Metagenomic analyses of bacteria on human hairs

12 Pages·2014·1.03 MB·English

by  Silvana R Tridico

Tridicoetal.InvestigativeGenetics2014,5:16
http://www.investigativegenetics.com/content/5/1/16
RESEARCH Open Access
Metagenomic analyses of bacteria on human
hairs: a qualitative assessment for applications in
forensic science
Silvana R Tridico1,2*, Dáithí C Murray1,2, Jayne Addison1, Kenneth P Kirkbride3 and Michael Bunce1,2
Abstract
Background: Mammalian hairs are one ofthemost ubiquitous types of trace evidence collected inthe course of
forensic investigations. However, hairs that are naturally shed or that lack rootsare problematic substrates for DNA
profiling; thesehair typesoftencontain insufficient nuclear DNA to yield short tandem repeat (STR) profiles.Whilst
therehave been a numberof initial investigations evaluating thevalue ofmetagenomics analyses for forensic
applications (e.g. examination ofcomputer keyboards), therehave been no metagenomicevaluations of human
hairs—a substrate commonly encountered during forensic practice.This present study attempts to address this
forensic capability gap, by conducting a qualitative assessment into the applicability of metagenomic analyses of
human scalp and pubic hair.
Results: Forty-two DNA extracts obtained from human scalp and pubic hairs generated a total of 79,766 reads,
yielding 39,814 reads post control and abundance filtering. The resultsrevealed the presence of unique
combinations of microbialtaxa that can enable discrimination between individuals and signaturetaxa indigenous
to female pubic hairs. Microbial data from a single co-habiting couple added an extra dimension to thestudy by
suggesting that metagenomic analyses might be of evidentiary value insexual assault cases when other associative
evidence is not present.
Conclusions: Of allthe data generated inthis study, the next-generationsequencing (NGS) data generated from
pubic hair held themost potential for forensic applications. Metagenomic analyses ofhumanhairs mayprovide
independent data to augment other forensic results and possibly provideassociation between victims of sexual
assault and offender when other associative evidence is absent. Basedon results garnered in thepresent study, we
believe that with further development, bacterial profiling ofhair will become a valuable addition to theforensic
toolkit.
Keywords: Forensic, Metagenomics,Bacteria, Scalp hairs, Pubic hairs, Sexual assaults,Next-generation sequencing,
16S DNA
Background of sequences per sample facilitated assessments of micro-
Over the last decade, the development of bacterial bialcommunitiesbetweenbodysitesandindividuals[2,3].
culture-independent approaches (metagenomics), based The increased sequencing power stimulated the develop-
on 16S rRNA genes (hereafter referred to as 16S), se- ment of robust computational programmes capable of
quences hasbecomethe cornerstoneofmicrobialecology processinglarge,complexsequencingdatasets[4]anden-
[1]. The advent of next-generation sequencing (NGS) abled phylogenetic analyses of human and environmental
technologies and platforms capable of generating millions genomes[5,6].
Studies on the human microbiome (the collective ge-
*Correspondence:[email protected] nomes present in the human body) suggest that there are
1VeterinaryandLifeSciences,MurdochUniversity,Perth,WA6150,Australia significant differences in bacterial composition not only
2TraceandEnvironmentalDNAlaboratory,DepartmentofEnvironmentand
between different body sites but also between individuals
Agriculture,CurtinUniversity,Perth,WA6845,Australia
Fulllistofauthorinformationisavailableattheendofthearticle
©2014Tridicoetal.;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreative
CommonsAttributionLicense(http://creativecommons.org/licenses/by/4.0),whichpermitsunrestricteduse,distribution,and
reproductioninanymedium,providedtheoriginalworkisproperlycredited.TheCreativeCommonsPublicDomain
Dedicationwaiver(http://creativecommons.org/publicdomain/zero/1.0/)appliestothedatamadeavailableinthisarticle,
unlessotherwisestated.
Tridicoetal.InvestigativeGenetics2014,5:16 Page2of12
http://www.investigativegenetics.com/content/5/1/16
[3,5,7].Thepotentialthatindividualsmayharbourunique assess the viability of metagenomic analyses of hairs in a
bacterialspeciesisofsignificancetoforensicinvestigations. forensiccontext.Thethreeaimsoftheresearchreported
For centuries, associative hair evidence relied solely on here weretoassess:
comparative microscopy based on qualitative features
such as colour and pigmentation [8-10]. The advent of 1) Whetherhumanscalpand pubichairscanbe
PCR in the mid-1980s initiated a paradigm shift in the differentiated onthebasisoftheir 16Smicrobial
forensic examination of hairs. For the first time, DNA composition
profiles could complement qualitative microscopical 2) Whetherindividuals canbedifferentiated onthe
observations [11]. However, the success of the highly basisofmicrobialtaxacolonisingscalp andpubic
discriminatory short tandem repeat (STR) profiling is hairs
dependent on hairs bearing anagen roots (actively grow- 3) Whetherbacterial 16Sprofilesonhairshaftsare
inghairs)thatare rich in nuclearDNA (nuDNA) and, to stable overtime
a lesser extent, hairs that are in the quiescent (catagen)
growth phase [12]. However, the majority of hairs recov- Overall, the objective of this initial study was to estab-
ered in forensic investigations are shed hairs (i.e. those lish whether further development of the technique is
in their telogen phase); these hairs have ceased to grow warranted.
