ComparativeGenomicsandTranscriptomicsofPropionibacteriumacnesElzbieta
Brzuszkiewicz1, January Weiner2, Antje Wollherr1, Andrea Thu rmer1,
Jennifer Hu peden1, Hans
B.Lomholt3,MogensKilian3,GerhardGottschalk1,RolfDaniel1,Hans-JoachimMollenkopf4,ThomasF.Meyer5,HolgerBru
ggemann5*1Go ttingenGenomicsLaboratory, Instituteof
MicrobiologyandGenetics, Georg-August-UniversityofGo ttingen, Go
ttingen, Germany, 2Departmentof Immunology,Max Planck Institute for
Infection Biology, Berlin, Germany, 3Department of Medical
Microbiology and Immunology, Aarhus University, Aarhus, Denmark,
4Core FacilityMicroarray, MaxPlanckInstituteforInfectionBiology,
Berlin, Germany, 5DepartmentofMolecularBiology,
MaxPlanckInstituteforInfectionBiology, Berlin, GermanyAbstractThe
anaerobic Gram-positive bacterium Propionibacterium acnes is a
human skin commensal that is occasionally associatedwith
inflammatory diseases. Recent work has indicated that evolutionary
distinct lineages of P. acnes play etiologic roles
indiseasewhileothers areassociatedwithmaintenanceof
skinhomeostasis. Toshedlight onthemolecular basis fordifferential
strain properties, we carried out genomic and transcriptomic
analysis of distinct P. acnes strains. We sequencedthe genome of
the P. acnes strain 266, a type I-1a strain. Comparative genome
analysis of strain 266 and four other P. acnesstrains revealed that
overall genome plasticity is relatively low; however, a number of
island-like genomic regions, encodingavarietyof
putativevirulence-associatedandfitnesstraitsdifferbetweenphylotypes,
asjudgedfromPCRanalysisof acollection of P. acnes strains.
Comparative transcriptome analysis of strains KPA171202 (type I-2)
and 266 during exponentialgrowthrevealedinter-straindifferences
ingeneexpressionof transport systems andmetabolicpathways.
Inaddition,transcript levels of genes encoding possible virulence
factors such as dermatan-sulphate adhesin, polyunsaturated fatty
acidisomerase, iron acquisition protein HtaA and lipase GehA were
upregulated in strain 266. We investigated differential
geneexpressionduringexponential andstationary growthphases. Genes
encodingcomponents of theenergy-conservingrespiratory chain as well
as secreted and virulence-associated factors were transcribed
during the exponential phase,
whilethestationarygrowthphasewascharacterizedbyupregulationof genes
involvedinstressresponses andaminoacidmetabolism. Our data
highlight the genomic basis for strain diversity and identify, for
the first time, the actively transcribedpart of the genome,
underlining the important role growth status plays in the
inflammation-inducing activity of P. acnes. Wearguethat
thedisease-causingpotential of different P. acnes strains is not
onlydeterminedbythephylotype-specificgenomecontentbutalsobyvariablegeneexpression.Citation:
Brzuszkiewicz E, Weiner J, Wollherr A, Thu rmer A, Hu peden J, et
al. (2011) Comparative Genomics and Transcriptomics of
Propionibacterium acnes. PLoSONE6(6): e21581.
doi:10.1371/journal.pone.0021581Editor:MalcolmJamesHorsburgh,
UniversityofLiverpool, UnitedKingdomReceivedFebruary8, 2011;
AcceptedJune3, 2011; PublishedJune27, 2011Copyright:
2011Brzuszkiewicz et al.This isanopen-accessarticle
distributedunder the terms ofthe
CreativeCommonsAttributionLicense,whichpermitsunrestricteduse,
distribution, andreproductioninanymedium, providedtheoriginal
authorandsourcearecredited.Funding:
FundingwasprovidedbytheMaxPlanckSociety.
Thefundershadnoroleinstudydesign, datacollectionandanalysis,
decisiontopublish, orpreparationofthemanuscript.Competing
Interests: The authors have read the journals policy and have the
following conflict: Holger Bru ggemann is an academic editor of
PLoS ONE. Thisdoesnotaltertheauthors adherencetoall
thePLoSONEpoliciesonsharingdataandmaterials.*E-mail:
brueggemann@mpiib-berlin.mpg.deIntroductionTheGram-positivebacteriumPropionibacteriumacnesisconsideredas
a skin commensal, which preferentially resides in
sebaceousfollicles. The bacteriums presence on the human skin is
suggested tobe beneficial, for instance due to the ability to lower
the skin pH byacidic fermentation products, thereby protecting the
follicles againstcolonizationbyharmful pathogens [1]. However,
several lines ofevidence indicate that, under certain conditions,
P. acnes can act as anopportunisticpathogen[1,2]. Theinvolvement of
P. acnes intheformationandseverityof acnevulgaris is
widelyaccepted, albeitprecisemechanisticinsight remainsscarce.
Moreover, P. acnes hasbeen detected in various opportunistic
infections such as endocarditisandosteomyelitis,
andinseverepost-surgical infections, e.g. afterimplantation of
prosthetic heart valves and shunts [3,4]. Initially, thebacteriums
presence was considered as contamination or secondaryinvasion;
morerecently, however, thereisgrowingawarenessthatP. acnes could be
an etiological agent of at least some of these diseases[4]. This
assumption has been bolstered by immunologicalobservationsinvitro,
demonstratingthatP.acnespossessesextensiveimmunostimulatory
activity that triggers the secretion of proin-flammatory cytokines
and chemokines, the activation of thecomplement system and the
stimulation ofT-cells [5,6,7,8]. P. acnesstrains belonging to
different phylotypes are reported to differ in
theirimmunostimulatory activity; for instance, strains vary in
their abilityto trigger human-beta-defensin 2 in keratinocytes, and
in their effectondifferentiationandviabilityof
sebocytes[9,10,11,12]. Todate,bacterial traits and factors
responsible for triggering and modulatinghost cell
responseshavenotbeenuncovered.P. acnesstrainswere
categorizedasphylotypes IA,IB,II
andIIIaccordingtosequencecomparisonoftheirtlyandrecAgenes[13].Morerecently,
amultilocus sequencetyping(MLST)
approach,basedonninehousekeepinggenes[14],
hasbeenusedtofurtherdiscriminatestrains,
resultingintheidentificationof 57sequencetypes(ST)
from210strainsanalyzed. Again, threedivisionswereidentified (I, II
and III); division I was further subdivided into I-1a,PLoSONE |
www.plosone.org 1 June2011 | Volume6 | Issue 6| e21581I-1b and I-2.
Subdivision I-1a comprised significantly moreisolates associated
with moderate to severe acne, while strainsfromother (sub)divisions
wereisolatedmoreoftenfromhealthyskinor opportunisticsoft
tissueinfections.
SubdivisionI-1aalsoincludedtheepidemiccloneST18andits descendents;
interest-ingly, 60%of all ST18 strains have been isolated
fromacnepatients, suggesting that ST18 strains possess a particular
virulencepotential [14].Due to the relative scarcity of available
P. acnes genome sequences,thegeneticbasisfortheheterogeneityof P.
acneshasnot yet beenstudied on the genomic level. Here we sequenced
strain 266, a type I-1a strainbelonging toST18, andusedits genome
sequence forcomparativegenomicanalyses. Weshowthat
themaindifferencesbetweengenomesofdifferentP.acnesphylotypesarelocatedwithinfour
large genomic regions with island-like characteristics, as well as
afewsmallergenomicregions.Moreover,
wenotedsubtledifferencesgenerated by point mutations and
potentially also by phase variation,affecting in particular the
expression of adhesins. We also carried outcomparative
transcriptomic analysis of two P. acnes strains andmonitored growth
phase-dependent transcription. Our data providesinsight in the
metabolic pathways utilized by P. acnes and suggests thatdifferent
strains of P. acnes employ distinct energy-conservingstrategies;
moreover, strain and growth phase differences in theexpression
ofvirulence-associatedtraitswere
uncovered.ResultsandDiscussionPhylotypingofsequencedP.
acnesandgeneral featuresofthegenomesCurrently (January 2011), three
P. acnes genomes have beencompletelysequenced(Table1):
strainKPA171202(KPA) [15],strainSK137andthehumanisolate266.
Thelatter strainwasisolated froma pleuropulmonary infection.
According to therecentlyestablishedMLSTscheme[14], strainKPAbelongs
todivision I-2 (ST34), whereas SK137 and 266 are both I-1a
strains,albeitbeingdifferentSTs(Table1).Strain266isanST18strain,which
represents an evolutionary lineage comprising
isolatesfrequentlyassociatedwithmildtosevereacne.AdditionalP.acnesgenomes
are in the gap closure/finishing phase, including
thestrainsJ139(II/new ST),J165 (I-1a/ST18)and SK187(I-1a/newST),
whichalongsidestrainSK137(I-1a/newST)
and67otherP.acnesstrains(draftassembly)serveasreferencegenomesfortheHumanMicrobiomeProject
[16]. TheP. acnes genomes haveasimilar GC content (60%) and size,
on average 2,509kb, with littlevariation (standard deviation:
30.2kb). The genome of P.freudenreichii is 105kb larger than the
average P. acnes genomeandhasanelevatedGCcontent (67%); however,
thenumberofcodingsequences(CDS) issimilartoP.
acnes[17].Comparativegenomics:
strain-specificgenecontentandgenomicregionswithisland-likefeaturesBi-Blast
anddirect
genomecomparisonsidentifiedstrain-specificregionsinthegenomesofKPA,266,SK137,SK187andJ139.AllsequencedP.
acnes isolates share alarge core genome (Figure 1).According toa
Bi-Blast analysis, 711%of the CDSinthe fivegenomes are
strain-specific. Comparative analysis identified four
largegenomicregions withintheKPAgenome, whichareabsent
fromsomeoralloftheotherP.
acnesgenomes(Figure1A).Theseregionswerepredictedasgenomicislands,
possiblyacquiredbyhorizontaltransfer. In general, most, but not
all, predicted islands were specific toa certain P. acnes strain
orphylotype (Figure S1, Table S1).The first region in the KPAgenome
comprises the genesPPA0846-PPA0876, which encode many hypothetical
proteinsandproteins withsimilarities to those
involvedinbacteriocin/lanthioninebiosynthesis.
