1.46 MB·English

by  Cristina Riquelme

ApplMicrobiolBiotechnol
DOI10.1007/s00253-016-7932-7
ENVIRONMENTALBIOTECHNOLOGY
Biotechnological potential of Actinobacteria from Canadian
and Azorean volcanic caves
CristinaRiquelme1&MariadeLurdesEnesDapkevicius1&AnaZ.Miller2&
ZacharyCharlop-Powers3&SeanBrady3&CohordMason4&NaowaratCheeptham4
Received:6May2016/Revised:7September2016/Accepted:12October2016
#Springer-VerlagBerlinHeidelberg2016
Abstract Cavesareregardedasextremehabitatswithappro- (Azores)andsubsequentcharacterizationoftheirantibacterial
priate conditions for the development of Actinobacteria. In andenzymaticactivities.Multipleenzymaticandantimicrobi-
comparison with other habitats, caves have not yet been the alactivitieswereidentifiedfrombacterialoftheArthrobacter
targetofintensivescreeningforbioactivesecondarymetabo- and Streptomyces genera demonstrating that actinomycetes
litesproducedbyactinomycetes.Asaprimaryscreeningstrat- from volcanic caves are promising sources of antibacterial,
egy, we conducted a metagenomic analysis of the diversity antibiofilmcompoundsandindustriallyrelevantenzymes.
andrichnessofakeygenerequiredfornon-ribosomalpeptide
(NRP)biosynthesis,focusingoncave-derivedsedimentsfrom Keywords Caves .Actinobacteria .Metagenomics .
twoCanadiancaves(alavatubeandalimestonecave)tohelp Antimicrobialactivity.Enzymaticactivity
uspredictwhetherdifferenttypesofcavesmayharbordrug-
producing actinobacteria. Using degenerate PCR primers
targeting adenylation domains (AD), a conserved domain in Introduction
the core gene in NRP biosynthesis, a number of amplicons
were obtained that mapped back to biomedically relevant Primary and secondary metabolites from Actinomycetales
NRP gene cluster families. This result guided our culture- (akaactinomycetes)areimportantsourcesofindustriallyrel-
dependent sampling strategy of actinomycete isolation from evantcompounds(Lam2006;Donadioetal.2010;Miaoand
thevolcaniccavesofCanada(BritishColumbia)andPortugal Davies2010;Manivasaganetal.2014).Theisolationofcom-
pounds from these bacteria has increasingly shown to yield
moreunknowncompoundssuggestingtheneedforexploring
alternativesourcesofbiodiversitywithinsuchprolificbacteria
Electronicsupplementarymaterial Theonlineversionofthisarticle
(doi:10.1007/s00253-016-7932-7)containssupplementarymaterial, (Katzetal.2015).GoodfellowandFiedler(2010)havesug-
whichisavailabletoauthorizedusers. gestedminingunderexploitedhabitatsforrareActinobacteria
andtheirmetabolitesasastrategytocircumventthisproblem.
* NaowaratCheeptham Rare Actinobacteria from different types of soils were de-
[email protected]
scribed as a potential source of novel metabolites (Tiwari
and Gupta 2013; Guo et al. 2015). Caves, particularly those
1 FoodScienceandHealthGroup(CITA-A),Departamentode
ofvolcanicorigin,remainanunderexploitedreservoirofbac-
CiênciasAgrárias,UniversidadedosAçores,Angrado
terialmetabolitesofpotentialindustrialrelevance(Cheeptham
Heroísmo,Açores,Portugal
etal.2013;MontanoandHenderson2013).
2 InstitutodeRecursosNaturalesyAgrobiologíadeSevilla,Consejo
Bothculture-independentandculture-dependentstrategies
SuperiordeInvestigacionesCientíficas(IRNAS-CSIC),
Sevilla,Spain are important tools in natural product research (Milshteyn
3 LaboratoryofGeneticallyEncodedSmallMolecules,The etal.2014).Culture-independenttechniquesmayprovidecru-
RockefellerUniversity,NewYork,NY,USA cial information on biodiversity and thus help avoiding the
4 DepartmentofBiologicalSciences,FacultyofScience, most common pitfall in natural product research, i.e., redis-
ThompsonRiversUniversity,Kamloops,BC,Canada covery (Katz et al. 2015). Scanning electron microscopy
ApplMicrobiolBiotechnol
(SEM) has also been used to provide a first, morphology- protocols may need to be shifted to include testing against
based, view on the bacterial diversity in cave habitats biofilms. Biofilms are communities of bacteria encased in a
(Northupetal.2011;Riquelmeetal.2015a). self-synthesizedpolymericmatrixthatmayadheretobioticor
Toassessthebacterialdiversityofcaveenvironments,we abiotic surfaces (Hall-Stoodley et al. 2004). This mode of
chose to investigate non-ribosomal peptides (NRPs), a large growthprotectsbacteriafromeradicationbydesiccation,nu-
and diverse family of bacterial metabolites. NRPs have di- trientdeprivation,and antibiotictreatment (GaddyandActis
verse pharmacological properties including antibiosis, anti- 2009). Bacteria associated with biofilms may be up to 1000
fungal, and antitumor activities (Newman and Cragg 2012), times more resistant to antibiotics compared to planktonic
and they are synthesized by large modular mega-enzymes cells,enablingcellstopersistdespiteintensiveantibioticther-
callednon-ribosomalpeptidesynthetases(NRPSs).NRPSen- apy (Mah and O’Toole 2001; Gaddy and Actis 2009).
zymesarecomposedofdiscretemodulesconsistingofhighly Additionally,therehavebeenmanystudiesontheactivation
conservedproteindomainsthatcatalyzetheincorporationof of natural compounds with either full spectrum sunlight or
singleaminoacidsintoagrowingpeptidechain(Walsh2004). ultraviolet light. These studies include the activation of bio-
Withtheexceptionofrarecasesofconvergentevolution,bio- cidalcompoundssuchasplantextracts,cowurine,andeven
synthetic gene clusters that encode closely related structures some antibiotics (Cheeptham and Towers 2002; Upadhyay
appear to have arisen from vertical evolutionary changes etal.2010;Yuanetal.2011).
