Published online 19 Decembe 2015 Nucleic Acids Resea ch, 2016, Vol. 44, No. 8 3595–3609
doi: 10.1093/na /gk 1483
OxyR-dependen o ma ion o DNA me hyla ion
pa e ns in Op ABOFF and Op ABON cell lineages o
Salmonella en e ica
Ignacio Co a1, Boyke Bunk2,3, Ca h in Sp ¨
oe 2,3,J¨
o g O e mann2,3, Ch is oph K ¨
onig4and
Josep Casades´
us1,*
1Depa amen o de Gen´
e ica, Uni e sidad de Se illa, Facul ad de Biolog´
ıa, Apa ado 1095, 41080 Se illa, Spain,
2Leibniz Ins i u e DSMZ-Ge man Collec ion o Mic oo ganisms and Cell Cul u es, 38124 B aunschweig, Ge many,
3Ge man Cen e o In ec ion Resea ch (DZIF), Pa ne Si e Hanno e -B aunschweig, 38124 B aunschweig, Ge many
and 4Paci ic Biosciences, 1380 Willow Rd, Menlo Pa k, CA 94025, USA
Recei ed Oc obe 5, 2015; Re ised Decembe 7, 2015; Accep ed Decembe 8, 2015
ABSTRACT
Phase a ia ion o he
Salmonella en e ica op AB
ope on gene a es a bac e ial lineage wi h s anda d
lipopolysaccha ide s uc u e (Op ABOFF) and a lin-
eage wi h sho e O-an igen chains (Op ABON). Reg-
ula ion o Op AB lineage o ma ion is ansc ip-
ional, and is con olled by he LysR- ype ac o OxyR
and by DNA adenine me hyla ion. The
op AB
eg-
ula o y egion con ains ou si es o OxyR bind-
ing (OBSA-D), and ou me hyla able GATC mo i s
(GATC1–4). Op ABOFF and Op ABON cell lineages dis-
play opposi e DNA me hyla ion pa e ns in he
op AB
egula o y egion: (i) in he Op ABOFF s a e, GATC1
and GATC3a e non-me hyla ed, whe eas GATC2and
GATC4a e me hyla ed; (ii) in he Op ABON s a e,
GATC2and GATC4a e non-me hyla ed, whe eas
GATC1and GATC3a e me hyla ed. We p o ide e i-
dence ha such DNA me hyla ion pa e ns a e gen-
e a ed by OxyR binding. The highe s abili y o he
Op ABOFF lineage may be caused by binding o OxyR
o si es ha a e iden ical o he consensus (OBSAand
OBSc), while he si es bound by OxyR in Op ABON
cells (OBSBand OBSD) a e no . In suppo o his
iew, amelio a ion o ei he OBSBo OBSDlocks he
sys em in he ON s a e. We also show ha he GATC-
binding p o ein SeqA and he nucleoid p o ein HU a e
ancilla y ac o s in
op AB
con ol.
INTRODUCTION
Fo decades, bac e iological esea ch was based on he
s udy o la ge popula ions o bac e ial cells in ba ch cul-
u es. This expe imen al app oach assumed ha he alue
o any pa ame e measu ed in he popula ion would e lec
a unimodal dis ibu ion a ound he a e age alue in indi-
idual cells. This may be ue o many cellula pa ame-
e s. Howe e , in he las wo decades single cell analysis
has shown ha clonal popula ions o bac e ia, e en when
g owing in homogeneous en i onmen s, can exhibi pheno-
ypic he e ogenei y be ween indi idual cells (1–3). In ce -
ain cases, pheno ypic he e ogenei y e lec s he occu ence
o bis abili y, he o ma ion o wo subpopula ions wi h dis-
inc pa e ns o gene exp ession (4,5). Pheno ypic di e si y
can be also gene a ed by e e sible ON-OFF swi ching o
gene exp ession a high equencies, a phenomenon known
as phase a ia ion (6,7). In bac e ial pa hogens, phase a i-
a ion o en occu s a loci ha encode en elope s uc u es
and may be iewed as a s a egy o gene a e p og ammed
polymo phism (6,7). Indeed, lineage o ma ion can help o
e ade he hos immune sys em and o p o ec bac e ial sub-
popula ions agains bac e iophage in ec ion, among o he
po en ial adap i e ad an ages (7).
The molecula mechanisms o phase a ia ion a e di-
e se. Some a e gene ic, such as si e-speci ic ecombina-
ion (8) and slipped-s and mispai ing in ac s o epe i-
i e DNA sequences (9). In o he cases, howe e , he o -
ma ion o bac e ial lineages has epigene ic o igin, wi h-
ou al e a ion o he DNA sequence (10–12). Some o he
bes known examples o epigene ic phase a ia ion in ol e
he o ma ion o he i able DNA adenine (Dam) me hyla-
ion pa e ns (10–12). The lis includes he pap ope on o
u opa hogenic Esche ichia coli, which encodes imb ial ad-
hesins o adhe ence o he u ina y ac epi helium (10,13),
he agn43 agg ega ion gene o E. coli (14) and he glyco-
syl ans e ase ope on g o Salmonella en e ica (15). In
all hese cases, a ansc ip ional egula o binds a egula-
o y egion ha con ains GATC si es, which become non-
me hyla ed because binding o he egula o hinde s Dam
me hylase ac i i y. The OFF and ON s a es o he phase
a ia ion locus hus di e in he me hyla ion s a e o c i -
*To whom co espondence should be add essed. Tel: +34 95 455 7105; Fax +34 95 455 7104; Email: [email protected]
C
The Au ho (s) 2015. Published by Ox o d Uni e si y P ess on behal o Nucleic Acids Resea ch.
This is an Open Access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License (h p://c ea i ecommons.o g/licenses/by-nc/4.0/), which
pe mi s non-comme cial e-use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed. Fo comme cial e-use, please con ac
jou nals.pe [email protected]
3596 Nucleic Acids Resea ch, 2016, Vol. 44, No. 8
ical GATC si es (12). Because DNA base me hyla ion o -
en p e en s o es ains binding o p o eins o DNA, non-
me hyla ion can inc ease binding o he egula o y p o ein,
hus gene a ing a posi i e eedback loop ha p opaga es he
epigene ic s a e (12). Howe e , in all phase a ia ion sys ems
bo h he OFF and ON s a es a e me as able, which pe mi s
pheno ypic swi ching a e a numbe o gene a ions (6,7).
