Phylogenetic Analysis and Antimicrobial Profiles of Cultured Emerging Opportunistic Pathogens

18 Pages·2017·1.34 MB·English

International Journal of
Environmental Research
and Public Health
Article
Phylogenetic Analysis and Antimicrobial Profiles of
Cultured Emerging Opportunistic Pathogens
(Phyla Actinobacteria and Proteobacteria) Identified
in Hot Springs
JocelynLeonieJardine1,AkebeLutherKingAbia2 ID,VuyoMavumengwana1and
EuniceUbomba-Jaswa1,3,* ID
1 DepartmentofBiotechnology,UniversityofJohannesburg,Doornfontein,2094Johannesburg,SouthAfrica;
[email protected](J.L.J.);[email protected](V.M.)
2 AMBIOEnvironmentalManagement,DepartmentofBiotechnology,VaalUniversityofTechnology,
AndriesPotgieterBlvd,PrivateBagX021,Vanderbijlpark1911,SouthAfrica;[email protected]
3 WaterResearchCommission,PrivateBagX03Gezina,Pretoria0031,SouthAfrica
* Correspondence:[email protected];Tel.:[phone]
Received:30June2017;Accepted:7September2017;Published:15September2017
Abstract:Hotspringwatermayharbouremergingwaterborneopportunisticpathogensthatcancause
infectionsinhumans. Wehaveinvestigatedthediversityandantimicrobialresistanceofculturable
emerging and opportunistic bacterial pathogens, in water and sediment of hot springs located in
Limpopo,SouthAfrica.Aerobicbacteriawereculturedandidentifiedusing16SribosomalDNA(rDNA)
genesequencing.ThepresenceofLegionellaspp.wasinvestigatedusingreal-timepolymerasechain
reaction. Isolatesweretestedforresistancetotenantibioticsrepresentingsixdifferentclasses:β-lactam
(carbenicillin),aminoglycosides(gentamycin,kanamycin,streptomycin),tetracycline,amphenicols
(chloramphenicol, ceftriaxone), sulphonamides (co-trimoxazole) and quinolones (nalidixic acid,
norfloxacin). Gram-positive Kocuria sp. and Arthrobacter sp. and gram-negative Cupriavidus sp.,
Ralstonia sp., Cronobacter sp., Tepidimonas sp., Hafnia sp. and Sphingomonas sp. were isolated,
all recognised as emerging food-borne pathogens. Legionella spp. was not detected throughout
the study. Isolates of Kocuria, Arthrobacter and Hafnia and an unknown species of the class
Gammaproteobacteriawereresistanttotwoantibioticsindifferentcombinationsofcarbenicillin,
ceftriaxone,nalidixicacidandchloramphenicol. Cronobactersp. wassensitivetoalltenantibiotics.
Thisstudysuggeststhathotspringsarepotentialreservoirsforemergingopportunisticpathogens,
includingmultipleantibioticresistantstrains,andhighlightsthepresenceofunknownpopulations
ofemergingandpotentialwaterborneopportunisticpathogensintheenvironment.
Keywords: emerging opportunistic pathogens; hot springs; antibiotic resistance; phylogenetic
analysis;Actinobacteria;Proteobacteria
1. Introduction
Precious water resources are threatened by pathogen pollution due to human activities, and
waterborne diseases are well described and investigated globally, including in South Africa [1–3].
Well-known waterborne bacterial pathogens include Escherichia coli, Campylobacter, Vibrio cholera,
Helicobacter, Shigella, Pseudomonas, non-tuberculosis mycobacteria and Legionella, while protozoa
infections also include Acanthamoeba and Naegleria fowleri [2]. Most waterborne diseases result in
diarrhoea, and globally more than 2.2 million children die each year as a result. In South Africa,
diarrhoeaistheleadingcauseofmortalityinchildrenunderfiveyearsold[4].
Int.J.Environ.Res.PublicHealth2017,14,1070;doi:10.3390/ijerph14091070 www.mdpi.com/journal/ijerph
Int.J.Environ.Res.PublicHealth2017,14,1070 2of18
Although hot springs are generally associated with healing practices like balneotherapy or
hydrotherapy[5,6],theyalsocarryapotentialforinfectionbymicroorganisms,asindicatedbythe
isolationoftheabovementionedpathogensfromswimmingpools[7],andthermalbathsassociated
withhotsprings[8,9]. Thepotentialofinfectionappearstobelinkedtothelevelofhumanactivity
andcontact,andtheseorganismsarenotthedominantdisease-causingmicroorganismsinthewaters
ofpristinehotsprings, i.e., waterthathashadnopreviouscontactwithhumanoranimalactivity.
Infections from hot springs are rare and sporadic and mostly associated with Legionella [10] and
free-livingamoeba(AcanthamoebaandNaegleriafowleri)[11].Otherprotozoa,suchasVittaforma[12]and
fungiOchroconisgallopava[13],havealsobeenreportedtocauseinfectionsassociatedwithhotsprings.
Theterms“emergingandopportunistic”pathogensrequirecurrentdefinitionandreviewsince
theyaredescribedrelativetotimeandtheassociatedvulnerablepopulations. Anemerginghuman
pathogenisdefinedastheetiologicalagentofaninfectiousdiseasewhoseincidencehasincreasedinthe
past20yearsandwillprobablydosointhefuture. Inseveralearlierreviewsonemergingwaterborne
pathogens,LegionellaandEnterobactersakazakii[1]werementioned.Thelatter,laterrenamedCronobacter
sakazakii,wasisolatedinbabyformulaandisanemergingpathogenofinfantsandneonates[14–16].