Thisregionisalsopresent inSK187(I-1a), but absent fromthe other P.
acnes genomes. Theregionisflankedbymobilegeneticelements:
the59endiscomposedof
atra-likeregionreminiscentofconjugativeplasmids,andthe39 endby a
transposase-encoding gene. This underlines its possiblehorizontal
acquisition as predicted by bioinformatics analysis(FigureS1).
Several CDS(PPA0859-PPA0865) withinthisislandwere identified in a
bioinformatics search for genes involved in thebiosynthesisof
thiazolylpeptides[18]. Thiopeptideantibioticsarepotent inhibitors
of proteinsynthesis inGram-positive bacteria.They are synthesized
from a precursor peptide that is subsequentlytransformed by
posttranslational modifications into the finalmacrocyclic
structure. Two precursor peptides are encoded
intheP.acnesgenome,locatedup-anddownstreamofthepredictedPPA0866
gene; one precursor possesses the 15-aa structuralpeptide of
berninamycin fromStreptomyces bernensis [18].
TheupstreamCDSexhibitsimilaritiestothebiosynthesispathwaysofnosiheptide
of Streptomyces actuosus (encoded by nos genes) [19],siomycin A of
Streptomyces sioyaensis (sio genes), nocathiacin ofNocardia sp.
(noc genes) and in particular TP-1161 of Nocardiopsis
sp.TFS65-07(tpagenes) [20];e.g.
PPA0859toTpaK/NocE/NosE/SioJ(2435%proteinidentity);PPA0860toTpaL/NocD/NosD/SioK(3133%identity)
and PPA0862 to
TpaC/NocG/NosG(2947%identity).Table1.Genomefeaturesofpropionibacteria.Strain
PhylotypeaLength GCContent CDSDNAcodingsequence GenBankP. acnes266
I-1a/ST18(IA) 2,494,578 60% 2,412 90% thisworkP. acnesKPA171202
I-2/ST34(IB) 2,560,265 60% 2,297 89% AE017283.1[15]P. acnesSK137
I-1a/newST(IA) 2,495,334 60% 2,352 88% CP001977.1P.
acnesSK187bI-1a/newST(IB) 2,510,934 59% 2,381 88% ADJM00000000P.
acnesJ165bI-1a/ST18(IA) 2,500,083 60% 2,403 88% ADJL00000000P.
acnesJ139bII/newST(II) 2,481,963 60% 2,364 88% ADFS00000000P.
freuden-reichiissp.ShermaniiCIRM-BIA1- 2,616,384 67% 2,375 86%
FN806773.1[17]amultilocussequenceanalysisaccordingto[14].
Inbrackets: typingaccordingtopreviousmethod[13].
ItisnoteworthythatthesubgroupIBdoesnotconstituteahomogeneousgroup;
thus,
strainSK187isnowallocatedtosubdivisionI-1a.bthesegenomesarecurrentlynotclosed:
J165, 62contigs; SK187, 37contigs; J139,
7contigs(January2011).doi:10.1371/journal.pone.0021581.t001GenomicsofPropionibacteriumacnesPLoSONE
| www.plosone.org 2 June2011 | Volume6 | Issue 6|
e21581GenomicsofPropionibacteriumacnesPLoSONE | www.plosone.org 3
June2011 | Volume6 | Issue 6|
e21581Thesecondregion(PPA1278-PPA1304) inthe KPAgenomeencodes a
gene cluster (PPA1284-PPA1292) with homology tonon-ribosomal
peptidesynthetases(NRPS) andgenesinvolvedinthe synthesis of cyclic
lipopeptides such as surfactin. This cluster isabsentinSK137(I-1a)
butpresentintheotherI-1astrains(266,SK187). Thus, this cluster
could serve as a marker to
furthersubdivideI-1astrains.Thereisalsoanindicationthatthisregionwas
acquired by horizontal transfer: the 59 end contains
PPA1280,whichencodesaParAdomainproteinwithsimilaritiestoknownplasmidpartitioningproteins,
indicatingthatthisgenomicregionmightbeacquiredviaplasmidinsertion(FigureS1).Thethirdgenomicregion(PPA1579-PPA1613),
specifictoKPA,encodes proteins with homologies to phage proteins,
thus is likely to bea cryptic phage. The phage-related genes
exhibit no homology to
thesequencedP.acnesbacteriophagePA6[21].Itisinsertedintoageneencoding
a type II restriction enzyme (PAZ_c16690 in strain 266);
theremaining gene fragments flank the island (PPA1578,
PPA1614).Several
genesofthefourthregion(PPA2055-PPA2092)arefoundonlyinstrainKPA.
Thegenes encodetwotransport systems, onepossiblyspecificfor
dipeptides andtheother for cobalt, andgenesencoding
poly-/oligosaccharide-degrading enzymes such as a
putativeb-glucanase(PPA2056), aputativea-L-fucosidase(PPA2070),
andaputative a-galactosidase (PPA2057).APCRanalysis was carriedout
toinvestigatehowthesefourlarge genomic regions are distributed in a
variety of strains (Table 2).It became apparent that the
presenceand absence of these regionsare consistent in strains
belonging to the same phylotype: All testedI-1a/ST18 strains
possessed only island-like region 2, whereas otherSTs of the
subdivision I-a possessed additional regions also;however,only
I-2strainscontainedallfourgenomicregions.IncontrasttoKPA,thegenomesoftheI-1astrains266,SK137and
SK187 possess only a few strain-specific genomic clusters(Figure
1B). Although the large island-like regions of strain KPA areabsent
in the genome of strain 266 (with the exception of the NRPSgene
cluster, region 2), a signature of genome plasticity (aberrant
GCcontent), can still be detected in the vicinity of these genomic
regions,indicating that these regions are hotspots of genomic
flexibility/rearrangements. Some regions are restricted to type
I-1a genomes;
forexample,PAZ_c08910-PAZ_c09070iscommontostrains266andSK137 but is
absent in the genomes of the other three strains (TableS1).
Thisregionreplacesregion1of strainKPA; besideshypothe-tical
proteins it encodes an efflux ABCtransport systemand
aFigure1.ComparisonoffiveP. acnesgenomes.(A)All
CDSofthereferencegenomeofstrainKPA171202areshowninthetwoouterrings(cyan,
depending on strand location). The subsequent rings depict Bi-Blast
results for each CDS of the reference genome in the following
genomes(exterior to interior): 266, SK187, SK137, J139 and P.
freudenreichii (the unfinished genome of P. acnes J165 was not
taken into consideration due toinsufficient quality of CDS
prediction). The innermost ring depicts the GC content variation
from the mean (60%) of the reference genome. CDS of
theKPAgenomewhichhaveaBi-Blasthitwith.=25%proteinsequenceidentityinthesubjectgenomearedepictedingrey,
thosewithaproteinsequenceidentity,25%areshowninred. (B) Thegenomeof
strain266wastakenasthereferencegenome(twoouter ringsingreen).
Thesubsequent rings, depicting Bi-Blast hits of each CDS of the
reference genome are (exterior to interior): KPA, SK187, SK137,
J139 and P.
freudenreichii.Thenumbers1to9representmaingenomicislandsidentifiedinthereferencegenomes(seeTableS1).doi:10.1371/journal.pone.0021581.g001Table2.PCR-detectionofisland-likegenomicregionsinP.
acnesisolates.1(a) 1(b) 2(a) 2(b) 3(a) 3(b) 4(a) 4(b)strain
phylotype266 + + I-1a/ST18SK137 I-1a/newSTSK187 + + + +
I-1a/newSTKPA171202 + + + + + + + + I-2/ST34J139 + + + +
II/newST1.4.L1 + + I-1a/ST1812.1.L1 + + I-1a/ST1832.1.L1 + +
I-1a/ST184.1.A1 + + I-1a/ST2012.1.R1 + + I-1a/ST2019.1.R1 + + + + +
+ I-1a/ST213.5.R1 + + + + I-1a/ST272.3.A1 + + + + + + + +
I-2/ST3521.1.L1 + + + + + + + + I-2/ST363.3.R1 + + + + + + + +
I-2/ST3610.1.R1 + + II/ST5234.1.A1 + + II/ST54CCUG35547 + + + +
III/ST44CCUG35900 + + + + III/ST43P6 + + + + + + + + I-2/ST33Two
PCRs (a and b) per island were performed to check for the presence
of the four main genomic islands in the listed human isolates
(islands in SK137, SK187
andJ139werepredictedfromtheirgenomes).doi:10.1371/journal.pone.0021581.t002GenomicsofPropionibacteriumacnesPLoSONE
| www.plosone.org 4 June2011 | Volume6 | Issue 6|
e21581two-component system. Another region
(PAZ_c15380-PAZ_c15460,region8inFigure1B)
isspecifictothethreetypeI-1astrainsandencodes glycosyl hydrolases
such as a chitinase-like enzyme(PAZ_c15460) and an ABCtransport
systembelonging to thesubfamilyspecificforthetransportofdipeptides,
oligopeptidesandnickel. Apart fromthese fewdifferences, there are
noadditionalstrain-specificgenesinthegenomeofstrain266withareportedorpredictedfunction
in
virulence.Interestingly,strainSK137possessestwostrain-specificandisland-like
gene clusters not found in any other sequenced P. acnes (depicted
asA and B in Figure S1 and Table S1). The first cluster encodes
diversefunctions such as a recombinase, an ABC transport system,
and an N-acetylmuramoyl-L-alanineamidase(autolysin).
