ParP prevents dissociation of CheA from chemotacticsignaling arrays and tethers them to a polar anchorSimon Ringgaarda,b, Martha Zepeda-Riveraa,b, Xiaoji Wua, Kathrin Schirnerb, Brigid M. Davisa,b,and Matthew K. Waldora,b,c,1

aDivision of Infectious Diseases, cHoward Hughes Medical Institute, Brigham andWomen’s Hospital, Boston, MA 02115; and bDepartment of Microbiology andImmunobiology, Harvard Medical School, Boston, MA 02115

Edited by E. Peter Greenberg, University of Washington, Seattle, WA, and approved December 6, 2013 (received for review August 19, 2013)

Bacterial chemotaxis proteins are organized into ordered arrays. Inperitrichous organisms, such as Escherichia coli, stochastic assem-bly processes are thought to account for the placement of chemo-taxis arrays, which are nonuniformly distributed. In contrast, wepreviously found that chemotactic signaling arrays in polarly flag-ellated vibrios are uniformly polar and that array localization isdependent on the ParA-like ATPase ParC. However, the processesthat enable ParC to facilitate array localization have not been de-scribed. Here, we show that a previously uncharacterized protein,ParP, interacts with ParC and that ParP is integral to array locali-zation in Vibrio parahaemolyticus. ParC’s principal contribution tochemotaxis appears to be via positioning of ParP. Once recruitedto the pole by ParC, ParP sequesters arrays at this site by capturingand preventing the dissociation of chemotactic signaling protein(CheA). Notably, ParP also stabilizes chemotactic protein com-plexes in the absence of ParC, indicating that some of its activityis independent of this interaction partner. ParP recruits CheA viaCheA’s localization and inheritance domain, a region found only inpolarly flagellated organisms that encode ParP, ParC, and CheA.Thus, a tripartite (ParC–ParP–CheA) interaction network enablesthe polar localization and sequestration of chemotaxis arrays inpolarly flagellated organisms. Localization and sequestration ofchemotaxis clusters adjacent to the flagella—to which the chemo-tactic signal is transmitted—facilitates proper chemotaxis as wellas accurate inheritance of these macromolecular machines.

pole development | protein localization | protein inheritance |protein sequestration | diffusion and capture

Motile bacteria use chemotaxis to survey their environmentsand navigate in response to them. In particular, chemo-

tactic sensing and response systems enable bacteria to recognizechemical gradients in their surroundings and to direct themselvestoward favorable environments and away from detrimental ones.Chemotactic behavior is mediated by two-component signal-transduction pathways that detect such gradients and transmitthis information via a phosphorelay to the bacterial flagella.Unfavorable gradients, i.e., a decrease in attractants or an in-crease in repellants, generate a signal to change flagellar rota-tion, which results in a change in swimming direction and a netmovement toward more favorable conditions. This ability to re-act to changes in the external milieu can be essential for viabilityand competitiveness (reviewed in ref. 1).Chemoeffectors typically are detected by transmembrane

chemosensory proteins termed “methyl-accepting chemotaxisproteins” (MCPs). These receptors interact with a histidine ki-nase, CheA, whose interaction with MCPs is stabilized by anadaptor protein, CheW. Upon recognition of a decrease in at-tractant or increase in repellant, the MCPs induce CheA auto-phosphorylation. Phosphorylated CheA transfers its phosphategroup to the response regulator CheY, which diffuses freely inthe cytoplasm. Accumulation and binding of phosphorylatedCheY to the flagellar switch protein FliM increases the proba-bility of a change in flagellar motor rotation, resulting in move-ment toward more favorable conditions (Fig. S1A) (1).

Chemosensory proteins are found universally in large macro-molecular clusters known as “chemotactic signaling arrays” (2–5). MCPs, together with CheA and CheW, form the sensory coreof these clusters. Interactions between the cytoplasmic signalingdomains of the receptors are sufficient for the formation of re-ceptor arrays (6), although the formation is enhanced in thepresence of CheW or CheA (6–8). Recent studies have suggestedthat array formation in Escherichia coli is a stochastic process inwhich individual receptors are inserted randomly in the mem-brane, where they diffuse freely and either join existing arrays ornucleate new ones (9). This process results in a nonuniformdistribution of signaling arrays at cell poles and randomly alongthe cell length (10). In other organisms chemosensory arrays arelocalized only to the cell poles (11–14); however, the mechanismsbehind polar localization are incompletely understood.We recently reported that the principal chemotaxis proteins in

Vibrio cholerae, the Gram-negative bacterium responsible for thediarrheal disease cholera, display a markedly different distribu-tion from that observed in E. coli (15). Newborn cells contain asingle focus of chemotaxis proteins at their old pole. Subse-quently, as cells mature, a second focus develops at the new pole,resulting in bipolar localization. Uni-and bipolar targeting ofthese chemotaxis proteins is dependent on ParC, a representa-tive of a distinct family of ParA-like ATPases encoded withinchemotaxis operons of many polar-flagellated bacteria. Like thechemotaxis signaling proteins, ParC exhibits a cell cycle-dependentunipolar/bipolar distribution (15). The mechanisms that enableParC to control the localization of chemotactic signaling arrayshave not been described. Targeting of ParC to the pole is dependent

Significance

Targeting of cellular components to a particular site in a cell isoften a highly regulated process, even in cells as small asbacteria. Robust chemotactic signaling, which is used by motilebacteria to survey their environments and navigate in responseto them, requires appropriate cellular distribution of a largechemosensory apparatus. Here, we report how polarly flagel-lated vibrios ensure polar localization of their chemotacticmachinery by capturing signaling proteins at the pole. Polarlocalization is mediated by a tripartite protein interactionnetwork in which one protein prevents disassociation of a keysignaling component from chemotactic complexes and tethersthe complexes to a polar anchor. Polar tethering and localiza-tion are prerequisites for proper chemotaxis.

Author contributions: S.R., M.Z.-R., X.W., K.S., B.M.D., and M.K.W. designed research; S.R.,M.Z.-R., X.W., and K.S. performed research; S.R., M.Z.-R., X.W., K.S., B.M.D., and M.K.W.analyzed data; and S.R., K.S., B.M.D., and M.K.W. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: mwaldor@research.bwh.harvard.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1315722111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1315722111 PNAS | Published online December 30, 2013 | E255–E264

MICRO

BIOLO

GY

PNASPL

US

on the polar determinant, HubP, although HubP and ParC do notinteract directly (16).ParA-like ATPases are responsible for the spatiotemporal

positioning of several additional intracellular processes, in-cluding positioning of chromosomes and plasmids and identifi-cation of the cell division plane (reviewed in refs. 17 and 18).ParA-like proteins modulate the localization of cytosolic che-motaxis clusters and carboxysomes in Rhodobacter sphaeroides(19) and Synechococcus elongatus (20), respectively. The locali-zation and activity of ParA-like proteins typically is controlled byATP binding and hydrolysis, because ATP binding governs theircapacity to oligomerize and to interact with other proteins(reviewed in refs. 17 and 18). Notably, many ParA-like proteinsare encoded adjacent to genes encoding interacting proteins thatalso are required for their function (e.g., ParB). Like ParAproteins, these partners typically display a restricted and dynamicdistribution within the cell. In general, the role of the partnerprotein is to act as a bridge between ParA and its target and toregulate the enzyme’s ATPase activity.Here, we investigated the function of an unannotated open

reading frame, now designated parP, which is encoded down-stream of parC. ParP proved to be a ParC partner protein, and,like ParC, ParP colocalizes with polar chemotactic signalingarrays in Vibrio parahaemolyticus. Positioning of ParP is de-pendent on ParC, and localization of ParP appears to be ParC’s

principal function. ParP promotes array stability as well as po-sitioning and retention at the pole. Following its recruitment tothe pole by ParC, ParP sequesters arrays at this site by capturingand preventing the dissociation of CheA. Notably, in contrast tomany other ParA-associated proteins, we found that ParP hasactivity that is independent of its ParA-like partner. Even in theabsence of ParC, ParP prevents the dissociation of CheA fromclusters of chemotaxis proteins. ParP maintains clusters of CheAby interacting with a localization and inheritance domain (LID)that is present only in CheA proteins with coresident ParP andParC and also can be bound by ParC. Thus, our data indicatethat a tripartite ParC–ParP–CheA interaction network promotesproper polar localization, sequestration, and inheritance of themacromolecular machine that is responsible for chemotacticsignaling.