(Fischbach et al. 2008). It is therefore possible to use the The main goal of this research was the screening of
phylogeneticdifferencesobservedbetweenindividualNRPS actinobacterial isolates collected in volcanic caves from
domainsequences as a surrogate for predicting relationships Azores (Portugal) and British Columbia (Canada) for their
amongentireNRPgeneclusters(Charlop-Powersetal.2014; antibacterialandenzymaticactivities,thustappingintoastill
ChangandBrady2014;Owenetal.2015).Usingthisgeneral unexploited potential source of useful bacterial metabolites.
approach,investigationofthediversityandrichnessofNRP To be able to meet such a goal, a preliminary investigation
biosynthesisincave-derivedsedimentscanbeachieved. was conducted on the functional differences observed be-
Caveconditionsmayenhancetheproductionofhydrolytic tween individual non-ribosomal peptide biosynthesis
enzymes and antimicrobial compounds (Davies 2006; (NRPS)domainsequences viaanassignmentofadenylation
Nakaew et al. 2009; Hibbing et al. 2010; Cheeptham et al. domains(AD)andknownantimicrobialproducinggeneclus-
2013).Bacterialhydrolyticenzymesareofinterestforarange ters to determine whether potential antimicrobial agent pro-
ofdifferentindustrialapplicationsandareanexpandingmar- ducers can be found in deeperand morepristine cave areas.
ket area (Adrio and Demain 2014). Gelatinase, chitinases, MetagenomicsandSEMdatawereusedtoguidethesampling
cellulases, xylanases, pectinases, inulinases, xylanases, effort in obtaining actinomycete isolates, and isolation
phytases, and DNases find applications in waste upcycling, targeted different activities, i.e., enzymatic profile, broad-
bioethanol production, biotechnology, household care, cos- spectrum antibacterial activity, antibiosis against planktonic
metic, pharma, textile, paper, and pulp industries, as well as cells, and biofilms of P. aeruginosa (with or without UV
inhumanandanimalnutrition(Mazottoetal.2011;Sanchez treatment).
andDemain2011;Prakashetal.2013;Swartjesetal.2013).
In particular, actinomycetes are a promising source of
biocatalysts (Miao and Davies 2010; Prakash et al. 2013). Materialsandmethods
However, cave actinomycetes have rarely been screened for
theirenzymaticactivities(Tomovaetal.2013). Samplingsites
Naturalproductscreeningisregardedasthemostpromis-
ing line for novel antibiotic discovery, which is needed to Samplesofsedimentsandcoloredmicrobialmatsdeveloping
counteractthelossofeffectivenessofthepresentlyavailable oncavewallsandceilingswerecollectedfromtwoCanadian
chemotherapeuticchoices(Kirst2013;Silver2015).Incom- caves and 12 Portuguese volcaniccaves: (i) Helmcken Falls
parison withother habitats, caveactinomycetes havenot yet Cave,whichisavolcaniccaveinWellsGrayProvincialPark
been the target of intensive screening efforts regarding their in Clearwater, British Columbia, Canada (Cheeptham et al.
antibacterialactivityagainstpathogenicbacteria,withonlya 2013);(ii)RaspberryRisingCave,whichisalimestonecave
few reports published in mainstreamjournals inthe lastfew inMountTuppersysteminGlacierNationalPark,Revelstoke,
years(Duangmaletal.2012;Cheepthametal.2013;Montano BritishColumbia(Canada);(iii)FurnadoLemos(GL),Gruta
andHenderson2013;RuleandCheeptham2013).Theirfind- dosMontanheiros(GM),GrutadaRibeiradoFundo(GRF),
ingsdemonstrateapromisinglyhighprobabilityofdiscover- andGrutadasTorres(GT)inPicoIsland(Azores,Portugal);
ing a novel compound with different modes of action. and (iv) Gruta das Agulhas (GA), Gruta dos Buracos (GB),
However,insomecases,suchaswhenassessingantibacterial Gruta dos Balcões (GBL), Gruta da Branca Opala (GBO),
activity against Pseudomonas aeruginosa, screening Gruta do Natal (GN), Gruta da Terra Mole (GTM), Gruta
ApplMicrobiolBiotechnol
dos Principiantes (GP), and Galeria da Queimada (GQ) in wasprecipitatedfromtheresultingsupernatantwiththeaddi-
TerceiraIsland(Azores,Portugal). tion of 0.6 volumes of isopropyl alcohol. This precipitated
DNA, known as environmental DNA or BeDNA,^ was col-
Scanningelectronmicroscopy(SEM) lectedbycentrifugation(×4000g,30min),washedwith70%
ethanolandresuspendedinaminimumvolumeofTE(10mM
Sediment samples from Helmcken Falls Cave were freeze- Tris,1mMEDTA[pH8]).CrudeeDNAwaspassedthrough
dried, fixed with osmium fumes as described in Cheeptham tworoundsofcolumnpurificationusingthePowerCleansys-
et al. (2013), and observed at the University of British tem(MoBio,Carlsbad,CA).PurifiedeDNAwasthendiluted
Columbia (UBC) BioImaging Facility, using a Hitachi to30ng/μlandarchivedforuseinPCRreactions.
S4700 Field Emission SEM (Hitachi, Tokyo, Japan).