Thesecondgenomicregion(20.8kb)
includesageneclusterencodingproteinsrelatedtostreptolysin
biosynthesis (HMPREF0675_3181-3188); CDS
withsignificanthomology(2631%proteinsequenceidentity)toSagBCDofstreptococcalspeciescanbefoundhere(HMPREF0675_3183-85)[22],
as well as an efflux ABCtransport systemwith similarity(34%and
35%protein sequence identity, respectively) to
MtrA(HMPREF0675_3187) and MtrB (HMPREF0675_3188). This systemis
essential forresistancetotheantitumoragent mithramycininitsproducer
Streptomyces argillaceus [23]. Thus, it is likely that strain
SK137isabletoproduce atoxinsimilartostreptolysin
andprotectsitselfbyemploying an efficient export
system.Smallerstrain-specificregions exist inall genomes
(TableS1);for example, the KPA genome harbors the gene cluster
PPA0372-PPA0382 (region 6 in Figure 1A), which differs
considerablybetweenKPAand I-1astrains. This regionencodes among
othersahyalorunatelyase(PPA0380), whichisanenzymepredictedtocleave
hyaluronan, a non-sulfated glycosaminoglycan and a
majorcomponentoftheextracellularmatrixofconnectivetissues.In
additionto the presence andabsence of strain-specific genes,the
genomes were found to harbor genes carrying
potentialframeshiftmutations.Theorthologsof34genesfromstrainKPAareframeshiftedinstrain266,whereasthegenomeofstrainKPApossesses
only 14 frameshifted open reading frames (ORFs), whoseorthologs in
strain 266 are intact (data not shown). This indicates
aslightlyacceleratedgenelossinstrain266,whichmightbeduetoreductive
genome evolution. Among the frameshifted genes severalputative
virulence determinants were found, including
twomutatedgenesinstrain266encodingputativelysophospholipases(PPA0594,
PPA1425). By contrast, a lipase (PAZ_c18730) of strain266is
frameshiftedinKPA; this lipase is 45%identical tothecharacterized
glycerol-ester hydrolase A (GehA, PPA2105,PAZ_c21800) [24].
Substantial differences also exist in
genesencodingputativeadhesinssuchasproteinswiththrombospondintype3repeats(TableS1).Figure2.
Transcriptionofthefourlargeisland-likegenomicregionsofP. acnes.
Thetranscriptional profilesofthemainfourgenomicislands are shown.
The upper panels (A) represents the expression profiles in strain
KPA compared to strain 266. The yellow bar depicts the
genomicregions that are absent in strain 266 (islands 1, 3 and 4).
The first profile depicts the log2 intensities (brown, KPA; green,
266), and the second profilethelog2ratios(blue, higherinKPA; red,
higherin266). Notethatthepredictednon-ribosomal
peptidebiosynthesisgenecluster(PPA1284-1292,greenbar) of
island2isupregulatedinstrainKPA, albeit present alsoinstrain266.
Thelower panel (B) depictsthegrowthphase-dependenttranscription of
the genomic regions in strain KPA. Again, the first profile depicts
the log2 intensities (lilac, exponential phase; cyan, stationary
phase),andthesecondprofilethelog2ratios(blue, higher inexponential
phase; red, higher instationaryphase). Notethat
thepredictedthiopeptidesynthesis gene cluster (PPA0859-PPA0866,
orange bar) of island 1 is strongly transcribed and upregulated in
the exponential growth phase of
KPA.Theblackarrowsmarktheboundariesofeachisland-likegenomicregion.doi:10.1371/journal.pone.0021581.g002GenomicsofPropionibacteriumacnesPLoSONE
| www.plosone.org 5 June2011 | Volume6 | Issue 6|
e21581WholegenomegeneexpressionanalysisofP.
acnesGenomicdifferencesconstituteonepossiblereasonfordifferen-tial
strain properties; however, this gives only a superficial picture
asregulatory events occurring on the transcriptional and
translationallevel are ignored. Therefore, we also determined the
transcriptomeof P. acnes, focusingontwosets of experiments: (i)
determiningdifferential
expressionintwophylogeneticallydistinctstrains, and(ii)
recordinggrowthphase-dependentgeneexpression changes.Microarray
technology was used for these experiments,
utilizingwholegenomemicroarraysthatcomprised39,094oligomersthatwere
derivedfromthe genomeof strainKPA. All
datacanbeaccessedathttp://bioinfo.mpiib-berlin.mpg.de/pacnes/.Transcriptional
differences between two phylogeneticallydistinctstrainsofP. acnesTo
obtain additional insight into differing lifestyle and/orvirulence
properties of P. acnes strains, we determined the extent
oftranscriptional variationbetweentwostrains. Thetranscriptomesof
strains KPA(I-2, ST34) and266(I-1a, ST18), grownunderidentical
conditions (brainheart infusion(BHI) broth,
anaerobicconditions)tomidandlateexponentialgrowthphases(OD6000.3and0.6,
respectively) were compared. About 100genes of
thestrain-specificgenecontent of
strainKPA(locatedmostlyintheisland-likegenomic regions 1, 3and4)
weretranscribedintheexponentialgrowth phase,and were, therefore,
part of the activegenome(Figure2).
Insubsequentanalysesstrain-specificgenesofthe two genomes were
excluded so that gene expression differencesinthecommongenepool
alonecouldbedetermined. This wasachieved by considering the data
from oligomers of the microarraythat matched100%tobothgenomes
(75.7%(n =29,587) of alloligomers). Theexpressionof 119commongenes
differed(fold-change cutoff 2) between the two strains grown to
mid-exponentialgrowthphase,i.e.37and82genesupregulatedinKPAand266,respectively(Table
S2A). Inthe late exponential growthphase,316 genes, present in both
genomes, were deregulated (TableS2B). It was evident that
approximately 27% of these differentiallytranscribed genes were
also deregulated between the mid-exponential and stationary growth
phase in strain KPA,
thuswereexpressedinagrowthphase-dependentmanner(seesectionGrowth
phase-dependent transcription). This could indicatethat
thetwostrainsdifferintheirgrowthproperties. Indeed, therecorded
growth curves showed that strain 266 had a slightlyshorter doubling
time andreachedthe stationary phase earlierthanstrainKPA,
whengrowninBHImedium(datanotshown).Growthphase-independenttranscriptional
differencesbetweenstrainsKPAand266affectmetabolicpathwaysandsecretedfactorsVarious
biological processes associated with differentially expressedgenes
betweenstrains KPAand266wereidentifiedusingKEGG(Table S3A).
Substantial expression differences involved genesimplicated in
energy metabolismand protein biosynthesis; thesewere also
transcribedina growthphase-dependent manner (seebelow). When growth
phase-independent expression
differencesbetweenthetwostrainswereconsidered,
genesencodingtransportsystemswereanoverrepresentedclass: at least
fiveABCtransportsystems, two phosphotransferase systems and three
other transportersweredifferentiallytranscribedinthetwostrains.
Another class
ofderegulatedgenesencodedthosewithvariousmetabolicfunctions;strain266apparently
possesses anextendedanaerobic metabolicversatility as compared to
KPA, in particular employing fermentativepathways to utilize amino
acids. For instance, genes
encodingpathwaycomponentstodegradehistidinetoglutamate(PPA2166/PPA2167),
toprocess threoninetoglycine(PPA0402/403)
andtointerconvertaspartate,homoserineandlysine(PPA0642,PPA1258,PPA1470,
PPA1998) were all upregulated in strain 266. In
addition,transcriptionof genes encodingenzymes that
providetheelectronacceptorfor anaerobic respiration, fumarate, was
affected: arginino-succinatelyase(PPA1346)
andaspartate-ammonialyase(PPA0094)were both 3.5-fold upregulated in
266 (Table S2B, Figure 3).
Anotherpeculiarityofstrain266wastheincreasedexpressionoftwolactatedehydrogenases
(PPA0012, PPA0887), whose genes are not clustered.Thus,
lactateseemedtobeamajor carbonsourceof strain266,possibly taken up
via L-lactate permease (PPA0166). The latter geneis clustered with
genes of unknown functions (PPA0166-0169), whichare 6- to 7-fold
upregulated in strain 266; the corresponding proteinsare predicted
to be iron-sulfur cluster containing proteins, andtherefore may be
involved in electron transport processes. The genecluster,
PPA0463-0465, which encodes proteins related to myo-inositol
utilization, was also strongly induced, indicating that P. acnes266
can use myo-inositol asa carbon
andenergysource.Regardingpossiblevirulencefactors,
thegehAgene(PPA2105)encodingthesecretedtriacylglycerol
lipase[24]isamongtheup-regulated genes in strain 266. In addition,
the genes for thesecretedfactorsendoglycoceramidase(PPA0644)
andtheadhesinPA-25957(PPA2127)are4-and6-foldupregulatedinstrain266,respectively.
The latter has been characterized as a host
cell-surfaceattachmentproteinwithdermatansulfate-bindingactivity[25].
By contrast, the gene for the Christie-Atkins-Munch-Petersonfactor
4(camp4, PPA1231), andthe secretedlysozymeM1 (PPA1662) are 8- and
12-fold upregulated
instrainKPA,respectively.Thisisinstrikingagreementwiththefindingsofourpreviousproteomicstudy,whichshowedthatstrain266exhibited(i)
reducedlysozymeM1secretion, (ii) increasedsecretionof thelipase
GehAand (iii) exclusive secretion of PPA2127, whereasCAMP4 factor
was secreted only by KPA[26]. In addition,expression of htaA
(PPA0786, PA-4687), encoding an ironacquisition protein, was 5-fold
increased in strain 266 as comparedtoKPA; HtaAhas
beenshowntobeamajor
immunoreactiveproteinofP.acnes[25].Thegeneencodingthewell-characterizedpolyunsaturated
fatty acid isomerase PAI (PPA1039), whichcatalyzes the
isomerizationof linoleic acidto 10,12-conjugatedlinoleicacid(CLA)
wasalso6-foldupregulatedin266.