ResultsParP, a Protein Found Within the Chemotaxis Operons of PolarlyFlagellated γ-Proteobacteria. Analysis of the V. parahaemolyticuschemotaxis operon revealed an ORF (vp2226) encoding a proteinof unknown function now designated parP (Fig. S1B). Interest-ingly, parPs always are found immediately downstream of genesencoding the ParA-like ATPase ParC, which mediates properpolar recruitment of chemotactic signaling arrays (15). Thus, asfor parC (15), parP is found only within chemotaxis operons of

wt

cheWparC

parP

parCparP

Rel

ativ

e sw

imm

ing

diam

eter

0

0.5

1.0

1.5A

Rev

ersa

ls /

sec

0

0.5

1.0

1.5B

% o

f cel

ls w

ith n

o ch

ange

in s

wim

min

g di

rect

ion

0

20

40

60C

% o

f cel

ls w

ith

non-

pola

r cl

uste

rs

0

10

20

30

40

50

% o

f cel

ls w

ith

aber

rant

clu

ster

s

No clusters Mislocalized clusters

H

0

20

40

60

80

I

F GFluorescence

Localization of chemotaxis proteins in parP

Cell length (µm)Phase contrast /

Fluorescence

Yfp

-Che

A

012345

1.7 2.7 3.7

012345

1.8 2.8 3.8 4.8

mC

herr

y-M

CP

012345

1.6 2.6 3.6 4.6

Yfp

-Che

W

Dis

tanc

e of

clu

ster

s fr

om c

ell p

ole

(µm

)

mC

herr

y-M

CP

012345

1.5 2.5 3.5 4.5

Yfp

-Che

W

012345

1.5 2.5 3.5 4.5

Yfp

-Che

A

012345

1.5 2.5 3.5

Fluorescence

Localization of chemotaxis proteins in wild-type

Cell length (µm)Phase contrast /

Fluorescence

Dis

tanc

e of

clu

ster

s fr

om c

ell p

ole

(µm

)

D E

Midcell

Cell pole

Fig. 1. ParP is required for polar localization of chemotactic signaling arrays. (A) Swimming assay of wild-type and mutant V. parahaemolyticus in 0.3% soft-agar plates. Bar graph shows the swimming diameter of mutant strains relative to wild type. Data are shown as mean and SEM. (B and C) Positional trackingof swimming wild-type and mutant V. parahaemolyticus using streaming acquisition. (B) Average rate of reversal of the swimming direction. (C) Percentageof cells with no change in swimming direction. (D–G) Fluorescence microscopy showing the intracellular localization of CheW, CheA, and MCP in wild-typeV. parahaemolyticus (D and E) and a ΔparP strain (F and G). (E and G) Graphs depict the distance of CheW, CheA, and MCP clusters from the cell poles asa function of cell length. (H) Column bar graph depicting the percentage of cells with nonpolar clusters. (I) Stacked bar graph showing the percentage of cellswith aberrant clusters, e.g., no clusters or mislocalized clusters. Data are shown as mean and SEM. (Scale bars: 2 μm.)

E256 | www.pnas.org/cgi/doi/10.1073/pnas.1315722111 Ringgaard et al.

polarly flagellated non-Enterobacteriaceae γ-proteobacteria(Fig. S1C). Consistent with a connection between ParPs andchemotaxis, the C-terminal halves of ParP proteins contain aCheW-like domain. The N-terminal halves of ParPs are veryproline rich but are variable in sequence and length; only a shortstretch of ∼10 amino acids close to the extreme N terminus isconserved among ParP proteins (Fig. S2).

ParP Is Required for Optimal Chemotaxis and Swimming Behavior. Toexplore whether ParP is involved in chemotaxis, we analyzed thechemotactic capacity of a ΔparP strain compared with wild-typeV. parahaemolyticus, either as individual cells using video trackingor collectively on soft-agar plates. On plates, the swim diameter ofmutants lacking parP was reduced by ∼25–30% compared withwild-type cells. In contrast, the swim diameter for ΔcheW cells,which have completely lost the ability to chemotax, was reducedby 95% (Fig. 1A). The reduced swim diameter was not caused byslower growth of the mutant strains (Fig. S3A) and could becomplemented by ectopic expression of YFP-ParP (Fig. S3B). Inthe single-cell tracking experiment, mutants lacking parP reversedtheir swimming direction much less frequently than wild-typeV. parahaemolyticus (Fig. 1B). Furthermore, only 10% of wild-typecells but ∼45–50% of mutant cells did not reverse their swimmingdirection during the time of the experiment (Fig. 1C). Thus, ourdata indicate that ParP is required for optimal chemotaxis andproper swimming behavior. Cells lacking parC or both parP andparC behaved similarly to the ΔparP cells (Fig. 1 A–C), suggestingthat ParP and ParC might work in the same pathway and, conse-quently, that ParP, like ParC, might be involved in localization ofchemotaxis signaling arrays.

ParP Mediates Proper Polar Localization and Inheritance of ChemotacticSignaling Arrays. To visualize the subcellular distribution of che-motaxis proteins in V. parahaemolyticus, we fused CheW, CheA,and an MCP (VP2629) to fluorescent proteins. In wild-type cellsall fusion proteins displayed two distinct localization patterns: uni-or bipolar localization (Fig. 1D, green and yellow arrowheads).The two patterns varied with cell length: Young (short) cellsshowed unipolar localization, whereas older (longer) cells showedbipolar localization (Fig. 1E), indicating that, as in V. cholerae(15), chemotaxis proteins were present initially at the old (flag-ellated) pole and then transitioned to a bipolar distribution asV. parahaemolyticus progressed through the cell cycle.Strikingly, deletion of ParP resulted in mislocalization of sig-

naling proteins (Fig. 1F, white arrowheads), with nonpolar arrayslocalized randomly along the cell length (Fig. 1G) in ∼25–30% ofcells (Fig. 1H). Furthermore, cells often divided without estab-lishing a bipolar localization pattern (Fig. 1F, red arrowheads),and in consequence ∼25–30% of mutant cells completely lackedarrays (Fig. 1 F, yellow arrowhead, and I). Thus, in the absenceof ParP, ∼50–60% of cells either lack arrays or contain mis-localized arrays (Fig. 1I), indicating that ParP plays a pivotal rolein the proper localization and inheritance of polar chemotacticsignaling arrays. A similar deficiency in localization of chemo-taxis proteins was observed in V. cholerae lacking ParP (Fig. S3 Cand D).