The swi ching equencies a e idiosync a ic o each phase
a ia ion locus, and may a y depending on cul u e condi-
ions (12).
The op AB locus o S. en e ica se o a Typhimu ium,
p e iously anno a ed as STM2209-STM2208,isa
Salmonella-speci ic locus ha encodes cy oplasmic
memb ane p o eins in ol ed in con ol o O-an igen chain
leng h (16). The op A and op B genes o m a bicis onic
ansc ip ional uni , which is ansc ibed om a canon-
ical, ␴70-dependen p omo e unde he con ol o he
LysR- ype ac o OxyR (16). Exp ession o op AB is
phase a iable, and S. en e ica ba ch cul u es con ain
subpopula ions o Op ABOFF and Op ABON cells. Each
subpopula ion ha bo s a dis inc ype o O-an igen, and
Op AB-media ed modi ica ion ende s Op ABON cells
esis an o bac e iophages ha use he O-an igen as e-
cep o (17). Howe e , he Op ABON subpopula ion shows
sensi i i y o se um, educed capaci y o p oli e a e in
mac ophages and a enua ion in he mouse model (16,17).
In his wo k we desc ibe he epigene ic mechanism e-
sponsible o he o ma ion o Op ABOFF and Op ABON
cell lineages in S. en e ica. Each lineage shows a dis inc pa -
e n o GATC me hyla ion a he op AB egula o y egion.
We p esen e idence ha such pa e ns a e gene a ed by di -
e en ial OxyR binding a he op AB egula o y egion. We
also show ha he GATC-binding p o ein SeqA and he nu-
cleoid p o ein HU a e ancilla y ac o s o op AB lineage
o ma ion. Finally, we p esen a model o op AB phase a i-
a ion, pa ly based upon expe imen al e idence, pa ly in-
spi ed by li e a u e da a and con aining some specula i e
elemen s as well.
MATERIALS AND METHODS
Bac e ial s ains, bac e iophage, media and cul u e condi-
ions
The s ains o S. en e ica used in his s udy (Supplemen-
a y Table S1) belong o se o a Typhimu ium, and o ig-
ina e om s ain ATCC 14028. Fo simplici y, S. en e ica
se o a Typhimu ium is ou inely abb e ia ed as S. en e -
ica.E. coli CC118 ␭pi [phoA20 hi-1 spE poB a gE(Am)
ecA1 (␭pi )]andE. coli S17–1 ␭pi [ ecA p o hsdR RP4–2-
Tc::Mu-Km::Tn7(␭pi )] we e used o di ec ed cons uc-
ion o poin mu a ions. E. coli M15 [pREP4] (Qiagen, Va-
lencia, CA, USA) was used o 6×His-OxyRC199S p oduc-
ion. Plasmid pTP166 (18) was kindly p o ided by Ma in
G. Ma inus, Uni e si y o Massachuse s, Wo ces e , MA,
USA.
Be ani’s lysogeny b o h (LB) was used as s anda d
liquid medium. Solid LB con ained aga a 1.5% inal
concen a ion. G een pla es (19) con ained me hyl blue
(Sigma-Ald ich, S Louis, MO, USA) ins ead o aniline
blue. The indica o o moni o ing ␤-galac osidase ac-
i i y in pla e es s was 5-b omo-4-chlo o-3-indolyl-␤-D-
galac opy anoside (X-gal; Sigma-Ald ich, 40 ␮g/ml). An-
ibio ics we e used a he concen a ions desc ibed p e i-
ously (20). To g ow oxyR s ains on LB aga , 75 ␮lo a
10 mg/ml ca alase solu ion (Sigma-Ald ich, S Louis, MO,
USA) we e sp ead on he su ace o he pla es.
T ansduc ional c osses using phage P22 HT 105/1in 201
(21) we e used o cons uc ion o s ains wi h al e ed ch o-
mosomal ma ke s. The ansduc ion p o ocol has been de-
sc ibed elsewhe e (22). To ob ain phage- ee isola es, ans-
duc an s we e pu i ied by s eaking on g een pla es. Phage
sensi i i y was es ed by c oss-s eaking wi h he clea -
plaque mu an P22 H5.
The oligonucleo ides used in his s udy ha e ei he been
desc ibed p e iously (16) o a e lis ed in Supplemen a y Ta-
ble S2. Gene dis up ion was achie ed using plasmids pKD3,
pKD4 and pKD13 (23) and oligonucleo ides PS1,PS2
o PS4. Ve i ica ion o he cons uc s was achie ed using
oligonucleo ides E1 and E2. An ibio ic esis ance casse es
in oduced du ing s ain cons uc ion we e excised by e-
combina ion wi h plasmid pCP20 (23).
Di ec ed cons uc ion o poin mu a ions
Mu a ion o GATC si es wi hin he op AB egula o y e-
gion was achie ed using p e iously desc ibed p ocedu es
and oligonucleo ides (16). Addi ional p ime s a e included
in Supplemen a y Table S2. An ibio ic esis ance casse es
om pKD3 and pKD4 we e in oduced in op AB::lac and
op AB::g p backg ounds using oligonucleo ides delGATC-
PS1 and delGATC-PS2, espec i ely (16). The esul ing
s ains we e used as in e media es in he cons uc ion o
poin mu a ions. Mu a ion o OxyR binding si es was
achie ed in he same way using p ime s labeled OxyRB and
OxyRD (Supplemen a y Table S2).
␤-galac osidase assays
Bac e ial cul u es we e g own in LB un il s a iona y phase
(O.D.600 ∼4). Le els o ␤-galac osidase ac i i y we e as-
sayed using he CHCl3-sodium dodecyl sul a e pe meabi-
liza ion p ocedu e (24). All da a a e a e ages and s anda d
de ia ions om mo e han h ee independen expe imen s.