and contain little or no nuDNA [13]. STR profiling of
these hair roots typically yields trace amounts of, often Methods
degraded, human DNA and can require the use of low- Samplecollection
template DNA strategies and the complications that Bacterial communities, associated with human scalp and
accompanysuchapproaches[14].Intheseinstances,mito- pubic hair, were surveyed using a multiplex barcoded
chondrial DNA (mtDNA) analysis is routinely conducted. sequencing approachfrom seven healthy Caucasianindi-
However, due to its common matrilineal inheritance and viduals of both sexes (two of whom were in a de facto
haploidnature,mtDNAtypingyieldsmodestexclusionary relationship), ranging in age from 23 to 53 years old.
capability, which lacks the statistical power afforded by The health status of each volunteer was self-reported
STRprofiling[15].However,lowyieldsof humannuDNA with each individual stating that antibiotics were not
fromforensichairsamplesdoesnotequatetotheabsence taken at least 8 months prior to the collection of hairs
of othersourcesofDNAthatcouldassistin theindividu- used in the study. Each individual self-collected a num-
alisation of hair. Indeed, metagenomic analyses of hairs ber of hairs cut from the scalp and pubic areas, at three
unsuitable for nuDNA profiling may provide a microbial time points, initial collection in addition to 2 and
fingerprint to augment other forensic results such as 5 months thereafter, referred to as T0, T2 and T5 re-
mtDNA analyses. This would not involve extra or add- spectively. Replication is important in NGS amplicon
itionalextractionprocedures,asDNAisolationprocedures sequencing workflows—due to the investigative nature
forhumanDNAwillalso‘collect’microbialDNA. of this study and limited availability of resources, we
Conventional forensic hair examination, using either selected to investigate multiple time points (temporal
morphologicalormoleculartechniques,iscontingentupon replicates) in lieu of multiple extractions at each time
the deposition and recovery of hairs; however, despite points(samplingreplicates).
Locard’s adage that ‘every contact leaves a trace’ [16], this Each volunteer was provided with a hair collection kit
may not always be the case. Research in relation to the consisting of labelled clip-seal plastic bags, sterilised scis-
transfer of pubic hairs in forensic investigations involving sors, ethanol wipe, latex disposable gloves and disposable
sexualassaultcasesdiscoveredlimitedtransfer(4%)ofmale forceps.Hairswerecutclosetotheskinandfromapproxi-
pubic hair to female genital area during sexual intercourse mately the same area at each time point; volunteers were
(SI) [17]. In addition, the present study demonstrated that asked to clean the scissors with ethanol wipes between
no female pubic hair transfer to male genital area took samplingtoavoidcontamination.Therationaleofsevering
place. the hairs close to the skin, rather than plucking the hairs
The utility of metagenomic analyses for forensic appli- wastoensurethebacterialtaxaidentifiedweremorelikely
cations has been explored since the inception of NGS; to originatefrom the hair shaft rather than the skin. Hairs
for example, Fierer et al. [18] conducted preliminary taken from the head were labelled as female scalp hair
work to explore the potential to link individuals to com- (FSH)ormalescalphair(MSH);similarly,femaleandmale
puter keyboardsandmiceonthebasis oftransferofskin pubic hairs were marked female pubic hair (FPH) or male
bacteria. However, one of the most ubiquitous of evi- pubic hair (MPH). Once sampled, these hairs were placed
dence types—humanhair—hasyet tobe evaluatedin the in separate labelled clip-seal plastic bags and stored at
context of forensic metagenomics. To the best of our roomtemperatureafterbeingcatalogued.Hairsweresam-
knowledge, this present study is the first to qualitatively pled and processed within 24 h of collection, and unused
Tridicoetal.InvestigativeGenetics2014,5:16 Page3of12
http://www.investigativegenetics.com/content/5/1/16
hairsampleswerereturnedtotheiroriginalpackagingand Table1Detailsofprimersusedintheamplificationof
stored at room temperature; these hairs were not further 16SRNAregionofbacterialmitochondrialgenome
sampled or examined. The effect of storage and storage Primersequence(5′to3′) Reference
conditionsonbacteriawasnotinthescopeofthispresent Forwardbacterialprimer GTGCCAGCMGCCGCGGTAA Turneretal.
study;howeverLauberetal.[19]investigatedtheeffectsof (Bact_16S_F515) [20]
storageconditionsonbacteriaandconcludedthatbacterial Reversebacterialprimer GGACTACHVGGGTWTCTAAT Caporasoetal.
communitycompositionisunaffectedintheshort-term. (Bact_16S_R806) [21]
Each volunteer was made aware of the nature of the
study and gave written, informed consent. Information of extracted DNA with primer concentration at 0.4 μM
regarding the sexual habits or orientations of the volun- (IDT) cycled for 95°C for 5 min followed by 50 cycles of
teers was not sought. The project was approved by, and 95°C for 30 s, 55°C for 30 s, 72°C for 45 s, with 1°C melt
conducted in accordance with, Murdoch University Hu- step and a 10-min final extension at 72°C. The optimal
man Research Ethics Committee Policies and Guidelines DNAconcentrationfreeofinhibitionwasusedforallsub-
(ProjectNumber2011/139). sequent analyses. Each hair sample bacterial extract had
CTvalueslessthan32PCRcyclesindicatingthepresence
DNAextractionandquantification of sufficient 16S template copy number for robust NGS
Three hairs from each body area were cut into approxi- ampliconsequencing.