Proper Polar Localization of ParP Is Dependent on ParC. As in V.cholerae, ParC also displayed cell cycle-dependent polar locali-zation (Fig. S4A) and mediated reliable localization and in-heritance of chemotaxis proteins in V. parahaemolyticus (Fig. S4B and C). Mislocalization and inheritance of signaling arrays ina ΔparC strain was similar to that observed for the ΔparP mutant(Fig. 1 H and I and Fig. S4 B and C). Localization defects in theParP- and ParC-deficient strains were corrected by comple-mentation with YFP-ParP and YFP-ParC, respectively (Fig. 1H),confirming that mislocalization of chemotaxis clusters resulted

from the absence of ParP or ParC and demonstrating that bothYFP fusion proteins are functional.Interestingly, YFP-ParP displayed cell cycle-dependent local-

ization similar to that of chemotaxis proteins and ParC (Fig. 2 Aand B). Time-lapse microscopy showed that ParP is localized tothe old (flagellated) pole in newborn cells (Fig. 2C, red asterisk)and later is recruited to the new pole (Fig. 2C, yellow asterisk),resulting in bipolar localization (Fig. 2C). Dual-labeling experi-ments with YFP-ParC/CFP-CheW or YFP-ParP/CFP-CheW (Fig.S4 D and G) showed that a higher proportion of cells possessedbipolar ParC or ParP than CheW (Fig. S4 E and H), suggestingthat ParC and ParP arrive at the new pole before CheW duringthe cell cycle. Plotting the distances of foci from the old cell poleas a function of cell length further indicated that recruitment ofParC/ParP precedes that of CheW (Fig. S4 F and I, dashed lines).In contrast, ParP and ParC appeared at the new pole simulta-neously (Fig. S4 J–L). This ordered development of the new poleis consistent with ParC/ParP driving recruitment of chemotacticsignaling arrays to the new pole before cell division.We also observed that uni- and bipolar localization of ParP

clusters, as well as their inheritance, was dependent on ParC(Fig. 2 D and E). In the absence of ParC, nonpolar ParP clusters(Fig. 2D, white arrowheads) were detected positioned at randomalong the cell length (Fig. 2E) in ∼25–30% of cells (Fig. 1 H andI). Furthermore, in the absence of ParC, cells did not establisha bipolar distribution of YFP-ParP before septation (Fig. 2D, redarrowheads). Consequently, daughter cells did not all inherita ParP cluster at cell division, and ∼25–30% of mutant cellscompletely lacked clusters (Figs. 2D, yellow arrowheads, and1I). The fact that 50–60% of ΔparC cells were compromised forlocalization or existence of ParP clusters (Fig. 1I) indicates thatParC mediates reliable polar localization and inheritance of ParPclusters and hence that ParP acts downstream of ParC in guidingpolar localization of signaling arrays.

ParP Promotes Array Localization by Preventing Disassociation ofCheA. Fluorescence recovery after photobleaching (FRAP)experiments were carried out to dissect the roles of ParP andParC in localization of chemotactic signaling arrays. A singlefluorescent cluster of YFP-CheA was photobleached per cell,and the recovery of fluorescence intensity was measured asa function of time relative to time 0 (immediately afterbleaching). In wild-type cells, ∼40% of the initial fluorescenceintensity of CheA was recovered by 400 s after photobleaching.The absence of ParC did not influence CheA recovery dynamics,even in mislocalized clusters, but the absence of ParP hada striking effect on the accumulation of new CheA in the cluster(Fig. 3 A and B): In the ΔparP strain, only 15% of initial CheAintensity was recovered by 400 s, regardless of whether thebleached cluster was polar or lateral. Moreover, in wild-type andΔparC strains, recovery continued throughout the 400-s recordingperiod, but in the ΔparP strain, recovery reached steady state after∼50 s (Fig. 3B). Because the initial “on-rate” of CheA recovery(during the first ∼30 s after bleaching) is similar in wild-type,ΔparC, and ΔparP strains, the early plateau in CheA recoveryin the absence of ParP suggests that ParP modulates CheA’sdissociation (the “off-rate”) from signaling arrays. Analysis ofnonbleached clusters also revealed that CheA accumulated con-tinuously in clusters in the presence of ParP but not in its absence(Fig. 3C). Further localization studies, performed in E. coli,suggested that ParP acts directly upon CheA to promote CheAclustering. We used a DivIVA microscopy-based assay, adaptedfrom ref. 21, in which the first 60 amino acids of DivIVA fromBacillus subtilis [DivIVA1:60, sufficient for DivIVA membranelocalization in E. coli (22)] were fused to the N terminus of CFP-ParP. YFP-CheA alone localized diffusely in the E. coli cytoplasm(Fig. S5A); however, when YFP-CheA was coexpressed withDivIVA1:60-CFP-ParP [which yields predominantly polar ParP

Ringgaard et al. PNAS | Published online December 30, 2013 | E257

MICRO

BIOLO

GY

PNASPL

US

(Fig. S5A)], CheA was detected only in compact clusters at thecell poles along with ParP (Fig. S5B). These data indicate thatParP localizes chemotaxis proteins to the cell pole by binding to

CheA and that it promotes the gradual accumulation of chemo-tactic arrays at the pole by preventing dissociation of CheA fromprotein complexes.

C

Yfp

-Par

P

012345

1.5 2.5 3.5 4.5

Localization of Yfp-ParP in wild-typeA

D

Yfp

-Par

P

012345

1.7 3 4.3

Localization of Yfp-ParP in parC

Dis

tanc

e of

clu

ster

s fr

om c

ell p

ole

(µm

)

Fluorescence Cell length (µm)Phase contrast /

Fluorescence

B

E

Midcell

Cell pole

Time-lapse of Yfp-ParP in wild-type V. parahaemolyticus

Fig. 2. Cell cycle-dependent polar localization of ParP requires ParC. (A) Fluorescence microscopy showing the intracellular localization of YFP-ParP in wild-type V. parahaemolyticus. (B) Graph depicts the distance of YFP-ParP clusters from the cell poles as a function of cell length. (C) Time-lapse microscopy of YFP-ParP in wild-type cells. Numbers indicate minutes elapsed. Red asterisks indicate old (flagellated) cell poles. Yellow asterisks indicate new cell poles. (D)Fluorescence microscopy showing the intracellular localization of YFP-ParP in V. parahaemolyticus ΔparC. (E) Graph depicts the distance of YFP-ParP clustersfrom the cell poles as a function of cell length. (Scale bar: 2 μm.)

A

wild-type parPparC

0

0.2

0.4

0.6

Time (s)0 100 200 300 400

FRAP of Yfp-ParP

wild-typeparC

0

0.2

0.4

0.6

0 100 200 300 400Time (s)

FRAP of Yfp-ParC

wild-typeparP

0

0.2

0.4

0.6

0 50 100Time (s)

wild-typeparP

Time-course of Yfp-CheA in non-bleached clusters

0 100 200 300 4000.5

1.0

1.5

2.0

Time (s)

C D E

B FRAP of Yfp-CheAFRAP of Yfp-CheA

wt

parC

2s

parP

Bleach 400s

Fig. 3. ParP sequesters signaling arrays at the pole by preventing the disassociation of CheA. (A and B) FRAP experiments on YFP-CheA, in wild-type andmutant backgrounds. (A) Representative YFP-CheA clusters from FRAP experiments in wild-type, ΔparC, and ΔparP backgrounds. The cell is outlined at pre- andpostbleaching with a green dot representing the localization of the CheA cluster and a dashed circle showing the region bleached. (B) Graph depicting therelative fluorescence intensity of the bleached region during postbleach time-lapse series of YFP-CheA. (C) Fluorescence intensity of nonbleached CheA clustersduring time-lapse microscopy plotted as a function of time relative to the initial intensity. (D and E) Graphs depicting the relative fluorescence intensity of thebleached region during postbleach time-lapse series of YFP-ParC (D), and YFP-ParP (E) in wild-type and mutant backgrounds. Fluorescence intensity is plotted asa function of time. Values are plotted relative to the fluorescence intensity immediately before bleaching. Data are shown as mean and SEM.

E258 | www.pnas.org/cgi/doi/10.1073/pnas.1315722111 Ringgaard et al.