Calcula ion o phase ansi ion equencies
Phase ansi ion a es we e es ima ed as desc ibed by Eisen-
s ein (25). B ie ly, a s ain ha bo ing an op AB::lac usion
waspla edonLB+X-gal.A e 16hg ow ha 37
oC,
colonies displaying ON and OFF pheno ypes we e cho-
sen, esuspended in phospha e bu e ed saline (PBS) and
esp ead on new pla es. Phase ansi ion equencies we e
calcula ed using he o mula (M/N)/gwhe e M is he num-
be o cells ha unde wen a phase ansi ion, N he o al
numbe o cells sco ed, and g he o al numbe o gene a-
ions ha ga e ise o he colony.
Flow cy ome y
Bac e ial cul u es we e g own in LB a 37◦C un il exponen-
ial phase (O.D.600 ∼0.3). Cells we e hen dilu ed in PBS
o a inal concen a ion o ∼107/ml. Da a acquisi ion and
Nucleic Acids Resea ch, 2016, Vol. 44, No. 8 3597
analysis we e pe o med using a Cy omics FC500-MPL cy-
ome e (Beckman Coul e , B ea, CA, USA). Da a we e col-
lec ed o 100 000 e en s pe sample, and we e analyzed wi h
CXP and FlowJo8.7 so wa e. Da a a e ep esen ed by a do
plo ( o wa d sca e [cell size] e sus luo escence in ensi y
[op AB::g p exp ession]).
Cons uc ion o plasmid pIZ1885 (pQE30::oxyRC199S)
A DNA agmen con aining oxyRC199S (16) was ampli-
ied using oligonucleo ides His-oxyR-BamHI-5 and His-
oxyR-SalI-3, and cloned in o pQE30 (Qiagen, Valencia,
CA, USA) using he BamHI and SalI si es. The ecombi-
nan plasmid (pIZ1885) was e i ied by es ic ion analysis
and DNA sequencing.
Pu i ica ion o OxyR p o ein
Fo 6×His-OxyRC199S pu i ica ion, plasmid pIZ1885 was
ans o med in o E. coli M15 [pREP4] (Qiagen, Valencia,
CA, USA). M15/pIZ1885 was g own in LB b o h con-
aining ampicillin, and exp ession o 6×His-OxyRC199S was
induced wi h 1 mM isop opyl ␤-D- hiogalac opy anoside
(IPTG). A e 3 h o induc ion, cells we e cen i uged and
esuspended in 10 ml o lysis bu e (20 mM T is, 300
mM NaCl, 10 mM imidazole) pe g o pelle ed cells, and
we e lysed by sonica ion. The suspension was cen i uged
a 10 000 pm o 30 min and he supe na an con aining
he soluble ac ion o 6×His-OxyRC199S was ans e ed
o a HisT ap HP nickel a ini y ch oma og aphy column
(GE Heal hca e, Wauwa osa, WI, USA). The column was
washed wi h 4 ml o lysis bu e , 4 ml o washing bu e
(20 mM T is, 300 mM NaCl, 30 mM imidazole) and 4 ml
o he same bu e wi h 50 mM imidazole. P o ein elu ion
was pe o med wi h 3 ml o elu ion bu e (20 mM T is, 300
mM NaCl, 300 mM imidazole). Elu ion ac ions en iched
in 6×His-OxyRC199S we e selec ed and combined. Imida-
zole was emo ed by ans e ing o an Amicon R
ul a
cen i ugal il e (Me ck Millipo e, Da ms ad , Ge many)
and washing wi h s o age bu e (20 mM T is, 300 mM
NaCl, 10% glyce ol) o by dialyzing in cellulose memb anes
(Sigma-Ald ich, S Louis, MO, USA). 6×His-OxyRC199S
was ei he used immedia ely o ozen in liquid ni ogen and
s o ed a −80oC.
Gel mobili y shi assay
A DNA agmen con aining p edic ed OxyR binding
si es in he op AB egula o y egion and labeled wi h
6-ca boxy luo escein (6-FAM) was p epa ed by poly-
me ase chain eac ion (PCR) ampli ica ion using p ime s
FAMGATCla go-5 and FAMGATCla go-3 (Supplemen a y
Table S2). The PCR p oduc was pu i ied wi h he Wiza d R

SV Clean-Up Sys em (P omega). The en R con ol ag-
men was p epa ed using p ime s en R-Fo -Dnase and
en R-Re -Dnase (26), and was kindly p o ided by Elena
Espinosa. Thi y i e nanog am we e used o each eac-
ion. The FAM-labeled p obe was incuba ed a oom em-
pe a u e o 30 min wi h inc easing concen a ions o pu-
i ied 6×His-OxyRC199S in a inal olume o 20 ␮l wi h
1×OxyR binding bu e [25 mM T is–HCl pH 7.5, 50
mM KCl, 5 mM MgCl2,5%glyce ol,50␮g/ml bo ine
se um albumin (BSA), 1 mM DTT, 1 ␮g/ml poly(dI-dC)].
P o ein–DNA complexes we e subjec ed o elec opho esis
a 4◦C in a 5% non-dena u ing polyac ylamide gel in T is-
glycine-e hylenediamine e aace ic acid (EDTA) bu e (25
mM T is–HCl pH 7.5, 380 mM glycine, 1.5 mM EDTA).
The gel was hen analyzed in a FLA-5100 Scanne (Fuji-
ilm, Tokyo, Japan).
DNA me hyla ion in i o
PCR agmen s we e me hyla ed in i o using Dam me hy-
lase (New England Biolabs, Ipswich, MA, USA) acco d-
ing o he manu ac u e ’s ins uc ions and subsequen ly di-
ges ed wi h MboI (New England Biolabs). The undiges ed
p oduc was pu i ied using he Wiza d R
SV Clean-Up sys-
em (P omega, Madison, WI, USA).