Ontheotherhand,opportunisticpathogensaredescribedaspathogenscausinginfectionsthatexploit
opportunities not normally available, such as a host with a weakened immune system, an altered
microbiota (such as a disrupted gut flora), or breached integumentary barriers [17]. Pathogens
thatarebothopportunisticandemerging,includingPseudomonasandRalstonia[16]Kocuria[18,19],
Sphingomonas[20]andCupriavidus[21],havebeendescribedinmorerecentpublications.
Opportunisticpathogens,bytheirnature,arenotclear-cutharmfulorharmless,andinfections
are rather a reflection of the host’s state of health and immunity. Commonly, these bacteria are
ubiquitousintheenvironmentandareclassifiedasthelowestlevelofbiohazard,i.e.,Level1[22].
They can, and do cause infections in debilitated or immunocompromised individuals (including
neonates, infants, the elderly, those infected with the human immunodeficiency virus/acquired
immune deficiency syndrome (HIV/AIDS) and cystic fibrosis patients) and include 27 genera as
listed by Berg et al. [23]. They also commonly cause nosocomial infections [24,25]. Eight percent
ofacutegastrointestinalinfectionsintheUSAarefromunknowncauses[26],andDewaaletal.[4]
alsodescribedthecausativeagentsofasubstantialproportionofglobaldiarrhealoutbreaks,which
remainunspecified. Theseopportunisticandemergingpathogenicbacteriamaybethemissinglinkin
waterbornediseasesofunknownaetiology. Casestudiesandoutbreaksduetotheseopportunistic
pathogensarelistedinTableS1(SupplementaryMaterials),showingthediversityofpossibleinfections,
globallocationofinfectionsandriskfactorsassociatedwiththehost’scondition. Itiswellknownthat
thesebacteria,includingC.sakazakii[13],Hafnia[27],Sphingomonas[27,28]andLegionella[29],canbe
isolatedfromenvironmentalsamplessuchassoilandwater. Cupriavidussp. hasbeenisolatedfrom
drinkingwater[30],andGulbenkianiasp. fromwastewater[31]. Notonlydotheyexistubiquitously
intheenvironment,butinterestinglytheyarealsofoundintherhizospheremicrobiome,andonor
withinplants,andplayaroleintheproductionofplantgrowthfactors[16,23].
A variety of bacterial species has been isolated from hot springs. Most cultured bacteria
reportedfromhotspringsinotherstudies[32–34]havebeendescribedasbelongingtothephylum
Firmicutes, genus Bacillus, due to their endospore-forming ability, thermotolerance and relatively
simplenutritionalrequirements[35]. However,inmetagenomicstudies,theGram-negativephylum
ProteobacteriaisequallypredominantinthemicrobiomeasthephylumFirmicutes[36].Inperiodically
publishedreportsonbacteriaisolatedfromhotsprings,threeclassesoftheProteobacteria,namely
Alphaproteobacteria,BetaproteobacteriaandGammaproteobacteria,havebeendescribed,including
Hafnia,Ralstonia,Tepidimonas,SphingomonasandSilanimonas(FigureS1andTableS1;Supplementary
Materials). It appears that these non-spore-forming bacteria are also able to withstand harsh
environmentalconditionssuchasextremetemperature,pH,salinityandradioactivity. Ithasbeen
reportedthatthefollowingbacteriathriveunderextremeconditions: Ralstonia[37],Legionella[38],
Int.J.Environ.Res.PublicHealth2017,14,1070 3of18
Kocuria[39,40],Tepidimonas[41]andCronobacter[42]. ExceptforLegionellaandPseudomonas,therehave
beennoreportsassociatedwithinfectionsbytheabovementionedbacteriafromhotsprings.
Theimpactofemergingandopportunisticpathogensislargelydeterminedbythehealthand
susceptibilityofthepopulationutilisingthewaterresourcesandindevelopingcountries,thisincludes
AIDS/HIV positive individuals. Globally, in 2016 the prevalence of AIDS/HIV was 0.5% with
3.7millionpeopleinfectedoutofaworldpopulationof7.5billion[43]. ThepopulationofSouthAfrica
isvulnerabletoopportunisticinfectionsforseveralreasons. In2016inSouthAfrica,theprevalence
of HIV was 12.7% [44]. Of the 539,714 deaths recorded by the country, 27.9% were HIV-related.
Approximately10.4%ofthepopulationis<5yearsold,whileabout8%is>60yearsold. Consequently,
18%ofthepopulationfallintotheveryyoungandelderlyfractions. Malnutritionalsoweakensthe
immunesystemandisamajorconcerninthecountrywherethegrowthofoneinfivechildrenunder
fiveyearsofageisreportedlystuntedduetomalnutrition[45].
Using bacterial cultivation and culture-independent techniques, the aim of this study was to
isolate culturable emerging opportunistic bacteria from hot springs located in Limpopo Province
and to determine the role these bacteria may play on water safety and health. Findings were
discussed in the context of potential opportunistic infections in a population of individuals who
maybeimmunocompromisedforseveralreasonsandwhofinditnecessarytouselocalgroundwater
forvariousdomesticactivities. Furthermore, sinceitisestimatedthat<1%ofmicrobeshavebeen
cultured from hot springs [46], any additional information on the diversity of culturable bacteria
isolatedfromhotspringswilladdtocurrentknowledge.
2. MaterialsandMethods
2.1. SamplingSites
Most of the South African hot springs are located in the Limpopo Province. Five hot springs
(Tshipise,Siloam,Mphephu,LekkerrusandLibertas)wereselectedbasedontemperatureandease
ofsampling.
Thegeographicalandphysicochemicalcharacteristicsofthehot-waterspringshavepreviously
been reported by Olivier et al. [47]. Temperature and pH were measured in situ using the YSI
ProfessionalPlusElectrode(XylemWaterSystems,Inc.,RyeBrook,NY,USA).Dissolvedoxygen(DO)
wastoolowtobemeasured(belowthedetectionlimitoftheinstrument). Table1providestheglobal
positioningsystem(GPS)coordinatesandtheconditionsofthehot-springsamplingsitesinLimpopo
Provincethatweresampled.