CLAsarereportedtohavebeneficialeffectsforthehumanhost
duetotheirapparentantioxidantandanti-cancerproperties[27].Taken
together, the transcriptomes of the two investigatedstrains
showedmarkeddifferences, inparticular withregardtometabolicpathways
but alsoinputativehost-interactingproper-ties.
Thesedatacouldprovideamolecularbasisfortheobserveddifferences of P.
acnes phylotypes interms of their inflammatorypotential
[10,11,12,28]. It alsosupports theprevious assumptionthat certain
P. acnes strains, i.e. type I-1a strains, are more prone
tocauseskindisorders thanothers, i.e. typeI-2strains.
However,futureworkisrequiredtoelucidatethefactorsresponsibleforthisdifferential
behavior,
andtheunderlyingmechanismsinvolved.Growthphase-dependenttranscriptionVirulence
properties of most pathogenic bacteria differ
betweentheirgrowthphases. It hasbeenspeculatedthat
stressresponses,oftenboostedinthe onset of the stationary phase,
might be acontributing factor to the immunostimulatory properties
of P. acnes[29]. Thus, we recorded growth
phase-dependenttranscription
bydeterminingthegeneexpressiondifferencesbetweenanexponen-tial
(OD600 0.3) and a stationary (OD600 0.8) growth phase cultureof
strain KPA, grown under anaerobic conditions in BHI
medium.25%(601genes) ofall
genesofthegenomewereatleast1.5-foldderegulated betweenthe two
growthphases. Table S4 lists
allGenomicsofPropionibacteriumacnesPLoSONE | www.plosone.org 6
June2011 | Volume6 | Issue 6| e21581genes deregulatedat least
2-fold: 137genes upregulatedintheexponential growthphase(EPgenes),
and122genesupregulatedinthestationaryphase(SPgenes).57%oftheEPgenescouldbemappedtoKEGGpathways,
whereasonly31%of theSPgeneswere mapped, indicating an elevated
presence of unknownfunctions inthestationaryphase(Table S3B).
Themost highlyenriched pathways among the EP genes were ribosome
(24genes), oxidative phosphorylation (17 genes), pyrimidine
andpurine metabolism (12 genes), protein export/secretion (6
genes)and glycolysis/gluconeogensis (5 genes); this reflects, as
expected,highly active metabolism, replication machinery and
proteintranslationduringtheEP. Bycontrast,
theSPwascharacterizedbytheupregulationof
genesinvolvedinaminoacidmetabolism(alanine, arginine, aspartate,
glutamateandprolinemetabolism;12genes) anddiverse transport
functions (ABCtransporters;
8genes).Growthphase-dependentuseofenergy-conservingroutesAlthoughP.
acnesisregardedasananaerobe, it canalsogrowunder lowoxygen
tensions. The methylmalonyl-CoApathway(Wood-Werkman cycle) of
propionate formation employed byP. acnes and other Propionibacteria
grown under
anaerobicconditionshasbeenwellstudied;themainenergy-conservingstepis
accomplished by electron transport to fumarate, a processtermed
fumarate respiration, and catalyzed by the fumaratereductase
(SdhABC) [30]. To date, however, knowledge regardingother
energy-conserving strategies of P. acnes in response tochanging
conditions and stress, in particular varying oxygenconcentrations,
remainsfragmentary.Here, thetranscriptomerevealedthat
despitebacterial growthunder anaerobicconditions, expressionof
thewholerespiratorychain could be detected. Strong up-regulation of
the NADHdehydrogenase/complex I (PPA1922-1936) was detectedintheEP,
indicating a necessity to re-oxidize NADH, which accumulatesduring
glycolysis and the catabolism of other substrates. Upregula-tion of
genes encoding the cytochrome bd oxidase (PPA0175, cydB,PPA0176,
cydA), a respiratory endoxidase, which, in Escherichia coli,favors
survival under low oxygen tensions [31], was detected in theEP. By
contrast, the cytochrome c oxidase (PPA0701, coxB,PPA0702,coxA,
PPA1561,cyoE) wasslightlystronglyexpressedinthe SP. Interestingly,
cytochrome c reductase (PPA0710-0712) andoxidase are absent fromthe
genome of P. freudenreichii [17],indicatingthat P. acnes has
agreaterabilitytoreact tochangingoxygenconditions. We alsonotedthe
differential expressionofalternative forms of the succinate
dehydrogenase/fumarateFigure 3. Growth phase-dependent utilization
of amino acid-metabolizing pathways in P. acnes. The schematic
representation includesamino acid-metabolizing enzymes whose genes
were strongly deregulated between the exponential (green) and
stationary (yellow) growth
phases.Theargininedeiminasepathway(PPA0582-0585) is anexponential
phasetrait, whereas thepathwaysfor theinterconversionof
glutamatetoornithine(PPA1347-1350) andof
glutamine/aspartatetodihydroorotate(PPA0997-1000)
arestationarygrowthphasetraits. Pleasenotealsotheconnected
fumarate-producing reactions selectively upregulated in strain 266
(in grey). Fumarate respiration is a major source of energy
conservationin P. acnes; most likely, reducing equivalents are
delivered by the NADH dehydrogenase (PPA1922-1936). The
enzyme-attributed numbers correspondto the gene nomenclature of the
KPA genome. For simplicity reasons inorganic and organic phosphate
cosubstrates were omitted from the
scheme.doi:10.1371/journal.pone.0021581.g003GenomicsofPropionibacteriumacnesPLoSONE
| www.plosone.org 7 June2011 | Volume6 | Issue 6|
e21581reductase(SdhABC): thegenecluster PPA1437-39(sdhABC)
wasstronglyupregulatedintheEP, whereas
PPA0950-52(sdhABC)wasslightlystronglyexpressedintheSP.Thesedataindicatethat
P. acnes employs oxidativephosphor-ylation to conserve energy,
employing alternative electrontransport routes, and possibly
utilizing different electron acceptors.Alikely scenario is that
reducing equivalents (in the formofNADH) are fed via complex I into
the respiratory chain in the EP.Electrons are transferred via
cytochrome b to SdhABC, which
actsasafumaratereductase,andusesfumarateasaterminalelectronacceptor.
Thus, protontranslocationcouldbe achievedbytwomembrane-bound
systems (complex I and SdhABC); ATP isgenerated via a FOF1 ATP
synthase (PPA1238-1245), which is alsopartiallyupregulatedintheEP.
Itseemsplausibletoassumethatthe Wood-Werkman cycle and the
respiratory chain areinterconnectedvia fumarate reductase. The key
enzyme of thecycle, methylmalonyl-CoA carboxytransferase
(PPA2007/2008), ishighly expressed in both growth phases. Due to
the use of the Gas-Pak anaerobic system for culturing P. acnes in
this study, we cannotexclude the possibility that residual
oxygenwas present, whichcould be used by the cytochrome bd oxidase.
However, expressionof genes of therespiratorychaindidnot differ
betweenP. acnesgrown in the Gas-Pak system from bacteria grown
under a definedand constant anaerobic atmosphere (data not shown).
Theemployment of terminal electronacceptors other thanoxygen,such
as nitrate, is very likely; indeed, a nitrate
reductase(PPA0507-PPA509) was strongly upregulated in the EP. It
remainsunclear why the expression of a different fumarate
reductase(SdhABC)andacytochromecoxidaseisupregulatedintheSP.One
possible explanation is that this branch of the aerobicrespiratory
chain is part of a stress response in the SP that preparesthe
bacteriatoreact tochangingenvironmental conditions,
i.e.increasingoxygenconcentrations.Metabolicpeculiarities:
deregulationofargininemetabolismAmong the biological processes that
were most stronglyderegulatedbetweenthetwogrowthphases, pathways of
aminoacidmetabolismwereprominentlyaffected(TablesS3B,S4).Forexample,
agenecluster (PPA0582-0585),
withhomologytotheargininedeiminase(ADI)pathwaywas5-to7-foldupregulatedinthe
EP (Figures 3 and 4). This pathway catalyzes the conversion
ofarginine to ornithine via citrulline, employing the
enzymesarginine deiminase (PPA0583), ornithine
carbamoyltransferase(PPA584) andcarbamate kinase (PPA0585), thereby
generatingNH3and ATP. There is also anarginine/ornithine
antiporter(PPA0582) encoded in this cluster. The arginine repressor
ArgR isencoded just downstream(PPA0586). This systemcould be
apowerful way for P. acnes tocounteract the acidificationof
theculture mediuminthe course of propionate formation,
akintoStreptococcussuisandsourdoughlacticacidbacteria,
whichusetheADIpathwaytofacilitatesurvival inacidicconditions[32].
TheADIpathwayhasalsobeenreportedtofunctionasanadditionalenergy
conserving pathway during anaerobic growthof
BacilluslicheniformisandPseudomonasaeruginosa[33,34].By contrast,
genes involved in arginine biosynthesis such asargininosuccinate
synthase (PPA2201), convertingaspartate andcitrulline
intoargininosuccinate, andthe cluster PPA1347-1350were upregulated
in the SP (Figure 3). The latter encodes
enzymescatalyzingtheconversionof glutamatetoornithine.
Othergenesinvolvedinaminoacidmetabolismandupregulatedinthe
SPweretwocopies of aglutaminesynthetase(PPA0664,
PPA0671,30%proteinsequence identity), whichconverts glutamate
intoglutamine, andtheclusterPPA0997-1001,
encodingasystemtointerconvert glutamine and (dihydro)orotate. In
conclusion,markedderegulationof
enzymeslinkedtotheureacyclecanbedetectedbetweentheEPandtheSP.DeregulationofgenesinvolvedinheatshockandotherstressresponsesHeat
shockproteins havebeenimplicatedinthevirulenceofP. acnes, for
example in the induction of a
proinflammatoryresponseinkeratinocytes [29,35]. Thegenes
encodingcytosolicFigure 4. The expressionprofile of the arginine
deiminasegeneclusterofP. acnes.