Samplesofsmallcavewallchipscoveredwithmicrobialmats First round PCR amplification Degenerate primers
fromAzoreanvolcaniccavesweremounteddirectlyonSEM targeting conserved regions of the AD [A3F (5′-
samplestubs,sputtercoatedwithAu-Pdfilm,andexamined GCSTACSYSATSTACACSTCSGG) and A7R (5′-
onaJEOL5800SEM,asdescribedinHathawayetal.(2014). SASGTCVCCSGTSCGGTA) (Ayuso-Sacido and Genilloud
2005)] were used to amplify gene fragments from crude
NRPSbiosyntheticconserveddomainpyrosequencing eDNA. Forwardprimers contained a 454 sequencing primer
andassessmentofNRPbiodiversityofsedimentsamples (CGTATCGCCTCCCTCGCGCCATCAG) followed by a
fromCanadiancaves unique 8 bp barcode, and the reverse primer contained the
454adaptorsequencefollowedbythedegenerateA7Rprimer
ExperimentalsetupFivesedimentsamplesfromtwodiffer- sequence (5′-CTATGCGCCTTGCCAGCCCG
ent Canadian caves and one forestsoilsample (as a control) CTCAGSASGTCVCCSGTSCGGTA-3′). The PCR reaction
were collected and subjected to DNA extraction; the DNA consisted of 25 μl of FailSafe PCR Buffer G (Epicenter,
extracted directly from each sample was column purified, Madison, Wisconsin), 1 μl recombinant Taq Polymerase
andthecrudeenvironmentalDNAextractswereusedastem- (Bulldog Bio, Portsmouth, NH), 1.25 μl of each primer
plates in PCR reactions with barcoded degenerate primers (100 mM), 14.5 μl of water, and 6.5 μl of purified eDNA.
designed to target NRPS AD sequences. The resulting AD PCR conditions for the first round of PCR were as follows:
amplicons, containing site-specific barcodes, were pooled 95°Cfor4minfollowedby40cyclesof94°Cfor0.5min,
and sequenced using the Illumina Miseq platform. Raw se- 63.5 °C for 0.5 min, 72 °C for 1 min, and finally 72 °C for
quencereads were filteredfor quality, de-barcoded toassign 5min.
readstoasourceenvironment,andthenclusteredatadistance
of 5 % to generate A-domain operational taxonomic units Second round PCR amplification The first round of PCR
(OTUs)foreachcavesite.TheseOTUtableswerethenused primerscreatesanampliconthatisreadyfor454sequencing.
to assess AD richness, predict the presence of known gene However, we sought to adapt our work to the Illumina plat-
cluster families and compare gene cluster content between form by using a second round of PCR to attach adaptor se-
samples.Thisexperimentalsetupisdetailedbelow. quences(Cimermancicetal.2014).Followingtheexampleof
arecentpaper(Fadroshetal.2014),wedesignedfourforward
CavesedimentcollectionFoursedimentsamples(designated andfourreverseprimersthatcontainedtheMiSeqP5andP7
Cave07-10)werecollectedfromvariouslocationsinthemain sequences, respectively, a 10 bp barcode, a 1–4 bp spacer
cavernoftheHelmckenFallsCave(Fig.1). sequence,and18basepairsofhomologytothe454adaptors
Asabaselinecomparison,aforestsoilsamplecollectedat onthefirstprimer(SupplementaryTableS1).ADamplicons
themouthofHelmckenCavewasalsoincluded(sample11). fromthefirstroundofPCRwerepooledandcleanedwithone
One sample (designated Cave06) was taken from the round of Agencourt Ampure XP beads (Beckman Coulter,
RaspberryRisingCave. Brea, CA) according to the manufacturer’s protocol at a
DNA/beadratioof0.6.Fourpoolsofampliconswereampli-
SedimentDNAextractionDNAwasextractedfromsamples fiedwithfoursetsof454-to-Illuminaexchangeprimers,using
usinga simplifiedversion ofourpreviouslypublished DNA thefollowingcycleprotocol:95°Cfor5minfollowedby8
isolationprotocols(Brady2007;Charlop-Powersetal.2015). cyclesof95°Cfor03s,55°Cfor30s,72°Cfor1min,and
Themodifiedprotocolisasfollows:25gofeachsamplewas finally 72 °C for 5 min. DNA from four reactions were
incubatedina50-mlconicaltubeat70°Cfor2hin25-mlof cleaned using another round of Ampure beads at a DNA/
lysisbuffer(2%sodiumdodecylsulfate[w/v],100mMTris- bead ratio of 0.6, verified for size-homogeneity using the
HCl, 100 mM EDTA, 1.5 M NaCl, 1 % cetyl trimethyl- TapeStation (Agilent), pooled at equimolar concentrations,
ammonium bromide [w/v]). Large particulates were then re- andrunontheMiSeqsequencingmachineusingv3chemistry
moved bycentrifugation(×4000g, 30min), and crude DNA and2×300cycles.
ApplMicrobiolBiotechnol
Fig.1 MapoftheHelmcken
Fallsvolcaniccave(Canada)with
thepositionsofthesampling
sites.SamplesdesignatedCave07
to10werecollectedfromwithin
thecaveandwereusedforNRPS
DNAsequencingwhilesample
11wasofforestoriginand
collectedoutsidethecave.