DNase I oo p in ing
DNA p obes con aining he op AB p omo e and
he ups eam egula o y egion, labeled wi h 6-
ca boxy luo escein (6-FAM) a he opposi e ends, we e
p epa ed by PCR ampli ica ion using he p ime pai s
FAMGATCla go-5 +FAMGATCla go-3 and seqGATC-5
+FAMGATCla goconFAM-3. Dam-me hyla ed e sions
o he p obes we e p epa ed as desc ibed abo e. DNase
I oo p in ing was pe o med as desc ibed elsewhe e (27)
wi h mino modi ica ions. DNase I oo p in ing eac ions
we e pe o med in 15 ␮l eac ion olumes con aining 1×
OxyR binding bu e and 2 ␮M6×His-OxyRC199S.The
binding eac ion was allowed o equilib a e a oom em-
pe a u e o 30 min. A o al o 1 ␮l (0.05 uni s) o DNase
I (Roche Fa ma, Ba celona, Spain) was hen added, mixed
gen ly and incuba ed a 37◦C o 5 min. The eac ion was
s opped by addi ion o 2 ␮l EDTA 100 mM ollowed by
igo ous o exing and he mal dena u a ion a 95◦C o
10 min. Diges ion p oduc s we e desal ed using Mic oSpin
G-25 columns (GE Heal hca e, Wauwa osa, WI, USA)
and analyzed on an ABI 3730 DNA Analyze along wi h
GeneScan 500-LIZ size s anda ds (Applied Biosys ems,
Fos e Ci y, CA, USA).
SMRT R
sequencing
Cul u es o S. en e ica we e en iched o Op ABON cells
i needed (17). SMRTbellTM empla e lib a ies we e p e-
pa ed acco ding o he ins uc ions om Paci ic Bio-
sciences (Menlo Pa k, CA, USA), ollowing he p oce-
du e and checklis o 1 kb empla e p epa a ion and se-
quencing. B ie ly, o p epa a ion o 600 bp lib a ies, 4
␮g o genomic DNA we e shea ed in mic oTubes using
adap i e ocused acous ics (Co a is, Wobu n, MA, USA).
Size ange was moni o ed on an Agilen 2100 Bioana-
lyze om Agilen Technologies, San a Cla a, CA, USA.
DNAs we e end- epai ed and liga ed o hai pin adap e s
applying componen s om he DNA Templa e P ep, Pa-
ci ic Biosciences, Menlo Pa k, CA, USA. SMRTbellTM em-
pla es we e exonuclease- ea ed o emo al o incomple e
eac ion p oduc s. Condi ions o annealing o sequencing
p ime s and binding o polyme ase o pu i ied SMRTbellTM
3598 Nucleic Acids Resea ch, 2016, Vol. 44, No. 8
empla es we e assessed wi h he Paci ic Biosciences’ Bind-
ing Calcula o . Six mo ies we e aken o bo h s a es on he
PacBio RSII (Paci ic Biosciences, Menlo Pa k, CA, USA)
using P4-C2 chemis y a 2 h collec ion ime. Secondly, s a-
iona y phase cul u es we e en iched o Op ABON cells
and lib a ies we e p epa ed as gi en abo e. In his case i e
mo ies we e aken using P4-C2 chemis y a 3 h collec ion
ime.
Resul ing da a we e mapped o he comple e genome
sequence (GenBank accession numbe CP001363.1) o
S. en e ica subsp. en e ica se o a Typhimu ium s ain
ATCC 14028, using he BLASR algo i hm (28) as imple-
men ed in Paci ic Biosciences’ SMRT R
Po al 2.1.0 wi hin
he ‘RS Modi ica ion and Mo i Analysis.1’ p o ocol ap-
plying de aul pa ame e se ings. Acco ding o he se up
o he expe imen he seconda y analysis jobs we e named
‘Op ABOFF’ and ‘Op ABON’. Besides he global me hyla-
ion pa e n, he me hyla ion s a us o ou GATC si es up-
s eam o he op AB ope on was in e ed using SMRT R

View, in es iga ing he ch omosomal posi ions 2 361 489
and 2 361 490 (GATC1), 2 361 439 and 2 361 440 (GATC2),
2 361 416 and 2 361 417 (GATC3) and 2 361 366 and 2 361
367 (GATC4). Resul s a e shown in supplemen a y .cs iles
S1 and S2 (Op AB-OFF basemod summa y and Op AB-
ON basemod summa y, espec i ely).
Sou he n blo
Genomic DNA was isola ed by phenol ex ac ion and
e hanol p ecipi a ion om s a iona y cul u es in LB
(O.D.600 ∼4). A o al o 16 ␮g o each DNA sample we e
diges ed wi h HaeIII and AccI (New England Biolabs, Ip-
swich, MA, USA), pu i ied and di ided in o ou ac ions,
h ee o which we e subsequen ly diges ed wi h DpnI, MboI
o Sau3AI (New England Biolabs). A e diges ion he sam-
ples we e un in a 2% TAE-aga ose gel a 100 V o 2 h. A -
e elec opho esis, he DNA was dena u ed by ea men
o he gel in acid condi ions (0.25 M HCl, wo washes 15
min each), ollowed by alkaliniza ion (0.5 M NaOH, 1.5 M
NaCl) and neu aliza ion (0.5 M T is, 1.5 M NaCl, pH 7.5;
wo washes, 30 min each). The gel was hen washed in SSC
10×bu e (1.5 M NaCl, 150 mM isodium ci a e, pH 7)
and he DNA was ans e ed by acuum o an Ame sham
Hybond-N+ memb ane (GE Heal hca e, Wauwa osa, WI,
USA) using a model 785 Vacuum Blo e (Bio-Rad, He -
cules, CA, USA). The DNA in he memb ane was hen im-
mobilized by UV c osslinking. A adioac i e p obe was p e-
pa ed by PCR using dCTP [␣-32P] (Pe kin Elme , Wal ham,
MA, USA) and oligonucleo ides 2208mu 1DIRnue o and
2208mu 4INVnue o (16). A e he PCR eac ion, non-
inco po a ed nucleo ides we e emo ed by ea men in a
Sephadex G-25 column (illus a Mic oSpin G-25 columns,
GE Heal hca e, Wauwa osa, WI, USA) ollowing he in-
s uc ions o he manu ac u e . P io o hyb idiza ion he
double-s anded DNA p obe was dena u ed by hea ing a
95◦C o 3 min, ollowed by incuba ion on ice. Hyb idiza-
ion wi h he p obe was pe o med o e nigh a 42◦Cinhy-
b idiza ion bu e (0.5 M sodium phospha e pH 7.2, 10 mM
EDTA, 7% sodium dodecylsulpha e (SDS)). Excess p obe
was emo ed wi h washing bu e (40 mM sodium phos-
pha e pH 7.2, 1% SDS) a 38◦C ( h ee washes, 30 min each).
The memb ane was de eloped using a FLA-5100 Scanne
(Fuji ilm, Tokyo, Japan).