Themosthighlyderegulatedgenesintheentire P. acnes transcriptome are
shown here. The gene clusterPPA0582-0585 encodes the arginine
deiminase pathway, andPPA0580/PPA0581encodestressresponsegenes.
NotetheopposingexpressionofPPA0580/0581andPPA0582-0585.
Theupperpanels(A)represents the expression profiles in strain KPA
compared to strain 266.Thefirst
profiledepictsthelog2intensities(brown, KPA; green, 266),and the
second profile the log2 ratios (blue, higher in KPA; red, higher
in266). The lower panel (B) depicts the growth-phase
dependenttranscriptionof this genomic regioninstrainKPA. The first
profiledepicts thelog2intensities (lilac, exponential phase; cyan,
stationaryphase), and the second profile the log2ratios (blue,
higher inexponentialphase; red,
higherinstationaryphase).doi:10.1371/journal.pone.0021581.g004GenomicsofPropionibacteriumacnesPLoSONE
| www.plosone.org 8 June2011 | Volume6 | Issue 6| e2158170-kDa
heat-shock protein (Hsp70) chaperones were highlyexpressed in both
growth phases (DnaK, PPA2040;
DnaJ,PPA2038;GrpE,PPA2039);thesameistrueforthetriggerfactor(PPA1575),
a protein with peptidyl-prolyl cis-trans isomeraseactivity, andthe
first proteinto encounter nascent polypeptidechains as they emerge
fromthe ribosome. By contrast, thechaperonins GroEL(two genes:
PPA0453 and PPA1772, 65%proteinsequenceidentity) andGroES(PPA1773)
werederegulat-edbetweenthegrowthphases,i.e.6to11-foldupregulatedintheEP(TableS4).TheRNAlevelsofthesethreegenesexhibitedthestrongestgrowthphase
dependencyamongallgenes.GroEL of P.acnes has been implicated in the
stimulation of
cytokineproduction,suchasinterleukin(IL)-1-alpha,tumornecrosisfactor(TNF)-alphaor
granulocyte/macrophagecolony-stimulatingfac-tor(GM-CSF)
inkeratinocytes[35].In the SP 7.5-fold upregulation of hsp20
(PPA0737,
18kDaantigen2)wasdetected;thisgeneencodesanalpha-crystallin-likeprotein
homologous to HspX(Acr, Rv2031c) of Mycobacteriumtuberculosis [36];
HspXis an immunodominant antigen that
isrecognizedbythemajorityofpatientswithactivetuberculosis.Inaddition,
other genes related to the heat shock response
wereupregulatedintheSP,includingthoseencodingaDnaJhomolog(PPA0916),
aDnaKfamilyprotein(PPA1098), aPAC2(protea-someassemblychaperone)
familyprotein(PPA1099), andthreeHsp100 chaperones (ClpB (PPA2021),
ClpC (PPA0278) and ClpX(PPA1571)).Several other genes that encode
proteins related to stressresponses
anddefensewerealsoinducedintheSP,
forexamplePPA0580andPPA0581.TheseareclustereddirectlyupstreamofthehighlyinducedEPgenes
PPA0582-0585, whichencodetheADI pathway. PPA0580is
anosmoticallyinducible(OsmC)-likeprotein. OsmC-family proteins
apparently function as peroxir-edoxins [37]. PPA0581 encodes a
protein with homology to Hsp31of E. coli, which has chaperone
activity [38].
Interestingly,PPA0580andPPA0581werethemoststronglyderegulatedgenesinall
microarrayexperiments(Figure4).Upregulation of the genes PPA1449
(HicB-like protein) andPPA1450 (hypothetical P. acnes specific
protein) was also acharacteristicof theSP. HicBispart of acluster,
HicAB, whichis predictedtobe aRNA-targetingtoxin-antitoxin(TA)
systempresentinavarietyof bacteria[39]; hicABtranscriptioninE.
coliwasshowntobestronglyinducedbynutrientstarvation[40].Anadditional,
larger SP gene cluster is PPA0387-0394, whichincludes
aproteinsimilar toalkalineshockproteins (PPA0387),andRpoE(PPA0392),
encodingthestress responsesigmafactors24. Othergeneclusters
upregulatedintheSPwere: PPA1973-1975, encodinganitric-oxide(NO)
reductase(NorB, PPA1975),whichcouldfunctionas
anNOdetoxifyingsystemtoeliminateNO produced by the host immune
system; PPA1548-1550,encoding proteins similar to SufB (PPA1549)
and SufD(PPA1548), which are part of the iron-sulfur cluster
scaffoldcomplex SufBCD, which functions under iron starvation
andoxidativestressconditions[41].Growthphase-dependentexpressionofputativevirulencefactorsTwomachineriesforproteinsecretion,theSec-signalrecognitionparticle
(Sec-SRP) and the twin arginine translocation (Tat) systems,were
strongly expressed by P. acnes, as shown by the expression
levelsofsecF(PPA1163),secE(PPA1892),yajC(PPA1161),tatB(PPA0643)andtatC(PPA1378).TheSec-SRPsystem,butnottheTatsystem,was
more strongly expressed during the EP than the SP, in particularthe
components secA (PPA1333), secF (PPA1163) and ftsY(PPA1447)and also
signal peptidase I (PPA1434). Previously, we identifiedsecreted
factors of P. acnes by a proteomic approach [26].
Strikingly,wefoundthatnotonlywerecomponentsof
theSec-SRPandTatsystemsstronglyexpressedbut alsosecretedfactors,
mostof whichwere upregulated during the EP; these encode
endo-glycoceramidase(PPA0644), p60 (PPA0721), lysophospholipase
(PPA2142), rarelipoprotein A (PPA2175), GroEL (PPA1772) and several
hypotheticalproteins (PPA0532-534, PPA2141, PPA2239). Regarding the
CAMPfactors,thegenescamp1,
2and4hadsignificantexpressionlevels;camp4 (PPA1231) ranked among
the most strongly expressed genes,independentof thegrowthphase.
camp1(PPA1340) waspreferen-tially expressed in the EP, as was camp2
(PPA0687). CAMP2 and 4have been shown to be secreted proteins, and
CAMP1 is bothsecreted and surface-associated [26,42]. By contrast,
increasedexpressionofgenesencodingsecretedfactorsintheSPwasscarce,with
the exception of genes encoding a hyaluronidase (PPA0380) andthe
immunoreactive adhesin PPA2127 (PA-25957). Albeit
highlyexpressed,PPA2127 isnot translatedin strain
KPA,possiblyduetogeneticmutations[25,26].ConclusionsP. acnes is a
life-long inhabitant of the skin of most human bodies.Assuch,
thisorganismmusthavedevelopedefficientstrategiestoadapt to the
microenvironment within sebaceous follicles. Thetranscriptome of P.
acnes as reported here provides valuableinsights into the lifestyle
of this skin commensal. The data indicatethat the organismhas
several metabolic options inadditiontopropionic acid fermentation.
Energy conservation employssubstrate-level phosphorylation (e.g.
ADI pathway) as well asrespiration, utilizing fumarate and possibly
also nitrate as terminalelectronacceptors. Givenits
persistenceinsebaceous follicles ofthehumanskin, thequestionarises
as tohowP. acnes adapts itsmetabolicprofileinthefaceof
alteredsebaceous glandactivityduringpuberty, whenandrogen-dependent
stimulationof glandsecretion affects oxygen and nutrient (i.e.
sebum) levels in thefollicular microenvironment. Our data indicate
that P. acnes is ableto carry out aerobic respiration, thereby
employing alternativeterminal oxidases. It is not apparent why only
nanoaerophilicconditions but not higher oxygen tensions are
favorable
forgrowth;theorganismpossessestheimportantantioxidantdefensesystems,
andemploys aset of additional stress
responsesystems,whosefunctionshavenotbeenstudiedinthisorganism.Cell
cultureexperimentsshowedthathostresponsestoP.
acnesvarieddependingonthebacterialstrainused[10,11,12].Wealsonoticed
in our previous work substantial differences betweenP. acnes
strains withrespect totheireffects onepithelial
prostatecells[28]:strain266(I-1a,ST18)infectionofRWPE1cellsledtosignificant
deregulationof 260 host cell genes, the majority ofwhich were
inflammation-associated genes, whereas the KPAstrain (I-2, ST34)
impacted on only 97 gene transcripts (fold-change.3or,-3),
indicatingthat
strain266possessesahigherimmunostimulatoryactivitythanKPA.
Inthepresent work, weinvestigated the basis for differences between
strains KPA and 266on genomic and transcriptional levels. Genome
comparisoncannot easily explainthe phenotypic differences
betweenthesetwostrainsregardingtheirinflammatorypotential;thegenomeofstrain
266 does not contain additional known or putative virulencefactors,
which are correspondingly absent from strain KPA.Instead, several
genomic regions were identified that encode KPA-specific functions;
these regions differed also among the majorphylotypes of P. acnes.
