Dis inguishing he Signals o Gingi i is and Pe iodon i is in
Sup agingi al Plaque: a C oss-Sec ional Coho S udy in Malawi
Liam Shaw,
a,b
Ulla Ha junmaa,
c
Ronan Doyle,
a
Simeon Mulewa,
d
Da ie Cha lie,
d
Ken Male a,
d
Robin Calla d,
a
A. Sa ah Walke ,
e
F ancois Balloux,
Pe Asho n,
c,g
Nigel Klein
a
Ins i u e o Child Heal h, UCL, London, Uni ed Kingdom
a
; Cen e o Ma hema ics and Physics in he Li e Sciences and Expe imen al Biology, UCL, London, Uni ed
Kingdom
b
; Cen e o Child Heal h Resea ch, Uni e si y o Tampe e and Tampe e Uni e si y Hospi al, Tampe e, Finland
c
; Uni e si y o Malawi College o Medicine,
Blan y e, Malawi
d
; MRC Clinical T ials Uni a UCL, London, Uni ed Kingdom
e
; UCL Gene ics Ins i u e, UCL, London, Uni ed Kingdom
; Depa men o Paedia ics, Tampe e
Uni e si y Hospi al, Tampe e, Finland
g
ABSTRACT
Pe iodon al disease anges om gingi al in lamma ion (gingi i is) o he in lamma ion and loss o oo h-suppo ing issues
(pe iodon i is). P e ious esea ch has ocused mainly on subgingi al plaque, bu sup agingi al plaque composi ion is also
known o be associa ed wi h disease. Quan i a i e modeling o bac e ial abundances ac oss he na u al ange o pe iodon al se-
e i ies can dis inguish which ea u es o disease a e associa ed wi h pa icula changes in composi ion. We assessed a c oss-
sec ional coho o 962 Malawian women o pe iodon al disease and used 16S RNA gene amplicon sequencing (V5 o V7 e-
gion) o cha ac e ize he bac e ial composi ions o sup agingi al plaque samples. Associa ions be ween bac e ial ela i e
abundances and gingi i is/pe iodon i is we e in es iga ed by using nega i e binomial models, adjus ing o epidemiological ac-
o s. We also examined bac e ial cooccu ence ne wo ks o assess communi y s uc u e. The main di e ences in sup agingi al
plaque composi ions we e associa ed mo e wi h gingi i is han pe iodon i is, including highe bac e ial di e si y and a g ea e
abundance o pa icula species. Howe e , e en a e con olling o gingi i is, he p esence o subgingi al pe iodon i is was as-
socia ed wi h an al e ed sup agingi al plaque. A small numbe o species we e associa ed wi h pe iodon i is bu no gingi i is,
including membe s o P e o ella,T eponema, and Selenomonas, suppo ing a mo e complex disease model han a linea p o-
g ession ollowing gingi i is. Cooccu ence ne wo ks o pe iodon i is-associa ed axa clus e ed acco ding o pe iodon i is ac oss
all gingi i is se e i ies. Species including Fili ac o alocis and Fusobac e ium nuclea um we e cen al o his ne wo k, which sup-
po s hei ole in he coagg ega ion o pe iodon al bio ilms du ing disease p og ession. Ou indings con i m ha pe iodon i is
canno be conside ed simply an ad anced s age o gingi i is e en when only conside ing sup agingi al plaque.
IMPORTANCE
Pe iodon al disease is a majo public heal h p oblem associa ed wi h o al bac e ia. While ea lie s udies ocused on a small num-
be o pe iodon al pa hogens, i is now accep ed ha he whole bac e ial communi y may be impo an . Howe e , p e ious high-
h oughpu ma ke gene sequencing s udies o sup agingi al plaque ha e la gely ocused on high-income popula ions wi h good
o al hygiene wi hou including a ange o pe iodon al disease se e i ies. Ou s udy includes a la ge numbe o low-income pa -
icipan s wi h poo o al hygiene and a wide ange o se e i ies, and we we e he e o e able o quan i a i ely model bac e ial
abundances as unc ions o bo h gingi i is and pe iodon i is. A signal associa ed wi h pe iodon i is emains a e con olling o
gingi i is se e i y, which suppo s he concep ha , e en when only conside ing sup agingi al plaque, pe iodon i is is no sim-
ply an ad anced s age o gingi i is. This sugges s he u u e possibili y o diagnosing pe iodon i is based on bac e ial occu -
ences in sup agingi al plaque.
Pe iodon al disease is a majo public heal h p oblem, pa icu-
la ly in low-income se ings like sub-Saha an A ica (1). Aside
om i e e sible oo h loss, ch onic pe iodon i is may also in-
c ease he isk o ad e se sys emic condi ions (2), such as ca dio-
ascula disease (3) and p e e m bi h; howe e o p e e m bi h,
di e en s udies ha e epo ed con lic ing esul s (4). The associ-
a ion be ween pe iodon i is and sys emic disease may be due o
bo h inc eased sys emic in lamma ion and o ansloca ion o
bac e ia in o he bloods eam (5). Despi e i s impo ance, he mi-
c obial ecology o pe iodon al disease in di e en o al habi a s
emains incomple ely unde s ood. S udies o he o al mic obiome
in pe iodon al disease ypically ocus on small popula ions in de-
eloped coun ies wi h ad anced den al heal h ca e sys ems,
which may no be ep esen a i e o he na u al his o y o pe i-
odon al disease in he absence o ea men (6).
In pe iodon al disease, he immune sys em esponds wi h in-
lamma ion o o al bio ilms (7). A e an ini ial ocus on iden i y-
ing pa icula pe iodon al pa hogens (8), i is now widely accep ed
Recei ed 9 June 2016 Accep ed 25 July 2016
Accep ed manusc ip pos ed online 12 Augus 2016
Ci a ion Shaw L, Ha junmaa U, Doyle R, Mulewa S, Cha lie D, Male a K, Calla d R,
Walke AS, Balloux F, Asho n P, Klein N. 2016. Dis inguishing he signals o
gingi i is and pe iodon i is in sup agingi al plaque: a c oss-sec ional coho s udy
in Malawi. Appl En i on Mic obiol 82:6057–6067. doi:10.1128/AEM.01756-16.
Edi o : A. M. Spo mann, S an o d Uni e si y
Add ess co espondence o Liam Shaw, [email p o ec ed].
Supplemen al ma e ial o his a icle may be ound a h p://dx.doi.o g/10.1128
/AEM.01756-16.
Copy igh © 2016 Shaw e al. This is an open-access a icle dis ibu ed unde he
e ms o he C ea i e Commons A ibu ion 4.0 In e na ional license.
c ossma k
Oc obe 2016 Volume 82 Numbe 19 aem.asm.o g 6057Applied and En i onmen al Mic obiology
on Oc obe 17, 2016 by TAMPERE UNIVERSITY LIBRARY DEPARTMENT OF HEALTH SCIENCESh p://aem.asm.o g/Downloaded om
ha o al bac e ial communi ies unde go a shi o dysbiosis (9)
and ha he p esence o pa icula disease-associa ed species may
exace ba e he in lamma o y eac ion o commensal bac e ia (10).