Table1.Geographicallocationofsamplingsitesandsamplingconditions.
SamplingSite GPSLocation pH Temperature(◦C) Comments
Tshipise 22◦36.521(cid:48)S30◦10.345(cid:48)E 8.63 55.2 Opentoairinenclosedsection
Siloam 22◦53.667(cid:48)S30◦11.7718(cid:48)E 9 69 Exitingfrompipeonprivateproperty
Mphephu 22◦54.225(cid:48)S30◦10.83(cid:48)E 7.07 42.4 Opentoairinenclosedsection
Lekkerrus 24◦28.04(cid:48)S28◦33.1(cid:48)E 7.46 43.5 Pipelineconveyswaterintopool;
flowmanuallycontrolled
Libertas 24◦27(cid:48)36”S28◦34(cid:48)11”E 7.44 52.1 Waterpumpedatsource
GPS:globalpositioningsystem.
ExceptforSiloam,theotherhotwaterspringshaveallbeendevelopedintoresortsforhuman
recreationalpurposes. Watersamplesweretakenduringthespringseasonof2014(September2014).
From all sites, water samples were collected in sterile 1 L Schott glass bottles (DWK Life Sciences,
Mainz,Germany)andsedimentsamples(exceptforLekkerrus)insterileplastic50mLFalcontubes
(LifeTechnologies,Johannesburg,SouthAfrica)andprocessedseparately. Sedimentsampleswere
collected0.5mbelowthewaterline.Allsamplesweretakendirectlyfromthehotspringwaterwithout
anyhumancontact,exceptforsamplingatLekkerruswherethewatersampleswerecollectedfrom
Int.J.Environ.Res.PublicHealth2017,14,1070 4of18
apipethatrandirectlyintotheswimmingpool. Theflowwithinthepipewasmanuallycontrolled
dependingondemand,thatrangedfromonceadaytoonceaweek. Samplesweretransportedtothe
laboratoryinacoolerboxkeptat4◦C,andprocessedwithin72hofcollection.
2.2. IsolationofBacteria
Forisolationofbacteriafromthewatersamples,a100-mLsamplealiquotwaspassedthrougha
0.22µmmembranefilter(MilliporenitrocelluloseGSWP04700,Merck,Modderfontein,SouthAfrica)
and the membrane filters were then placed on the surface of five different agar media prepared
aspermanufacturer’sinstructions: nutrientagar,actinomyceteisolationagar,minimalLuriaagar,
cyanobacterialagarandpotatodextroseagar(HiMedia,Mumbai,India). MinimalLuriaagarwas
madeupas10%oftherecommendedconcentrationnormallyusedtocreatealownutrientmedia
simulatingthelownutrientsinthewatersamples. Atotaloffourplatespermediawereinoculated.
Thefilterswereremovedbeforetheplateswereincubatedaerobicallyfor48hat37 ◦Cand55 ◦C.
Forisolationofbacteriafromsedimentsamples,thestreakplatemethodwasused;agarplateswere
streakedusinganinoculationlooptospreadoutthesamples,andsinglecolonieswerepickedforpure
cultures. Allmediawereincubatedwithoutinoculationconcurrentlyasacontrolformediasterility
andcontamination. Allcolonieswithdistinctanddifferentmorphologyweresub-culturedatleast
threetimesuntilpurecultureswereobtainedandculturesweremaintainedonnutrientagarslants
at4◦C.Sincethiswasastudyonmicrobialdiversityandnotaquantitativeinvestigation,colonies
werepickedbasedondifferencesincolonymorphology. Theisolationconditionsfor16isolatesare
describedinTable2.
Table2.ListofActinobacteriaandProteobacteriaisolateswithisolationconditionsgiven.
IsolateNo. Site IsolationTemperature(◦C) Sample IsolationMedia
57T Tshipise 37 Water Nutrientagar
58T Tshipise 37 Water Actinomyceteisolationagar
87T Tshipise 25 Water MinimalLuriaagar
61T Tshipise 25 Water Cyanobacterialagar
72T Tshipise 37 Water Nutrientagar
80Lk Lekkerrus 37 Water Nutrientagar
79M Mphephu 37 Water Nutrientagar
44M Mphephu 53 Sediment Nutrientagar
55M Mphephu 37 Water Potatodextroseagar
37Lb Libertas 53 Water Actinomyceteisolationagar
42T Tshipise 53 Sediment Nutrientagar
59Lk Lekkerrus 37 Water MinimalLuriaagar
5T Tshipise 53 Water MinimalLuriaagar
27M Mphephu 53 Water Actinomyceteisolationagar
69Lk Lekkerrus 25 Water Cyanobacterialagar
31Lk Lekkerrus 53 Water Nutrientagar
2.3. DNAExtraction,16SrDNAGeneSequencingandPhylogeny
TheDNAwasextractedbyre-suspendingbacterialcoloniesinsterilephosphatebuffersaline,then
boiledfor10minandcentrifugedat10,000rpmfor10mininaC2500PrismAir-cooledMicrocentrifuge
(LabnetInternational,Edison,NJ,USA)[46]. Thesupernatantwasusedforpolymerasechainreaction
(PCR).Universalprimers8F,27F,300Fand1472R[48]forthecomplete1400bp16SribosomalDNA
(rDNA)genefragmentwereobtainedfromInqabaBiotechnology(Pretoria,SouthAfrica)andused
indifferentcombinationstoobtainaPCRproductfordirectsequencing. ThePCRtubepersample
contained9.5µLofwater,12.5µLofPCRMasterMix(2×)(DreamTaqGreenK1081;ThermoFisher
Scientific,Johannesburg,SouthAfrica),1µLofeachprimerat10µM,and1µLofgenomicDNAat