They were actively transcribedinstrainKPA,
indicatingtheirfunctionality, andencodefitness functionssuchas
additional defensesystems against other
(Gram-positive)bacteriaandadditional
nutrientuptakeanddegradationsystems.GenomicsofPropionibacteriumacnesPLoSONE
| www.plosone.org 9 June2011 | Volume6 | Issue 6|
e21581Apossibleexplanationfortheelevatedinflammatorypotentialofstrain266comparedtostrainKPAcouldbededucedfromtheexpression
profiles of a number of predicted or characterized
host-interactingfactorsofP.acnes.Forinstance,thegeneencodingthelipaseGehAis
upregulatedinstrain266; GehAis suspectedtohydrolyze
sebumtriacylglycerides, resulting in the release ofglycerol and
free fatty acids [24,43]. Released fatty acids arethought
tobeinflammatory; theyfavorductal hypercornificationand increase
adhesion between P. acnes and cells of the hair
follicle,promotingcolonizationof P. acnes [43]. Another exampleis
theupregulationofthehtaAgeneinstrain266;HtaA(PA-4687)isanironacquisitionprotein,
whichhas beenshownto be
stronglyimmunoreactive[25].Besidesphylotype-specificvariation,
somepropertiesofP. acnes,e.g. biofilm formation, are apparently
independent of the phylotype[9]. We suspect that strain differences
between members of the samephylotype are, at least in part, due to
phase variation, e.g. affectingthe expressionof the
dermatan-sulphate adhesins PPA2127andPPA2210; however, it is likely
that other factors are also important.Taken together, our data
support the assumption that the severity ofacne vulgaris andotherP.
acnes-associateddiseases depends upontherespective phylotype of the
causative strain. An opportunistic scenariofor the formation of
inflammatory acne could thus be envisaged: (i) thepresence of a P.
acnes strain with elevated virulence potential, preferablya member
of the phlyogenetic subdivision I-1a expressing phase-variable
adhesins, and (ii) favorable conditions for its active growth,
i.e.nutrient (e.g. sebum) availability and anaerobic or
nanoaerophilicconditions after comedo formation, and (iii) a
predisposed host with anunbalanced immunological response to P.
acnes.NoteArecent publication by McDowell et al (Microbiology, 2011
Apr 21)presented an alternative MLST typing scheme for P. acnes.
Accordingtothisschemethestrainsinourstudyareclassifiedasfollows:KPA,typeIB1(ST10);
266, typeIA-CC6(ST25); SK137, typeIA-CC6(ST11); SK187, type IA
(ST39) and J139, type II
(ST40).MaterialsandMethodsStrainsandgrowthconditionsP.acnesstrain266waschosenforsequencing;itisatypeI-1a/ST18
strain, which was isolated from a pleuropulmonary infection(kindly
providedby Oliver KnappandMichel Popoff,
PasteurInstitute,Paris,France). For transcriptomeexperiments
strain266andKPA171202(I-2/ST34) wereused; thegenomeof
thelatterstrain was previously sequenced [15]. 14 additional P.
acnes strains,listed in table 2, were used for PCRdetection of four
largegenomic regions. The phylotypes of these humanisolates
werepreviously identified using the MLST scheme and
database(http://pacnes.mlst.net) developedbyLomholt
andKilian[14].All P. acnes strains wereculturedonBrucellaagar
plates underanaerobicconditionsat 37uCforthreedays.
Forliquidcultures,plate-grownbacteria were resuspended and washed
inbrain heartinfusion(BHI) broth(Sigma-Aldrich); BHI brothwas
inoculatedwith P. acnes (OD6000.01) and cultures were grown to
midexponential (OD600 0.3), late exponential (OD600 0.6) or
stationaryphases(OD6000.8)at37uCunderanaerobicconditionsusingtheGas-PakTMsystem
(Oxoid). To avoid inconsistent growthconditions, all
reportedgrowthexperiments were
performedinBHImediumfromthesamebatch. Inaddition,
forcomparativeexperiments (strain 266 versus strain KPA,
exponential phaseversus stationary growthphase) the culture
mediumwas
takenfromthesameautoclavedliquidmediumstock.Toavoidvariableanaerobic
conditions in comparative experiments due to the usageof
theGas-Paksystem, P.
acnesstrainswereculturedinthesameGasPakcontainer,
guaranteeingidentical growthconditions.DNAisolationandPCRDNA from
P. acnes was isolated using the MasterPureTMGramPositive DNA
Purification Kit (Epicentre). For the PCR detection
offourlargegenomicregionsinP.acnestheprimerslistedbelowwereusedtoamplify400500bpsof
thecorrespondinggenomicregion(twoprimerpairs foreachregion).
PCRamplifications
weredoneusingPlatinumTaqDNApolymerase(Invitrogen). Samples
wereinitially heated at 95uC for 3min, followed by 30 cycles
consisting of1minat95uC, 30sat55uC,
and45sat72uC.IS1_PPA0849_forACACACCGGCGTCACTTTCA, IS1_PPA0849_rev
ATGTT-GAGGGCGGTGACGTT; IS1_PPA0860_for CCGCTATTCG-CAATCGCTTC,
IS1_PPA0860_rev GAGAGCCGGAACCGA-GAACA; IS2_PPA1280_for
GGATCGGGTCGTCTTGTTGG,IS2_PPA1280_rev CAGCAACCTGGGCGAACTCT;
IS2_PPA-1288_for GCAGCGGTCTCTGACCAACA, IS2_PPA1288_rev
T-CCCCGAGAGCCAATCAGAG;IS3_PPA1585_forCGTGACT-GGCTGCTGCATCT,
IS3_PPA1585_rev CGTTGCCGTCCA-AGTGTTTG;
IS3_PPA1612_forGCCGCCCTAGGCAAGAAA-CT, IS3_PPA1612_rev
GCAGGCGAGCAAGCTGGTAT; IS4_P-PA2056_for CGACATTTCGCGGGACTACC,
IS4_PPA2056_r-ev ACGACCTCGTGCTTCCCAAC; IS4_PPA2070_for
TACTG-CGCCGCTCAACTCAC,IS4_PPA2070_revCATCGTGCACC-CTCATCCAC.Genomesequencing,
assemblyandgapclosureA pyrosequencing approach was used for
whole-genome sequenc-ing of P. acnes strain 266. Chromosomal DNA
was nebulized, a single-strandedtemplate
DNAlibrarypreparedandsequencedusingaGenome Sequencer FLX Instrument
and Titanium chemistry(Roche Applied Science); the library
preparation and
DNAsequencingwereperformedaccordingtomanufacturersprotocols(RocheAppliedScience).
Thesequencedata(247,277reads) wereassembled using the Newbler
Assembler (454 Life Sciences). 15contigs were obtained (.500
bases); a 36-fold coverage was obtained.Editing of sequences was
done with the GAP4 software as part of
theStadenpackage[44].Tosolveproblemswithmisassembledregionscaused
by repetitive sequences and to close remaining sequence
gaps,standard PCR, combinatorial multiplex PCR, and primer walking
onrecombinant plasmids were applied. The genome sequence
reportedinthis paper has beendepositedinthe GenBank database
withaccession number CP002409.BioinformaticsCodingsequences (CDS)
andopenreadingframes (ORFs) werepredicted with YACOP [45], using
the ORF finders Glimmer, CriticaandZ-curve. The output was
verifiedandeditedmanually usingcriteria such as the presence of a
ribosome binding site, GC frame plotanalysis, and similarity to
known protein-encoding sequences.Annotationwas
doneinatwo-stepapproach. Initially, all
proteinswerescreenedagainstSwiss-Protdataandpubliclyavailableproteinsequences
fromother completed genomes. By protein
sequencecomparisonwiththePfam, GenBank, ProDom,
COGandPrositedatabases, all predictions were verified or modified
manually using theERGOsoftwarepackage[46],
licensedbyIntegratedGenomicsTM.Completegenomecomparisonsweredonewithaproteinsequence-based
bidirectional BLAST approach. For genome-wide comparisonsof
nucleotide sequences the Artemis Comparison Tool (ACT) was
used[47]. Sequence homologies were onlymentionedinthis
studyforproteinswithan aminoacididentityof .25%andan
overlapofthequery and subject sequence of .90%. To identify genomic
islands theIslandViewer was used, acomputational tool that
integrates threeGenomicsofPropionibacteriumacnesPLoSONE |
www.plosone.org 10 June2011 | Volume6 | Issue 6| e21581different
genomicislandpredictionmethods, i.e. IslandPick, Island-Path-DIMOB,
and SIGI-HMM[48]. For the analysis and mapping ofpathways of P.
acnes KEGG and KEGG MAPPER were used (http://www.genome.jp/kegg/);
several pathways were manually curated.RNAisolation,
quantificationandqualitycontrolTotal RNAwas
isolatedbyamodifiedTRIzolHreagent RNApreparation method
(Invitrogen). Briefly, a P. acnes culture washarvested by
centrifugation and the pellet (,109bacteria)
wasresuspendedin1mlTRIzol.Theresuspensionwasaddedtoatubewithglassbeads(LysingMatrixB,MPBiochemicals),andprocessedtwice
for 25s at speed 6.5 in a FastPrep FP120 shaker (Savant).
Aftercentrifugation(5min, 80006g)
theupperphasewastransferredtoamicrocentrifuge tube and mixed with
200 ml chloroform. Subsequent-ly, the steps as detailed in the
manufacturers protocol for the TRIzolHreagent werecarriedout
(Invitrogen). All centrifugationsteps weredone at 4uC. The amount
of RNA was determined by
OD260/OD280measurementusingaNanoDropH1000spectrophotometer(Kisker).The
RNA size, integrity and the amount of total RNA were
measuredwithaBioanalyzer2100(Agilent Technologies) withaRNANano6000
microfluidics
kit.MicroarraydesignP.acnesKPA171202microarraysweredesignedwithOligoWiz[49]
as 4x44Kcustommicroarrays including all ORFs andintergenic regions
(IGR) fromthe whole genome of P. acnesKPA171202. ORF specific
probes were designed as senseoligonucleotides of the coding strand.