Sample11wasusedasabaseline
control
Atwo-stepPCRapproachwasemployedsothatwecould Similarity analyses Ecological distances between samples
continuetousethebarcoded454PCRprimerswehaveused were calculated using the Jaccard distance metric, a metric
in the analysis of other metagenomes (Fadrosh et al. 2014). thatassessthepercentageofOTUssharedbetweentwosam-
ThesecondroundofPCRaddedaninvariantsequencethatis plesasafractionoftheOTUsinbothsamples (OTUA&B)/
required for Illumina sequencing. As this second PCR step (OTUA+OTUB-OTUA&B](Oksanenetal.2015).Network
only runs for 8 cycles and is consistent with other protocols plots were generated in Phyloseq (McMurdie and Holmes
forattachinganinvariantprimerforunbiasedsequencing,we 2013),andthedistanceswerealsodisplayedinmatrixformat.
expectthatitintroducedverylimitedifanyadditionalampli- InanefforttoidentifybiomedicallyrelevantNRPgenecluster
ficationbias. familiespresentinbacteriafromcavesediments,weusedthe
Raw reads for all samples are deposited online at the eSNaPD algorithm to analyze the sequence data obtained
National Center for Biotechnology Information. Sequence fromeachsample(Reddyetal.2014).eSNaPDusesabasic
readaccessionsandsampleinformationcanbefoundassoci- BLASTsearchalgorithmandacuratedsetofgeneclustersto
ated with the BioProject PRJNA324563 entitled Cave Soil identify OTUs that are uniquely related, at high sequence
AdenylationDomains.Theaccessionnumbersforthesamples identity, to known gene cluster families (Owen et al. 2015).
inthisstudyaresummarizedinSupplementaryTableS2. Using 454 sequencing reads, eSNaPD was empirically vali-
datedtohaveanapproximately80%successrateforcorrectly
Processing Illumina Miseq data Raw reads from the assigning amplicons to known gene cluster families. The
forward-facing primer were filtered for quality and assigned Illuminadatageneratedinthisstudyweresubjectedtoasim-
tosamplesusingamultistageprocess.Low-qualitybasecalls ilareSNaPDthresholdforanalyzing454sequencingdata.
were trimmed, excised from the reads using the phred algo-
rithm in seqtk, after which samples were de-barcoded using Rarefaction analysis To assess the total AD diversity of
Qiime (version 1.9) (Caporaso et al., 2010). All reads were amplicons obtained from the samples, we performed a rare-
trimmed to 250 bp or discarded if they were shorter. These faction analysis where the OTU table was subsampled at a
readswerethenclusteredat97%identityusingUSEARCH depth ranging from 1 to 700,000. At each sampling depth,
(version 7) (Edgar 2013), and chimeric sequences were re- theOTUsforeachsamplewererandomly,independentlysub-
moved using the default clustering tool. The consensus se- sampledtentimes.Foreachsamplingevent,theChao1esti-
quences at 97 % were used to re-cluster at 95 %, and the mateofdiversitywascalculated(Chao1984;ChaoandShen
resultingB5%^ADOTUtableswereusedforallsubsequent 2003),andthemeanofwhichisdisplayed.Sampleswerealso
rarefactionanddiversityanalyses. sampled evenly at a single fixed depth (346,870 reads; the
ApplMicrobiolBiotechnol
smallest number of reads per sample), and the Shannon and target strains streaks in, at least, their central portion where
Simpson Diversity measurements were calculated for each they crossed the actinobacteria streak was considered as a
sample(Shannon1948;Simpson1949). positive result (inhibition of the target strain by the
actinobacterialculture).
Assignment of AD to known gene clusters AD amplicon
readswereassignedtoknownbiosyntheticgeneclustersusing Screening for enzymatic activity Eighteen actinobacterial
a modification of the environmental Survey of Natural isolates from Gruta dos Buracos (1), Gruta dos Balcões (4),
Product Diversity (eSNaPD, http://esnapd2.rockefeller.edu/) Gruta dos Montanheiros (8), and Gruta das Torres (5) were
bioinformatic algorithm (Reddy et al. 2014). For Miseq for- screened for their enzymatic activities. Amylase, pectinase,
ward reads, we used 250 bp of the 300 bp amplicon and a inulinase,cellulase,xylanase,chitinase,andDNaseweretest-
conservativee-valueBLASTthresholdofe-80toassigndo- ed using the culture media proposed by Babavalian et al.
mainstoageneclusterfamily. (2013), but omitting NaCl. Phytase activity was assessed
using the culture medium proposed by Quan et al. (2001),
whereas the screening for gelatinase activity was performed
Screeningforenzymaticandantimicrobialactivity
ontheculturemediumproposedbynoticedTerzic-Vidojevic
fromAzoreanvolcaniccaves
etal.(2009).Incubationwascarriedoutat15°Cfor7days.
Actinobacteria isolation. One hundred and forty-eight
Identification of the active isolates Selected isolates
actinobacterialisolateswereobtainedfromthewallsandceil-
displayingenzymaticand/orantimicrobialactivitywereiden-
ings of volcanic caves from the Terceira and Pico Islands
tifiedbysequencingofthe 16S rRNAgene. Genomic DNA
(Azores, Portugal). Sampling was carried out by touching
was extracted from cultures grown in nitrate broth with the
speleothems, oozes, bacterial mats, and apparently non-
UltraClean Microbial DNA isolation kit (MoBio, Carlsbad,
colonized surfaces with the cotton tip of a sterile swab. California) using manufacturer’s protocol. The 16S rRNA
These swabs were used to inoculate the surface of half-
genewasamplifiedbyPCRwithuniversalprimers,8forward
strengthR2A(½R2A)agar.Theinoculatedplateswereincu- (5′-AGAGTTTGATCCTTGGCTCAG-3′) and 1492 reverse
bated aerobically at temperatures close to those in the caves (5′-GCYTACCTTGTTACGACTT-3′) using Amplitaq.