X X X

X X X X X X X X X X

X X X X X X X X X

X X X X X X X X X X

X X X X X X X X X X

X

X X X

X X X X X X X X X

X X X

X X X X X X X X X

X

X X X

X X X

X X X X X X X X X

X X X X X X X

X X X X X X

X

A B

Che

W

Par

C

Che

A

Che

A:1

/606

Che

A:6

06/7

45

Che

A:3

54/7

45C

heA

:354

/606

Che

A:1

/110

Che

A:1

/354

Che

A:1

09/3

54C

heA

:109

/606

Che

Y

Che

A:2

16/6

06C

heA

:216

/354

Che

A:2

16/2

91C

heA

:109

/215

Che

A:1

/215

Par

P

CheA:1/110

CheA:109/354

CheA:354/606

CheW

ParC

CheA

CheA:1/606

CheA:606/745

CheA:354/745

CheA:1/354

CheA:109/606

CheY

CheA:216/606CheA:216/354CheA:216/291CheA:109/215

CheA:1/215

ParP

Blue color indicates the LID domain

D

CheW

ParC

CheA

ParP

Vector

E

CheW

Che

A

Par

C

ParP

CheY

P1 P3 P4 P5P2

LID

ParC/ParP/CheA-LIDinteraction network

responsible for polar localization

Network responsible for chemotactic signaling

MCP

P1 P3 P4 P5P2

CheA:109/354CheA:1/606

CheA:1/110CheA

CheA:1/354

CheA:606/745CheA:354/606CheA:354/745

CheA:109/606

CheA:1/215

CheA:109/215CheA:216/291CheA:216/354CheA:216/606

C

Fig. 4. A domain in CheA that coevolved with ParP and ParC mediates a tripartite interaction network. (A) BTH analysis testing for interaction betweenCheW, CheA, ParC, and ParP. Formation of blue colonies indicates that a protein–protein interaction occurs. (B) Phylogenetic tree from an alignment of CheAproteins with an associated ParC and ParP (blue), with an associated ParC but no ParP (orange), and with no ParC or ParP present in the chemotaxis operon(green). CheA protein sequences from γ-proteobacteria without an associated ParC and/or ParP were chosen at random from a STRING analysis of ParC. (C)Schematic of CheA domain structure and CheA truncations used for the BTH analysis in D. P1–P5 refers to specific structural and functional domains of CheA:P1 contains the site of phosphorylation; P2 is required for interaction with CheY; P3 is required for dimerization; P4 is the histidine-kinase domain; and P5 ishomologous to CheW and is required for interaction with CheW and the MCPs. Allele names indicate the remaining region of CheA, e.g., CheA:1/110 retainsamino acids 1–110. (D) BTH analysis testing for interaction between ParC, ParP, CheW, CheY, CheA, and truncated versions of CheA. An “x” indicates that aninteraction was detected. (E) Schematic of the two protein interaction networks responsible for chemotactic signaling and polar localization of signalingarrays respectively. The two interaction networks are linked through CheA but interact with distinct domains within CheA that are specific for each pathway.LID indicates the ParP/ParC interaction domain of CheA.

Ringgaard et al. PNAS | Published online December 30, 2013 | E259

MICRO

BIOLO

GY

PNASPL

US

In addition to promoting polar CheA accumulation, our dataalso indicate that ParP promotes the accumulation of ParC at thecell poles in V. parahaemolyticus. A significant fraction of ParC islocalized diffusely rather than polarly in the absence of ParP(Fig. S5C), and the recovery dynamics of photobleached YFP-ParC in the ΔparP background were similar to those of CheA(Fig. 3D). Furthermore, as with CheA recovery in a ParC-deficient strain, YFP-ParP recovery was not influenced by theabsence of ParC (Fig. 3E). Thus, our data support a model inwhich ParP and ParC together direct polar localization of che-motactic signaling arrays, but ParC primarily identifies theirtarget site, whereas ParP sequesters arrays at this site by pre-venting the dissociation of CheA.

ParP, ParC, and CheA Form a Conserved Interaction Network. Weperformed bacterial two-hybrid (BTH) analyses of ParP, ParC,and chemotaxis proteins to investigate further the means bywhich these proteins are recruited to the cell poles. Theseexperiments indicated that ParP and ParC interact and that ParP

self-interacts (Fig. 4A). They also revealed that both ParC andParP interact with CheA but not with CheW (Fig. 4A). Theseresults further support the idea that ParP retains chemotaxisproteins at the pole via its interaction with CheA. Notably, theseresults also indicate that the well-conserved chemotactic signal-ing network has been amended to include a new subnetwork ofinteractions between ParP, ParC and CheA (Fig. 4E).

Identification of a Domain in CheA Proteins Required for Interactionwith ParP and ParC. Bioinformatic and phylogenetic analysis ofdiverse CheA sequences revealed that CheA proteins fromorganisms encoding both ParC and ParP homologs form theirown clade (Fig. 4B, blue), distinct from CheAs associated onlywith ParC (Fig. 4B, orange) and CheAs with no associated ParCor ParP (Fig. 4B, green). In particular, one region within theusual CheY interaction domain of CheA distinguished thesegroups (Fig. S6, red square), and we predicted that this regionmediates interactions with ParP and/or ParC. We mapped theregions of CheA that interact with its different partners using a

B C

ParC

CheA

ParP

CheA

ParP

ParC

Proline rich

Y16A W338A

parPA

Cfp

-Che

AY

fp-P

arP

W33

8AP

hase

con

tras

tE Localization of Yfp-ParP

variants in parP

% o

f cel

ls w

ith n

on-p

olar

or

lack

ing

clus

ters

100

20304050

A A

Localization of Cfp-CheA in strains encoding ParP variants

F

0

20

40

60

parP background

% o

f cel

ls w

ith n

on-p

olar

or

lack

ing

clus

ters

D

Cfp

-Che

AY

fp-P

arP

Y16

AP

hase

con

tras

t

Cfp-CheA and Yfp-ParPY16A in parP

0

250

500

750

1000

0 400 800 1200(AU)

Localization of CheA and ParPY16A in parP

Yfp

-Par

PY

16A

Cfp

-Che

A

G

0

200

400

600

800

1000

0 400 800 1200

Localization of CheA and ParPW338A in parP

Cfp

-Che

AY

fp-P

arP

W33

8A

(AU)

I

H Cfp-CheA and Yfp-ParPW338A in parP

Fig. 5. Distinct interactions between ParP, ParC, and CheA drive polar localization of signaling arrays. (A) Schematic of ParP domains and the location ofY16A and W338A within ParP. (B) BTH analysis of interactions between ParP, ParPY16A, ParC, and CheA. (C) BTH analysis of interactions between ParP,ParPW338A, ParC, and CheA. (D and H) Dual-labeling microscopy showing the localization of YFP-ParPY16A and CFP-CheA (D) and YFP-ParPW338A and CFP-CheA (H) in V. parahaemolyticus ΔparP. (E) Bar graph depicting the percentage of cells with nonpolar or absent YFP-ParP, YFP-ParPY16A, and YFP-ParPW338Aclusters in V. parahaemolyticus ΔparP. (F) Bar graph depicting the percentage of cells with nonpolar or absent CFP-CheA clusters in wild-type, parPY16A, andparPW338A backgrounds. (G and I) Graphs depicting the coordinates of YFP-ParPY16A and CFP-CheA (G) and (YFP-ParPW338A and CFP-CheA (I) in a field ofcells in a colocalization study in V. parahaemolyticus ΔparP.

E260 | www.pnas.org/cgi/doi/10.1073/pnas.1315722111 Ringgaard et al.

series of CheA truncations (Fig. 4C) and found that the expectedregions (4, 23–25) mediated CheA’s interactions with CheY,CheW, and itself (Fig. 4 C–E). Strikingly, we found that the re-gion specific to CheAs with an associated ParP and ParC(CheA:216/291) was necessary and sufficient for interaction withboth ParP and ParC (Fig. 4D, blue). Collectively, these resultssuggest that during the coevolution of CheA, ParP, and ParC,a new LID in CheA emerged (Fig. 4E). Noticeably, as in the caseof ParC (15) and ParP, the CheA LID was present only in bac-teria with polar flagella, in which it apparently is beneficial toplace chemotactic signaling arrays in close proximity to the fla-gellum. Bipolar localization of arrays likely is established late inthe cell cycle to ensure correct inheritance of arrays, so that allcells contain polar arrays at the old pole (the flagellar site) im-mediately after cell division.