mately 1 cm lengths and placed into 1.5 ml Eppendorf
tubes. The contents of each tube were digested overnight Fusion-tagged16SV4Amplicongeneration
using 1 ml of hair digest buffer containing: 10 mM Tris Bacterial F515 and R806 (Table 1) 16S primers (targeting
pH 8 (Sigma, St. Louis, Mo, USA), 10 mM NaCl (Sigma), theV4region)usedintheinitialqPCRextractscreen,giv-
5 mM CaCl , (Sigma), 2.5 mM EDTA pH 8 (Invitrogen, ing a size variable product minus primers of ~250 base
2
Carlsbad,CA,USA),1mg/mlProK(Amresco,Solon,OH, pairs,weremodifiedintofusionprimersforthegeneration
USA), 40 mM DTT (Thermo Fisher Scientific, Waltham, of amplicon products for subsequent sequencing. Each
MA, USA) 2%SDS(Invitrogen)and milliQ water (Sigma) fusion primer consisted of a GS FLX Titanium (Lib-A)
to make the remaining volume. The samples, including adapter A or B on the 5′ end followed by a unique 6 bp
extraction and laboratory environmental controls, were multiplex identifier (MID) tag and the template specific
secured on rotating arms (to ensure total immersion) and forwardorreverseprimeratthe3′endoftheprimer[22]
digestedovernightina55°Coven. giving a final size variable product of ~350 bp including
All samples were then centrifuged for 2 min at primers and additions. A single-step, uniquely tagged fu-
13,000 rpm. To concentrate the DNA, a total of 600 μl of sion PCR approach was employed to minimise the con-
supernatantwastransferredtoVivaspinultrafiltrationspin taminationassociatedwiththemultiplePCRstepsusedin
columns with a 30,000 MW cutoff (Sartorius Stedim Bio- NGSworkflows[20].
tech, Göttingen, Germany) and centrifuged at 30,000 rpm Foreachtimepoint,eachextractwasassignedaunique
to leave 50–100 μl of supernatant. Concentrated super- 6 bp MID-tagged fusion primer in preparation for ampli-
natant was subsequently combined with five volumes of consequencing.MID-taggedampliconsweregeneratedin
PBbuffer(Qiagen,Valencia,CA,USA)andtransferredtoa triplicate (i.e. PCR replicates) in 25 μL reactions contain-
Qiagen silica spin column and centrifuged at 13,000 rpm ing: 1X PCR Gold Buffer (Applied Biosystems), 2.5 mM
for 1 min. Two wash steps followed (Qiagen AWI buffer MgCl (Applied Biosystems), 0.1 mg/ml BSA (Fisher Bio-
2
and AWII buffer) prior to elution of DNA from the spin tech,WA, Australia),0.25mM ofeachdNTP(Astral Sci-
column with 60 μl of 10 mM Tris-Cl pH 8 buffer. The entific, NSW, Australia), 0.4 μM of forward and reverse
DNA extracts were subsequently quantified via real-time primer (IDT), 0.2 μL1 unit of Taq DNA polymerase
quantitative polymerase chain reaction (qPCR; Applied (AmpliTaq Gold™), 1:80,000 (final concentration) of SYBR
Biosystems StepOne, Foster City, CA, USA) using SYBR Green ‘gel-stain’ (LifeTechnologies, S7563, Carlsbad, CA,
greenandthebacterial16SF515/Bact16S(V4loop)_R806 USA) and DNA extract. The same processes were per-
primers(Table1). formed on PCR negative/reagent controls for each PCR
Extracts were analysed using qPCR for neat extracts in platerun,pre-andpostadditionofDNAextracts.
addition to 1/10 and 1/100 dilutions, in order to deter- All PCR amplicons were purified using the Agencourt
mineifextractionsweresuccessfulandtoidentifysamples AMPure XP™ Bead PCR Purification protocol (Beckman
with low-template DNA (defined as those with CT Coulter Genomics, Danvers, MA, USA). Solely for the
values >32). The possible presence of PCR inhibitors purposeofsequencingcoverage,purifiedampliconswere
was also determined by qPCR. The 16S qPCR assay was electrophoresed on 2% agarose gel to obtain roughly
conducted in 25 μl reactions using a 2X ABI Power equimolar ratios of each sample. Where extraction/en-
SYBRmastermix(AppliedBiosystems)togetherwith2μl vironmental controls or PCR negative/reagent controls
Tridicoetal.InvestigativeGenetics2014,5:16 Page4of12
http://www.investigativegenetics.com/content/5/1/16
showed positive qPCR results, these were also pooled, Followingtaxonomic identification andPCoAconstruc-
purified and sequenced. The final pooled library was tion, the core microbiome for each sex/somatic origin
quantified using qPCR to determine the appropriate vol- (SSO) grouping (i.e. FSH, FPH, MSH and MPH) was de-
umeoflibrarytouse for emulsionPCR(emPCR) prior to termined using QIIME. The ‘core’ microbiome was de-
amplicon sequencing on the GS Junior™ (described in fined in accordance with that established by Shade et al.