1–10ng. A“wateronly/noDNA”controlwasincludedineachrun,asano-templatecontrol(NTC),
toensurecontaminationhadnotoccurred. ThePCRassayswererunonaBio-RadMyCycler(Bio-Rad,
Int.J.Environ.Res.PublicHealth2017,14,1070 5of18
Rosebank,SouthAfrica). Thethermalcycleprofilewasasfollows: initialdenaturationat94◦Cfor
5min,40cyclesof94◦Cfor30s,50◦Cfor30sand72◦Cfor60s,followedby72◦Cfor10minfor
the final extension, and held at 4 ◦C until the machine was switched off. The PCR products were
run on 1% tris-acetate-ethylenediaminetetraacetic acid (TAE) agarose gels at 80–100 V for 60 min
togetherwithmolecularweightmarkers(SM1113middlemarkers;ThermoScientificWaltham,MA,
USA),stainedwithethidiumbromideandvisualisedandphotographedwiththeBio-RadGelDocEZ
Imager(Bio-Rad). ThePCRproductswerethensequencedwiththeABIBigDyeTerminatorv3.1cycle
sequencingkit(AppliedBiosystems,Foster,CA,USA)accordingtothemanufacturer’sinstructions,
and run on the ABI capillary sequencer at the African Centre for DNA Barcoding, University of
Johannesburg. If primer 8F failed, only partial sequencing was obtained. A contiguous sequence
wasconstructedwithforwardandreversesequencingdataresultinginafragmentofapproximately
1400bp,withDNABaserSequenceAssemblerv4(2013)(HeracleBioSoft,Arges,Romania).
SequenceswerecomparedwiththoseintheNCBIdatabase(GenBank)usingtheBLASTSequence
Analysis Tool [49], and EzTaxon-e [50]. Isolates with a >99% match to the published sequences
were identified to the species level, and those with a >97% match were identified to the genus
level [51]. The highest similarities (as percentage similarities) and accession numbers are given
in Table S2 (Supplementary Materials). Sequences of type strains were obtained from GenBank
andincludedinphylogeneticanalysistoconfirmidentification. AlignmentsweremadebyClustal
Omega,amultiplesequencealignmentprogramme(EuropeanBioinformaticInstitute,Cambridge,
UK). Neighbour-joining phylogenetic trees of a 947 bp fragment for Actinobacteria and a 510 bp
fragmentforProteobacteriawereconstructedusingtheSeaViewsoftwareprogram[52]. Methanogenic
bacteria(GenBankDQ372975.1),werefoundtobetheoutgroupintheProteobacteriatree,whilethe
Actinobacteriatreewasunrootedforbetterresolution. Statisticalconfidenceinbranchingpointswas
determinedbybootstrapanalysis.CompleteandpartialsequencesweresubmittedtoGenBank,andthe
accessionnumbersareindicatedinTableS2(SupplementaryMaterials). GenBankaccessionnumbers
ofthetypestrainsusedinthephylogenetictreesarelistedinTableS3(SupplementaryMaterials).
Theconsensussequenceswerecomparedtothoselistedintwodatabases,GenBankandEzTaxon-e.
The highest similarities (as percentage similarities) and accession numbers are given in Table S2
(SupplementaryMaterials). Values>95%suggestamatchingenus,whileavalueof>99%suggestsa
matchinspecies.Similaritiesof86–95%canonlybeidentifiedtoafamilylevel[51].Partialsequencesof
the16SrDNAsequenceswereusedforActinobacteria(947bp)andProteobacteria(510bp)phylogeny,
allowing alignment to the shortest fragment of the group. Actinobacteria sequences of a 947 bp
fragmentofthe16SrDNAwereanalysedasanunrootedparsimonytree(SeaView)withbootstrapping,
afteralignmentusingMultipleSequenceComparisonbyLog-Expectation(MUSCLE).
2.4. AssessmentofthePresenceofLegionellaspp. UsingReal-TimePCR
A300-mLsamplealiquotofhotspringwaterwaspassedthrougha0.22µmfilter,washedoffwith
phosphatebufferedsaline,centrifugedat10,000rpminaC2500PrismAir-cooledMicrocentrifuge,and
theDNAextractedwithZRfungal/bacterialDNAMiniPrepkit(D6005)(ZymoResearchCorp.,Irvine,
CA,USA).ThepresenceofLegionellaspp. inthewatersampleswasinvestigatedusingreal-timePCR
onaCorbettLifeScienceRotor-Gene6000Cycler(Qiagen,Hilden,Germany). PrimersusedinthePCR
assaywereJFP(5(cid:48)-AGGGTTGATAGGTTAAGAGC-3(cid:48))andJRP(5(cid:48)-CCAACAGCTAGTTGACAT
CG-3(cid:48))[53]. ThePCRassaywasruninatotalvolumeof20µLcontaining10µLof2×SensiFASTHRM
mix(Bioline,Luckenwalde,Germany),0.2µLofeachprimer(eachatafinalconcentrationof0.2µM),
5.6µLofnuclease-freewaterand4µLoftemplateDNA.TemplateDNAfromL.pneumophilaATCC
211-33-2(AmericanTypeCultureCollection(ATCC),Manassas,VA,USA)wasusedaspositivecontrol.
To test the detection limit of the PCR reaction, the DNA concentration of the positive control was
determinedusingaNanoDropHach6200Spectrophotometer(Hach,Loveland,CO,USA),andthen
seriallydilutedupto10−6. Thesamplesandthepositivecontrolwererunintriplicate.