(15 °C for all caves, except the plates from Montanheiros Reactions and amplification werecarried out in a 50-μl vol-
cave, which were incubated at 11 °C), until visible growth
ume,asdescribedinHathawayetal.(2014).Ampliconswere
was observed(3–5 days).Actinobacteriawereselectedfrom
cleaned and purified using the Qiagen PCR cleanup kit
theobservedgrowthineachplateonthebasisofcolony(typ-
(Qiagen, Germantown, Maryland). Isolates were sequenced
ical of actinomycetes in shape, structure, and odor) and cell
usingtheBigDyeTerminatorKit(AppliedBiosystems)with
morphological characteristics (Gram-positive filaments) and primers 46 forward (5′-GCYTAAYACATGCAAGTCG-3′)
purified by repeatedly streaking out onto ½ R2A. and 1409 reverse (5′-GTGACGGGCRGTGTGTRCAA-3′)
Arthrobacter isolates that had been sequenced in a previous
andprecipitatedwithsodiumacetateandethanol.Sequences
work(unpublishedresults)werealsoincluded.Stockcultures
werereadbytheABI3130Sequencer(AppliedBiosystems).
were prepared on ½ R2A agar slants and kept at −4 °C for
Searchingfor the closest relatives ofthe sequences was per-
furthertesting.
formedusingthenucleotideBLASTalgorithm(Altschuletal.
1997). Sequences were aligned by SINA aligner (1.2.11)
ScreeningforantimicrobialactivityThe148actinobacterial
(Pruesseetal.2012).Sequencesimilaritymatriceswerebuilt
isolatesobtainedwerescreenedfortheirantimicrobialactivity for each isolate and its BLAST best hit’s sequence using
against Proteus sp. (collection of the Microbiology
BioEditsequenceeditor(version7.2.5)(Hall1999).Alistof
Laboratory–CITA-A), Salmonella typhimurium ATCC
GenBank accession numbers is provided in Supplementary
14028, Staphylococcus aureus (ATCC 9144 and 29,523),
TableS3.
Escherichia coli ATCC 25922, Pseudomonas aeruginosa
ATCC 2785, Listeria monocytogenes ATCC 7466, and
Listeria innocua ATCC 33090, by an agar diffusion assay, ScreeningforantibiofilmactivityinaCanadianvolcanic
using the cross-streak technique (Guo et al. 2010). In short, cave
theactinobacterialisolatesweregrownonalayerof½R2Aat
11–15°,for3–5days,asasinglestreakacrossthediameterof Three bacterial isolates (CM_A1A3, CM_PM58B, and
the agar surface. An overlay of plate count agar (PCA) was CM_RA003)wereoriginallyisolatedfromsedimentsamples
carefullypouredoverthebasis½R2A.Aftersolidifying,the collectedfromthesamplesite08(Fig.1)fromtheHelmcken
targetbacteriawereperpendicularlystreaked.Platesweresub- FallsvolcanicCave(BritishColumbia, Canada)and usedin
sequentlyincubatedat37°Cfor24h.Lackofgrowthonthe this study due to their ability to inhibit biofilms of
ApplMicrobiolBiotechnol
Pseudomonasaeruginosa(Mason2015).Thethreecavebac- TSB,V8,R2A,ISP2,anddeionizedwater.Thepositivecon-
teriawereusedtoinoculatetesttubescontainingeachofthe trolswere2%Virkonsolutionandtheantibioticstetracycline
three different broth media: V-8 juice, International HCl and ciprofloxacin HCl. A 1 % v/v inoculum of
Streptomyces Project #2 (ISP2), R2A fermentation media P.aeruginosaat1.0McFarlandwastransferredto200mlof
(Cheepthametal.2013).Thesebrothmediawereautoclaved sterile molten TSA, and then poured into the sterile Thermo
at121°Cfor15minpriortoinoculation.Theinoculatedtest Scientific™Nunc™SquareBioAssaydishesandallowedto
tubeswerethenculturedat25°Cfor10dayswith100rpmof solidify. The impregnated disks were then placed onto the
agitation in an orbital shaker (New Brunswick Scientific bioassayagarplates. Theplateswereincubated at35°Cfor
Innova 42). Then, 800-μl samples of the cell-free extracts 18h;afterincubation,theinhibitoryzonesweremeasured.
fromeachtubeweretakenatdays2,4,6,8,and10andstored
at−20°Cuntilused.
Results
Screening of cell-free extracts against P. aeruginosa
biofilmsCulturing,exposure,andrecoveryofP.aeruginosa Morphologicalobservations
biofilms were done using the MBEC P&G assay device
(Innovotech, Canada) according to BStandard Test Method SEMrevealedmicrobialstructureswithmorphologicalchar-
for Testing Disinfectant Efficacy against P. aeruginosa acteristics typical of Actinobacteria in the sediment samples
Biofilm using the MBEC Assay^ (ASTM E2799-12, 2012). andcoloredmicrobialmatsfromtheCanadianandPortuguese
Thecell-freeextractsweresplitintotwogroups:UVandnon- volcaniccaves(Fig.2).MostoftheActinobacteria-likemats
UV-treatedsamples.TheUVtreatmentconsistedofexposing comprisedmassesofsporeswithhairysurface(Fig.2a,b)or
the 96-well microplate with the supernatant to UV light spiny surface (Fig. 2c, d). They were arranged inclusters of
(254nm)inaLabconcoClassIIbiologicalsafetycabinet,at single spores (Fig. 2b, d) or spore chains (Fig. 2c, e, f).
room temperature. The negative controls were sterile TSB, Generally, the spores were of globose shape (Fig. 2d), but
V8, R2A, ISP2, and deionized water. The positive controls ovoid and rod-shaped spores were also frequently found
were 2 % Virkon solution and the antibiotics tetracycline (Fig. 2c, e). Spores with spiny surface were observed with
HCl and ciprofloxacin HCl. The measurement of cell-free variablespinesizes(Fig.2d,e).