Murray et al. [21], using reaction conditions in Murray [32],allOTUsthatoccurintwoormore(i.e.themajority)
et al. [23]). All emPCR, bead recovery and amplicon se- oftherecordedtimepointsforeachoftheSSOgroupings.
quencingprocedureswerecarriedoutaccordingtoRoche Any OTU’s occurring in only one of three time points is
GSJunior™protocolsforampliconsequencing(LibA). classed as ‘transient’ (Tr). In addition to this, the number
of OTUs that were unique to an individual was deter-
Bioinformaticanalysis mined;theseweredefinedasallOTUsoccurringsolelyin
Amplicon sequence reads obtained from the GS-Junior™ that individual across at least two time points irrespective
(hereafter referred to as sequences) were sorted into of whether it was found to be a core OTU in the above
batches based on MID tags assigned to each extract SSOgroupings.UponidentificationofpersonalisedOTUs,
allowing for no mismatch in MID tag DNA sequences. itwasdeterminedwhetherornotthesaidOTUsoccurred
Additionally, template-specific 16S bacterial primer se- in pubichair, scalp hair or in both.Finally, the numberof
quenceswereannotatedandtrimmedfromallsequences OTUs shared solely by two individuals was identified to
allowing for no mismatch in base composition or primer examine whether the number of OTUs between the
sequence length. Sequences that failed to meet these cri- cohabiting couple was greater in comparison to other
teria were discarded. The aforementioned steps were non-co-habitingparticipants.
conductedusingGeneious™v7.0.6[24].
Once batched and trimmed, sequence FASTA files were Results and discussion
importedintoQIIMEV1.8.0[25]andmergedintoa single Forty-two pools of DNA extracts obtained from human
FASTA file. Chimeric sequences were identified and re- scalpandpubichairswereusedtointerrogatetheirmicro-
movedonaperindividualsamplebasisusingtheusearch61 bialcompositionbynext-generationsequencing.Atotalof
[26]denovomethodpassing–split_by_sampleid.Following 79,766 reads were generated, yielding 39,814 reads post
this, operational taxonomic units (OTUs) were identified control and abundance filtering. On average,the coverage
using an open reference OTU picking method using per sample was 1,899 reads pre-filtered and 948 post-
usearch61 with a 97% clustering identity, using the most filtered. Whilst this depth of coverage is less than ideal
abundant sequence within each OTUas the representative given the advancement in NGS technology (i.e. Illumina
sequence and the Greengenes 13.8 database release [27]. and Ion Torrent platforms), these 454 data are still suffi-
RepresentativesequencesforeachOTUwerealignedusing cient to explore the potential of hair microbial forensics
PyNAST[28]againsttheGreengenes13.8pre-aligneddata- forfuturedevelopment.Likeallnovelforensictechniques,
base, the alignment filtered and phylogeny built using Fas- metagenomic analyses of hairs will ultimately require ro-
tTree[29]inQIIME.Additionally,anyOTUsfoundwithin bust evaluation and validation to ensure that these ana-
the controlsamplesofspecifictimepoints(i.e.T0,T2,T5) lyses are fit for purpose and able to withstand scientific
were removed from samples contained within the respect- scrutiny. Part of this validation should take into consider-
ive time point. Following the removal of control OTUs, ation: replication (spatial, temporal and PCR replicates);
eachindividualsamplewasfilteredtoremovelowabundant persistence of hair bacteria not only once they are trans-
OTU clusters. In each case, all singleton OTUs were dis- ferred or deposited (during contact and stability during
cardedandanyOTUwhoseabundancewasbelow0.2%,an storage) and prevention of contamination during pro-
estimatederrorrateassociatedwith454sequencing[30],of cessing hairs in the laboratory. Budowle et al. [33] out-
the total number of filtered sequences in that sample was line and discuss in detail the future validation criteria
removed. OTUs remaining post-filtering were taxonomic- for metagenomic analyses in relation to microbial
allyidentifiedusingtheBLASTnoptionwithinQIIME’sas- forensic applications, which they believe will require
signtaxonomyscriptagainsttheGreengenes13.8database. international participation. However, such an undertak-
Moreover, OTUs were taxonomically assigned using RDP ing is beyond the scope of this initial evaluation into
[31] and UCLUST [26] options, again against the Green- just one of many applications of forensic metagenomic
genes13.8database[27]. investigations.
To determine at a gross level if there was clustering of Therearemanywaystopresentmetagenomicdatasuch
samplesaccording tosexand/orsomaticorigin,aprinci- as generated here; the sections below explore the data
pal co-ordinate analysis (PCoA) plot was constructed using PCoA, taxonomy and OTU’s focusing on the value
using filtered sequences negating whether or not se- of the data in forensic applications. OTUs taxonomically
quenceswerepart ofthe coremicrobiome. assigned using RDP or UCLUST options revealed little to
Tridicoetal.InvestigativeGenetics2014,5:16 Page5of12
http://www.investigativegenetics.com/content/5/1/16
no difference in assignment to the rank of family. For individuals sharing living or communal spaces [34] or
this reason, all assignments refer to BLAST taxonomic duringcontactsportsinwhichMeadowetal.[35]observe
assignments. ‘Our results are consistent with the hypothesis that the
human skin microbiome shifts in composition during
Principalcoordinatesplot activities involving human to human contact’. The results
Of all the data generated in this study, the NGS data we present here suggest that the pubic hair microbiome
generated from pubic hair held the most potential for might be quite stable, even during cohabitation, but it
forensic applications. A general dichotomy was observed might be shifted dramatically during sexual intercourse
between taxa (OTUs) harboured on male and female for some time. This present study is the first to suggest
pubichair shafts(Figure 1). cross-transference of pubic/genital microbial taxa as a
In general, males were clustered close to the PC2 axis resultofintercourse.Althoughfurtheranalysesneedtobe
along the PC1 axis whilst females were more evenly conducted, this initial finding bodes well for future foren-
spread along the PC1 axis and further from the PC2 axis sicapplicationsinvolvingsexualcrimes.
