Flagella-independent surface motility in Salmonellaenterica serovar TyphimuriumSun-Yang Park, Mauricio H. Pontes, and Eduardo A. Groisman1

Howard Hughes Medical Institute and Department of Microbial Pathogenesis, Yale School of Medicine, Boyer Center for Molecular Medicine, New Haven, CT06536-0812; and Yale Microbial Sciences Institute, West Haven, CT 06516

Edited by Howard C. Berg, Harvard University, Cambridge, MA, and approved December 31, 2014 (received for review December 1, 2014)

Flagella are multiprotein complexes necessary for swimming andswarming motility. In Salmonella enterica serovar Typhimurium,flagella-mediated motility is repressed by the PhoP/PhoQ regula-tory system. We now report that Salmonella can move on 0.3%agarose media in a flagella-independent manner when experienc-ing the PhoP/PhoQ-inducing signal low Mg2+. This motility re-quires the PhoP-activated mgtA, mgtC, and pagM genes, whichspecify a Mg2+ transporter, an inhibitor of Salmonella’s own F1FoATPase, and a small protein of unknown function, respectively.The MgtA and MgtC proteins are necessary for pagM expressionbecause pagM mRNA levels were lower in mgtA and mgtC mu-tants than in wild-type Salmonella, and also because pagM expres-sion from a heterologous promoter rescued motility in mgtA andmgtCmutants. PagM promotes groupmotility by a surface protein(s),as a pagM-expressing strain conferred motility upon a pagM nullmutant, and proteinase K treatment eliminated motility. ThepagM gene is rarely found outside subspecies I of S. enterica andoften present in nonfunctional allelic forms in organisms lackingthe identified motility. Deletion of the pagM gene reduced bacte-rial replication on 0.3% agarose low Mg2+ media but not in lowMg2+ liquid media. Our findings define a form of motility thatallows Salmonella to scavenge nutrients and to escape toxic com-pounds in low Mg2+ semisolid environments.

magnesium | PagM | PhoP/PhoQ | MgtA | MgtC

Cells have the ability to move on a variety of surfaces (1). Thisability enables cells to search for nutrients, to avoid toxic

compounds, to colonize new niches, and to form complex struc-tures such as biofilms (2, 3). Bacterial motility has been classifiedinto distinct types based on the energetic requirements and thestructural elements involved (4, 5). For example, rotation of themultiprotein complex known as flagella can propel bacteria for-ward in a process that requires the proton motive force. Twitchingmotility is a surface movement mediated by the extension and re-traction of surface appendages known as type IV pili, and glidingmotility is an active surface movement that occurs along the longaxis of the cell via focal adhesion complexes. By contrast, slidingmotility is a passive surface translocation powered by growth andfacilitated by a surfactant.Salmonella enterica serovar Typhimurium exhibits two forms

of flagella-mediated motility on semisolid agar media: swarmingand swimming (6). Swarming motility entails morphological dif-ferentiation into swarmer cells, takes place on 0.4–0.7% agarsurfaces, and requires an energy-rich carbon source such as glu-cose even in nutrient-rich media (6). Whereas swarming con-stitutes a multicellular group behavior on a surface, swimmingis an individual bacterial movement experienced in liquid andlow (0.1–0.3%) agar (2). Flagella-mediated motility is under thecontrol of positive and negative regulators that respond to dif-ferent environmental and cellular signals. For example, the na-ture of the carbon source is critical because the cAMP-receptorprotein promotes transcription of the flagellar master regulatorflhDC operon (7), and the RNA binding regulator CsrA furthersstability of the flhDC mRNA (8). By contrast, the PhoP/PhoQregulatory system down-regulates flagella-mediated motility by

repressing transcription of flagellar genes (9) and by decreasingthe membrane potential (10).The PhoP/PhoQ system consists of the DNA binding tran-

scriptional regulator PhoP (11) and the sensor PhoQ, whichresponds to low extracytoplasmic Mg2+ (12), mildly acidic pH(13), and certain antimicrobial peptides (14) by promoting theactive (i.e., phosphorylated) form of PhoP. Transcription of ∼5%of the Salmonella genes is under PhoP control (15). The PhoP-activated mgtA gene specifies a Mg2+ transporter that enhancesPhoP-P levels by transporting Mg2+ away from the periplasm,where it acts as an inhibitory signal for PhoQ (16). Becausetranscription elongation into themgtA coding region is controlledby the Mg2+-respondingmgtA leader RNA, the MgtA protein andthe resulting higher levels of PhoP-P are produced only whencytoplasmic Mg2+ levels drop below a certain threshold (17, 18).This regulatory architecture defines a two-tier structure amongPhoP-activated genes based on whether they require mgtA formaximal expression (16). This architecture allows Salmonella todelay transcription of a subset of the PhoP regulon until the cy-tosolic conditions triggering MgtA production are met.The PhoP-activated genes most dependent on the MgtA

protein specify proteins of unknown function and, like the mgtAgene, are not required for virulence in an animal (19–21).Therefore, we hoped that the investigation of mgtA-dependentphenotypes might reveal a novel aspect(s) of Salmonella’s life-style. We now report that the MgtA-dependent pathway governsa form of surface migration that does not appear to involveflagella or fimbriae. We establish that this surface migration isdependent on the PhoP-activated MgtA-dependent pagM gene.The rare occurrence of pagM outside the species S. entericasuggests that the uncovered motility is a property exclusive to this