Distinct Regions in ParP Mediate Interaction with ParC and CheA.Thereis a particularly high conservation of eight amino acids near theN terminus of ParP proteins (Fig. S2). Substitution of one of these(Y16) for alanine (ParPY16A) markedly reduced the resultingmutant’s interaction with ParC in the two-hybrid assay but not itsinteractions with CheA or wild-type ParP (Fig. 5 A and B).Furthermore, in contrast to YFP-ParP, YFP-ParPY16A wasnot recruited to DivIVA1/60-CFP-ParC in a microscopy-basedinteraction assay (Fig. S7A). Together, these data provide evidencethat the highly conserved region at the N terminus of ParP proteinsconstitutes their ParC interaction domain.Additionally, a stretch of five amino acids (R336, P337, W338,

L339, and G341) in the C-terminal CheW-like domain of ParP(Fig. S2) was 100% conserved among ParPs; therefore theseamino acids were chosen as good candidates for mutagenesis.Analyses of mutants revealed that replacement of W338 withalanine completely abolished interaction between the mutantParP (ParPW338A) and CheA in the two-hybrid assay but left itsinteraction with ParC and its capacity to self-interact unaffected(Fig. 5 A and C). These results suggest that ParP’s CheW-likedomain is responsible for its interaction with CheA and indicatethat distinct regions within ParP mediate its interactions withParC and CheA. Notably, W338 is not contained within the re-gion of ParP’s CheW-like domain that enables CheW’s in-teraction with CheA (Fig. S2) (24). Thus, although parP appar-ently arose via duplication of cheW, subsequent evolution both ofParP and CheA has markedly altered the nature of the ParP/CheA interaction.

Interaction Between ParC and ParP Is Required for Polar Localizationof ParP and Chemotaxis Signaling Arrays. To test the significance ofinteractions between ParC and ParP, we assessed the effect ofthe Y16A point mutation on localization of YFP-ParPY16A,CFP-CheA, and CFP-CheW in a ΔparP background. Strikingly,loss of ParC–ParP interaction resulted in mislocalized clusters ofYFP-ParPY16A, CFP-CheA, and CFP-CheW (Fig. 5D and Fig.S7B, white arrowheads). Clusters of all three proteins formed ataberrant (nonpolar) sites or were completely absent in ∼50% ofcells (Fig. 5 E and F and Fig. S7C), a defect in localization andinheritance similar to that observed in ΔparC or ΔparP back-grounds (Fig. 1I). The YFP-ParPY16A/CFP-CheA and YFP-ParPY16A/CFP-CheW clusters colocalized in almost 100% ofcells (Fig. 5G and Fig. S7D), presumably because the mutantParP still can interact with CheA. Thus, ParP’s interaction withParC is pivotal for reliable recruitment of polar ParP clusters,and abrogation of this interaction significantly impairs V. para-haemolyticus’ capacity to retain and sequester clusters of che-motaxis proteins at its cell poles.

Interaction Between ParP and CheA Is Required for Polar Localizationof CheA and Signaling Arrays. Unlike ParPY16A, ParPW338Adisplayed wild-type–like uni- and bipolar localization in a ΔparP

background (Fig. 5 E and H), consistent with its unimpairedcapacity to interact with ParC. However, coexpression of YFP-ParPW338A/CFP-CheA and YFP-ParPW338A/CFP-CheW inthe absence of wild-type ParP resulted in aberrantly localized(nonpolar) CheA and CheW clusters or the absence of CheA andCheW clusters in ∼50% of cells (Fig. 5 F and H and Fig. S7 C andE), a deficiency in localization and inheritance similar to thatobserved in a parP background (Fig. 1I). Consequently, colocal-ization of ParPW338A with CheA and CheW was reduced sig-nificantly compared with wild-type (Fig. 5I and Fig. S7F). Thus,

F Localization of Cfp-ParP in strains encoding mutant ParCs

0

20

40

60%

of c

ells

with

non

-pol

ar

or la

ckin

g cl

uste

rs

Localization of Cfp-ParP in strains encoding mutant ParCsCfp-ParP Overlay

parC

K15

Qpa

rCG

11V

E

G HLocalization of Cfp-CheW in strains encoding mutant ParCs Localization of Cfp-CheW in

strains encoding mutant ParCs

0

20

40

60

Strain background

Strain background

% o

f cel

ls w

ith n

on-p

olar

or

lack

ing

clus

ters

Cfp-CheW Overlay

parC

K15

Qpa

rCG

11V

ATP concentration (mM)

mM

AT

P h

ydro

lyze

d pe

rm

inut

e pe

r m

ol p

rote

in

0.0 0.5 1.0 1.5 2.0 2.52.5

3.0

3.5

4.0

4.5

5.02 mM ATP

4 mM ATP

ParC concentration (µM)0 2 4 6 8 10

0

1

2

3

4 2 mM ATP

4 mM ATP

BA

D Localization of Yfp-ParC point mutants in parC

Yfp Overlay

Par

CK

15Q

Par

CG

11V

C

ParC

ParCG11V

ParCK15Q

Par

P

Che

A

Fig. 6. The ATP hydrolysis cycle of ParC regulates the polarity and locali-zation of ParC, ParP, and signaling arrays. (A and B) ATPase activity of His8-ParC as a function of ATP concentration (A) and protein concentration (B).(C) BTH analysis of interactions of ParC, ParCK15Q, and ParCG11V with ParPand CheA, respectively. (D) Localization of YFP-ParCK15Q and YFP-ParCG11Vin V. parahaemolyticus ΔparC. (E and G) Localization of CFP-ParP (E) and CFP-CheW (G) in V. parahaemolyticus strains encoding parC variants. (F and H) Bargraphs depicting the percentage of cells with nonpolar or absent CFP-ParP (F)and CFP-CheW (H) clusters in V. parahaemolyticus strains encoding parC variants.

Ringgaard et al. PNAS | Published online December 30, 2013 | E261

MICRO

BIOLO

GY

PNASPL

US

our data indicate that the interaction between ParP and CheA’sLID is critical for reliable polar recruitment and sequestration ofCheA (and, in consequence, of chemotaxis clusters) as well as forthe proper inheritance of clusters at cell division.

ParC’s ATP Hydrolysis Cycle Regulates Its Polar Localization andInteraction with ParP and CheA. Because the localization and ac-tivity of ParA-like proteins typically is controlled by ATP bindingand hydrolysis, we analyzed the effect of ParC’s ATP hydrolysiscycle in promoting polarity and interaction with its partnerproteins. Initially, using purified ParC, we confirmed that ParCpossessed ATPase activity in vitro (Fig. 6 A and B). ParC hasa weak intrinsic ATPase activity of Vmax = 4.3 mM ATPhydrolyzed·min−1·mol−1 protein and a Km = 20 μM ATP, whichis similar to that of other Walker-type ATPases (15, 26). Basedon previous characterization of ParA mutants, we also engi-neered point mutations in ParC that are expected (i) to abolishATP binding and lock ParC in the Apo-monomer state (ParCK15Q);or (ii) to enable ATP binding but inhibit hydrolysis and allow ParCto cycle between existence as an ATP-bound monomer and a2ATP-bound dimer (ParCG11V). Notably, BTH analysis suggestedthat both mutations impair ParC’s ability to interact with its partnerproteins. ParCK15Q exhibited essentially no interaction with eitherParP or CheA, and ParCG11V exhibited a markedly reduced(relative to wild-type ParC) interaction with ParP, and no in-teraction with CheA (Fig. 6C). However, analysis of the localizationof mutant YFP-ParCs revealed that only ParCK15Q displayed ab-errant localization; this mutant exhibited a largely diffuse distri-bution throughout the cytoplasm (Fig. 6D). In contrast,localization of YFP-ParCG11V was similar to that of wild-typeParC (Fig. 6D). These data suggest that ATP binding, but nothydrolysis, is required for polar targeting of ParC in V. para-haemolyticus, as previously was observed in V. cholerae (15).