than the males. Data relating to two individuals, who An additional advantage is that compared to other
were a cohabiting couple, presented some interesting body areas such as the skin, gastrointestinal tract (GIT)
results. The red dots in the yellow ellipse at high PC1 and mouth, fewer bacterial species seem to comprise the
represents the taxa present on the female partner of the vaginal microbiome [36]. The advantage of simpler com-
coupleatT0andT2whilstthetwoorangedotsenclosed munities and fewer taxa in the vaginal microbiome is
by small blue circles at low PC1 represent the taxa from one that may facilitate forensic investigations by provid-
the male partner at T0 and T2. The lilac circle encloses ingresultsinatimely manner.
one red dot(taxa from thefemaleat T5) andone orange Theclearmicrobialdistinctionsbetweenpubichairsfrom
dot (taxa from the male at T5). Microbial taxa extracted the sexes may largely be attributable to the prevalence of
from the male and female at this time point were more Lactobacillusspp.inthefemalepubichairsamplesandthe
similar to each other than to their other previous time absence of these bacteria in the male samples (excepting
points (T0, T2). Discreet enquiries revealed, unlike the the co-habiting male at T5) (Figure 1). Additionally, male
preceding time points; the couple in question had en- pubic hair microbial taxa were clustered along axis PC2
gaged in sexual intercourse prior to the collection of T5 suggesting that these taxa (OTUs) were common to the
hair samples. It is noteworthy that intercourse had taken malemicrobiota.
place 18hpriortothe collectionofpubic hairsandboth In contrast, female pubic hair bacterial taxa showed
individuals had showered in the interim period. Cross- elongation along axes PC1 and PC2. The elongation of
transference of bacteria during intercourse may account dataalongPC1maybeattributabletofemalesharbouring
for the variation in taxa observed. Cross-transference, differentlactobacillispecies(Tables2and3).However,the
or shedding of skin micro flora, is not uncommon for concomitant elongation of data along axis PC2 suggests
Figure1Principalcoordinateplots(PCoA).Clusteringofmicrobialtaxafromeachindividualateachcollectiontimepoint.Thelilaccirclerepresents
post-SIbacterialsequences,whilstthepaleblueandyellowcirclesrepresentnon-SIbacterialsequences—bothcirclesrelatesolelytotheco-habiting
couple.PanelArepresentspubichairmicrobialtaxafrommale(orange)andfemale(red)participants.PanelBrepresentsscalphairmicrobialtaxafrom
male(green)andfemale(blue)participants.PanelCrepresentsmicrobialtaxapresentinmaleandfemalescalpandpubichairsamples.
Tridicoetal.InvestigativeGenetics2014,5:16 Page6of12
http://www.investigativegenetics.com/content/5/1/16
Table2Sharedtaxafrompairwisecomparisonsofalldatalocatedinscalp(Sc)and/orpubic(Pu)hairs
Individual 1(female) 2(male) 3(male) 4(female) 5(female) 6(female) 7(male)
1(Female) Xanthomonadaceae Dialister Lactobacillusiners 0 0 0
(Sc) spp.(Pu) (Pu)
Prevotellaspp.(Pu)
Peptinophilusspp.
(Sc/Pu)
2(male) 0 0 Betaproteobacteria Actinomycetales(Pu), Dietziac(Sc)
(Sc/Pu) Neisseriaceae(Sc)
Neisseriaceae(Sc) Mycoplanaspp. Knoelliasubterranea
(Sc/Pu) (Sc/Pu)
Corynebacteriaceae
(Pu)
Lactobacillusspp.
(Pu)X4
3(male) Bifidobacterium 0 0 Corynebacteriaceae
(Pu/Sc) (Sc)
Anaerococcus(Pu) Paracoccus(Sc/Pu)
4(female) Lactobacillusspp. Limnohabitans 0
X2(Pu) spp.(Sc)
Rhodobacteriaceae
(Sc)
5(female) 0 0
6(female) Corynebacterium
spp.(Sc/Pu)
Aggregibacter
segnis(Pu)
7(male)
Thecohabitingcouple(bold)sharemoretaxa,includingmultiplelactobacillispecies,thanotherindividuals.
the presence of secondary differences, differences that Therefore, in general, pubic hair microbiomes appear to
maybeduetothepresenceofpersonalisedtaxa(Table3). be less influenced by environmental bacteria than scalp
The PCoA plot of male and female scalp hair micro- hairs and possible harbour more niche specific bacteria.
biota over the 5-month time period did not demonstrate Zhou et al. [37] support this premise by demonstrating
any significant clustering (Figure 1). This is most likely that (in comparison to other areas of the body) vaginal
attributable to male and female scalp hairs harbouring microbiotaconsistedoflessstablebacteria(i.e.moretran-
similar bacterial taxa. However, some of the female taxa sient bacteria) and showed lower alpha diversity (i.e. low
are slightly spread out along axis PC1 suggesting that species richness), supporting the premise of pubic hair
there may be some variation in microbial taxa in the harbouringnichespecificbacteria.