Significance

We identified a form of surface motility in the bacterium Sal-monella enterica serovar Typhimurium that is activated in lowMg2+ by PhoP/PhoQ, a regulatory system that hinders flagellaexpression and activity. PhoP furthers motility by promotingexpression of the pagM gene, which specifies a small protein ofunknown function, and also of two other genes that create thecytosolic conditions necessary for full pagM expression. LowMg2+-promoted motility is a group behavior exclusive to asubset of S. enterica serovars harboring a particular allele ofpagM. The pagM gene is present in nonfunctional allelic formsin certain S. enterica serovars and rarely found outside thegenus Salmonella. Not required for virulence, PagM-mediatedmotility helps survival outside animal hosts.

Author contributions: S.-Y.P. and E.A.G. designed research; S.-Y.P. performed research;M.H.P. contributed new reagents/analytic tools; S.-Y.P. and E.A.G. analyzed data; andS.-Y.P. and E.A.G. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: eduardo.groisman@yale.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1422938112/-/DCSupplemental.

1850–1855 | PNAS | February 10, 2015 | vol. 112 | no. 6 www.pnas.org/cgi/doi/10.1073/pnas.1422938112

species, allowing it to explore surroundings under conditionsunfavorable to flagella-mediated motility.

ResultsThe PhoP/PhoQ System Promotes Surface Motility on Low Mg2+

Agarose Media. To examine the migratory behavior of wild-typeS. enterica serovar Typhimurium strain 14028s under PhoP/PhoQ-inducing conditions, we grew bacteria in N-minimal liquidmedia with 10 μM Mg2+ for 4.5 h and then spotted the bacteriaon the same media with different agarose concentrations (0.2–0.6%). The distance migrated was inversely correlated with theagarose concentration, and at 0.6% agarose, no movement wasdetected (Fig. S1).In contrast to the wild-type strain, a phoP mutant displayed

no motility on 0.3% agarose N-minimal media with 10 μM Mg2+

(Fig. 1A). Moreover, neither wild-type Salmonella nor the phoPmutant were motile on 0.3% agarose with 10 mM Mg2+

N-minimal media (Fig. 1A), presumably because the PhoP/PhoQsystem is not active in high Mg2+ (11). The identified motilitydoes not appear to be swarming motility because the PhoP/PhoQsystem represses flagella-mediated motility (9, 10), and onewould expect a phoP mutant to move more than the wild-typestrain. In addition, swarming motility requires glucose, and themedia had glycerol as the carbon source.We ruled out the participation of flagella in the identified

motility because mutants defective in flagella rotation (motA),hook protein (flgE), flagellin subunit (fliC), or the master flagellaregulator (flhDC) moved similarly to the parental strain on 0.3%agarose N-minimal media with 10 μM Mg2+ (Fig. 1B). By con-trast, all four mutants were not motile or less motile than thewild-type strain if the agarose concentration was lowered to 0.1%(Fig. 1B), as expected for a process dependent on flagella. ThefliC single mutant exhibited intermediate swimming behaviorpossibly because Salmonella harbors two flagellin genes, fliC andfliB (6), and only one was inactivated.Fimbriae or pili are surface appendages required for twitching

and gliding motility in certain bacterial species (22, 23). TheS. enterica serovar Typhimurium genome has 13 fimbriae operons(24). Individual mutations in each of the 13 fimbriae operons(fimA, pefA, stfA, bcfD, stbB, lpfB, csgB, stdA, stiA, stcA, sthA,safA, and stjC) had no effect on bacterial migration on 0.3%agarose low Mg2+ media (Fig. 1C). The results suggest that theSalmonella PhoP/PhoQ system controls a previously unidentifiedform of surface motility. (Note that the participation of thefimbriae cannot be entirely ruled out until a strain deleted for all13 fimbriae operons is constructed and evaluated.)