IGR, 39- and
59-overlappingprobestoORFsweredesignedassenseandreversecomplementprobes
to the +-strand. Overlapping probes to ORFs weregenerated by
sequence stretches comprising -45 bases downstreamor
+45basesupstreamandthefirstorlast45basesof theORF,respectively. An
oligo aim length of 60 bases with a minimal
lengthof50baseswasusedtogetherwithmaxhomology(avoidselfhit)97, 80%
cross hybridization maximum length (avoid self hit) and arandom
primed position preference score within OligoWiz.Dependent
ontheseparameters andthesequenceinput length,eachORFandIGRwas
coveredbyseveral specificoligonucle-otideprobes. Intotal
39,094specificP. acnes KPA171202probesequences were uploaded to
eArray (Agilent
Technologies,https://earray.chem.agilent.com/earray/) as ChIP
applicationwith a customer specified (ordered according
chromosomallocation)
featurelayout.ForcomparativeexpressionprofilingofthetwoP.
acnesstrains266andKPA, all oligomersof
themicroarraywereidentifiedbyBLASTNwhich matched 100%between both
genomes; onlythose29,587oligomers(75.7%)
wereusedtocalculateexpressiondifferencesbetweenthetwoP.
acnesstrains.MicroarrayexpressionprofilingMicroarrayexperimentswereperformedasdual-colorhybrid-izations.Inordertocompensatespecificeffectsofthedyesandtoensure
statistically relevant data analysis, a color-swap dye-reversalwas
performed[50]. RNAlabelingwas performedwiththetwocolor Quick Amp
Labeling Kit (Agilent Technologies) usingFullSpectrumMultiStart
PrimerforT7IVTRNAAmplification(BioCat GmbH) as randomT7 labeling.
Inbrief, mRNAwasreverse transcribed and amplified using a
FullSpectrumMulti-Start-T7-promotor primer and was labeled either
with Cyanine 3-CTPor Cyanine 5-CTP. Alternatively, the total
RNAsampleswere amplified with the TransPlex Whole
TranscriptomeAmplification Kit (Sigma-Aldrich) and labeledwith
BioPrime PlusArray CGHIndirect Genomic Labeling System(Invitrogen)
asWTA BioPrime indirect. In brief, a library was generated by
standdisplacementreactionusingphi-29polymeraseandquasi-randomprimersfollowedby17PCRcycles.
Theresultant
cDNAlibrarywaslabeledwithKlenowpolymeraseinanindirectreactionusinganimoallyl
nucleotides andsubsequent chemical NHS-ester Cy-dyecoupling.
Afterprecipitation, purificationandquantification,labeled samples
(cRNAfor randomT7 labeling or cDNAforWTABioPrime indirect labeling)
were hybridizedseparatelytothe 4x44K custom-commercial microarrays
according to thesuppliers protocol (Agilent Technologies). Scanning
of micro-arrays was performedwith5
mmresolutionandextendedmodeusingahighresolutionmicroarraylaserscanner(G2505,
AgilentTechnologies). Rawmicroarray image datawere
extractedandanalyzed with the Image Analysis/Feature Extraction
softwareG2567AA (Version A.10.5.1.1, Agilent Technologies).
Theextracted MAGE-ML files were further analyzed on
reporterandsequencelevelswiththeRosettaResolverBiosoftware,
Build7.2.2.0 SP1.31. Ratio profiles comprising single
hybridizationswere combined in an error-weighted fashion to create
ratioexperiments. A 2fold change expression cut-off for
ratioexperiments was applied together withanti-correlationof
ratioprofiles rendering the microarray analysis highly significant
(P-value
.0.01),robustandreproducible.Additionally,dataderivedfromthedifferent
labelingprocedures (randomT7labelingandWTABioPrimeindirectlabeling)
werecomparedandcombined.Onlyresultsderivedfrombothlabelingmethodswereconsideredas
relevant for further analysis. The data presented in
thispublication have been deposited in NCBIs Gene ExpressionOmnibus
(GEO, http://www.ncbi.nlm.nih.gov/geo/) and
areaccessiblethroughGEOSeriesaccessionnumberGSE26738.Inaddition,themicroarrayresultscanbevisualizedathttp://bioinfo.mpiib-berlin.mpg.de/pacnes/.
All oligomers are
plottedaccordingtotheirexpressionlevels(absoluteintensitiesaswell
aslog-ratios/fold-changes). Genomic deletions in strain 266
aremarked as yellowboxes. For a full legend see:
http://bioinfo.mpiib-berlin.mpg.de/pacnes/icons/legend.png.SupportingInformationFigureS1
Genomicislandsof
sequencedPropionibac-teriagenomes.Themeta-programIslandviewer(http://www.pathogenomics.sfu.ca/islandviewer/query.php)
was used to pre-dict genomic islands. Predicted genomic islands are
colored withinthe circular image based on the following tools:
SIGI-HMM,orange;IslandPath-DIMOB,blue; integrated,red. Black
lineplot:GC content (%). The numbers and letters of the
islandscorrespond to figure 1 and table S1. Strain-specific
genomicregions (listed in table S1) often, but not always,
correspond to thepredictedislands.Theisland19
inthegenomeofstrainSK137isdissimilartotheisland1ofstrainKPA,
seetableS1.(DOC)Table S1 Genomic differences between P. acnes
strains.Listedareall regions .2genes whichdiffer (deletions,
insertions,replacements, differences) between KPAand 266 (A),
KPAand SK137(B), and 266 and SK137 (C). In brackets: size of each
genomic region(inkb). Colorcode: green, specificregions inKPA;
yellow, specificregionsin266; red, specificregionsinSK137; grey,
presentinbothgenomes, but with substantial differences (identity
,50% orframeshifted). The numbers and letters in the first column
correspondto the genomic regions depicted in figures 1 and
S1.(DOC)Table S2 Gene expression changes between two
P.acnesstrainsgrownanaerobicallyto(A) mid-exponen-tial phase or to
(B) late-exponential phase in
BHIGenomicsofPropionibacteriumacnesPLoSONE | www.plosone.org 11
June2011 | Volume6 | Issue 6| e21581medium. The data shownis
basedonly onoligomers of
themicroarraywhichmatched100%inthetwogenomes
KPAand266.Theintensitieswerecalculatedastheaverageintensityfromoligomersthatcoveredtherespectivegene.
Onlysignificantlyde-regulated genes with fold changes .2 or ,-2 are
listed. For furtherdetails, seetheGEOentryGSE26738or goto:
http://bioinfo.mpiib-berlin.mpg.de/pacnes/.(XLS)Table S3 A:
Biological processes associatedto
genesdifferentiallyregulatedintheP.acnesstrainsKPAand266.All119and316genes(seeTableS2)thatwerede-regulated(fold-change
of .2 or ,-2) in the two strains grown to mid-exponential and late
exponential growth phases, respectively,
wereconsideredinthisanalysis. Intheexponential
growthphase(OD0.3),82geneswereup-regulatedinstrain266,40ofwhich(49%)were
KEGG-mappable; inKPA, 37were up-regulated, only 7(19%) were
KEGG-mappable. In the late exponential
growthphase(OD0.6),194geneswereup-regulatedinstrain266,74ofwhich
(38%) were KEGG-mappable; in KPA, 122 genes were up-regulated, 57
of which (46%) were KEGG-mappable.
KEGGMAPPER(http://www.genome.jp/kegg/mapper.html) was usedto map
the corresponding genes to biological processes.
Onlyprocesseswithat least2hits arelisted. B: Biological
processesassociatedtogenesup-regulatedinP.acnesKPAgrownto
exponential or stationary growth phase. All genes with
afold-changeof .2or ,-2wereconsideredinthis analysis.
137geneswereexponentialphase(EP)genes,78ofwhich(57%)wereKEGG-mappable.122
genes were stationaryphase (SP) genes;38of which(31%)
wereKEGG-mappable. Onlyprocesses
withatleast2hitsarelisted.(DOC)Table S4 Gene expression changes
between exponentialandstationaryphase P. acnes KPA171202. P. acnes
wasgrowninBHImediumtoOD0.3(exp) andOD0.8(stat) underanaerobic
conditions. The intensities were calculated as theaverage intensity
of all oligomers that covered the respective gene.Only
significantly de-regulated genes with fold changes .2 or
,-2arelisted.ForfurtherdetailsseetheGEOentryGSE26738orgoto:
http://bioinfo.mpiib-berlin.mpg.de/pacnes/.(XLS)AcknowledgmentsWethankMeikeSorensenandStefanieOffschankaforexcellent
technicalassistance, LesleyA. OgilvieandKateHolden-Dyeforcritical
readingofthemanuscript.AuthorContributionsConceived and designed
the experiments: HB. Performed the experiments:EB JHHB. Analyzed
the data: EB JWATHJMHB. Contributedreagents/materials/analysis
tools: AWHBLMK. Wrotethepaper: HB.Assistedinthedesignofthestudy:
GGRDTFM.References1. CogenAL, Nizet V, GalloRL(2008)
Skinmicrobiota: asourceof diseaseordefence?BrJDermatol 158:
442455.2. Dessinioti C, Katsambas AD(2010) Theroleof
Propionibacteriumacnes inacnepathogenesis: factsandcontroversies.
ClinDermatol 28: 27.3. JakabE,ZbindenR, GublerJ, RuefC,
vonGraevenitzA, etal. (1996)
SevereinfectionscausedbyPropionibacteriumacnes:
anunderestimatedpathogeninlatepostoperativeinfections. YaleJBiol
Med69: 477482.4. Soderquist B, Holmberg A, Unemo M(2010)
Propionibacterium acnes as anetiological agent of
arthroplasticandosteosyntheticinfectionstwocaseswithspecificclinical
presentationincludingformationof drainingfistulae. Anaerobe16:
304306.5. GrahamGM, Farrar MD, Cruse-Sawyer JE, HollandKT,
InghamE(2004)ProinflammatorycytokineproductionbyhumankeratinocytesstimulatedwithPropionibacteriumacnesandP.
acnesGroEL. BrJDermatol 150: 421428.6. JappeU, InghamE,HenwoodJ,
HollandKT(2002)
Propionibacteriumacnesandinflammationinacne:P.acneshasT-cellmitogenicactivity.BrJDermatol146:202209.7.
JugeauS,TenaudI,KnolAC,JarrousseV,QuereuxG,etal.(2005)Inductionoftoll-likereceptorsbyPropionibacteriumacnes.
BrJDermatol 153: 11051113.8. KimJ, Ochoa MT, Krutzik SR, Takeuchi
O, Uematsu S, et al. (2002)Activation of toll-like receptor 2 in
acne triggers inflammatory cytokineresponses. JImmunol 169:
15351541.9.