RESULTS
Bo h he absence and he o e exp ession o Dam me hylase
inc ease op AB exp ession and abolish phase a ia ion
Genes unde Dam me hyla ion con ol all in o wo ca e-
go ies. One includes genes in which me hyla ion and non-
me hyla ion p o ide opposi e signals (12). An example is
he aJ gene o he Salmonella i ulence plasmid, which
is ep essed by GATC me hyla ion (29). In his class o
genes, exp ession o he dam gene om a mul icopy plasmid
does no al e he wild- ype pheno ype (29). In o he genes,
howe e , a plasmid-bo ne dam gene does al e he gene ex-
p ession pa e n. This phenomenon is usually an indica-
ion ha Dam dependen ansc ip ional con ol is mo e
complex, and in ol es he o ma ion o Dam me hyla ion
pa e ns (combina ions o me hyla ed and non-me hyla ed
GATC si es) (12). To asce ain whe he op AB belonged
o he ‘simple’ o he ‘complex’ class o Dam me hyla ion-
dependen genes, he e ec o in oducing a dam gene ca -
ied on plasmid pTP166 was assayed. The esul s we e as
ollows:
(i) In a wild- ype backg ound, an op AB::lac ansla-
ional usion showed phase a ia ion, and o med
whi e (Op ABOFF) and blue (Op ABON) colonies in
he p esence o X-gal. In a dam backg ound, phase
a ia ion was abolished, and all colonies we e Lac+
(Op ABON). Plasmid pTP166 yielded an in e media e
pheno ype (Figu e 1A), sugges ing ha o ma ion o
he Op ABOFF and Op ABON subpopula ions migh
in ol e he es ablishmen o a DNA me hyla ion pa -
e n in he GATC si es o he op AB con ol egion,
a he han me hyla ion o non-me hyla ion o he ull
se o GATC si es. A simila phenomenon occu s in he
g ope on (15)whichis ep essedinadam backg ound
while in oduc ion o a cloned dam gene esul s in an
in e media e pheno ype.
(ii) Exp ession o op AB::lac was also moni o ed by ␤-
galac osidase assays (Figu e 1B). Lack o Dam me hy-
la ion inc eased exp ession o he op AB ope on as p e-
iously desc ibed (16). In oduc ion o he dam gene
ca ied on he pTP166 plasmid yielded an in e medi-
a e op AB exp ession le el, as in he colonies desc ibed
abo e.
(iii) Exp ession o an op AB::g p ansc ip ional usion
was moni o ed by luo escence analysis (Figu e 1C).
A majo Op ABOFF subpopula ion and a mino
Op ABON subpopula ion we e de ec ed in he wild-
ype. In a dam backg ound, a single popula ion in he
ON s a e was obse ed, in acco dance wi h he esul s
ob ained wi h a op AB::lac usion. In he p esence o a
cloned dam gene (pTP166), a single popula ion wi h in-
e media e le els o exp ession was de ec ed and a shi
owa d he ON s a e emained isible (Figu e 1C).
Al oge he , he abo e obse a ions sugges ed ha DNA
me hyla ion pa e ns migh be o med a he op AB con ol
egion. This egion, loca ed ups eam o he op AB p o-
Nucleic Acids Resea ch, 2016, Vol. 44, No. 8 3599
Figu e 1. Regula ion o op AB exp ession by Dam me hyla ion and o ma ion o Op AB subpopula ions. (A) Visual obse a ion o phase a ia ion on LB
+ X-gal pla es in Salmonella en e ica s ains ca ying an op AB::lac usion in he wild- ype, a dam mu an and a s ain ha o e p oduced Dam me hylase
(ATCC 14028/pTP166). (B) A e ages and s anda d de ia ions o ␤-galac osidase ac i i y o he same s ains. (C) GFP luo escence dis ibu ion in a s ain
ca ying an op AB::g p usion in he same backg ounds. Da a a e ep esen ed by a do plo ( o wa d sca e [cellula size] e sus luo escence in ensi y
[op AB::g p exp ession]). All da a we e collec ed o 100 000 e en s pe sample.
mo e , con ains ou GATC si es sepa a ed by 46, 19 and
46 n and cen e ed a he −172.5, −122.5, −99.5 and −49.5
posi ions ups eam o he ansc ip ion s a si e (Figu e 2A
and Supplemen a y Figu e S1). F om now on, hese GATC
si es will be e e ed o as GATC1 o GATC4, he la e be-
ing closes o he −35 module o he op AB p omo e (Sup-
plemen a y Figu e S1).
Roles o indi idual op AB GATC si es in he o ma ion o
Op ABOFF and Op ABON cell lineages
To s udy he con ibu ion o each GATC si e o op AB eg-
ula ion, mu a ions we e in oduced by si e-di ec ed mu age-
nesis. The mu a ions we e designed o change GATC si es
so ha hey would no longe be a subs a e o Dam me hy-
la ion. Because OxyR is essen ial o op AB exp ession (16),
al e a ion o consensus sequences was a oided inside pu a-
i e OxyR binding si es. CATC si es we e hus in oduced
in place o GATC si es, and e e y combina ion o mu a ed
and non-mu a ed GATC si es was p oduced.
The e ec o GATC mu a ions on op AB exp ession was
i s analyzed by compa ing he ␤-galac osidase ac i i y o
an op AB::lac ansla ional usion in dam+and dam back-
g ounds (Figu e 2B). Rele an obse a ions we e as ol-
lows:
(i) Mu a ion o GATC1and GATC3hadasmalle ec
on egula ion by Dam me hyla ion, al hough he abso-
lu e alues o ␤-galac osidase ac i i y we e highe . Mu-
a ion o GATC2 esul ed in diminished egula ion by
Dam me hyla ion. When GATC4was mu a ed, con ol
by Dam me hyla ion showed an in e ed pa e n (ex-
p ession was highe in a dam+backg ound).
(ii) As a gene al ule, combina ions o wo o mo e mu-
a ions seemed o ha e an addi i e e ec . A ema k-
able case was he combina ion o mu a ed GATC2and
GATC4which exace ba ed he in e sion o egula ion
by Dam me hyla ion caused by mu a ion o GATC4
alone. I is no ewo hy ha mu a ions in GATC1,
GATC2and GATC3 oge he did no abolish Dam-
dependen egula ion, whe eas a single mu a ion in
GATC4in e ed he pa e n o Dam-dependen egu-
la ion.