extractantibiofilm/antimicrobialactivitywasdonewithdilu-
tions and viable cell counts. A preliminary surviving cell NRPSbiosyntheticconserveddomainsandNRP
count was done using spot plate technique on large square biodiversityofCanadiancavesedimentsamples
plates(ThermoScientific™Nunc™SquareBioAssaydishes
with 250 ml of TSA) to assess the antibiofilm/antimicrobial SamplesimilarityToassesstheoverallrelationshipofNRPS
activity.Theplatewasdividedintosections,and10μlofeach systemsintheHelmckenFallsandRaspberryRisingCaves,
samplewasspotplatedontheagar.TheTSAplateswerethen we isolated eDNA from six sediments and amplified the
incubatedat35°Cfor18h.Afterincubation,eachspotonthe adenylation domains from those samples. After sequencing
TSAplateswasinspectedtodetermineifanysampleshadno the amplicons, we assessed the similarity between samples
growthorreducedgrowth.Suchsamplesweretobesubjected by clustering the sequences into OTUs at 95 % similarity
to further measurement. Once identified, the samples were and comparing them using the Jaccard distance metric
diluted from10−1 to10−4. Then, 10mlofeachdilutionwas (Fig.3a).TheJaccarddistanceisaratiooftheOTUsshared
spotplatedontoTSAplatesandincubatedat35°Cfor18h. by two samples as a fraction of the total OTUs that both
After 18 h, the CFU/ml of each sample was calculated and samples contain, which provides a simple way of assessing
thensubjectedtoa2samplettestwithpooledvariances(after AD overlap between amplicon datasets. The closest Jaccard
testingdatafornormalityandequalvariance)todetermineif distances observed between any two samples were obtained
the difference was significant. A P value of <0.05 was con- forsamplesCA07andCA09fromtheHelmckenFallsCave,
sideredsignificant. where5.95%oftheADOTUswereshared.Overall,samples
within the Helmcken Falls Cave cluster more to each other
Kirby-Bauerdiskdiffusionassaytodetermineplanktonic thantotheRaspberryRisingCavesample(CA06)(Fig.3a).
antimicrobial activity The standard Kirby-Bauer (KB) disk TheRaspberryRisingCave(limestone)sampleshowedvery
diffusion assay (Bauer et al. 1966; Wilkins et al. 1972) was limitedrelationshiptoanyofthesamplesfromtheHelmcken
used to determine the antimicrobial activity of the cave iso- Falls Cave (volcanic), sharing less than 0.6 % of its OTUs
latesagainstplanktonicP.aeruginosa.Thesteriledisksused withanyoftheHelmckenFallsCavesamples.
were8mmAdvantecpaperdisks(Tokyo,Japan).Eightymi-
croliters of each cell-free extracts was impregnated on the RarefactionanddomaindiversityanalysisThenumberof
disksinreplicatesofthree.Thenegativecontrolsweresterile ADOTUspresentineachcavesamplewaspredictedfromthe
ApplMicrobiolBiotechnol
Fig. 2 Scanning electron microscopy images showing several with spiny surface; d cluster of individual globose spores with spiny
morphologicalfeaturesofActinobacteria-likestructuresfromCanadian surface fromGruta das TorresCave(Pico Island, Azores, Portugal); e
andAzoreancavesamples.aDensemassofsporeswithhairysurfaceand rod-shaped spores with spiny surface developed in a long chain in
interwoven filaments found in Gruta dos Montanheiros (Pico Island, samples from Helmcken Falls Cave; f chains of spores with hairy
Azores,Portugal);bmassofsporeswithspinyandhairysurfacefrom surface found within microbial mats from Gruta da Terra Mole
HelmckenFallsCave(BritishColombia,Canada);cActinobacteria-like (TerceiraIsland,Azores,Portugal).Scalebars,2μm
matfromHelmckenFallsCavecomposedofchainsofrod-shapedspores
OTUtabledatausingtheChao1diversitymetric(Fig.3b).AD Screeningforenzymaticandantimicrobialactivity
richnesspredictionsforHelmckenFallsCavesamplesrange fromAzoreanvolcaniccaves
from~5000to10,000OTUspersample,whiletheRaspberry
Rising Cave sample is predicted to contain approximately Onehundredforty-eightisolatesobtainedfromthewallsand
2500 unique AD OTUs. The Chao1 diversity metric can be ceilingsofvolcaniccavesfromtheTerceiraandPicoIslands
heavily biased by singleton sequences, and therefore, addi- (Azores, Portugal) were identified as actinomycetes on the
tional diversity estimate experiments using the Shannon and basis of their colony and cell morphology. A considerable
Simpson diversity metrics were performed (Fig. 3c). Using proportion of the tested isolates (27 isolates; 18.1 %)
either of these measures of richness, the relative richness displayedantibacterialactivityagainstatleastoneofthetarget
among samples is similar, and the Raspberry Rising Cave bacteria under study (Table 1). The active isolates were ob-
sampleispredictedtohavelowerADsequencediversitythan tained from Furna do Lemos, Gruta dos Balcões, Gruta da
any other sample. Interestingly, in these analyses, very little Branca Opala, Gruta dos Montanheiros, Gruta das Torres,
differences are observed between the AD diversity estimate and Gruta da Terra Mole. No active isolates were obtained
for the samplecollected atthe mouth ofthe HelmckenFalls from Gruta dos Buracos and Gruta da Ribeira do Fundo
Caveandthesamplescollectedwithinthiscave. (SupplementaryTableS4).