The wo main ea u es o pe iodon al disease a e gingi al in lam-
ma ion (gingi i is) and he o ma ion o pe iodon al pocke s
(pe iodon i is). While i is clea ha gingi i is always p ecedes
pe iodon i is (11), gingi i is does no always p og ess o pe io-
don i is (12), sugges ing ha hese condi ions may no simply
ep esen di e en s ages o a con inuous spec um o disease.
While he e is some e idence ha a s eady con inuous p og ession
may be expec ed (13), mos models in ol e acu e bu s s o exac-
e ba ion and longe pe iods o emission (14,15).
Despi e his knowledge, s udies o o al bac e ia in pe iodon al
disease o en ail o cap u e he ull ange o pe iodon al condi-
ions, om heal h h ough gingi i is o pe iodon i is. In sup ag-
ingi al plaque, in pa icula , compa ing only heal hy subjec s wi h
subjec s su e ing om pe iodon i is may lead o bac e ial associ-
a ions being a ibu ed o pe iodon i is alone, despi e he ac ha
hey may also be p esen in subjec s wi h gingi i is. To explain he
p og ession o disease and iden i y ac o s uniquely a ibu able o
pe iodon i is, i is necessa y o compa e subjec s ac oss he ull
ange o pe iodon al se e i ies. In i sel , his is no a no el concep ,
wi h many p e ious s udies in es iga ing bac e ial associa ions
wi h disease using checke boa d DNA-DNA hyb idiza ion (16–
18). Ea lie s udies we e a ge ed a a small numbe o bac e ial
species ( ypically a ound 40). The ad en o high- h oughpu 16S
RNA gene amplicon sequencing has acili a ed he imp o ed
analysis o he o al bac e ial di e si y in he o al ca i y (19,20),
iden i ying a ound 1,000 species ha may be p esen (10) and
showing ha samples om he mou h ypically ha e highe alpha
di e si ies han hose om o he body si es (21,22). Recen s ud-
ies ha e used such amplicon sequencing o cha ac e ize subgingi-
al plaque ac oss a ange o pe iodon al condi ions, inding di -
e ences be ween subjec s wi h gingi i is and pe iodon i is (23,
24). Wo k on sup agingi al plaque has been less common due o
he ac ha i does no ha e a di ec link o in lamma ion and he
subsequen loss o a achmen in pe iodon i is. I he e o e e-
mains ambiguous whe he , o sup agingi al plaque, pe iodon i-
is can be simply conside ed an ad anced s age o gingi i is o i
he e a e de ec able di e ences in bac e ial composi ion.
To add ess his ques ion, we in es iga ed bac e ial abundances
in sup agingi al plaque using quan i a i e modeling ha akes
in o accoun gingi i is (quan i ied by bleeding on p obing [BoP])
and pe iodon i is (quan i ied by pe iodon al pocke dep h) in a
c oss-sec ional coho o 962 Malawian women who had ecen ly
gi en bi h (25).
We used nega i e binomial models ha we e o iginally de el-
oped o ansc ip ome sequencing (RNA-seq) expe imen s (26),
making use o absolu e (i.e., unno malized) ead coun s o a oid
losing in o ma ion—a downside o o he s a is ical app oaches
applied o ma ke gene da a like a e ying (27). A e i ing a
nega i e binomial dis ibu ion o he coun da a o a gi en spe-
cies, he mean o his dis ibu ion was hen used as he ou pu o a
gene alized linea model wi h a loga i hmic link using expe imen-
al a iables (e.g., disease se e i y) as inpu s, which allowed he
iden i ica ion o di e en ially abundan species. This app oach
conside s bac e ial species o be independen , bu in eali y, o al
bac e ia exis in complex polymic obial bio ilms (28). The e o e,
we also applied a cooccu ence analysis o pe iodon i is-associ-
a ed bac e ia o iden i y impo an membe s o he communi y.
In summa y, we aimed o iden i y he e ec s o pe iodon i is
on sup agingi al plaque a e con olling o gingi i is se e i y,
sepa a ing and dis inguishing he signals o hese wo ea u es o
pe iodon al disease.
MATERIALS AND METHODS
S udy popula ion. Women who we e analyzed in his s udy we e pa ic-
ipan s in he iLiNS-DYAD-M ial (In e na ional Lipid-Based Nu ien
Supplemen s s udy g oup, en olling mo he -child dyads in Malawi;
ClinicalT ials egis a ion no. NCT01239693) (25). This was a andom-
ized con olled ial ha in es iga ed he e ec s o he ollowing h ee
nu i ional supplemen s on bi h ou comes: lipid-based nu ien supple-
men (LNS), mul iple mic onu ien s (MMN), o i on- olic acid (IFA).
Women we e eligible o en ollmen in he ial i hey we e p egnan o
⬍20 weeks, ⬎14 yea s old, had no ch onic illnesses equi ing equen
medical ca e, had no alle gies, had no e iden p egnancy complica ions
(edema, blood hemoglobin o ⬍50 g/li e , sys olic blood p essu e o ⬎160
mm Hg, o dias olic blood p essu e o ⬎100 mm Hg), no ea lie pa ici-
pa ion in he same ial, and no concu en pa icipa ion in any o he ail.
A o al o 1,391 p egnan women we e en olled be ween Feb ua y
2011 and Augus 2012 a an ena al clinics a wo hospi als (Mangochi and
Malindi) and wo heal h cen e s (Lungwena and Namwe a) in Mangochi
dis ic , Malawi. All women we e sel - epo ed nonsmoke s and we e
gi en wo cou ses o p e en i e mala ia ea men wi h sul adoxine-py i-
me hamine (SP; h ee able s o 500 mg sul adoxine and 25 mg py ime h-
amine o ally), one a en ollmen and one be ween he 28 h and 34 h
ges a ional week. A e gi ing bi h, 1,229 women comple ed an o al
heal h examina ion, consis ing o a clinical examina ion and a pano amic
X- ay o he jaws. A o al o 1,024 women had his examina ion wi hin 6
weeks o deli e y o a single in an (mo he s o wins we e excluded) and
we e included in u he analysis. A e excluding women wi hou a su-
p agingi al sample (n⫽59) and hose wi h an unknown HIV s a us (n⫽
3), 962 women emained o ou c oss-sec ional analysis.
Classi ica ion o pe iodon al disease. Gingi i is was measu ed by he
numbe o den al a ch sex an s wi h bleeding on p obing (BoP) ou o six,
wi h h ee sex an s on each jaw (le , middle, and igh ). Fo pe iodon i is
classi ica ion, each oo h was examined o e idence o deepened den al
pocke s, bo h clinically and adiologically. A oo h was de ined as ha ing
pe iodon i is i ei he a ⱖ4-mm pocke was measu ed in clinical exami-
na ion o a e ical bony pocke was iden i ied a leas a he ce ical oo
le el adiologically. A woman was de ined as ha ing pe iodon i is i she
had a leas h ee ee h wi h pe iodon i is o a leas one den al a ch sex-
an wi h ho izon al bone loss (a leas a he ce ical le el). The exami-
na ion and classi ica ion me hods a e explained in de ail elsewhe e (29).