The o e all conclusion om hese expe imen s was ha
all ou GATC si es a e in ol ed in Dam-dependen con ol
o op AB exp ession, and ha he GATC4si e may ha e an
especially p ominen ole.
E en hough dis up ion o OxyR binding si es had been
a oided, GATC mu a ions a ec ed op AB exp ession i e-
spec i e o he p esence o absence o DNA me hyla ion, as
obse ed in a dam backg ound (Figu e 2B). In he absence
o Dam me hyla ion, mu a ions in GATC1and GATC3in-
c eased op AB exp ession whe eas mu a ions in GATC2
and GATC4 esul ed in lowe op AB exp ession. To sep-
a a e such e ec s om hose o Dam me hyla ion i sel ,
he ␤-galac osidase ac i i y o op AB::lac in a wild- ype
backg ound was pu in ela ion o he ␤-galac osidase ac-
i i y in a dam backg ound (Figu e 2C). This ep esen a-
ion leads o he in e es ing conclusion ha mu a ions in
GATC2and GATC4ac i a e op AB exp ession. Mu a ions
in GATC1and GATC3show li le e ec on hei own be-
cause op AB exp ession is low in he wild- ype, bu hey e-

3600 Nucleic Acids Resea ch, 2016, Vol. 44, No. 8
Figu e 2. E ec o mu a ions in he op AB GATC si es on op AB exp es-
sion. (A) Diag am o he op AB egula o y egion, wi h he GATC si es
and he OxyR binding si es ou lined. (B) A e ages and s anda d de ia-
ions o ␤-galac osidase ac i i y o s ains ca ying an op AB::lac usion
in a wild- ype backg ound (black ba s) and in a dam backg ound (whi e
ba s). Mu a ed GATC si es a e indica ed by numbe s 1–4. (C)Rela i e␤-
galac osidase ac i i y o he op AB::lac usion in he same s ains (ac i i y
in he wild- ype di ided by ac i i y in a dam backg ound).
p ess op AB exp ession when combined wi h ac i a ing mu-
a ions in GATC2and/o GATC4. Hence, he GATC si es
in he op AB egula o y egion can be en a i ely di ided
in wo pai s: me hyla ion o pai GATC1+GATC
3seems
o be associa ed wi h he Op ABON s a e while me hyla ion
o pai GATC2+GATC
4seems o be associa ed wi h he
Op ABOFF s a e.
Analysis o luo escence using an op AB::g p ansc ip-
ional usion (Figu e 3) allowed us o dis inguish whe he
he di e ences in op AB exp ession in GATC mu an back-
g ounds e lec ed di e ences in gene exp ession o di e -
ences in he sizes o he Op ABON and Op ABOFF subpop-
ula ions. The main obse a ions we e as ollows:
(i) In he wild- ype, he Op ABON subpopula ion com-
p ised ∼0.18% cells.
(ii) Mu a ion o GATC4caused a d as ic inc ease in he
size o he Op ABON subpopula ion. Mu a ions in
GATC1,GATC
2and GATC3had a smalle e ec ,
which was mo e clea ly seen when hey we e combined
wi h each o he and/o wi h a mu a ion in GATC4.
(iii) Two subpopula ions we e s ill dis inguished when
h ee GATC si es we e mu a ed, p o ided ha ei he
GATC3o GATC4 emained unal e ed. The ela i e
size o he Op ABOFF and Op ABON subpopula ions
was howe e di e en in each case, wi h a p edominan
Op ABOFF subpopula ion when GATC4 emained un-
al e ed and a p edominan Op ABON subpopula ion
when GATC3 emained unal e ed.
(i ) Mu a ion o bo h GATC3and GATC4elimina ed sub-
popula ion o ma ion ega dless o he p esence o
mu a ions in GATC1and GATC2, and yielded an
Op ABON popula ion.
These obse a ions a e consis en wi h he gene exp es-
sion analyses epo ed abo e, and pe mi o in e p e he
gene exp ession esul s in e ms o subpopula ion o ma-
ion. Mu a ion o GATC4caused he mos d as ic inc ease
in he p opo ion o Op ABON cells, he eby con i ming
ha me hyla ion o he GATC4si e may ha e a ele an ole
in he o ma ion o he Op ABOFF subpopula ion. Inc ease
o Op ABON subpopula ion was likewise obse ed when a
mu a ed GATC4was combined wi h o he mu a ed GATC
si es (Figu e 3).
OxyR binds he op AB egula o y egion
Fou pu a i e OxyR binding hal -si es a e ound in he eg-
ula o y egion o op AB cen e ed in he −148, −116, −75
and −43 posi ions (Supplemen a y Figu e S1), and sha ing
10, 8, 10 and 7 n espec i ely wi h he 10-n consensus se-
quence (16). The OxyR binding hal -si es ups eam o he
op AB p omo e will be om now on e e ed o as OBSA o
OBSD, he la e being immedia ely ups eam o he op AB
−35 p omo e module (Figu e 2A). Assuming a helical pe-
iodici y o 10.5 bp (30), he OBS a e p edic ed o be spaced
by one, wo and one helical u ns, which means ha all he
OxyR binding si es may be on he same ace o he DNA
helix. The dis ance be ween OBSAand OBSB, and be ween
OBSCand OBSDas well, is canonical o binding o he e-
duced o m o OxyR (31). GATC2and GATC4o e lap wi h
OBSBand OBSD, espec i ely (Figu e 2A).
To es whe he OxyR binds he op AB egula o y e-
gion, an elec opho e ic mobili y shi assay (EMSA) was
ca ied ou using pu i ied OxyR p o ein (Figu e 4A). To
a oid uncon olled oxida ion o OxyR and because i was
p e iously shown ha he oxida ion s a e o OxyR is no
ele an o op AB egula ion (16), we used a mu an e -
sion o he OxyR p o ein, OxyRC199S, which canno be ox-
idized bu e ains he p ope ies o he educed o m o
OxyR (31,32). Pu i ied 6×His-OxyRC199S p o ein (hence-
o h named OxyR o simplici y) was hus used. A DNA
agmen con aining he ou egula o y GATC si es and
he ou OxyR binding hal -si es was p oduced using a 6-
FAM-labeled oligonucleo ide and was incuba ed wi h in-
c easing concen a ions o OxyR. Binding was unambigu-
ously de ec ed. A DNA agmen om he egula o y e-
gion o an un ela ed gene (en R) was used as a nega i e con-
ol, and binding was no de ec ed (Figu e 4A).