HolmbergA,LoodR,MorgelinM,SoderquistB,HolstE,etal.(2009)BiofilmformationbyPropionibacteriumacnes
is acharacteristicof invasiveisolates. ClinMicrobiol Infect15:
787795.10. NagyI,PivarcsiA,KoreckA,SzellM,UrbanE,etal.(2005)
DistinctstrainsofPropionibacteriumacnes
induceselectivehumanbeta-defensin-2andinterleukin-8expression in
human keratinocytes through toll-like receptors. J Invest
Dermatol124: 931938.11. Tanabe T, Ishige I, Suzuki Y, Aita Y,
Furukawa A, et al. (2006) Sarcoidosis andNOD1 variation with
impaired recognition of intracellular Propionibacterium
acnes.BiochimBiophysActa1762: 794801.12. Nagy I, Pivarcsi A, Kis K,
Koreck A, Bodai L, et al. (2006) Propionibacterium acnesand
lipopolysaccharide induce the expression of antimicrobial peptides
andproinflammatorycytokines/chemokinesinhumansebocytes.MicrobesInfect8:21952205.13.
McDowell A, Valanne S, RamageG, TunneyMM, GlennJV, et al.
(2005)Propionibacteriumacnes types I andII represent
phylogeneticallydistinct groups.JClinMicrobiol 43: 326334.14.
Lomholt HB, KilianM(2010) Populationgenetic analysis of
Propionibacteriumacnes
identifiesasubpopulationandepidemicclonesassociatedwithacne.
PlosOne5: e12277.15.
BruggemannH,HenneA,HosterF,LiesegangH,WiezerA,etal.(2004)
Thecomplete genome sequence of Propionibacterium acnes, a commensal
of human skin.Science305: 671673.16. NelsonKE, WeinstockGM,
Highlander SK, WorleyKC, CreasyHH, et al.(2010) Acatalogof
referencegenomes fromthehumanmicrobiome. Science328: 994999.17.
Falentin H, Deutsch SM, Jan G, Loux V, ThierryA, et al.(2010) The
completegenomeof Propionibacteriumfreudenreichii CIRM-BIA1,
ahardyactinobacteriumwithfoodandprobioticapplications. PlosOne5:
e11748.18. BrownLCW,AckerMG,
ClardyJ,WalshCT,FischbachMA(2009)Thirteenposttranslational
modificationsconvert
a14-residuepeptideintotheantibioticthiocillin.ProcNatl AcadSci
USA106: 25492553.19. Yu Y, Duan L, Zhang Q, Liao R, Ding Y, et al.
(2009) Nosiheptide
biosynthesisfeaturingauniqueindolesideringformationonthecharacteristicthiopeptideframework.
ACSChemBiol 4: 855864.20.
EngelhardtK,DegnesKF,ZotchevSB(2010)Isolationandcharacterizationofthe
gene cluster for biosynthesis of the thiopeptide antibiotic
TP-1161. ApplEnvironMicrobiol 76: 70937101.21. FarrarMD,HowsonKM,
BojarRA,West D,TowlerJC,etal.(2007)
GenomesequenceandanalysisofaPropionibacteriumacnesbacteriophage.JBacteriol
189:41614167.22. Datta V, Myskowski SM, Kwinn LA, ChiemDN, Varki N,
et al. (2005)Mutational analysis of the group A streptococcal
operon encodingstreptolysin
Sanditsvirulenceroleininvasiveinfection. Mol Microbiol 56:
681695.23. Fernandez E, LomboF, Mendez C, Salas JA(1996)
AnABCtransporter isessential for resistance tothe antitumor agent
mithramycininthe producerStreptomycesargillaceus. Mol GenGenet251:
692698.24. Miskin JE, FarrellAM, Cunliffe WJ, Holland KT (1997)
Propionibacteriumacnes, aresident of lipid-rich human skin,
produces a 33 kDa extracellular lipase encodedbygehA.
Microbiology143: 17451755.25.
LodesMJ,SecristH,BensonDR,JenS,ShanebeckKD,etal.(2006)Variableexpression
of immunoreactive surface proteins of Propionibacterium
acnes.Microbiology152: 36673681.26. HollandC, Mak TN, Zimny-Arndt
U, SchmidM, Meyer TF, et al. (2010)Proteomic identification of
secreted proteins of Propionibacteriumacnes. BMCMicrobiol 10:
230.27. LiavonchankaA, HornungE, FeussnerI, RudolphMG(2006)
Structureandmechanismof the Propionibacteriumacnes
polyunsaturatedfatty acidisomerase.ProcNatl AcadSci USA103:
25762581.28. Fassi Fehri L, Mak TN, Laube B, BrinkmannV, Ogilvie
LA, et al. (2011)Prevalence of Propionibacterium acnes in diseased
prostates and its inflammatory andtransforming activity on prostate
epithelial cells. Int J Med Microbiol 301: 6978.29. Farrar MD,
Ingham E, Holland KT (2000) Heat shock proteins
andinflammatoryacnevulgaris:
molecularcloning,overexpressionandpurificationofaPropionibacteriumacnesGroELandDnaKhomologue.FEMSMicrobiolLett191:
183186.30. Wood HG(1981) Metabolic cycles in the fermentation by
propionic acidbacteria.
In:EstabrookRW,SreraP,eds.Currenttopicsincellularregulation-1981.
NewYork: AcademicPress.
pp255287.GenomicsofPropionibacteriumacnesPLoSONE | www.plosone.org
12 June2011 | Volume6 | Issue 6| e2158131.
JonesSA,ChowdhuryFZ,FabichAJ,AndersonA,SchreinerDM,etal.(2007)Respiration
of Escherichia coli in the mouse intestine. Infect Immun
75:48914899.32. Gruening P, Fulde M, Valentin-Weigand P, Goethe R
(2006) Structure,regulation, and putative function of the arginine
deiminase system of Streptococcussuis. JBacteriol 188: 361369.33.
Maghnouj A, deSousaCabral TF, StalonV, Vander WauvenC(1998)
ThearcABDC gene cluster, encoding the arginine deiminase pathway of
Bacilluslicheniformis, andits activationbytheargininerepressorargR.
JBacteriol 180:64686475.34. Gamper M, Zimmermann A, Haas D (1991)
Anaerobic regulation oftranscription initiation in the arcDABC
operon of Pseudomonas aeruginosa.JBacteriol 173: 47424750.35.
Wilcox HE, Farrar MD, Cunliffe WJ, Holland KT, Ingham E (2007)
ResolutionofinflammatoryacnevulgarismayinvolveregulationofCD4+T-cellresponsestoPropionibacteriumacnes.
BrJDermatol 156: 460465.36. Hu Y, Movahedzadeh F, Stoker NG, Coates
AR(2006) Deletion of theMycobacterium tuberculosis
alpha-crystallin-like hspX gene causes increased
bacterialgrowthinvivo. InfectImmun74: 861868.37. ShinDH, Choi IG,
BussoD, JancarikJ, YokotaH, et al. (2004) StructureofOsmC from
Escherichia coli: a salt-shock-induced protein. Acta Crystallogr D
BiolCrystallogr60: 903911.38. Mujacic M, Baneyx F (2006) Regulation
of Escherichia coli hchA, a
stress-induciblegeneencodingmolecularchaperoneHsp31.Mol Microbiol
60: 15761589.39. MakarovaKS,GrishinNV,KooninEV(2006)
TheHicABcassette,aputativenovel,
RNA-targetingtoxin-antitoxinsysteminarchaeaandbacteria.
Bioinfor-matics22: 25812584.40. Jorgensen MG, Pandey DP, Jaskolska
M, Gerdes K (2009) HicA of Escherichia colidefinesa novel family
oftranslation-independentmRNA interferasesin bacteriaandarchaea.
JBacteriol 191: 11911199.41. Chahal HK, DaiY,Saini
A,Ayala-CastroC,OuttenFW(2009)TheSufBCDFe-S scaffold complex
interacts with SufA for Fe-S cluster transfer. Biochemistry48:
1064410653.42. ValanneS, McDowell A, Ramage G,TunneyMM, Einarsson
GG, etal. (2005)CAMP factor homologues in Propionibacterium acnes:
a new protein familydifferentiallyexpressedbytypesIandII.
Microbiology151: 13691379.43. GribbonEM, CunliffeWJ,
HollandKT(1993) Interactionof
Propionibacteriumacneswithskinlipidsinvitro. JGenMicrobiol 139:
17451751.44. StadenR, JudgeDP, BonfieldJK(2003)
ManagingsequencingprojectsintheGAP4 environment. In: Krawetz SA,
Womble DD, eds. Introduction tobioinformatics. Totowa, USA:
Atheoretical andpractical approach. HumanaPress.45. TechM, Merkl
R(2003) YACOP:
Enhancedgenepredictionobtainedbyacombinationofexistingmethods.
InSilicoBiol 3: 441451.46. OverbeekR, LarsenN, Walunas T, DSouzaM,
PuschG, et al. (2003)
TheERGOgenomeanalysisanddiscoverysystem.NucleicAcidsRes31:164171.47.
CarverTJ, RutherfordKM, BerrimanM, RajandreamMA, Barrell BG,
etal.(2005) ACT: theArtemisComparisonTool. Bioinformatics21:
34223423.48. Langille MG, BrinkmanFS(2009) IslandViewer:
anintegratedinterface forcomputational identification and
visualization of genomic islands. Bioinformatics25: 664665.49.
Wernersson R, Juncker AS, Nielsen HB (2007) Probe selection for
DNAmicroarraysusingOligoWiz. NatProtoc2: 26772691.50. Churchill GA
(2002) Fundamentals of experimental design for cDNAmicroarrays.
NatGenet32Suppl: 490495.GenomicsofPropionibacteriumacnesPLoSONE |
www.plosone.org 13 June2011 | Volume6 | Issue 6| e21581