hairs of these individuals. The distribution and compos-
ition of the microbial communities colonising scalp and
pubichair isdiscussedinfurtherdetail below. Pubichairmicrobiota
Male pubic hairs could be readily distinguished from
female pubic hairs on the basis of their respective micro-
Hairmicrobiota biota. Lactobacillus was the most prevalent taxon that
Bacteria colonising male and female scalp and pubic hair clearly differentiated male and female pubic hair micro-
samples are classed as either ‘core’ or transient (Tr) bac- biota (Figure 2). Whilst the prevalence of Lactobacillus
teria (Figure 2, see the ‘Methods’ section). In relation to spp.inthe vaginaand vaginalsecretionsiswellestablished
the numberof OTUs extracted from scalp and pubic hair [38-40], this present study is the first to discuss these bac-
microbiomes, far less bacterial sequences were lost post teria colonising pubic hairs, and general pubic area, in the
control filtering for pubic microbiomes in comparison to context of probative value in forensic investigations. Flem-
scalp hair. Pubic hairs in general contained more OTUs ingandHarbison[41]suggestedthepresenceoftwoLacto-
than scalp hair (approximately 50 male OTUs/55 female bacillus spp. (Lactobacillus crispatus and Lactobacillus
for scalp hairs c.f. approximately 73/76 for pubic hairs). gasseri) as suitable forensic markers to identify vaginal
Tridicoetal.InvestigativeGenetics2014,5:16 Page7of12
http://www.investigativegenetics.com/content/5/1/16
Table3Personalised(unique)bacterialtaxacolonisingmaleandfemalescalpandpubichair
Somaticorigin Bacterialtaxa Naturalhabitat
(sex/individual)
Pubic(F.Ind.1) Lactobacillusspp.×1a(Lactobacillaceae) Mostprevalentgenerainhumanvagina;
alsopresentingastro-intestinal
tract(GIT)[39]
Scalp(F.Ind.1) Neisseriaceae Normalhumanfloraof
oro-nasopharynx[39]
Scalp(M.Ind.2) Nocardioidaceae Soilandaquatichabitats[44]
Scalp(M.Ind.2) Streptococcussobrinus(Streptococcaceae) Implicatedindentalpathologies[39]
Pubic(M.Ind.3) Corynebacterium×2spp.a(Corynebacteriaceae) Majorinhabitantsofskinflora[39]
Pubic(M.Ind.3)spp. Tissierellaceaefam.nov. Intestine,vagina,oralcavity;
somegeneralenvironmental[46]
Pubic(M.Ind.3) Anaerococcusspp.a(Tissierellaceaefam.nov) Humannasalcavity,skinandvagina[47]
Pubic(F.Ind.4) Campylobacteraceae Humanoralcavity,intestinaltractsand
environmental[48]
Scalp(F.Ind.5) Rhodocylcaceae Activatedsludgeandwastewater[49]
Scalp(F.Ind.5) Micrococcaceae Widelydistributedinenvironment;
commensalonhumanskin[39]
Pubic(F.Ind.5) Lactobacillus×3spp.a(Lactobacillaceae) Prevalentandabundantinhumanvagina;
alsopresentinGIT[39]
Pubic(F.Ind.6) Gardnerellaspp.a(Bifidobacteriaceae) Implicatedinbacterialvaginosis(BV)[39]
Pubic(F.Ind.6) Enterobacteriaceae Manyspeciesharmlesssymbiontsof
thehumanGIT[39]
Pubic(F.Ind.6) Methylobacteriaceae Ubiquitousinnature,opportunistic
pathogens[50]
Pubic(F.Ind.6) Pasteurellaceae Respiratory,alimentaryandreproductive
tracts[51]
Pubic(F.Ind.6) Pseudomonasspp.a(Pseudomonadaceae) Transientsonhumanskin[39]
Pubic(M.Ind.7) Aurantimonadaceae Marineenvironment[52]
Scalp(M.Ind.7) Brachybacteriumspp.a(Dermabacteraceae) Varietyofenvironments[53]
Scalp/Pubic(M.Ind.7) Gordoniaceae Majorityenvironmental[54],some
humanpathogens[55]
Scalp/Pubic(M.Ind.7) Rhodobacteriaceae Aquatichabitats[56]
Scalp/Pubic(M.Ind.7) Sphingomonadaceae Widespreadinnature;presentinwater[57,58]
Datashowssexandsomaticoriginofhairsthatharbouredpersonalisedbacterialtaxa,aswellasthenaturalhabitatsofthetaxa.
aBacteriaidentifiedtogenusorspeciesareregardedasthemostdiscriminatory.Assuch,thesetaxamaybeofsignificanceinforensicinvestigations.
secretions.However,microbialdatagarneredinthispresent maybeattributabletolactobacilliconferring‘antimicrobial
study suggest that a NGS metagenomic approach may be protection’ to the vagina by preventing colonisation by
preferable to those that target specific species. The variety other microorganisms [38]. Li et al. [42] also found that
of Lactobacillus spp. detected in pubic hairs from the in comparison to other body areas, the vaginal micro-
female cohort consisted of 11 OTUs (taxa) in total; three biome is less transient (i.e. more stable). This stability
Lactobacillusspp.wereuniquetoFemale5,oneLactobacil- was apparent in the differences between the number of
lus spp. occurred in Female 1, and four Lactobacillus spp. OTUs detected in the scalp and pubic hair controls;
were uniquely between the cohabiting couple. In addition, there were significantly less OTUs present in controls
two Lactobacillus spp. were uniquely shared between F4 from the pubic hairs in comparison to the scalp hairs.
andF5,andoneOTUwasuniquelysharedbetweenF4and Post control filtering for FSH and MSH samples there
F1(Tables2and3). were33%and43%(respectively)ofsequencesleft.Incom-
Compared to male pubic hairs, female pubic hairs parison, for FPH and MPH there were 70% and 72%
harboured fewer transientbacteria (Figure 2); the number (respectively) sequences left, post-filtering. The disparity
of bacterial sequences comprising transient bacteria of betweenthetwosomaticoriginssuggeststhatthebacterial
female pubic hairs was approximately half the number of taxa in scalp hair extracts had a high proportion of
those found in male pubic hair (Table 4). This disparity environmentalbacteriathatreadilyappearincontrols.