Sample collec ion. Sup agingi al den al plaque samples we e col-
lec ed by swabbing he gingi al ma gin o each oo h wi h a s e ile plas ic
swab s ick wi h a nylon ibe ip (mic oRheologics no. 552; Coban,
B escia, I aly). A e ans e in a cold box wi h ice packs o a labo a o y,
swabs we e s o ed in c yo ials a ⫺20°C be o e being ans e ed o
⫺80°C.
DNA ex ac ion and sequencing. We used Illumina compa ible p im-
e s (785F, GGATTAGATACCCBRGTAGTC, and 1175R, ACGTCRTCC
CCDCCTTCCTC) (30) ha ampli y he V5 o V7 egion o he 16S RNA
gene o gene a e a sequencing lib a y (31). Each sample was ampli ied
wi h dual indexes on he o wa d and e e se p ime . All ba codes and
adap e sequences used ha e been p e iously published (32). Each eac-
ion mix u e was se up wi h 1⫻Molzym PCR bu e (Molzym), 200 ␮M
deoxynucleoside iphospha es (dNTPs) (Bioline), 0.4 ␮M o wa d and
e e se p ime wi h ba code a ached, 0.025 ␮M MolTaq (Molzym), and
5␮l o empla e DNA and PCR g ade wa e (Bioline) o make a inal
eac ion mix u e olume o 25 ␮l. Cycling pa ame e s we e as ollows:
94°C o 3 min, 30 cycles o 94°C o 30 s, 60°C o 40 s, and 72°C o 90 s,
and one inal ex ension a 72°C o 10 min.
Samples we e pu i ied and pooled in o an equimola solu ion using
he SequalP ep no maliza ion pla e ki (Li e Technologies) and u he
Shaw e al.
6058 aem.asm.o g Oc obe 2016 Volume 82 Numbe 19Applied and En i onmen al Mic obiology
on Oc obe 17, 2016 by TAMPERE UNIVERSITY LIBRARY DEPARTMENT OF HEALTH SCIENCESh p://aem.asm.o g/Downloaded om
cleaned using AMPu e XP beads (Beckman Coul e ), bo h pe he man-
u ac u e ’s ecommenda ions. A e quan i ica ion using Qubi 2.0 (Li e
Technologies), he lib a y was dilu ed and loaded in o he MiSeq eagen
ca idge a 10 pM. MiSeq uns we e se o gene a e 250-bp pai ed-end
eads and wo 12-bp index eads o each sample.
Taxonomic classi ica ion. Sequenced eads we e me ged, demul i-
plexed, and quali y il e ed (minimum a e age Ph ed sco e o ⬎25) using
QIIME 1.8.0 (33). Closed- e e ence ope a ional axonomic uni s
(OTUs) we e picked a a 98.5% simila i y agains he Human O al Mic o-
biome Da abase (HOMD) 13.2 (20) using USEARCH 6.1.544 (34)in
QIIME 1.8.0 (33) wi h pa allel_pick_o us_usea ch61_ e .py. We used a
98.5% sequence simila i y because his is he h eshold used o de ine axa
in HOMD, as i app oxima ely co esponds o species-le el clus e s o
mos o al bac e ia (20). This app oach iden i ied 664 bac e ial OTUs co -
esponding o 13,049,932 eads. The mean numbe o eads pe sample
was 13,565 ⫾6,833.
Closed- e e ence OTU picking su e s om a numbe o issues, in-
cluding sensi i i y o he o de o e e ence sequences when sequences a e
iden ical o e he egion conside ed (35). This is a pa icula p oblem
when sequences a e simila ; he e exis o al bac e ia ha ha e ⬎99%
sequence simila i y in gi en egions o he 16S RNA gene bu occupy
sepa a e o al habi a s (36). Fo his eason, we also pe o med minimum
en opy decomposi ion (MED) on eads. MED is an unsupe ised e sion
o he oligo yping pipeline (37), which allows a g ea e esolu ion o mi-
c obial di e si y by pa i ioning sequences based on si es wi h high posi-
ional en opy in a e e ence- ee manne (36).
A e me ging o e lapping eads, he a e age sequence leng h was 369
bases. We il e ed sequences wi h an expec ed e o o mo e han 1 using
as q_ il e in VSEARCH 1.11.1 (38). We hen disca ded all sequences
ha we e sho e han 350 bases o longe han 380 bases bu pe o med
no o he quali y il e ing (e.g., leng h unca ion) because MED assumes
ha leng h a ia ion is biologically meaning ul. We an MED 2.1 on
14,449,794 sequences (in o ma ion on he eads disca ded a each s age is
a ailable in he supplemen al ma e ial). Because we wan ed o be able o
de ec a e sequences, we se he minimum subs an i e abundance pa-
ame e (M) o 1,444 (0.1% o he o al numbe o eads) and he maxi-
mum a ia ion allowed wi hin a node (V) o 3. All o he pa ame e s we e
se o hei de aul alues. We assigned axonomy o MED phylo ypes
using he Global Assignmen o Sequence Taxonomy (GAST) (39) wi h
VSEARCH 1.11.1 eplacing USEARCH.
S a is ical analyses. (i) Di e si y. We i ed a mul i a ia e linea e-
g ession model o p edic species ichness (obse ed numbe o species)
and he Shannon index (a measu e o ichness and e enness) using gingi-
i is, pe iodon i is, and he a iables lis ed in Table 2 o 811/962 samples
wi h comple e da a and ⬎5,000 eads. Richness and Shannon index we e
a e aged o e 100 i e a ions o a e ying o 5,000 eads pe sample. Back-
wa ds s epwise educ ion by he Akaike in o ma ion c i e ion (AIC) (40)
was used o selec he inal model.
(ii) Di e en ial abundances. We used DESeq2 1.6.3 (26) in Phyloseq
o model abundances. DESeq2 uses nega i e binomial gene alized linea
models o compa e he absolu e numbe o eads o each axon be ween
ca ego ies (27). Gingi i is was included as a con inuous a iable (BoP
anging om 0 o 6) and pe iodon i is as a bina y ac o . The model also
con ained e ms con olling o po en ial con ounde s (s udy si e, nu i-
ional in e en ion, HIV s a us, and sequencing un). P alues we e co -
ec ed o mul iple es ing using he Benjamini-Hochbe g p ocedu e (41).
Full DESeq esul s o gingi i is and pe iodon i is a e a ailable in he
supplemen al ma e ial (Da a Se S1).
(iii) Co ela ion ne wo ks. To acili a e a highe esolu ion o he
ne wo k o pe iodon i is-associa ed bac e ia, we selec ed all MED phylo-
ypes ha had ep esen a i e sequences wi h ⬎98.5% sequence simila i y
o pe iodon i is-associa ed HOMD OTUs. We calcula ed pai wise Spea -
man co ela ion coe icien s be ween hese MED phylo ypes ac oss sam-
ples. We used he Spa CC p ocedu e o es ima e co ela ions om com-
posi ional da a using log- a io ans o med abundances (42) wi h de aul
pa ame e s (20 in e ence i e a ions and a co ela ion s eng h exclusion
h eshold o 0.1). To calcula e pseudo P alues ( wo-sided es ), we
shu led he da a se s o each g oup 100 imes and epea ed he p oce-
du e, emo ing co ela ions ha we e no signi ican (P⬍0.05, no mul-
iple es ing co ec ion). Ne wo ks o s ong co ela ions, de ined as being
ou side o he 95% con idence in e al (CI) o he mean co ela ion be-
ween nodes (mean ⫹1.96 ⫻s anda d de ia ion [SD], e.g., 0.405 o he
ne wo k in Fig. 4a) we e isualized as ne wo ks wi h qg aph 1.3.1 (43)
using he F uch e man-Reingold algo i hm o node placemen (44).