Polar Localization of ParP and Chemotactic Signaling Arrays IsDependent on ParC’s ATP Hydrolysis Cycle. We also analyzed theimpact of ParC’s ATPase cycle on the localization of ParP andchemotactic signaling arrays (using CheW as a marker protein).Consistent with its diffuse localization and inability to interactwith ParP, ParCK15Q was unable to recruit ParP and CheW tothe cell poles (Fig. 6 E and G, white arrowheads). Approximately50% of cells expressing ParCK15Q contained nonpolar ParP orCheW clusters or lacked clusters altogether (Fig. 6 F and H).Intriguingly, despite its presence at the pole, YFP-ParCG11Vsimilarly was unable to recruit either ParP or CheW to this site(Fig. 6 E and G, white arrowheads). Approximately 50% of cellsexpressing ParCG11V exhibited either nonpolar or no detectableclusters (Fig. 6 F and H), similar to cells lacking ParC (Fig. 1 Hand I and Fig. S4B) and cells expressing ParCK15Q (Fig. 6 F andH). Collectively, these data suggest that distinct stages in ParC’sATP hydrolysis cycle are required for ParC polar localizationand ParC-dependent localization of partner proteins and thatrecruitment of partner proteins is an active process, powered byATP hydrolysis.

DiscussionBacterial chemotaxis is mediated by a macromolecular machinethat consists of a large array of chemoreceptors and chemo-signaling proteins. Distribution of these arrays within somebacterial species (e.g., E. coli) is thought to result from stochasticassembly processes. Here, we present data indicating that invibrios with polar flagella the localization and retention of theseorganelles at the cell pole is spatially and temporally controlledby interactions between three proteins: the ParA-like ATPaseParC; ParP, which is the product of an adjacent gene with nopreviously known function; and CheA, which is a conservedchemotactic signaling protein. ParP and ParC interact with eachother and with CheA’s LID, a region that is found only in polarly

flagellated organisms that encode ParC and ParP proteins (Fig.4), and all three proteins self-interact. Both ParC and ParPpromote proper polar localization of arrays (Figs. 1 and 2). ParCprimarily governs array localization by mediating the polar lo-calization of ParP, whereas ParP serves both to retain CheA insignaling arrays, presumably stabilizing them, and to sequesterarrays at the cell pole. Disruption of ParP’s interaction withCheA’s LID or its interaction with ParC results in mislocaliza-tion of signaling arrays (Fig. 5). Sequestration of chemotaxisclusters at the pole (adjacent to the flagella) facilitates properchemotaxis (Fig. 1 A–C) as well as accurate inheritance of thesemacromolecular machines (Fig. 1 H and I).ParP likely promotes the localization/stabilization of chemo-

tactic signaling arrays via multiple processes. ParP’s ability tointeract with the CheA LID, coupled with its capacity to auto-associate, presumably maintains CheA molecules in close prox-imity to each other. This interaction with and clustering of CheA,which can occur at polar or lateral positions, also should allowParP to influence the localization of indirectly associated pro-teins, such as CheW and MCPs, which always are found ina complex with CheA (1–5, 10, 27), and thereby stabilize che-motactic signaling arrays (Figs. 3 and 5). In addition, ParPinteracts with ParC, which enables polar recruitment of ParP,and consequently of the associated chemotactic signaling arrays.Notably, the polar localization of ParP and ParC is interde-pendent, perhaps because ParP also stabilizes polar clusters ofParC (Fig. 3D). This polar retention of ParC could be a directconsequence of the interaction between the two proteins; how-ever, it also may be facilitated by an interaction between ParCand CheA. Although the ParC–CheA interaction apparently isinsufficient to induce polar localization of chemotaxis proteins inthe absence of ParP, it nonetheless may aid in corralling ParCand minimizing its diffusion away from the pole when ParP ispresent. Finally, it is possible that ParP may aid in tethering ParCto the polar anchor protein HubP.Because the localization and activity of ParA-like proteins

typically is regulated by their ATP hydrolysis cycle, we analyzedthe role of ParC’s ATP binding and hydrolysis in polarity andinteraction with ParP and CheA. Our data suggest that distinctstages in ParC’s ATP hydrolysis cycle drive its own polar local-ization and that of its partner proteins and chemotactic signalingarrays (Fig. 6). ParC’s polar localization appears to be dependenton its capacity to bind (but not hydrolyze) ATP and thus likelyrequires dimerization, which typically is associated with ATPbinding for Walker-type ATPases (17, 18). In contrast, ParC-mediated localization of ParP and other components of che-motaxis signaling arrays appears to require both ATP bindingand hydrolysis, because ParCG11V is unable to mediate properpolar localization of ParP and CheW despite the presence ofParCG11V at the cell poles (Fig. 6).Processes regulated by ParA-like proteins often are regulated

by or are dependent on associated proteins that modulate ParAATPase activity (17, 18, 26, 28–31). It is possible that ParPsimilarly modulates ParC’s enzymatic activity; to date, we havebeen unable to purify ParP to test this possibility. However, it isimportant to note that, in contrast to many other ParA-associ-ated proteins (such as plasmid-borne ParBs), at least a subset ofParP activity is independent of its ParA-like partner. We ob-served that ParP prevents the dissociation of CheA from clustersof chemotaxis proteins (Fig. 3 A and B) even in the absence ofParC. Thus, ParP’s role in cluster stabilization is separable fromits contribution to cluster localization. It is unusual for a ParAinteraction partner to possess its own distinct function in addi-tion to regulating ParA ATPase activity and acting as an adaptorbetween the ParA-like protein and its target.ParC/ParP/CheA systems likely function to promote polar lo-

calization and sequestration of chemotaxis proteins in the otherpolarly flagellated γ-proteobacteria that encode these proteins

E262 | www.pnas.org/cgi/doi/10.1073/pnas.1315722111 Ringgaard et al.

(e.g., Fig. 4). Interestingly, the distribution and inheritance ofcytosolic (nonpolar) chemotaxis clusters also has been shown tobe mediated by a ParA-like protein (PpfA) and an adjacentlyencoded protein, transducer-like protein T (TlpT), in R. sphaeroides(19). PpfA belongs to a clade of ParA-like proteins that are in-volved in plasmid segregation, a clade distinct from that ofParCs, which form their own group of ParA-like proteins (15).TlpT is a cytoplasmic chemoreceptor that also regulates PpfAfunction and localization, likely by modulating PpfA ATP hy-drolysis and thereby regulating PpfA’s association with the nu-cleoid. PpfA is thought to ensure proper inheritance of cytosolicchemotaxis clusters via a mechanism similar to that used byParAs for plasmid segregation (32). Notably, TlpT, like ParP,appears to have originated as a core component of the chemo-tactic apparatus. ParP, like TlpT, promotes inheritance of che-motactic machinery (19); however, it differs in that it also aids inpositioning this machinery near its downstream effector (theflagellum) and stabilizes the chemotactic signaling complexes viaits association with CheA.In E. coli, which lacks ParC and ParP homologs, clustering of

chemotaxis proteins is believed to occur stochastically, indepen-dent of a specific targeting mechanism. MCPs likely are insertedat random sites throughout the cell membrane and then aresubject to diffusion-mediated clustering that results in the re-cruitment of additional chemotaxis proteins (6, 9, 33–36). Be-cause V. parahaemolyticus chemotaxis proteins form randomlylocalized clusters in the absence of ParC and ParP, it is likely thatsuch a diffusion-mediated mechanism also contributes to theassembly of the V. parahaemolyticus chemotactic machinery.Given the high frequency of polar arrays in E. coli, stochastic pro-cesses also likely account for the ∼50% of V. parahaemolyticus cellsthat contain polar chemotaxis arrays in the absence of ParC orParP. However, the presence of ParC and ParP apparently convertsthis otherwise stochastic process of array assembly and localizationto an anchored “diffusion and capture” process (37, 38). Whenpositioned at the pole by ParC, ParP captures and sequesterschemotaxis complexes by preventing the release of CheA. Thetripartite ParC–ParP–CheA interaction network enables reliablepolar localization, sequestration, and inheritance of chemotaxisarrays. Thus, a well-established cellular system has been amen-ded by increasing CheA’s connectivity through network rewiring.By redefining the topology of the chemotactic interaction network,evolution has enabled the emergence of new complex processes,regulatory mechanisms that enable spatiotemporally controlledpolar localization and subsequent inheritance of the macromo-lecular machine responsible for chemotactic signaling.