Fromourfindings,Escherichiacoliwasthemostfrequent-
ly inhibited target bacterium (21 isolates), whereas only one
Tracking known natural product gene cluster families in isolate (Streptomyces mauvecolor GM32B4) inhibited
cavesamplesBasedonthemetricsstudied,wewereableto Pseudomonasaeruginosa.S.mauvecolorGMB32B4wasal-
map between 0.1 and 0.3 % of reads obtained from our so the only isolate that was active against all of the tested
amplicon data to known gene clusters (Fig. 3d). In the pathogens, making it a promising source of broad spectrum
eSNaPD analysis, we detect AD sequences from both caves antibioticsthatshouldbeinvestigatedfurther.
that have high sequence identity to a variety of gene cluster Eight other isolates were active against most (7) of the
families known to encode for non-ribosomal peptides of di- tested target bacteria. None of them were active against
verse biomedicalimportance(Fig. 3e). Figure3e shows that P.aeruginosaandListeriainnocua.Twoofthemwereiden-
volcanic cave samples overall showed higher hits with the tifiedasS.avidiniiandtwoasS.spiroverticillatus(Table2).
identifiedADsequencesfromtheeSNaPDanalysisthanthose MostidentifiedisolatesbelongedtotheStreptomycesgenus,
oflimestonecavesample.Also,incomparisonwithacontrol withonlytwoidentifiedasArthrobacter(Table2).Itwaspos-
sample(sample#11:forestsampleoutsidethevolcaniccave), sible to assign all the isolates to a species, with five of them
two samples (samples#08 and 09 which were deeper in) identifiedasS.nojiriensis,threeasS.spiroverticillatus,threeas
showedhighernumberoftheADsequencessimilartothose S.avidinii,twoasA.nicotinovorans,andoneasS.mauvecolor
knownantibioticsfromtheeSNaPDanalysis. (Table2).
ApplMicrobiolBiotechnol
a b
c e
d
Fig. 3 Metagenomic analysis of cave sediment A-domains: a OTU Shannonand Simpson methods. D) Consensus sequences from 95 %
tables generated from PCR-amplifiedand sequenced A-domains were OTUs were subjected to eSNaPD analysis and the fraction of reads
used to calculate ecological distance between cave samples using the assignedtoaknownclusterisdisplayedforeachsample.eNumberof
Jaccard distance metric. b A-domain diversity estimates using the OTUsthatmaptoaknowngeneclustersusingeSNaPD
Chao1 diversity metric. c A-domain diversity estimates using the
Psychrotrophicenzymeactivities(15°C)weretestedin18 abletodegradephytate.Thirteenofthetestedisolatesdegrad-
actinobacterialisolates(Table3).Alltestedisolatesdisplayed edmost(5–8)ofthetestedsubstrates.Allisolateswereableto
at least four of the tested enzymatic activities, but none was degradegelatinandinulin.Degradationofpectin(17isolates),
ApplMicrobiolBiotechnol
Table1 Antibacterialactivityofactinobacterialisolatesobtainedfrom Table 2 Closest relatives of bioactive actinomycetes isolated from
volcaniccavesfromAzores(Portugal)againstSalmonellaTyphimurium volcaniccavesfromtheAzores(Portugal)
ATCC14028(ST),EscherichiacoliATCC25922(EC),Pseudomonas
aeruginosaATCC2785(PA),Proteussp.(PT),Listeriamonocytogenes Isolatecode Closestrelative Identity %Coverage
ATCC 7466 (LM), Listeria innocua ATCC 33090 (LI), and
Staphylococcusaureus(ATCC9144and29,523;SA1andSA2) GB10C Arthrobacternicotinovorans 0.995 100
GBL11D2 Streptomycesnojiriensis 0.994 100
Isolatecode PT ST SA1 SA2 EC PA LM LI
GBL1B Arthrobacternicotinovorans 0.994 99
GBL11D2 − − − − + − − + GBL5B Streptomycesnojiriensis 0.996 100
GBL12A − − − - + − − + GBL5C Streptomycesnojiriensis 0.996 99
GBL5B + − − − + − − − GM32B4 Streptomycesmauvecolor 0.991 99
GBL5C + − − − + − − − GM35A2 Streptomycesspiroverticillatus 0.978 100
GBL7A1 − − − − + − − + GM35A4 Streptomycesnojiriensis 0.990 99
GBO33B − − − + − − − + GM35A7 Streptomycesspiroverticillatus 0.981 100
GBO33F − − − + − − − + GM35B11 Streptococcusnojiriensis 0.986 99
GL02A4 − + + + + − − − GM47C3 Streptomycesspiroverticillatus 0.987 100
GM32B4 + + + + + + + + GT12A3 Streptomycesavidinii 0.972 100
GM32C1 − −+ +− − − − GT12A7 Streptomycesavidinii 0.993 99
GM35A2 + + + + + − + + GT12B3A Streptomycesavidinii 0.972 99
GM35A4 − − − − + − − −
GM35A5 + + + + + − + +
GM35A6 − − − − + − − − assessment of which cell-free extracts showed any signs of
GM35A7 + + + + + − + + antibiofilm activity. In case of CM_A1A3, antibiofilm activity
GM35B11 - − + + + − − + wasobservedonlyintheV8juicemedium,whileCM_PM58B
GM35B14 − −+ − + − − − showedactivityinbothV8juiceandR2Amedia.CM_RA003,
in all three media, showed antibiofilm activity. Only
GM47C3 − − − − + − − +
CM_RA003 and CM_PM58B in V8 had antibiofilm effects
GT12A3 − −+ − − − − −
when exposed to UV light. After the preliminary screening,
GT12A7 + + + + + − + +
samplesshowinginhibitionwerefurtherenumeratedtocalculate
GT12B3A + + + + + − + −
thecolonyformingunitspermicroliter(CFU/ml).Theobserved
GT12B3B + + + + + − + +
CFU/mlforeachgrowthcontrol(R2A,ISP2,andV8)wascom-
GT12B3C + + + + + − + +
paredtotheTSBcontrolusinga2samplettesttoconfirmthat
GT24C6B + + + + + − + +
the media had no effect on the P. aeruginosa biofilms. The P
GTM1E2 − − − + − − − +
valueforeachcomparisonofmediatoTSBcontrolwasgreater
GTM5F − − − + + − − +
than0.05meaningthattherewasnosignificantdifference.