Accession numbe (s). Reads we e deposi ed in he Eu opean Nucle-
o ide A chi e unde s udy accession no. PRJEB15035 (see he supplemen-
al ma e ial o de ails).
RESULTS
Desc ip ion o coho . A o al o 962 Malawian women we e in-
cluded in ou analysis, wi h a mean age o 25.4 ⫾6.2 yea s. O
hese women, 140 (14.6%) had no pe iodon al disease, 822
(85.4%) had gingi i is (bleeding on p obing [BoP] sco e o ⱖ1),
and 307 (32.0%) had pe iodon i is (Table 1). Gingi i is and pe i-
odon i is we e signi ican ly co ela ed (Spea man’s ␳⫽0.44),
wi h he majo i y o women wi h pe iodon i is ha ing high le els
o gingi i is. Pe iodon i is and gingi i is we e mo e common in
women who we e olde , had lowe socioeconomic s a us, and had
ewe yea s o educa ion (Table 2; o modeling, see Table S1 in he
supplemen al ma e ial).
Plaque ichness and di e si y a e highe in mo e se e e gin-
gi i is and pe iodon i is. Ini ial explo a o y analysis wi h p inci-
pal-coo dina e analysis (PCoA) o dina ions showed ha , al-
hough he e was la ge a iabili y in communi y composi ion
ac oss sup agingi al plaque samples, he e was also a clea end
ela ed o gingi i is se e i y ha was obus o he analysis me hod
used (HOMD OTUs o MED phylo ypes) (Fig. 1). S a i ying by
pe iodon i is in he same way did no indica e isually clea di -
e ences.
A quan i a i e analysis o di e si y e lec ed his end. Gingi-
i is was associa ed wi h highe mic obial communi y ichness
(Fig. 2a) and Shannon indexes (Fig. 2b). Mic obial communi ies
did no ma kedly di e be ween heal hy women and hose wi h
low le els o gingi i is. Bo h gingi i is and pe iodon i is we e as-
socia ed wi h highe sup agingi al plaque ichness in a linea e-
g ession, con olling o demog aphic a iables (see Table S3a in
he supplemen al ma e ial). In he inal model p edic ing Shan-
non index, pe iodon i is was no e ained bu gingi i is was (see
Table S3b in he supplemen al ma e ial). Re e sing he analysis,
ichness was e ained in he inal model o p edic ing gingi i is
bu no pe iodon i is (see Table S2 in he supplemen al ma e ial).
Di e ences in bac e ial abundances wi h gingi i is. Di e -
en ial abundance analyses wi h DESeq2 (26,27) ound 118 OTUs
ha we e signi ican ly ( alse-disco e y a e [lsqbq]⬍0.05) asso-
cia ed wi h a g ea e se e i y o gingi i is (see Da a Se S1 in he
supplemen al ma e ial), making up 16.6% o he da a se in e ms
o eads. Con e sely, 47 OTUs we e associa ed wi h lowe se e i y
TABLE 1 B eakdown o all women by se e i y o pe iodon al disease
Pe iodon i is
No. o women acco ding o se e i y o pe iodon al
disease (no. den al a ch sex an s wi h BoP)
0 123 456
No 137 72 95 111 72 63 102
Yes 4 112327 5150145
Gingi i is and Pe iodon i is in Sup agingi al Plaque
Oc obe 2016 Volume 82 Numbe 19 aem.asm.o g 6059Applied and En i onmen al Mic obiology
on Oc obe 17, 2016 by TAMPERE UNIVERSITY LIBRARY DEPARTMENT OF HEALTH SCIENCESh p://aem.asm.o g/Downloaded om
TABLE 2 Demog aphic cha ac e is ics b oken down by se e i y o pe iodon al disease
BoP
(no.) Pe iodon i is
To al no. o
women Age (y [SD])
No. wi h posi i e
HIV es (%)
No. wi h
mala ia (%)
a
Mean
BMI (SD)
Mean educa ion
(y [SD])
No. wi h
anemia (%)
b
Socio-economic
s a us (SD)
c
No. o women
om each
si e
d
No. o women
ecei ing each
nu i ional
in e en ion
e
No o samples
on each
sequencing
un
0 No 140 23.4 (5.8) 27 (19.3) 37 (26.6) 22.7 (3.2) 5.6 (3.6) 36 (25.7) 0.38 (1.22) 36/37/18/49 43/53/44 47/49/41/3
1 No 72 23.9 (5.9) 7 (9.7) 16 (22.2) 22.6 (3.4) 5.1 (3.8) 12 (16.7) 0.19 (1.11) 25/9/17/21 32/19/21 34/26/12/0
Yes 11 31.6 (6.1) 1 (9.1) 1 (9.1) 22.7 (2.4) 4.4 (3.3) 3 (27.3) ⫺0.35 (0.62) 6/2/1/2 8/0/3 3/5/3/0
2 No 95 24.7 (6.2) 11 (11.6) 22 (23.2) 22.1 (2.6) 4.4 (3.6) 19 (20.0) 0.10 (1.10) 39/19/13/24 38/34/23 31/41/23/0
Yes 23 27.5 (6.2) 5 (21.7) 5 (21.7) 21.7 (2.0) 2.7 (3.3) 4 (17.4) ⫺0.16 (0.91) 13/1/4/5 5/11/7 9/7/7/0
3 No 111 24.4 (5.4) 11 (9.9) 32 (28.8) 21.7 (2.3) 4.3 (3.3) 21 (18.9%) ⫺0.12 (0.84) 41/22/22/26 40/34/37 36/34/39/2
Yes 27 26.5 (5.7) 4 (14.8) 3 (11.1) 22.2 (2.7) 3.6 (3.0) 6 (22.2) ⫺0.20 (0.91) 11/6/3/7 11/4/12 11/6/10/0
4 No 72 25.0 (6.4) 9 (12.5) 16 (22.2) 21.7 (2.2) 3.4 (3.0) 11 (15.3) ⫺0.16 (0.80) 28/16/10/18 16/26/30 26/28/18/0
Yes 51 26.9 (5.4) 8 (15.7) 11 (21.6) 21.8 (2.7) 3.3 (3.1) 7 (13.7) ⫺0.17 (0.81) 27/3/7/14 14/19/18 23/7/21/0
5 No 63 24.9 (5.2) 7 (11.1) 12 (19.0) 21.6 (2.4) 4.0 (3.6) 15 (23.8) ⫺0.16 (0.81) 22/11/9/21 22/23/18 26/13/24/0
Yes 50 26.6 (5.9) 5 (10.0) 7 (14.0) 21.8 (3.1) 2.4 (2.8) 5 (10.0) ⫺0.36 (0.61) 18/11/7/14 16/15/19 21/12/17/0
6 No 102 24.5 (5.5) 10 (9.8) 18 (17) 21.9 (2.3) 3.5 (3.0) 26 (25.7) ⫺0.20 (0.81) 36/24/16/26 33/41/28 18/46/36/2
Yes 145 28.3 (7.0) 30 (20.7) 28 (19.3) 22.1 (2.5) 2.9 (3.0) 32 (22.1) ⫺0.27 (0.74) 66/28/17/34 45/48/52 59/43/41/2
a
Mala ia was diagnosed wi h a apid diagnos ic es ob ained om a inge p ick.