Materials and MethodsGrowth Conditions and Media. If not otherwise mentioned, V. para-haemolyticus, V. cholerae, and E. coli were grown in LB medium or on LBagar plates at 37 °C containing antibiotics at the following concentrations:chloramphenicol 20 μg/mL for E. coli and 5 μg/mL for V. cholerae andV. parahaemolyticus; streptomycin 200 μg/mL; kanamycin 50 μg/mL; ampicillin100 μg/mL; and carbenicillin 50 μg/mL.

Strains and Plasmids. Strains and plasmids used and constructed for this studyare listed in Table S1. All V. parahaemolyticus strains used in this study arederivatives of the wild-type V. parahaemolyticus RIMD 2210633. E. coli strainDH5αλpir was used for cloning, and E. coli strain SM10λpir was used forconjugation of DNA into V. parahaemolyticus and V. cholerae (39). Con-struction of plasmids is detailed in SI Materials and Methods. The primers arelisted in Table S2.

Fluorescence and Time-Lapse Microscopy. Fluorescence microscopy of V. par-ahaemolyticus was carried out essentially as described in refs. 15 and 40:5 mL LB was inoculated with a colony of cells harboring the relevant plasmidand grown to OD600 = 0.1 at 37 °C; then expression of fluorescent fusionproteins was induced by adding L-arabinose to a final concentration of 0.2%.The cultures were incubated for an additional 2 h before microscopy.Fluorescence microscopy of E. coli was carried out essentially as described in

ref. 40: 5 mL LB was inoculated with 50 μL of an overnight culture, grown toOD600 = 0.3 at 37 °C; then expression of fluorescent fusion proteins was in-duced by adding L-arabinose to a final concentration of 0.2%. Cultures wereincubated for an additional 3 h before microscopy. For microscopy, cellswere mounted on pads of 1% agarose in 20% (vol/vol) PBS/20% (vol/vol) LBon microscope slides. Images were taken with Zeiss Axioplan 2 microscopeequipped with a 100× Plan lens and a Hamamatsu cooled CCD camera.Microscopy images were analyzed using ImageJ imaging software. Furtherdetails are given in SI Materials and Methods.

In experiments measuring the number of ParC, ParP, and/or chemotaxisprotein clusters, experiments were performed a minimum of three times. Foreach experiment 250–500 cells were counted. Clusters were enumerated byhand using the cell counter plug-in for ImageJ. Data were presented as themean of the total number of experiments ± SEM.

For FRAP microscopy, cells were treated and mounted on agarose pads asdescribed for time-lapse fluorescence microscopy. Microscopy was performedusing a Nikon eclipse Ti motorized inverted microscope with Perfect FocusSystem, and images were obtained with an Andor Technology iXon3 cooledCCD camera. For FRAP experiments, areas of interest were bleached withthree pulses using a 405-nm MicroPoint laser at 40% intensity. Images thenwere recorded at the indicated time intervals, and the fluorescence intensityof the bleached region was measured and plotted relative to the imagerecorded immediately prebleach. For analyses of the parP and parC mutantstrains, approximately equal numbers of lateral and polar foci were ana-lyzed; localization was not found to influence the rate or extent of recovery.

Measuring the Frequency of Reversal in Swimming Direction. A colony wasinoculated in 5 mL LB medium and incubated at 37 °C for 2 h. Cultures werediluted to OD600 = 0.01 in 5 mL LB and incubated at 37 °C until they reachedOD600 = 1.1. Then 1 μL was spotted on a microscope slide, covered witha LifterSlip cover glass (Erie Scientific Company), and used immediately formicroscopy. Images were recorded every 112 ms using the streaming ac-quisition function in the Metamorph software. Three individual experimentswere performed, and in each experiment the swimming of 20–25 cells wasanalyzed. Individual cells were tracked using the MTrackJ plug-in for ImageJimaging software. For each tracked cell, the number of reversals was enu-merated by hand, and the reversal frequency was calculated as the numberof reversals per second per cell. Then, for each experiment, the averagereversal frequency was calculated.

Swimming Assay in Soft-Agar Plates. V. parahaemolyticus was inoculated in5 mL LB at 37 °C and grown into early stationary phase. A toothpick wasdipped in the culture and pricked into semisolid LB agar plates solidifiedwith 0.3% agar. Plates were incubated at 30 °C overnight. Wild-type andmutant strains were grown on the same plates. The swimming diameter ofeach strain was measured and plotted relative to wild-type from thesame plate.

Bioinformatic Analysis. The genetic context of VP2226 (ParP) homologs wasanalyzed using the STRING database (41). Analysis was performed usingdefault settings. CheA protein sequences from γ-proteobacteria without anassociated ParC/ParP were chosen at random from a STRING analysis of ParC.Multiple alignments were performed using MUSCLE (42) at default settings.Alignments then were analyzed, and phylogenetic trees were generatedusing Jalview Average Distance BLOSOM62. Phylogenetic trees generated inJalview were displayed and colored using iTOL (43).

BTH Analysis. BTH assays were used to investigate protein–protein inter-actions. Proteins of interest were fused to T18 and T25 fragments of Bordetellapertussis adenylate cyclase in vectors pUT18C and pKT25, respectively. The BTHassay was carried out essentially as described in ref. 44. Plasmids expressingfusion proteins to T18 and T25 fragments were cotransformed into E. colistrain BTH101 and spotted on LB agar plates plus 80 μg/mL X-gal, 250 mMisopropyl β-D-1-thiogalactopyranoside, 50 μg/mL kanamycin, and 50 μg/mLcarbenicillin. Plates were incubated at 30 °C overnight and then weretransferred to ambient temperature.

ACKNOWLEDGMENTS. We thank members of the M.K.W. laboratory forcomments on the manuscript. This work was funded by Grant R37 AI-042347from the National Institute of Allergy and Infectious Diseases (to M.K.W.) andby the Howard Hughes Medical Institute (M.K.W.). S.R. was funded by a post-doctoral fellowship from the Villum Kann Rasmussen foundation. K.S. wassupported by a postdoctoral fellowship from the Deutsche Forschungsge-meinschaft. M.Z.-R. was supported by the Howard Hughes Medical InstituteExceptional Opportunities for Exceptional Students program.

Ringgaard et al. PNAS | Published online December 30, 2013 | E263

MICRO

BIOLO

GY

PNASPL

US

1. Sourjik V, Armitage JP (2010) Spatial organization in bacterial chemotaxis. EMBO J29(16):2724–2733.

2. Briegel A, et al. (2009) Universal architecture of bacterial chemoreceptor arrays. ProcNatl Acad Sci USA 106(40):17181–17186.

3. Briegel A, et al. (2012) Bacterial chemoreceptor arrays are hexagonally packed trimersof receptor dimers networked by rings of kinase and coupling proteins. Proc NatlAcad Sci USA 109(10):3766–3771.

4. Liu J, et al. (2012) Molecular architecture of chemoreceptor arrays revealed by cry-oelectron tomography of Escherichia coli minicells. Proc Natl Acad Sci USA 109(23):E1481–E1488.