Tridicoetal.InvestigativeGenetics2014,5:16 Page8of12
http://www.investigativegenetics.com/content/5/1/16
Figure2Microbialdataextractedfromscalpandpubichairs.Diagramsillustratingcoreandtransient(Tr)bacterialtaxaonmaleandfemale
scalpandpubichairsamples.
Scalphairmicrobiota
In contrast to the pubic hairs, scalp hair microbiota
showednocorrelationwiththesexofthedonor(Figure2).
Table4Numberofbacterial16Ssequencesfoundincore
Maleandfemalescalphairbacterialtaxaconsistedofnor-
microbiomescomparedwithtransientnumberof
mal human skin commensals, e.g. Anaerococcus spp., and
sequencesforeachsex/somaticoriginsampled
environmentally derived taxa, e.g. Knoellia subterranea,
Sex/somaticorigin Core Transient Total
bacteria bacteria bacteriaa many of which occurred in both male and female samples
(Table 5). In the present study, the most significant differ-
Femalescalphair 3,123 2,162 5,285
ence observed in male and female scalp hairs was the dis-
Femalepubichair 16,019 1,524 17,543
parateproportionsofthetransientbacterialtaxa(Figure2).
Malescalphair 4,838 1,220 6,058
Almosttwiceasmanytransientbacterialtaxawerepresent
Malepubichair 8,109 2,819 10,928 infemalescalphaircomparedtomales(Table4).Thismay
Femalepubichairharbouredlesstransientbacteria,butmorecorebacteria be due to the greater frequency of females grooming and/
thanmalepubichair. or washing and/or dyeing or bleaching their hair in com-
aTotalnumberofsequencesfoundtoremainoncesequencesfoundin
controlsandlowabundantclusterswereremoved. parison to males. Such grooming practices may prevent
Tridicoetal.InvestigativeGenetics2014,5:16 Page9of12
http://www.investigativegenetics.com/content/5/1/16
Table5Naturalhabitatsofsharedbacterialtaxaidentifiedbypairwisecomparisons
Sharedbacterialtaxa Habitat
Aggregibactersegnis(Pasteurellaceae) Normalhumanoralflora[51]
Anaerococcusspp.(Tissierellanov.fam.) Commensalhumanflora,opportunisticpathogensalsoenvironmental[46,47]
Betaproteobacteria Highordertaxon(Class).Memberslargelyenvironmental,alsoincludehumanpathogensand
commensals[60]
Bifidobacteriumspp.(Bifidobacteriaceae) Normalintestinalflora[39]
Corynebacterium×1spp.(Corynebacteriaceae) Keymembersassociatedaspartofskinflora[39]
Dialisterspp.(Veillonaceae) Implicatedinoralcavitydiseases[39]
Dietziaceae Environmentalandimplicationsasanemerginghumanpathogen[48]
Knoelliasubterranea(Intrasporangiaceae) Environmental[61]
LactobacillusinersandLactobacillusspp.×6 Partofasuiteoflactobacillithatinhabithumanfemalegenital-urinaryarea[40]
(Lactobacillaceae)
Mycoplanaspp.(Caulobacteraceae) Environmental[62]
Neisseriaceae Membersmaybehumancommensals(ofthemouth)orpathogenic[39]
Paracoccus(Rhodobacteriaceae) Environmental[49]
Peptinophilusspp.(Tissierellafam.nov.) Pathogenic[47]
Prevotellaspp.(Prevotellaceae) Oral,vaginalandGITcommensals[39]
Xanthomonadaceae Plantpathogen(opportunistichumanpathogen)[59]
ThetaxaandhabitatsrelatetothedataprovidedinTable2.
establishmentofmorestablebacterialcoloniesinfavourof Insidetheconfinesofthe16SV4region,withtheexcep-
less stable (transient) bacterial colonies. Irrespective of the tion of one male (co-habiting male at T5), all individuals
cause of this disparity, this observation cannotbe regarded harboured unique taxa on their pubic hairs (Figure 3). In
significantinrelationtoforensicinvestigations. additiontopersonalisedbacteriathatwerepartofthenor-
Costello et al. [3] identified two dominant 16S se- mal skin flora, e.g. Corynebacteriaceae, pubic hairs were
quences from scalp swabs: Propionibacterinae in which also colonised by environmentally derived bacterial taxa,
members are predominant bacteria in hair follicles and e.g.Methylobacteriaceae(Table3).
other sebaceous sites [39] and Streptophyta (a plant Hairs from scalp and pubic regions, for both sexes, in-
phylum). In contrast, the predominant bacterial taxa cludedsharedtaxathatarecommoninhabitantsofhuman
from hair shafts in this study were Corynebacteriaceae skin or scalp, e.g. Corynebacteria, or were environmental
and Tissierellacea fam.nov (‘new family’) (Figure 2). The in origin, e.g. Rhodobacteriaceae (Tables 2 and 5). At first
difference may be attributable to either environmental glance, the commonality of these bacteria may appear to
differences (i.e. different study sites) or the collection beofminimalprobativevalue;asdiscussedinapreceding
technique employed by Costello et al. [3] where swab- section, personalised features should be uncommon traits
bing the top of the head might have favoured the or features. However, common bacteria may harbour
removal of scalp/follicular bacteria (i.e. propionibacteria singlenucleotidepolymorphisms (SNPs)within their gen-
rather thanhair shaftbacteria). ome,whichmayfurtherdiscriminatebetweenindividuals.