Nucleic Acids Resea ch, 2016, Vol. 44, No. 8 3601
Figu e 3. GFP luo escence dis ibu ion in Salmonella en e ica s ains ca ying an op AB::g p usion and mu a ions in he op AB GATC si es. Mu a ed
GATC si es a e indica ed by numbe s 1–4. Da a a e ep esen ed by a do plo , and we e collec ed o 100 000 e en s pe sample.
OxyR p o ec s he op AB egula o y egion
To de ine he binding pa e n o OxyR o he op AB egu-
la o y egion, pu i ied OxyR was used in a oo p in ing as-
say pe o med using 6-FAM-labeled DNA agmen s and
DNase I (Figu e 4B). The same DNA agmen used in
he EMSA assays, con aining bo h he GATC si es and
p edic ed OxyR binding si es, was labeled a he al e na e
ends and used in pa allel expe imen s. Me hyla ed and non-
me hyla ed DNA p obes we e used, as well as a p obe in
which GATC si es 1–4 had been con e ed o CATC si es by
si e-di ec ed mu agenesis. The analysis con i med he abili y
o OxyR o bind he op AB egula o y egion in i o (Fig-
u e 4and Supplemen a y Figu e S2). Rele an obse a ions
we e as ollows:
(i) P o ec ion om DNase I diges ion was de ec ed in a
133 bp DNA span, albei wi h egional di e ences.
GATC1, GATC2, GATC3a e loca ed in he p o ec ed
egion. F agmen -speci ic binding pa e ns we e de-
ec ed and he o e all p o ec ion was less e icien when
he DNA p obe was ei he me hyla ed o GATC-less.
(ii) The OBSAand OBSCsi es we e ully p o ec ed, while
OBSBwas pa ially p o ec ed.
(iii) OBSD, which con ains he GATC4si e, was no p o-
ec ed.
The ele ance o hese obse a ions may be limi ed as
me hyla ed and non-me hyla ed DNA p obes we e used,
and e idence p esen ed abo e had sugges ed ha op AB
egula ion in ol ed bo h me hyla ed and non-me hyla ed
GATC si es (Figu e 1). Wi h his ca ea , oo p in ing ex-
pe imen s con i med he abili y o OxyR o bind he op AB
egula o y egion and de ined he DNA egion p o ec ed by
OxyR binding. An addi ional, in e es ing obse a ion was
ha OxyR p o ec ion ex ended ou side he OxyR binding
si es, as p e iously desc ibed o o he LysR- ype ac o s
(33–36)(seebelow).
3602 Nucleic Acids Resea ch, 2016, Vol. 44, No. 8
Figu e 4. Binding o 6xHis-OxyRC199S o he op AB p omo e egion.
(A) Elec opho e ic mobili y shi assay o 6xHis-OxyRC199S binding o
a DNA agmen con aining he op AB p omo e and he ups eam egu-
la o y egion. The egula o y egion o en R was used as a nega i e con ol.
(B) DNase I oo p in ing o 6xHis-OxyRC199S binding o DNA agmen s
con aining he op AB p omo e and egula o y egion wi h a 6-FAM label
in ei he he op o he bo om s and. Me hyla ed, non-me hyla ed and
GATC-less e sions o he agmen we e used.
Op ABOFF and Op ABON subpopula ions a e cha ac e ized
by in e se pa e ns o Dam me hyla ion
Single-molecule eal- ime (SMRT R
) sequencing esul s
showed ha >97% o he o al o 38 458 GATC si es
p esen in he genome o S. en e ica se o a Typhimu ium
a e me hyla ed, and ha non-me hyla ed si es a e he excep-
ion. Wi hin his se , se e al non-me hyla ed GATC si es
we e de ec ed ups eam o he op AB ope on. In o de o
analyze hem in mo e de ail, posi ion-speci ic base modi-
ica ion analyses we e pe o med. Addi ion o he i ulen
P22 H5 phage o a cul u e o S. en e ica esul s in selec-
ion o he Op ABON subpopula ion (17). Using his p o-
cedu e, a cul u e was en iched in Op ABON cells and he
me hyla ion s a e o he op AB GATC si es was analyzed
using SMRT R
sequencing (37). An o dina y cul u e, which
con ains >99% Op ABOFF cells (16,17), was also subjec ed
o SMRT R
sequencing. A o al o 246 373 (430 408) poly-
me ase eads wi h a mean polyme ase ead leng h o 10
010 (8516) bp and mean sequence co e age o 178×(329×)
we e ob ained o he Op ABON (Op ABOFF) SMRT se-
quencing. The esul s om posi ion-speci ic base modi ica-
ion analysis a e shown in supplemen a y .cs iles S1 and
S2, and can be summa ized as ollows:
(i) In an o dina y Op ABOFF cul u e, GATC1and GATC3
we e non-me hyla ed, whe eas GATC2and GATC4
we e me hyla ed (Table 1).
(ii) In he Op ABON cul u e, an in e se DNA me hyla ion
pa e n was ound: non-me hyla ion o GATC2and
GATC4and me hyla ion o GATC1and GATC3(Ta-
ble 1).
These obse a ions con i m ha es ablishmen o he
OFF and ON s a es o he op AB locus in ol es he o ma-
ion o DNA me hyla ion pa e ns, as in o he phase a ia-
ion loci unde Dam me hyla ion con ol (10,13,15).