5. Zhang P, Khursigara CM, Hartnell LM, Subramaniam S (2007) Direct visualization ofEscherichia coli chemotaxis receptor arrays using cryo-electron microscopy. Proc NatlAcad Sci USA 104(10):3777–3781.

6. Kentner D, Thiem S, Hildenbeutel M, Sourjik V (2006) Determinants of chemoreceptorcluster formation in Escherichia coli. Mol Microbiol 61(2):407–417.

7. Skidmore JM, et al. (2000) Polar clustering of the chemoreceptor complex in Es-cherichia coli occurs in the absence of complete CheA function. J Bacteriol 182(4):967–973.

8. Studdert CA, Parkinson JS (2005) Insights into the organization and dynamics ofbacterial chemoreceptor clusters through in vivo crosslinking studies. Proc Natl AcadSci USA 102(43):15623–15628.

9. Thiem S, Sourjik V (2008) Stochastic assembly of chemoreceptor clusters in Escherichiacoli. Mol Microbiol 68(5):1228–1236.

10. Sourjik V, Berg HC (2000) Localization of components of the chemotaxis machinery ofEscherichia coli using fluorescent protein fusions. Mol Microbiol 37(4):740–751.

11. Alley MR, Maddock JR, Shapiro L (1992) Polar localization of a bacterial chemore-ceptor. Genes Dev 6(5):825–836.

12. Maddock JR, Shapiro L (1993) Polar location of the chemoreceptor complex in theEscherichia coli cell. Science 259(5102):1717–1723.

13. Wadhams GH, Warren AV, Martin AC, Armitage JP (2003) Targeting of two signal trans-duction pathways to different regions of the bacterial cell. Mol Microbiol 50(3):763–770.

14. Bardy SL, Maddock JR (2005) Polar localization of a soluble methyl-accepting proteinof Pseudomonas aeruginosa. J Bacteriol 187(22):7840–7844.

15. Ringgaard S, Schirner K, Davis BM, Waldor MK (2011) A family of ParA-like ATPasespromotes cell pole maturation by facilitating polar localization of chemotaxis pro-teins. Genes Dev 25(14):1544–1555.

16. Yamaichi Y, et al. (2012) A multidomain hub anchors the chromosome segregationand chemotactic machinery to the bacterial pole. Genes Dev 26(20):2348–2360.

17. Gerdes K, Howard M, Szardenings F (2010) Pushing and pulling in prokaryotic DNAsegregation. Cell 141(6):927–942.

18. Lutkenhaus J (2012) The ParA/MinD family puts things in their place. Trends Microbiol20(9):411–418.

19. Thompson SR, Wadhams GH, Armitage JP (2006) The positioning of cytoplasmicprotein clusters in bacteria. Proc Natl Acad Sci USA 103(21):8209–8214.

20. Savage DF, Afonso B, Chen AH, Silver PA (2010) Spatially ordered dynamics of thebacterial carbon fixation machinery. Science 327(5970):1258–1261.

21. Ding Z, et al. (2002) A novel cytology-based, two-hybrid screen for bacteria applied toprotein-protein interaction studies of a type IV secretion system. J Bacteriol 184(20):5572–5582.

22. Lenarcic R, et al. (2009) Localisation of DivIVA by targeting to negatively curvedmembranes. EMBO J 28(15):2272–2282.

23. Miller AS, Kohout SC, Gilman KA, Falke JJ (2006) CheA Kinase of bacterial chemotaxis:Chemical mapping of four essential docking sites. Biochemistry 45(29):8699–8711.

24. Griswold IJ, et al. (2002) The solution structure and interactions of CheW fromThermotoga maritima. Nat Struct Biol 9(2):121–125.

25. Zhao J, Parkinson JS (2006) Mutational analysis of the chemoreceptor-coupling do-main of the Escherichia coli chemotaxis signaling kinase CheA. J Bacteriol 188(9):3299–3307.

26. Barillà D, Carmelo E, Hayes F (2007) The tail of the ParG DNA segregation proteinremodels ParF polymers and enhances ATP hydrolysis via an arginine finger-likemotif. Proc Natl Acad Sci USA 104(6):1811–1816.

27. Sourjik V (2004) Receptor clustering and signal processing in E. coli chemotaxis. TrendsMicrobiol 12(12):569–576.

28. Pratto F, et al. (2008) Streptococcus pyogenes pSM19035 requires dynamic assemblyof ATP-bound ParA and ParB on parS DNA during plasmid segregation. Nucleic AcidsRes 36(11):3676–3689.

29. Kiekebusch D, Michie KA, Essen LO, Löwe J, Thanbichler M (2012) Localized dimerizationand nucleoid binding drive gradient formation by the bacterial cell division inhibitorMipZ. Mol Cell 46(3):245–259.

30. Easter J, Jr., Gober JW (2002) ParB-stimulated nucleotide exchange regulates a switchin functionally distinct ParA activities. Mol Cell 10(2):427–434.

31. Hu Z, Lutkenhaus J (2001) Topological regulation of cell division in E. coli spatio-temporal oscillation of MinD requires stimulation of its ATPase by MinE and phos-pholipid. Mol Cell 7(6):1337–1343.

32. Roberts MAJ, Wadhams GH, Hadfield KA, Tickner S, Armitage JP (2012) ParA-likeprotein uses nonspecific chromosomal DNA binding to partition protein complexes.Proc Natl Acad Sci USA 109(17):6698–6703.

33. Thiem S, Kentner D, Sourjik V (2007) Positioning of chemosensory clusters in E. coliand its relation to cell division. EMBO J 26(6):1615–1623.

34. Greenfield D, et al. (2009) Self-organization of the Escherichia coli chemotaxis net-work imaged with super-resolution light microscopy. PLoS Biol 7(6):e1000137.

35. Wang H, Wingreen NS, Mukhopadhyay R (2008) Self-organized periodicity of proteinclusters in growing bacteria. Phys Rev Lett 101(21):218101.

36. Shiomi D, Yoshimoto M, Homma M, Kawagishi I (2006) Helical distribution of thebacterial chemoreceptor via colocalization with the Sec protein translocation ma-chinery. Mol Microbiol 60(4):894–906.

37. Rudner DZ, Pan Q, Losick RM (2002) Evidence that subcellular localization of a bac-terial membrane protein is achieved by diffusion and capture. Proc Natl Acad Sci USA99(13):8701–8706.

38. Deich J, Judd EM, McAdams HH, Moerner WE (2004) Visualization of the movementof single histidine kinase molecules in live Caulobacter cells. Proc Natl Acad Sci USA101(45):15921–15926.

39. Miller VL, Mekalanos JJ (1988) A novel suicide vector and its use in construction ofinsertion mutations: Osmoregulation of outer membrane proteins and virulencedeterminants in Vibrio cholerae requires toxR. J Bacteriol 170(6):2575–2583.

40. Ringgaard S, van Zon J, Howard M, Gerdes K (2009) Movement and equipositioning ofplasmids by ParA filament disassembly. Proc Natl Acad Sci USA 106(46):19369–19374.

41. Jensen LJ, et al. (2009) STRING 8—a global view on proteins and their functional in-teractions in 630 organisms. Nucleic Acids Res 37(Database issue):D412–D416.

42. Edgar RC (2004) MUSCLE: Multiple sequence alignment with high accuracy and highthroughput. Nucleic Acids Res 32(5):1792–1797.

43. Letunic I, Bork P (2011) Interactive Tree Of Life v2: Online annotation and display ofphylogenetic trees made easy. Nucleic Acids Res 39(Web Server issue):W475-8.

44. Ringgaard S, Löwe J, Gerdes K (2007) Centromere pairing by a plasmid-encoded type IParB protein. J Biol Chem 282(38):28216–28225.

E264 | www.pnas.org/cgi/doi/10.1073/pnas.1315722111 Ringgaard et al.