Among all mammals, the microbiota composition is
extensivelyconservedatthehightaxonomiclevelssuchas
Personalisedandsharedbacterialtaxa phylum or class. At these taxonomic levels, humans are
Forensic investigations seek to establish ‘common origin’ very similar to each other (and other mammals) but vari-
or‘sourceattribution’ofevidence,thatis,toestablishwith ationincreasesprogressivelyatthelowertaxonomiclevels.
reasonable scientific certainty that a particular individual Personalised taxa, whichallowdiscriminationbetweenin-
is the source of an evidentiary sample. In relation to bio- dividuals(thegoalofforensicapplications),arelikelytobe
logical evidence, this question may be addressed through detected at these lower taxonomic rankings. Personalised
the detection of individualising biological characteristics, taxa may be present in high or low abundance; detection
for example, a human DNA profile, characteristics which oflowabundancetaxamayonlybedetectedbyultra-deep
excludes other individuals as being the source. Ideally, sequencingoftheextractedbacterialDNA.Inthisregard,
these characteristics should not commonly occur within higher depthofcoverage affordedbyNGSplatforms such
thegeneralpopulationoronethatissolelyfoundinmales as Ion Torrent or Illuminamay be moreinformative than
orfemales. the 454 data presented here. As Ursell et al. [43] noted ‘it
Tridicoetal.InvestigativeGenetics2014,5:16 Page10of12
http://www.investigativegenetics.com/content/5/1/16
accurate and capable of withstanding scientific and legal
scrutiny.
Temporal stability data garnered in this study broadly
suggest that bacteria on scalp hairs may be more prone
to fluctuations in comparison to pubic hairs (in addition
to being more prone to environmental contaminants).
The data shows that, on average and post-filtering, ap-
proximately 17% (range 6%–25%) of pubic hair bacterial
OTUs were temporally stable across all time points;
whilst, on average, scalp hair harbour approximately 5%
(range 0%–13%) of bacterial OTUs (Table 6). These pre-
liminary findings suggest that pubic hair bacteria may be
more temporally stable than scalp hair bacteria and
therefore potentially of more probative value than scalp
hairbacteria.
Although temporal stability of an individual’s bacterial
taxa may appear to be an important prerequisite for
metagenomics to have forensic value, the most relevant
attributes will mostly likely be transference of bacteria
(during contact), persistence of bacteria post transfer
and storage conditions. Consider a case of unlawful sex-
ual intercourse (of an adult female), the most relevant
microbial data will be the taxa available for transfer at
the time of the assault (rather than what it was weeks,
months or days before or after) and the persistence of
the victim’s bacteria on the offender’s genitals/pubic area
(and vice versa). This, of course, is reliant upon collec-
tion of evidence from the victim and suspect(s) within
several hours of the time of the assault rather than
several days. Microbial data from the cohabiting couple,
albeit preliminary, are encouraging, in supporting the
suggestion of bacterial transfer and persistence following
sexual intercourse.
Conclusions
Despite the modest sample size, we believe that the data
in this qualitative assessment of metagenomic analyses
Figure3Personalisedmicrobialdata.Diagrammaticsummaryof
of hairs are sufficient to warrant further development of
uniquebacterialtaxafoundinmaleandfemalescalpandpubichair
samples.Maleindividualslefttoright:IndividualsM2(cohabiting this approach. For this approach to gain traction, there
male),M3andM7.Femaleindividualslefttoright:IndividualsF1,F4, is a need to refine molecular targets—the broad-brush
F5(cohabitingfemale)andF6. approach of the 16S V4 region looked at here is a good
starting point. Additional analyses may provide further
is important to realise that sampling depth may be critical information in relation to the microbial composition of
for distinguishing taxa that are absent from those that are ‘core’ microbiomes and their potential value in forensic
merelyrare’.Underthesecircumstances,itiscriticaltodis- investigations. However, there is ultimately a need to de-
countbacterialtaxapresentinallcontrolsamplesinorder velop amorefocused approach that targets, for example,
for the results to be not only robust but also scientifically population level differences within Lactobacillus spp. or
Table6Temporalstabilityofbacterialtaxa(OTUs)
FSH1 FPH1 MSH2 MPH2 MSH3 MPH3 FSH4 FPH4 FSH5 FPH5 FSH6 FPH6 MSH7 MPH7
TotalnumberofOTUs 28 18 31 36 19 45 21 32 25 32 20 39 27 36
TemporallystableOTUs 0 3 4 8 1 9 0 8 1 6 0 3 3 2
%TemporallystableOTUs 0 17 13 22 5 20 0 25 4 19 0 8 11 6
Bacterialtaxapresentinscalpandpubichairssampledatthreetimepointsoveraperiodof5months.