b
Anemia was de ined as a hemoglobin coun o ⬍110 g/li e .
c
A p oxy o socioeconomic s a us was c ea ed om a p incipal componen s analysis by combining in o ma ion on he building ma e ial o he house, main sou ce o wa e and elec ici y, sani a y acili ies, and main ype o cooking
uel used.
d
Women we e en olled a he ollowing ou si es: Lungwena, Malindi, Namwe a, and Mangochi, espec i ely.
e
Women ecei ed one o he ollowing h ee nu i ional in e en ions: IFA, MMN, o LNS, espec i ely.
Sup agingi al samples we e un on one o ou sequencing uns on Illumina MiSeq.
Shaw e al.
6060 aem.asm.o g Oc obe 2016 Volume 82 Numbe 19Applied and En i onmen al Mic obiology
on Oc obe 17, 2016 by TAMPERE UNIVERSITY LIBRARY DEPARTMENT OF HEALTH SCIENCESh p://aem.asm.o g/Downloaded om
(18.7% o he da a se ), implying ha gingi i is is no only ela ed
o bac e ial load bu also o he na u e o he mic obial commu-
ni y.
Figu e 3a and bshow he cumula i e abundances o heal h-
and gingi i is-associa ed OTUs, espec i ely, showing he p o-
g essi e na u e o changes wi h he deg ee o bleeding. Mos o he
pai wise compa isons o summed abundances o heal h- and gin-
gi i is-associa ed OTUs we e no signi ican ly di e en be ween
women wi h and wi hou pe iodon i is (K uskal-Wallis es , P⬎
0.05). Howe e , o women wi h pe iodon i is, he se e i y o gin-
gi i is was impo an , as he e we e mic obial di e ences be ween
women wi h and wi hou pe iodon i is o bo h mode a e gingi-
i is (BoP o 3; P⫽0.014) and se e e gingi i is (BoP o 6; P⫽
0.011). The mos signi ican ly gingi i is-associa ed OTU was Pep-
os ep ococcus s oma is, which was p esen in o e 75% o samples
ac oss se e i y ca ego ies and was an a e age o 1.45- old mo e
abundan (95% CI o 1.37 o 1.54) wi h a uni inc ease in BoP.
Di e ences in bac e ial abundances wi h pe iodon i is.
While gingi i is had a s onge associa ion wi h sup agingi al mi-
c obio a, he e we e also di e ences in mic obial communi y
composi ions wi h pe iodon i is (Fig. 3c and d). Se en y-one
OTUs we e signi ican ly (q⬍0.05) mo e abundan in women
wi h pe iodon i is (see Da a Se S1 in he supplemen al ma e ial),
making up 4.4% o he da a se in e ms o eads. Thi een OTUs
we e signi ican ly mo e abundan in he absence o pe iodon i is,
making up 3.6% o he da a se by eads. These heal h-associa ed
OTUs we e Lau opia mi abilis,Ro hia ae ia,S ep ococcus pyo-
genes,S ep ococcus mu ans, and se en membe s o Ac inomyces.
A he genus le el o pe iodon i is-associa ed OTUs, P e-
o ella (14 OTUs) and T eponema (10 OTUs) we e he mos ep-
FIG 1 The PCoA o dina ion o sup agingi al plaque samples shows an app oxima e end wi h gingi i is se e i y ha is obus o analysis me hods. PCoA
o dina ions based on B ay-Cu is dissimila i ies be ween samples o 626 HOMD OTUs (a, b) and 502 MED phylo ypes (c, d). Filled ellipses show mean alues
o each gingi i is se e i y, anging om 0 (yellow) o 6 (da k ed). In bo h cases, an app oxima e end is isible, despi e he noisiness o he da a se . Be o e
plo ing, samples we e a e ied o 5,000 eads o minimize he impac o sequencing dep h.
Gingi i is and Pe iodon i is in Sup agingi al Plaque
Oc obe 2016 Volume 82 Numbe 19 aem.asm.o g 6061Applied and En i onmen al Mic obiology
on Oc obe 17, 2016 by TAMPERE UNIVERSITY LIBRARY DEPARTMENT OF HEALTH SCIENCESh p://aem.asm.o g/Downloaded om

esen ed. Only one membe o he pa hogenic ed complex (8) was
signi ican ly associa ed wi h pe iodon i is, T eponema den icola.
The o he wo membe s (Po phy omonas gingi alis and Tanne ella
o sy hia) we e addi ionally no iden i ied as MED phylo ypes in
he da a se , which is possibly due o p ime misma ch (see he
discussion in he supplemen al ma e ial). Eubac e ium noda um,
p e iously iden i ied as clus e ing wi h he ed complex in su-
p agingi al plaque (45), was signi ican ly associa ed wi h pe i-
odon i is.
Di e ences in bac e ial abundances unique o pe iodon i is.
Fo y ou o 71 pe iodon i is-associa ed OTUs (56%) we e no
associa ed wi h gingi i is (see Table S4 in he supplemen al ma e-
ial). These axa we e a e; hei mean cumula i e abundance was
2.2%, wi h only six OTUs ha ing mean ela i e abundances o
⬎0.1%. The mos ep esen ed gene a we e P e o ella (9 OTUs),
T eponema (5 OTUs), and Selenomonas (4 OTUs).
The p esence o absence o pe iodon i is was no a signi ican
de e minan o he cumula i e abundances o hese OTUs o
women wi h he same le els o gingi i is (K uskal-Wallis es , P⬎
0.05), excep o women wi h a BoP o 4 (P⫽0.026).
The cooccu ence ne wo k o pe iodon i is-associa ed axa.
The abo e analysis conside s each OTU as independen , bu in
eali y, o al bac e ia exis in complex polymic obial bio ilms
whe e in e ac ions a e ex emely impo an (28). Cooccu ence
analysis can allow o he iden i ica ion o impo an membe s o
mic obial communi ies (46). We he e o e analyzed he cooccu -
ence ne wo ks o pe iodon i is-associa ed bac e ia ac oss all pe i-
odon al se e i ies.
A p elimina y ne wo k analysis o pe iodon i is-associa ed
OTUs ac oss pe iodon al se e i ies indica ed ha he ne wo k was
mo e connec ed in women wi h pe iodon i is ac oss gingi i is se-
e i ies (see Fig. S1 in he supplemen al ma e ial). Howe e , we
sough o con i m his cooccu ence pa e n wi h a highe esolu-
ion analysis. We he e o e selec ed all MED phylo ypes ha had
⬎98.5% simila i y o a pe iodon i is-associa ed OTU (see Ma e-
ials and Me hods). Eigh y-one MED phylo ypes had ep esen a-
i e sequences wi h ⬎98.5% simila i y o a pe iodon i is-associ-
a ed OTU (see Da a Se S2 in he supplemen al ma e ial).