GTM7C + − − + − − − +
Significantly,theisolateCM_PM58BculturedinV8juice
Shownonlytheisolatesthatwereinhibitoryagainstatleastonetarget medium demonstrated antibiofilm activity with and without
bacterium UVexposure(Fig.4).Thiswouldsuggestthattherewassome
activemetaboliteinthecell-freeextractaffectingthebiofilm.
starch(15isolates),xylan(11isolates),andDNA(12isolates) In contrast, there was one sample out of 9 treatments that
were frequent. Less than half of the isolates were found to demonstratedactivityafter exposuretoUV light,althougha
degradechitinandcellulose. previousstudyconductedbyRuleandCheeptham(2013)also
showedthatUVexposureforenhancing/deactivatingantimi-
ScreeningforantibiofilmactivityinCanadianvolcanic crobialeffectsisinconclusiveandthereforeneedsmorework
caves tounderstandmechanismsofUVexposuretothesepotential
cavebacterialmetabolites.
Three bacterial isolates (CM_A1A3, CM_PM58B, and
CM_RA003), obtained from the sediment samples of the
Helmcken Falls Cave, were used in this study because they Discussion
showedtheabilitytoinhibitbiofilmsofP.aeruginosainapre-
viousstudy(Mason2015).Thesethreeisolatesshowedinhibi- Morphologicalobservations
tory activity against P. aeruginosa biofilms on day 8. Three
different fermentation media (R2A, ISP2, and V8) were used Morphologically,Actinobacteriaareoneoftherichestgroups
in this study. The preliminary screening was used as a quick of bacteria, sharing characteristics with both bacteria and
ApplMicrobiolBiotechnol
Table3 EnzymeactivitiesinactinomycetesisolatedfromvolcaniccavesfromtheAzores(Portugal)
Isolatecode Isolateidentity Dnase Gelatinase Xylanase Chitinase Cellulase Amylase Inulinase Pectinase Phytase
GB10C Arthrobacternicotinovorans − + + + − + + + −
GBL11D2 Streptomycesspororaveus + + + + + + + + −
GBL5B Streptomycesspororaveus + + − − − − + + −
GBL5C Streptomycesspororaveus + + − − − − + + −
GBL1B Arthrobacternicotinovorans + + + + + + + − −
GM32B4 Streptomycesmauvecolor + + + + − − + + −
GM35A2 Streptomycesmauvecolor + + + + − + + + −
GM35A4 Streptomycesspororaveus + + + − − + + + −
GM35A5 Notidentified − + + − − + + + −
GM35A7 Streptomycesmauvecolor − + − − − + + + −
GM35B11 Streptomycesspororaveus + + − − − + + + −
GM35B14 Notidentified − + − − − + + + −
GM47C3 Streptomycesmauvecolor + + + − + + + + −
GT12A3 Streptomycesavidinii − + − ? − + + + −
GT12A7 Streptomycesavidinii − + + + − + + + −
GT12B3A Streptomycesavidinii + + − + − + + + −
GT12B3B Notidentified + + + − − + + + −
GT12B3C Notidentified + + + − + + + + −
fungi(Lietal.2016).Themorphologyofsporeshape,spore NRPSbiosyntheticconserveddomainpyrosequencing
surface, and spore structure is an important criterion for andassessmentofNRPbiodiversityofsedimentsamples
recognizing Actinobacteria by microscopy. Many microbial fromCanadiancaves
features observed by SEM in samples from Canadian and
Azorean caves were mainly composed of Actinobacteria, Metagenomicprofilingincavesofdifferentoriginanddimen-
suggesting that these bacteria are able to survive and sionmayprovide ameanstoprobethe relationshipbetween
develop within these volcanic caves, as demonstrated by cave-depthandthebacterialandchemicaldiversityofthecave
Hathaway et al. (2014) and Riquelme et al. (2015a, 2015b). sediments.Ashasbeenobservedinsimilaranalysesofsoils,
These observations have prompted us to furthering our re- only a small fraction of OTUs are shared among samples
search on cave-dwelling Actinobacteria in order to search collected even in very close proximity to each other (Reddy
fornovelantimicrobialcompoundsandenzymaticactivities. etal.2014).Inthisstudy,solelythesamplesCA07andCA09
fromtheHelmckenFallsCaveclusteredtoeach,possiblydue
tophysicalcharacteristicsofthesediments.UnlikeCA08and
CA10 (more like fine sand), the texture of CA07 and CA09
(coarse sand) sediments was observed to be very similar in
textureandgeologicalorigin.Inthisparticularcave,however,
all of the samples withinthe cave are roughly equidistant to
one another when using the 95 % jaccard distance metric,
hindering the assessement of relationships between cave-
depth and bacterial diversity. This observation is consistent
with the culture-based studies of cave microorganisms from
thesecaves,ahypothesiswesoughttotestbyisolatingorgan-
ismsfromcavesamples.Inpreviouswork,266bacterialiso-
latesweresuccessfullyisolatedfrom23samples(collectedat
about100mfromthecaveentrance)fromRaspberryRising
Cave(Golapkhanetal.2013).Eightyofthecaveisolateswere
screened and nine (11.25 %) showed antimicrobial activity
Fig.4 TheaveragesurvivalofPseudomonasaeruginosacells(inCFU/
against MDR-Staphylococcus aureus and seven against
ml)comparedtotheTSBcontrolafterexposuretoday8samples(n=3).
Theerrorbarsindicatestandarderrorofthemean Micrococcus luteus. The detailed identification of these