Int.J.Environ.Res.PublicHealth2017,14,1070 6of18
ReactionconditionsforthePCRwereoptimisedasfollows: initialactivationat95◦Cfor10min;
denaturationat95◦Cfor5s;annealingat57◦Cfor5s;andafinalextensionat72◦Cfor5s,fora
totalof40cycles. Thelastcyclewasfollowedbyasecondincubationperiodof5minat72◦C.The
secondincubationwasfollowedbyameltcurvepreparedbyrampingupthemeltingtemperature
from72◦Cto95◦Cataramprateof1◦Cateachstep,apre-meltholdof90sonthe1ststepfollowed
bya5-sholdoneachofthenextsteps. ARotor-Gene6000Cycler(Qiagen)wasusedtorunthePCR
assays. ReactionmixtureswithouttemplateDNAwereusedasNTCsineachreaction. Melting-curve
analyseswereperformedusingtheRotor-GeneReal-TimeAnalysisSoftware,Version6.1(Build93)
(CorbettLifeScience,Sydney,Australia).
2.5. AntibioticResistanceAssay
Thefollowing10antibiotics(obtainedfromMastDiagnostics,Merseyside,UK,ifnotindicated
otherwise)wereusedinthisstudy,representingsixdifferentclasses: β-lactam(carbenicillin,CAR),
aminoglycosides (gentamycin, GEN; kanamycin, KAN; streptomycin, STR), tetracycline (TET),
amphenicols(chloramphenicol,CHL;ceftriaxone,CEF),sulphonamides(co-trimoxazole,COT)and
quinolones(nalidixicacid,NA;norfloxacin,NOR),andassayedusingtheKirby-Bauerdiskdiffusion
assay [54] according to the criteria published by the Clinical Laboratory Standards Institute [55].
CommercialantibioticdiscsfromOxoid(ThermoFisherScientific)wereusedatthefollowingpotencies
in µg: gentamycin 10, tetracycline 30, co-trimoxazole 25, chloramphenicol 30, ceftriaxone 30 and
norfloxacin 10. Solutions of carbenicillin (100 µg), kanamycin (30 µg), streptomycin (10 µg) and
nalidixic acid (125 µg) were prepared from stock solutions (Sigma-Aldrich, Johannesburg, South
Africa)anddriedonsterileWhatmanNo17discs(ThermoFisherScientific)beforeplacementona
lawnofbacteria. Theinoculum(100µL)waspreparedfromanovernightcultureinnutrientbroth.
ThediscswereappliedtolawnsofbacteriaonMueller-Hintonagar(HiMedia,http://doua.prabi.fr/
software/seaview). Plateswerescoredafter24hat53 ◦Cor37 ◦C,dependingontheisolate. The
zonesofinhibitionweremeasuredandcomparedtotheCLSIstandardvalues. Themultipleantibiotic
resistance(MAR)indexwascalculatedastheratio(a:b)betweenthenumberofantibioticstowhich
theisolatewasresistant(a)andthetotalnumberofantibioticstested(b)[56]. MARwasdefinedasthe
resistanceofisolatestoatleastthreedifferentantibiotics.
2.6. GenBankAccessionNumbers
The16SrDNAgenesequencesofhotspringisolatesfromSouthAfricawereallocatedaccession
numbersanddepositedinGenbankasindicatedinTableS2(SupplementaryMaterials). Strainsof
Actinobacteria, Kocuria sp. and Arthrobacter sp. have accession numbers MF120234 and MF120235,
respectively. TwoalphaProteobacteriawereassignednumbersMF120236andMF120239. Accession
numbers: MF120227–MF120233andMF120237weregiventobetaProteobacteriaincludingGulbenkiania
mobilis,Cupriavidusgilardii,TepidimonasfonticaldiandRalstoniamannitolilytica. Fourstrainsofgamma
ProteobacteriaincludingHafniaalveihaveGenbankaccessionnumbersofMF144571–MF144573and
MF120238. AllreferencestrainsusedforanalysesarelistedinTableS3(SupplementaryMaterials)
togetherwiththeirassociatedGenbankaccessionnumbers.
3. Results
3.1. IsolationRatesofBacteria
Themajorityofbacteriawereisolatedfromwatersamples,althoughtwoisolates,namely42Tand
44M,werefromthesediment. Bacteriawerenotisolatedfromthewarmesthotspring,Siloam,at69◦C,
whilethesecondwarmestsite,Tshipise,at55◦C,hadthehighestnumberofisolates(Figure1).
Furthermore, the coldest water temperature at Mphephu at 42.7 ◦C did not yield the highest
numberofisolates. Bycomparingthedifferentmediausedforisolation,nutrientagarperformedthe
Int.J.Environ.Res.PublicHealth2017,14,1070 7of18
best,resultinginthehighestnumberofisolates,andpotatodextroseagar,whichisgenerallyusedfor
theisolationoffungi,wasthepoorestperformer(Figure2).
However,thisisonlyanobservationlimitedtothisspecificsamplingregimen. Otherisolation
patternsonmediamightbeobservedindifferentsamplingprogrammes. Furthermore,asBacillus
spp. andrelatedspecieswereisolatedfromthesameplates,isolatescouldhavecompetedforspace
andnutrients.Int. J. Environ. Res. Public Health 2017, 14, 1070 7 of 18
Int. J. Environ. Res. Public Health 2017, 14, 1070 7 of 18
8
atesolates678567
umber of isolNumber of is01234501234
N
Hot spring sample sites
Hot spring sample sites

Figure1.Distributionofthenumberofisolatesfromfivehotsprings,LimpopoProvince,SouthAfrica.
Figure 1. Distribution of the number of isolates from five hot springs, Limpopo Province, South
Figure 1. Distribution of the number of isolates from five hot springs, Limpopo Province, South
TemperatureaAnfdricpa. HTemopfewratautree ransda pmH pofl ewsataerr seamgpivlees narei ngivpena rine pnatrhenetsheesse.s.