OxyR p o ec s GATC si es om Dam me hyla ion in i o
OxyR has been p e iously desc ibed as a DNA
me hyla ion-blocking ac o , able o induce he o ma-
ion o non-me hyla ed GATC si es (15,38). To es
whe he OxyR has a simila DNA me hyla ion-blocking
abili y in he op AB ope on, he me hyla ion s a e o he
GATC si es in he op AB egula o y egion was es ed in
i o. Fo his pu pose, a Sou he n blo was pe o med using
genomic DNA ex ac ed om he wild- ype s ain and
om an oxyR mu an . The me hyla ion s a e o indi idual
GATC si es was in e ed om es ic ion analysis using
enzymes ha cu GATC sequences depending on hei
me hyla ion s a e (MboI, DpnI and Sau3AI). GATC1and
GATC3we e ound o be non-me hyla ed while GATC2
and GATC4we e ound o be me hyla ed in he wild- ype
s ain (Figu e 5). In con as , in an oxyR backg ound,
all ou GATC si es we e ound o be me hyla ed (Figu e
5). These obse a ions con i med ha OxyR has DNA
me hyla ion-blocking abili y in i o a he op AB egula o y
egion.
Mu a ions in he OBSBand OBSDOxyR binding si es abol-
ish phase a ia ion
O he ou OxyR binding hal -si es in he op AB egula o y
egion, OBSAand OBSCa e an absolu e ma ch (10 ou o
10 n ) o he consensus sequences de ined o OxyR binding
(31). In con as , OBSBand OBSDsha e only 8 and 7 ou
o 10 n wi h he consensus sequence, espec i ely. The ac
ha op AB phase a ia ion is skewed owa d he OFF s a e
led us o hypo hesize ha he deg ee o OxyR binding si e
pe ec ion played a ole in such bias. To es ou hypo he-
sis, 1 n change was in oduced in OBSBand wo nucleo ide
changes in OBSDso ha hei mu a ed e sions would sha e
9 ou o 10 n wi h he consensus sequence. Cons uc ion o
a pe ec consensus sequence was a oided since i would in-
e i ably des oy GATC2and GATC4.
The consequences o OBSBand OBSDDNA sequence
amelio a ion we e analyzed using op AB::g p (Figu e 6A)
and op AB::lac usions (Figu e 6B). Mu a ions in ei he
OBSBo OBSDabolished op AB phase a ia ion, yielding
a uni o m Op ABON popula ion. In he case o OBSB,a
single nucleo ide change led also o ull exp ession o he
Nucleic Acids Resea ch, 2016, Vol. 44, No. 8 3603
Table 1. DNA modi ica ion s a us acco ding o SMRT R
View o posi ion speci ic base-modi ica ion analysis ups eam o he op AB ope ona
Si e Genome posi ion Op ABOFF Op ABON
GATC12 361 489+ unmodi ied (1.35, 31) m6A (2.76, 59)
2 361 490−unmodi ied (1.22, 31) m6A (3.99, 46)
GATC22 361 439+ m6A (4.55, 55) unmodi ied (0.93, 49)
2 361 440−m6A (2.85, 54) unmodi ied (0.85, 37)
GATC32 361 416+ unmodi ied (0.78, 55) m6A (2.29, 45)
2 361 417−unmodi ied (0.43, 55) m6A (2.15, 36)
GATC42 361 366+ m6A (2.79, 45) unmodi ied (1.02, 52)
2 361 367−m6A (3.14, 37) unmodi ied (0.59, 47)
aIn e pulse du a ion a ios as well as s and-speci ic co e age alues a e gi en in pa en heses.
Figu e 5. Me hyla ion s a e o GATC si es in he op AB egula o y egion
in wild- ype and oxyR backg ounds. (A) Sou he n blo o genomic DNA
ob ained om wild- ype and oxyR cul u es and diges ed wi h HaeIII and
wi h AccI (con ol) and DpnI, MboI o Sau3AI. F agmen sizes a e indi-
ca ed in base pai s. (B) Diag am o he HaeIII-AccI agmen and pa e n
o agmen s ob ained.
ope on. The mu a ion in OBSDcaused a smalle inc ease in
exp ession and was epis a ic o he mu a ion in OBSB.
An in e p e a ion o hese obse a ions is ha OBSBand
OBSDDNA sequence amelio a ion may ‘ ap’ OxyR in he
Op ABON con igu a ion. In suppo o his iew, absence
o Dam me hyla ion had no e ec on op AB exp ession in
hese mu an backg ounds (Figu e 6). Hence, he p e e ence
o OxyR o ce ain OxyR-binding si es may be a key ac-
o in egula ion o op AB phase a ia ion, and al e na i e
binding o OxyR ups eam o he op AB p omo e may gen-
e a e he Op ABOFF and Op ABON subpopula ions.
SeqA con ibu es o he small size o he Op ABON subpop-
ula ion
SeqA was conside ed a po en ial ancilla y candida e o eg-
ula ion o op AB since i binds GATC si es (39) and is in-
ol ed in egula ion o o he phase a ia ion loci (40,41).
Thus we analyzed he e ec o a seqA mu a ion on op AB
exp ession and i s in luence on he o ma ion o Op AB
subpopula ions. A s ain ca ying a seqA null allele and
an op AB::lac usion o med da ke (Lac+) colonies on LB
+ X-gal han he wild- ype, and displayed equen sec o -
ing. None heless, wo g oups o di e en ly colo ed colonies
(ligh blue and da k blue) we e s ill dis inguishable (Figu e
7A), which allowed calcula ion o phase ansi ion equen-
cies. The OFF→ON ansi ion a e was ound o be 50- old
highe in a seqA backg ound (3.0 ×10−3compa ed wi h 6.1
×10−5in he wild- ype), whe eas he ON→OFF ansi ion
a es we e simila (3.1 ×10−2compa ed o 3.7 ×10−2in
he wild- ype). No su p isingly, he ␤-galac osidase ac i -
i y o an op AB::lac usion was ∼10- old highe in a seqA
backg ound (Figu e 7B).
Fluo escence assays showed ha mu a ion o seqA
caused an inc ease in he size o he Op ABON subpopula-
ion (Figu e 7C). The e ec was s onge in he p esence o
mu a ions in GATC1and/o GATC2, and o a lesse ex en
in GATC3(Supplemen a y Figu e S1). In e es ingly, when
GATC4was mu a ed, a mu a ion in seqA had an e ec op-
posi e o ha obse ed in he wild- ype: he Op ABON sub-
popula ion was educed (Supplemen a y Figu e S1). When
bo h GATC3and GATC4we e mu a ed, he seqA mu a-
ion did no ha e a signi ican e ec (Supplemen a y Fig-
u e S1). These esul s seem o indica e ha he main ole
o SeqA in he egula ion o op AB is he main enance o a