The s ongly connec ed cooccu ence ne wo k in women wi h
se e e gingi i is (BOP o 6) and pe iodon i is showed se e al ge-
nus-le el clus e s, including Selenomonas,Pep os ep ococcus, and
P e o ella (Fig. 4a). No ably, hese clus e s we e connec ed by a
small g oup o cen al bac e ia, including Fili ac o alocis (phylo-
ype 158) and se e al membe s o Fusobac e ium nuclea um wi h
phylo ypes classi ied axonomically as subspecies incen ii (phylo-
ypes 3163 and 622) and polymo phum (phylo ypes 618 and 619),
sugges ing hei oles in he coagg ega ion o pe iodon al bio ilms.
Ranking phylo ypes in he s ongly connec ed ne wo k acco ding
o hei be weenness cen ali y, which measu es he po en ial o
in luence on in o ma ion ans e in a ne wo k (47), he mos
connec ed phylo ype was F. nuclea um subsp. incen ii (phylo ype
3163) (see Table S5 in he supplemen al ma e ial). T. den icola was
no p esen in his ne wo k, bu when MED analysis was epea ed
wi h he minimum subs an i e abundance pa ame e educed by
a ac o o 10 o 0.01%, we ound ha i was placed in he ne wo k
in a cen al posi ion.
To con i m ha his al e ed communi y s uc u e was a dis in-
guishing ea u e o sup agingi al plaque be ween women wi h and
wi hou pe iodon i is, we clus e ed he co ela ion ma ices based
on Man el dis ances o each ca ego y o pe iodon al disease (Fig.
4b). Ne wo ks clus e ed by he pe iodon i is s a us o he women
in he g oup, which con i med ha he al e ed communi y s uc-
u e wi h pe iodon i is was de ec able e en in women wi h low
le els o gingi i is. Wi hin he pe iodon i is g oupings, ma ices
clus e ed by gingi i is se e i y.
DISCUSSION
In his s udy, we in es iga ed changes in he sup agingi al mic o-
biome associa ed wi h pe iodon al disease se e i y in a la ge c oss-
sec ional coho in Malawi. Ou main inding was ha e en
hough he composi ion o sup agingi al plaque is p ima ily asso-
cia ed wi h gingi i is, as quan i ied by bleeding on p obing, a he
han he p esence o absence o pe iodon i is, he p esence o pe i-
odon i is has de ec able associa ions wi h sup agingi al mic obi-
FIG 2 Mic obial communi y ichness and Shannon index inc ease wi h gin-
gi i is se e i y. Bo h ichness (numbe o obse ed species) (a) and Shannon
index (measu e o di e si y) (b) o sup agingi al plaque inc ease wi h gingi i-
is se e i y. Es ima es o each sample we e calcula ed by sampling wi h e-
placemen a a a e ac ion dep h o 5,000 sequences pe sample and a e aging
o e 100 i e a ions. The i ed line shows a local polynomial eg ession i
calcula ed using loess in R, wi h he g ay egion indica ing he 95% CI. A o al
o 138/965 samples we e excluded due o ha ing ewe han 5,000 sequences.
Changing he a e ac ion dep h did no a ec he conclusion ha gingi i is
se e i y was associa ed wi h an inc ease in bo h species ichness and Shannon
index.
Shaw e al.
6062 aem.asm.o g Oc obe 2016 Volume 82 Numbe 19Applied and En i onmen al Mic obiology
on Oc obe 17, 2016 by TAMPERE UNIVERSITY LIBRARY DEPARTMENT OF HEALTH SCIENCESh p://aem.asm.o g/Downloaded om
o a ha a e un ela ed o gingi i is. In pa icula , he di e ences
in cooccu ence pa e ns o axa be ween women wi h and
wi hou pe iodon i is suppo a mo e complex e iology o dis-
ease han a simple p og ession om heal h h ough gingi i is
o pe iodon i is.
Gingi i is and pe iodon i is we e associa ed wi h highe mi-
c obial communi y ichness and Shannon indexes, and his asso-
cia ion emained a e adjus men o demog aphic ac o s, in-
cluding age, body mass index (BMI), and socioeconomic s a us.
This inding is consis en wi h p e ious esea ch (48,49), wi h
highe di e si y meaning ha , in pe iodon al disease, he o al mi-
c obio a is added a he han exis ing axa unde going eplace-
men . This may co espond o p ima y ecological succession in a
new en i onmen al niche, as sugges ed by Abusleme e al. (50).
We ound ha many axa we e associa ed wi h gingi i is and
pe iodon i is. The abundance o he majo i y o hese axa in-
FIG 3 Summed pe cen age abundances o OTUs associa ed wi h dec eased gingi i is (a), inc eased gingi i is (b), absence o pe iodon i is (c), and p esence o
pe iodon i is (d) o each pe iodon al disease ca ego y. Fo plo ing pu poses, samples we e a e ied o 10,000 eads pe sample, esul ing in he emo al o
269/962 samples; his a e ac ion was no used in he selec ion o he OTUs, which was pe o med using DESeq2 on he whole da a se . One ou lie and wo
ou lie s in panels c and d, espec i ely, a e no shown due o imming he yaxis a a ela i e abundance o 30%.
Gingi i is and Pe iodon i is in Sup agingi al Plaque
Oc obe 2016 Volume 82 Numbe 19 aem.asm.o g 6063Applied and En i onmen al Mic obiology
on Oc obe 17, 2016 by TAMPERE UNIVERSITY LIBRARY DEPARTMENT OF HEALTH SCIENCESh p://aem.asm.o g/Downloaded om
c eased wi h gingi i is se e i y, and his pa e n was no in lu-
enced by he p esence o pe iodon i is. Fu he mo e, some
women wi hou gingi i is had simila summed pe cen age abun-
dances o disease-associa ed axa compa ed o women wi h se e e
gingi i is. I would appea ha ela i e bac e ial abundances alone
a e insu icien o explain he p esence o disease, which is consis-
en wi h a equi emen o o he ac o s, such as he hos in lam-
ma o y esponse, o cause disease.
Pe iodon i is-associa ed OTUs we e also iden i ied, including
known pe iodon al pa hogens like F. alocis,T. den icola,F. nuclea-
um, and P. s oma is, which is consis en wi h indings om o he
popula ions (28). OTUs, including membe s o P e o ella,T epo-
nema,and Selenomonas, we e no signi ican ly associa ed wi h
gingi i is se e i y, suppo ing he idea ha pe iodon i is is no jus
an ad anced phase o gingi i is and in ol es addi ional bac e ia.
Howe e , cumula i e abundances o pe iodon i is-associa ed
OTUs did no di e signi ican ly be ween women wi h and wi h-
ou pe iodon i is who had he same le els o gingi i is, which sug-
ges s ha abundances do no ully explain he disease.