Africa. Temperature and pH of water samples are given in parentheses.
8
8
ber of isolatesmber of isolates34567234567
mu2
uN1
N1
0
0
Media used for isolation
Media used for isolation

Figure2.CFiogumrep 2.a Croismopnariosofnn ouf nmumbbeerr ooff isioslaotleas toebtsaionebdt aonin deifdfereonnt mdeidfifae. rentmedia.
Figure 2. Comparison of number of isolates obtained on different media.
3.2. 16S rDNA Gene Sequencing
3.2. 16S rDNA Gene Sequencing
3.2. 16SrDNAGeneSequencing
As indicated in the Supplementary Materials (Tables S1–S3), isolates of Sphingomonas echinoides,
As indicated in the Supplementary Materials (Tables S1–S3), isolates of Sphingomonas echinoides,
Hafnia alvei, Tepidimonas fonticaldi, Gulbenkiania mobilis, Ralstonia mannitolilytica, Cupriavidus gilardii,
AsindicaHKteoacfdnuiraii ana ltvuterhif,a eTneeSpnisudisipm aopnndlae s Amforntehtirncoabtladacitr, eyGr ulMulbteenoalkutieasn rwiiaa emrleso bii(dlTiesn,a tRbifailleesdtso ntoSia 1 tmh–eaS nsn3pi)teo,cliiielsyst oilcelava, etCle,u swprohiaifvleidS iuspso hlgaiitlnearsgd ooiif,m onasechinoides,
Kocuria turfanensis and Arthrobacter luteolus were identified to the species level, while isolates of
Hafniaalvei,TeCpriodnoimbacotenr,a Ssilfaonnimtoincaas,l dTie,piGdimuolnbaesn aknida nZioaoglmoeao bwielirse ,idReantlisfiteodn oinalym tao nthnei tgoelniulys tliecvael,. CFouupr riavidusgilardii,
Cronobacter, Silanimonas, Tepidimonas and Zoogloea were identified only to the genus level. Four
isolates (55M, 59Lk, 61T and 72T) were unknown as they presented poor matches to the published
Kocuria turfaneisnolsaitses a(5n5Md, A59rLtkh, 6r1oTb aancdte 7r2Tl)u wteeroel uunsknwowenre as itdheeyn ptriefiseendtedt opootrh meatscphees ctoie tshe lpeuvbelislh,ewd hile isolates of
sequences. Sequences were submitted to GenBank (List S1; Supplementary Materials), and their
sequences. Sequences were submitted to GenBank (List S1; Supplementary Materials), and their
Cronobacter,Silraenleivmanotn acacse,ssTioenp induimmboernsa asrea linstdedZ ino tohge lSouepapwlemeernetairdy eMnatteirfiiaelsd. onlytothegenuslevel. Fourisolates
relevant accession numbers are listed in the Supplementary Materials.
(55M,59Lk,61Tand72T)wereunknownastheypresentedpoormatchestothepublishedsequences.
3.3. Phylogenetic Analysis
3.3. Phylogenetic Analysis
SequencesweresubmittedtoGenBank(ListS1;SupplementaryMaterials),andtheirrelevantaccession
Two isolates identified in this investigation, namely 57T (pink pigment) and 58T (yellow
Two isolates identified in this investigation, namely 57T (pink pigment) and 58T (yellow
numbersarelipsitgemdenitn), gthroeupSeud pwpithle Km. tuernfatnaenrsyis Manda Ate. rluitaeolslu.s, respectively (Figure 3), as predicted from the
pigment), grouped with K. turfanensis and A. luteolus, respectively (Figure 3), as predicted from the
BLAST results with GenBank and EzTaxon-e, given in the Table S2 (Supplementary Materials).
BLAST results with GenBank and EzTaxon-e, given in the Table S2 (Supplementary Materials).
Reference strains are included in the analysis (accession numbers listed in Table S3; Supplementary
3.3. PhylogenetRicefAerennacley sstriasins are included in the analysis (accession numbers listed in Table S3; Supplementary
Materials), and pathogenic isolates are Arthrobacter woluwensis, Arthrobacter cumminsii, Arthrobacter
Materials), and pathogenic isolates are Arthrobacter woluwensis, Arthrobacter cumminsii, Arthrobacter
mysorens, Kocuria rhizophila, K. marina, K. varians, K. kristinae and K. rosea [18]. Three unknown
Twoisolatmeyssoirdenes,n Ktioficuerida rihnizotphhiisla,i nKv. mesartiinga,a Kti.o vnar,iannas, mK.e klryist5in7aTe a(npdi nKk. rposiega m[18e]n. Tt)hraene dun5k8noTw(ny ellowpigment),
sequences from hot-spring isolates identified in other studies, and listed in the GenBank database,
sequences from hot-spring isolates identified in other studies, and listed in the GenBank database,
grouped withwKer.e tiuncrlufadnede.n Asritshroabnacdter ANC.CluP tfreoomlu hso,t srpersinpgse icnt iPvakeilsytan( gFriogupuerde w3it)h, thaes papthroegdenicict egrdoufpr om the BLAST
were included. Arthrobacter NCCP from hot springs in Pakistan grouped with the pathogenic group
of A. woluwensis. Uncultured Arthrobacter clone BR5clone TPB_GMAT_5_1 grouped with the cultured
resultswithGoef nAB. waonlukwenasnis.d UnEczulTtuarxedo Anr-ther,obgacitvere cnlonien BRth5celonTea TbPlBe_GSM2A(TS_u5_p1 gprloeumpede nwtitahr thye cMultautreedr ials). Reference
strainsareincludedintheanalysis(accessionnumberslistedinTableS3;SupplementaryMaterials),

and pathogen ic isolates are Arthrobacter woluwensis, Arthrobacter cumminsii, Arthrobacter mysorens,
Kocuriarhizophila,K.marina,K.varians,K.kristinaeandK.rosea[18]. Threeunknownsequencesfrom
hot-spring isolates identified in other studies, and listed in the GenBank database, were included.
ArthrobacterNCCPfromhotspringsinPakistangroupedwiththepathogenicgroupofA.woluwensis.
UnculturedArthrobactercloneBR5cloneTPB_GMAT_5_1groupedwiththeculturedArthrobacterGM37
isolatefromhotspringsinIndia,suggestingthattheycouldbecloselyrelated. KocuriaB38,anisolate
fromhotspringsinIndia,groupedwithK.flava.
Int. J. Environ. Res. Public Health 2017, 14, 1070 8 of 18
Arthrobacter GM37 isolate from hot springs in India, suggesting that they could be closely related.
Int.J.Environ.Res.PublicHealth2017,14,1070 8of18
Kocuria B38, an isolate from hot springs in India, grouped with K. flava.

FiFgiugruer e3.3 .UUnrnorootoetde dpaprasrimsimonoyn ytretree efofro rAActcintionboabcatcetreiari ashsohwowinign gthteh eplpalcaecmemenetn otfo ifsiosloaltaet e575T7 Twwithit h
KKocoucruirai aanadn disiosloaltaet e585T8T wwitiht hAArtrhtrhorboabcatcetre rwwitihth bbooootsttsrtarpap vvalauluese.s .
Following MUSCLE alignment (SeaView), a parsimony tree of the 16S rDNA gene sequences of
FollowingMUSCLEalignment(SeaView),aparsimonytreeofthe16SrDNAgenesequencesofthe
the Proteobacteria isolates was drawn using a partial sequence of 510 bp, with bootstrapping. A
Proteobacteriaisolateswasdrawnusingapartialsequenceof510bp,withbootstrapping.Amethanogenic
methanogenic archaeon (GenBank accession No. DQ372975.1) was the outgroup used. The results of
archaeon(GenBankaccessionNo.DQ372975.1)wastheoutgroupused.Theresultsofthisinvestigation
this investigation showed that three classes of the Proteobacteria were represented, namely
showedthatthreeclassesoftheProteobacteriawererepresented,namelyAlphaproteobacteria(n=2),
Alphaproteobacteria (n = 2), Betaproteobacteria (n = 8), and Gammaproteobacteria (n = 3). Reference
Betaproteobacteria (n = 8), and Gammaproteobacteria (n = 3). Reference type strains were included
type strains were included in the analysis (Table S3; Supplementary Materials) and the opportunistic
intheanalysis(TableS3;SupplementaryMaterials)andtheopportunisticpathogensareSphingomonas
pathogens are Sphingomonas paucimobilis, Cronobacter sakazakii, Hafnia alvei and H. paralvei, Tepidimonas
paucimobilis,Cronobactersakazakii,HafniaalveiandH.paralvei,Tepidimonasarfidensis,Ralstoniamannitolilytica
arfidensis, Ralstonia mannitolilytica and Ralstonia pickettii, Cupriavidus metallidurans and C. gilardii.
andRalstoniapickettii,CupriavidusmetalliduransandC.gilardii.
From published sequences in GenBank, four isolates that have not been cultured but sequenced
FrompublishedsequencesinGenBank,fourisolatesthathavenotbeenculturedbutsequenced
from hot springs in other studies were included in the phylogenetic tree to establish whether they
fromhotspringsinotherstudieswereincludedinthephylogenetictreetoestablishwhetherthey
were similar to isolates obtained in this study, or to determine whether their identity could be further
were similar to isolates obtained in this study, or to determine whether their identity could be
characterised.
furthercharacterised.
3.4. Detection of Legionella spp.
3.4. DetectionofLegionellaspp.
A total of four water samples were analysed in triplicate. All samples analysed were negative
Atotaloffourwatersampleswereanalysedintriplicate. Allsamplesanalysedwerenegative
for the 16S rDNA gene amplified. A real-time PCR assay for Legionella is given in Figure S2A
for the 16S rDNA gene amplified. A real-time PCR assay for Legionella is given in Figure S2A
(Supplementary Materials) while Figure S2B (Supplementary Materials) shows the high-resolution
(SupplementaryMaterials)whileFigureS2B(SupplementaryMaterials)showsthehigh-resolution
melt (HRM) curve analysis of the PCR reaction. The mean melting temperature of the positive
melt(HRM)curveanalysisofthePCRreaction. Themeanmeltingtemperatureofthepositivesamples
sawmapsl8e6s. 1w±as 08.61.1◦C ±. 0.1 °C.
3.35..5 A. Antnibtiiboitoicti cRReseissitsatnacnec e
IsIosloaltaetse sKKocoucruirai,a ,AArtrhtrhorboabcatcetre ranandd HHafanfniai asspppp. .wewree rererseisstiasntat nttot otwtwo oanatnibtiiboitoictsic sini nddififfefreernent t
cocmombibniantaitoinons sofo fcecfetfrtiraiaxxoonnee,, nnaalliiddiixxiicc aacciidd aanndd ccaarrbbeenniiccilillilnin,,a assi ninddiciactaetdedin inT aTbalbel3e. 3Is. oIslaotleatCer Conroobnaocbtaerctseprp .
spwpa. swsaesn sseitnivsietivtoe atoll atlel nteann atnibtiiboitoictsicws whihleileth theeu unnkknnoowwnni issoollaattee ((iissoollaattee 7722TT)) bbeelloonnggiinngg toto ththe efafmamilyil y
EEntnetreorboabcatcetreiraicaecaeea,e w,wasa salasloso rerseissitsatnant ttoto twtwoo aanntitbibioiotitcisc,s ,nnaammeelyly chchlolorarammpphheennicicool laanndd cceeftfrtiraiaxxoonnee. .