Wha we did obse e we e di e en cooccu ence pa e ns
ac oss disease ca ego ies o pe iodon i is-associa ed bac e ia,
which indica ed he p esence o a consis en communi y s uc u e
in women wi h pe iodon i is ac oss all gingi i is se e i ies. Cen al
nodes in his pe iodon i is-associa ed ne wo k included F. alocis
and se e al subspecies o F. nuclea um, which ac ed as hubs con-
nec ing di e en clus e s. Ne wo k analysis using be weenness
cen ali y anked F. nuclea um subsp. incen ii (phylo ype 3163)
as he mos cen al phylo ype in he s ongly connec ed cooccu -
ence ne wo k in women wi h se e e gingi i is and pe iodon i is.
These indings a e consis en wi h he p oposed oles o “b idging
bac e ia” ha con ibu e o he coagg ega ion o pe iodon al bio-
ilms (51). F. nuclea um has been shown expe imen ally o acili-
a e he su i al o obliga e anae obes in ae a ed en i onmen s
(52) and has been iden i ied as one o he impo an p ecu so s o
a achmen by la e colonize s in pe iodon al disease (51). F. alocis
has also been expe imen ally linked o he coagg ega ion o pe i-
odon al bio ilms (53,54) and co ela es wi h he g ea e in lam-
ma ion in pe iodon i is (24). Chen e al. also iden i ied a simila F.
alocis-cen e ed cooccu ence g oup o axa ha was en iched in
mul iple o al habi a s du ing pe iodon i is compa ed wi h hose
in heal hy con ols (49).
Limi a ions. The main s eng h o his s udy is ha we we e
able o include women wi h di e en se e i ies and combina ions
o pe iodon al disease, allowing us o dis inguish signals om gin-
gi i is and pe iodon i is. Howe e , ou obse a ions abou pe io-
don i is only apply o sup agingi al plaque, as we did no sample
om subgingi al plaque due o he di icul y o collec ing such a
la ge numbe o samples om a coho in a esou ce-limi ed se -
ing. Howe e , p e ious wo k has shown ha sampling sup agin-
gi al plaque s ill allows o he de ec ion o bac e ia associa ed
wi h pe iodon i is while being minimally in asi e and simple o
pe o m (55). Simila ly, we we e able o obse e changes in he
abundances o a e axa ha we e known o be associa ed wi h he
subgingi al plaque o pe iodon i is. Fo example, F e ibac e ium
as idiosum (HOMD iden i ica ion 360BH017), which accoun ed
o a mean o jus 0.009% o eads, was s ill signi ican ly mo e
abundan (2.5- old) in women wi h pe iodon i is, which is con-
sis en wi h he ecen inding o a highe abundance in subgingi-
al plaque when pe iodon i is was compa ed o gingi i is (23).
Ano he limi a ion was ha samples we e collec ed om ac oss
he mou h ins ead o localizing sampling o si es o speci ic in e -
es . The dis ibu ion o bac e ial species ac oss he mou h is
FIG 4 The cooccu ence ne wo k o pe iodon i is-associa ed bac e ia shows a dis inc communi y s uc u e wi h he p esence o pe iodon i is ac oss gingi i is
se e i ies. (a) The s ongly connec ed cen al cooccu ence ne wo k o pe iodon i is-associa ed bac e ia ac oss sup agingi al plaque samples om n⫽110
women wi h se e e gingi i is (BoP ⫽6) and pe iodon i is. Shown he e a e signi ican s ong pai wise Spea man co ela ion coe icien s (P⬍0.01; ␳⬎0.405),
calcula ed wi h Spa CC be ween MED phylo ypes wi h ⬎98.5% simila i y o pe iodon i is-associa ed HOMD OTUs (see Ma e ials and Me hods). Node colo
indica es axonomic genus, size is p opo ional o log- ans o med mean ela i e abundance, and edge weigh indica es he s eng h o he co ela ion. The ed
ci cle indica es he node wi h he highes be weenness cen ali y, classi ied axonomically as Fusobac e ium nuclea um subsp. incen ii. Node layou was
de e mined using he F uch e man-Reingold algo i hm in qg aph 1.3.1. Twen y- wo nodes wi hou any s ong co ela ions connec ing hem o he es o he
ne wo k (i.e., no edges wi h a ␳o ⬎0.405) we e emo ed du ing igu e p epa a ion. (b) Clus e ing using hclus in R o he co ela ion ma ices calcula ed in his
way o all se e i ies o pe iodon al disease. The pe iodon i is-associa ed cooccu ence ne wo k is mo e simila be ween women wi h pe iodon i is, ega dless o
gingi i is se e i y. Co ela ion ma ices we e no adjus ed o signi icance due o he di e en numbe s o women be ween g oups.
Shaw e al.
6064 aem.asm.o g Oc obe 2016 Volume 82 Numbe 19Applied and En i onmen al Mic obiology
on Oc obe 17, 2016 by TAMPERE UNIVERSITY LIBRARY DEPARTMENT OF HEALTH SCIENCESh p://aem.asm.o g/Downloaded om
known o be he e ogeneous, wi h sup agingi al plaque a si es
adjacen o deepened pe iodon al pocke s showing signi ican ly
highe coun s o pe iodon i is-associa ed species (45). Due o he
size o ou coho , we used a single swab, which was p obably
esponsible o he la ge amoun o a iabili y in ou da a se when
isualized in o dina ions (Fig. 1), and e ec i ely pooled all su-
p agingi al si es. This p ecluded an in es iga ion o he e ogenei y
be ween si es, bu de ec able associa ions wi h bo h gingi i is and
pe iodon i is we e s ill p esen e en wi h his app oach.
We ea ed gingi i is as a con inuous a iable bu pe iodon i is
as bina y. In eali y, pe iodon i is is a complex disease wi h a p ob-
lema ic classi ica ion (15), and i is likely ha ou simple ea -
men o pe iodon i is obscu es his complexi y. This may cause
bac e ial cooccu ence pa e ns in women wi h pe iodon i is o
appea s onge , as women wi h mo e se e e disease may ha e
g ea e abundances o associa ed bac e ial species.
Ou s udy is he la ges o be conduc ed so a in a sub-Saha an
popula ion, and ou esul s appea o be consis en , o he mos
pa , wi h p e ious wo k on bac e ial associa ions wi h pe iodon-
al disease (16,28,45,49,56). Howe e , i should be poin ed ou
ha ou popula ion was addi ionally no able in wo espec s. Fi s ,
all pa icipan s we e women who had ecen ly gi en bi h. P eg-
nancy, pa icula ly in i s ea ly o middle s ages, is known o be
linked o pe iodon al disease and po en ial changes in he o al
mic obiome (57), wi h an inc eased suscep ibili y o gingi i is
(58), al hough subgingi al le els o known pe iodon al pa hogens
may emain unchanged (59). Quali a i e di e ences be ween
pe iodon al pa hogens ound du ing p egnancy and pos pa um
ha e also been obse ed (60). I is no clea o how long a e
p egnancy he o al mic obiome emains al e ed, bu e idence ha
signi ican changes a e mainly de ec able in ea ly p egnancy (57)
and he consis ency o ou esul s wi h o he s udies sugges ha
e ec s emaining a e 6 weeks pos pa um a e small. Second, all
women in he s udy we e in e mi en ly gi en sul adoxine-py i-
me hamine (SP) a en ollmen and be ween he 28 h and 34 h
ges a ional week o mala ia p e en ion. Since sys emic an ibio -
ics can be gi en as a ea men o agg essi e pe iodon i is (61),
pa ien s who ha e ecei ed an ibio ic ea men in he p e ious 6
mon hs a e o en excluded om s udies o pe iodon i is. How-
e e , he sali a y mic obiome has been shown o be obus o
dis u bance by a week-long cou se o an ibio ics (62). Gi en ha
SP ea men was in e mi en , in ol ed an ibio ics no a ge ed a
pe iodon al bac e ia, and ook place a ound 2 mon hs be o e he
o al sampling, we belie e ha i is unlikely o ha e played an im-
po an ole bu ha e no di ec e idence o suppo his claim.