Int.J.Environ.Res.PublicHealth2017,14,1070 9of18
Table3.AntibioticresistanceprofilesofActinobacteriaandProteobacteriaisolatesagainsttendifferentantibiotics.
IsolateNo. Identification CAR100 GEN10 KAN30 STR10 TET30 CEF30 CHL30 COT25 NA30 NOR10
57T Kocuriaturfanensis 15 9 12 7 12 0 14 16 0 2
58T Arthrobacterluteolus 0 5 3 4 9 1 6 11 0 2
79M Hafniaalvei 0 2 3 1 4 0 5 6 3 7
80Lk Cronobactersp. 1 4 5 5 6 2 6 6 3 11
72T unknownEnterobacteriaceae 5 6 10 5 7 0 0 16 5 14
CAR:Carbenicillin;GEN:gentamicin;KAN:kanamycin;STR:streptomycin;TET:tetracycline;CHL:chloramphenicol;CEF:ceftriaxone;COT:co-trimoxazole;NA:nalidixicacid;
NOR:norfloxacin;allinµg.Valuesof0indicateresistancewhilenumericalvaluesarezonesofinhibitioninmillimetres.
Int.J.Environ.Res.PublicHealth2017,14,1070 10of18
4. Discussion
PhylumProteobacteria(mostlybetaandgammaProteobacteria)dominatesthemicrobiomeof
TshipiseandMphephu(upto30%),whilethephylumActinobacteriawasfoundtooccuratlessthan
1%[54]. However,metagenomicstudiesreportdiversityasproportionsofdifferentDNAsequences
withoutanyinformationontheviabilityofcells. Infact,Gram-positivespore-formingBacillusofthe
phylumFirmicutesarethemostpredominantbacteriaisolatedfromhotspringsglobally[32,33,35],
although there are sporadic reports of isolation by culture of Pseudomonas spp. of the phylum
ProteobacteriainAlgeria[57]andIceland[58].
4.1. PhylumActinobacteria
Fromsedimentsamples,twopigmentedmesophilicisolates,namely57Tand58T,wereidentified
tobeKocuriaturfanensisandArthrobacterluteus,respectively,usingthesearchtool,BLASTforGenBank
and EzTaxon-e (Table S2; Supplementary Materials) and confirmed using phylogenetic analysis
(Figure3).
Kocuriaspp. ASB107,closelyrelatedtoK.polarisandK.rosea,whichwasisolatedfromIranian
Ab-e-Siahhotsprings,showedresistancetogammaandultraviolet(UV)rayswithatemperature
growthrangeof0to37 ◦C[39]. AnotherisolateofK.roseaMG2fromthesameIranianhotspring
optimallygrewatpH9. 2andatatemperatureof28◦C.Itcouldalsotolerate15%NaClsalinity,high
doses of radioactivity and grew in the presence of 4% hydrogen peroxide [40], providing further
proof that members of this genus are capable of multiple-extreme resistance. An unpublished
Kocuriaspp.B38wasisolatedfromBakreshwarhotspringsinIndiaandwasincludedinthephylogeny
tree to establish whether there was any similarity between that one and the one isolated during
thisinvestigation. KocuriaB38groupedwithKocuriaflava,anairborneActinobacteria[59]GenBank
KC492107,withabootstrapvalueof90%,suggestingthatitispossibletoculturecloselyrelatedKocuria
isolatesthatwerepreviouslyreportedunculturedfromhotspringsatothersites.PathogenicKocuriasp.
areK.rhizophila,K.marina,K.rosealistedinreviews[18,60],andK.kristinae[61,62]andK.varians[63].
All,exceptforK.rosea,fellintoaseparategroup. EventhoughneitherK.turfanensisnorK.flavahas
beenreportedasopportunisticpathogens,itdoesnotexcludethepossibilitythattheyalsohavethe
abilitytoacquirevirulencefactors.
Similarly, radio-resistant Arthrobacter sp. has been isolated from hot springs in Japan [64]
and Tibet [65]. Three 16S rDNA gene sequences from Arthrobacter sp. isolated from hot springs
wereincluded inthephylogeneticanalysis(Figure3). Interestingly, Arthrobacter NCCP1348from
Pakistan(GenBankAccessionLC065375)wasrelatedtopathogenicA.woluwensis,whileArthrobactersp.
GM37AC3K2stronglygroupedwithunculturedArthrobactersp. cloneTPB_GMAT_AC3. Thislatter
observationwasnotsurprisingsincebothcamefromthesameinvestigationofhotspringsinIndia
(Table S3; Supplementary Materials). Three species of pathogenic Arthrobacter were included in
the phylogenetic analysis and A. woluwensis, and A. mysorens were closely related. Interestingly,
A.cumminsiigroupedclosedwiththepathogenicKocuriaspecies. Aliteraturesearchdidnotreveal
A.luteolusasanopportunisticandemergingpathogen; however,thetypestrainofA.luteoluswas
isolatedoriginallyfromhumanclinicalspecimens[66],andthereforethisdoesnotentirelyexclude
thisspeciesfrombeingapotentialopportunisticpathogen.
4.2. PhylumProteobacteria
Fourteen Proteobacteria were represented by three classes: Alphaproteobacteria (n = 2),
Betaproteobacteria (n = 8) and Gammaproteobacteria (n = 4) (Table S2; Supplementary Materials).
Ofthese,fourisolatescouldnotbeclassifieduptogenuslevelastheywere<97%similartoGenBank
databaseentries. Fourcouldbeidentifiedtothegenuslevel,andsixidentifiedtothespecieslevel.