Conclusion. This s udy ep esen s he la ges o da e in es i-
ga ing associa ions be ween sup agingi al plaque composi ion
and a ious se e i ies o pe iodon al disease in a low-income sub-
Saha an popula ion wi h limi ed o al hygiene. We ha e iden i ied
dis inc signals associa ed wi h gingi i is and pe iodon i is in su-
p agingi al plaque, wi h a dominan con ibu ion om gingi i is.
Fu u e p oposals o a diagnos ic es o pe iodon i is based on
sup agingi al plaque sampling, which may be use ul in low- e-
sou ce se ings, will need o ake his in o accoun . Ne wo k anal-
ysis o obse ed cooccu ence pa e ns was consis en wi h he
ole o b idging bac e ia like F. nuclea um and F. alocis in he
coagg ega ion o pe iodon al bio ilms p io o pene ance in o
subgingi al egions. Al hough some pe iodon i is-associa ed bac-
e ia we e also associa ed wi h gingi i is, he majo change wi h
pe iodon i is is in he ne wo k o cooccu ences. Viewed his way,
gingi i is se s he s age o pe iodon i is o de elop by p o iding
an en i onmen whe e pe iodon i is-associa ed axa can inc ease
in abundance and coagg ega e in o pa hogenic bio ilms ha may
hen pene a e o subgingi al egions. Mo e quan i a i e model-
ing o associa ions be ween o al bac e ia and a ious clinical ea-
u es o disease will be necessa y o unde s and hese complex
ela ionships and explo e he mic obial ecology o pe iodon i is.
ACKNOWLEDGMENTS
We hank he s udy pa icipan s, he local communi ies, he esea ch pe -
sonnel a he s udy si es, and he iLiNS P ojec ex ended esea ch eam o
hei assis ance wi h he s udy. Membe s o he iLiNS P ojec S ee ing
Commi ee and P ojec Manage Ma y A imond (h p://ilins.o g) p o-
ided echnical suppo .
FUNDING INFORMATION
This wo k, including he e o s o Liam Shaw, was unded by Enginee ing
and Physical Sciences Resea ch Council (EPSRC) (EP/F500351/1). This
wo k, including he e o s o Ulla Ha junmaa, Ronan Doyle, Simeon
Mulewa, Da ie Cha lie, Ken Male a, Pe Asho n, and Nigel Klein, was
unded by Uni ed S a es Agency o In e na ional De elopmen (USAID)
(AID-OAA-A-12-00005). This wo k, including he e o s o Ulla Ha jun-
maa, Ronan Doyle, Simeon Mulewa, Da ie Cha lie, Ken Male a, and Pe
Asho n, was unded by Bill and Melinda Ga es Founda ion (Bill & Me-
linda Ga es Founda ion).
This publica ion is based on esea ch unded in pa by he O ice o
Heal h, In ec ious Diseases, and Nu i ion, Bu eau o Global Heal h, U.S.
Agency o In e na ional De elopmen (USAID) unde e ms o Coope -
a i e Ag eemen No. AID-OAA-A-12-00005, h ough he Food and Nu-
i ion Technical Assis ance III P ojec (FANTA) managed by FHI 360.
Addi ional unding was p o ided by he Bill and Melinda Ga es Founda-
ion h ough a g an o he Uni e si y o Cali o nia, Da is. The indings
and conclusions con ained wi hin he a icle a e hose o he au ho s and
do no necessa ily e lec posi ions o policies o he Bill and Melinda
Ga es Founda ion, USAID, he U.S. go e nmen , o he o he unde s.
L.S. is unded by an EPSRC Cen e o Doc o al T aining s uden ship a
UCL CoMPLEX (EP/F500351/1). Resea ch was suppo ed by he Na-
ional Ins i u e o Heal h Resea ch Biomedical Resea ch Cen e a G ea
O mond S ee Hospi al o Child en NHS Founda ion T us and UCL.
REFERENCES
1. Pe e sen PE, Bou geois D, Ogawa H, Es upinan-Day S, Ndiaye C. 2005.
The global bu den o o al diseases and isks o o al heal h. Bull Wo ld
Heal h O gan 83:661–669.
2. Li X, Koll ei KM, T ons ad L, Olsen I. 2000. Sys emic diseases caused
by o al in ec ion. Clin Mic obiol Re 13:547–558. h p://dx.doi.o g/10
.1128/CMR.13.4.547-558.2000.
3. Yu Y-H, Chasman DI, Bu ing JE, Rose L, Ridke PM. 2015. Ca dio as-
cula isks associa ed wi h inciden and p e alen pe iodon al disease. J
Clin Pe iodon ol 42:21–28. h p://dx.doi.o g/10.1111/jcpe.12335.
4. Ide M, Papapanou PN. 2013. Epidemiology o associa ion be ween ma-
e nal pe iodon al disease and ad e se p egnancy ou comes—sys ema ic
e iew. J Pe iodon ol 84:S181–S194. h p://dx.doi.o g/10.1902/jop.2013
.134009.
5. Hajishengallis G. 2015. Pe iodon i is: om mic obial immune sub e -
sion o sys emic in lamma ion. Na Re Immunol 15:30–44. h p://dx.doi
.o g/10.1038/n i3785.
6. Baelum V, Scheu z F. 2002. Pe iodon al diseases in A ica. Pe iodon ol
2000 29:79–103. h p://dx.doi.o g/10.1034/j.1600-0757.2002.290105.x.
7. Van Dyke TE. 2008. The managemen o in lamma ion in pe iodon al
disease. J Pe iodon ol 79:1601–1608. h p://dx.doi.o g/10.1902/jop.2008
.080173.
8. Soc ansky SS, Ha ajee AD, Cugini MA, Smi h C, Ken RL. 1998.
Mic obial complexes in subgingi al plaque. J Clin Pe iodon ol 25:134–
144. h p://dx.doi.o g/10.1111/j.1600-051X.1998. b02419.x.
9. Jiao Y, Hasegawa M, Inoha a N. 2014. The ole o o al pa hobion s in
Gingi i is and Pe iodon i is in Sup agingi al Plaque
Oc obe 2016 Volume 82 Numbe 19 aem.asm.o g 6065Applied and En i onmen al Mic obiology
on Oc obe 17, 2016 by TAMPERE UNIVERSITY LIBRARY DEPARTMENT OF HEALTH SCIENCESh p://aem.asm.o g/Downloaded om