The PhoP-Activated mgtA, mgtC, and pagM Genes Are Required forSurface Motility. To identify the PhoP-regulated gene(s) re-sponsible for motility on 0.3% agarose low Mg2+ media, wetested the behavior of strains mutated in each of 19 differentPhoP-activated genes (mgtA, mgtB, mgtC, mig-14, pagC, pagK,pagM, pagN, pagO, pagP, pcgL, pgtE, phoN, pmrD, rstA, slyA,ugtL, virK, and yobG). Mutants lacking the mgtA, mgtC, or pagMgenes were as defective as the phoP null mutant, whereas a slyAmutant was partially defective, displaying a motility intermediatebetween that of the wild-type strain and the phoP mutant (Fig.S2). The other mutants displayed a wild-type behavior (Fig. S2).How do the mgtA, mgtC, and pagM genes promote surface

motility in low Mg2+ conditions? Given that the Mg2+ trans-porter MgtA is required for full pagM transcription (16) (Fig.2A), we reasoned that MgtA’s role in motility might be to pro-mote pagM expression. In agreement with this notion, expressionof the pagM gene from a heterologous promoter partially re-stored motility to the mgtA mutant, whereas the plasmid vectordid not (Fig. 2B).The mgtC gene specifies an inhibitor of the F1Fo ATPase (25)

and may also promote motility by allowing pagM expression

given that MgtC controls ATP levels and cytosolic pH (25). Inagreement with this notion, the mRNA levels of the pagM genewere fivefold higher in the wild-type strain than in the isogenicmgtC mutant following growth in low Mg2+ media for 4.5 h (Fig.2A). Moreover, the pagM-expressing plasmid, but not the vectorcontrol, restored complete motility to themgtC mutant (Fig. 2B).The pagM-expressing plasmid enhanced motility of the wild-typestrain (Fig. 2C), and of course, it rescued motility in the pagMmutant (Fig. 2B).That the mgtA and mgtC genes act upstream of pagM in the

identified motility was reinforced by two independent sets of

Fig. 1. The PhoP/PhoQ system promotes surface motility on low Mg2+ agarosemedia. (A) Surface motility of wild-type S. enterica serovar Typhimurium(14028s), phoP (MS7953s), mgtA (EG16735), mgtC (EL4), and pagM (SP244)strains grown as described in Materials and Methods and spotted onto 0.3%agarose low (10 μM) Mg2+ (HL) and high (10 mM) Mg2+ minimal media. (B)Surface behavior of wild-type S. enterica serovar Typhimurium (14028s) andmutants defective in the motA (MP7), fliC (SP255), flgE (SP370), or flhDC(EG11308) genes. Bacteria were grown as described in A and spotted onto 0.3%(Top) and 0.1% (Bottom) agarose. (C) Surface behavior of wild-type S. entericaserovar Typhimurium (14028s) and mutants defective in the fimA (SP347), pefA(SP371), stfA (SP372), bcfD (SP374), stbB (SP375), lpfB (SP376), csgB (MP362),stdA (SP379), stiA (SP380), stcA (SP381), sthA (SP382), safA (SP373), or stjC(SP378) fimbriae genes on 0.3% agarose HL. Motility was examined as describedin A. Photo image was captured after bacteria were grown at 37 °C for 20 h.Shown are the representatives of three independent experiments.

Park et al. PNAS | February 10, 2015 | vol. 112 | no. 6 | 1851

MICRO

BIOLO

GY

experiments. First, wild-type and pagM mutant strains producedsimilar amounts of mgtA mRNA and the same was true for themgtC mRNA (Fig. S3A). And second, plasmids expressing themgtA or mgtC gene from a heterologous promoter failed to re-store motility to the pagM mutant and behaved like the vectorcontrol (Fig. S3B). By contrast, the pagM-expressing plasmidrescued motility in the pagM mutant (Fig. S3B).To test the hypothesis that PagM-mediated motility requires

low Mg2+ solely to activate the PhoP/PhoQ system, we inves-tigated whether the pagM-expressing plasmid could restoremotility to a pagM motA double mutant in high Mg2+ media. Byconducting the experiment in a motA mutant background, wecould rule out a role for flagella in any observed motility. Ashypothesized, the pagM-expressing plasmid conferred motilityupon the pagM motA strain in high Mg2+, whereas the vectorcontrol did not (Fig. 2D). The pagM motA strain harboring thepagM-expressing plasmid displayed more surface motility in highthan in low Mg2+ (Fig. 2D). As Salmonella grows faster in highthan in low Mg2+ (26), our findings suggest there is a correlationbetween PagM-mediated motility and bacterial growth.

PagM-Mediated Motility Requires a Surface Protein(s) but Not CertainWetting Agents. The pagM gene encodes a 60 amino acid-longpeptide that includes a 28 amino acid-long predicted signal

sequence and a mature 32 amino acid-long product (Fig. S4).The amino acid composition of the mature PagM is unusual, as ithas seven glycines and three cysteines (Fig. S4). Protein locali-zation predictions suggest PagM is an extracytoplasmic peptide,and this prediction is supported by the alkaline phosphataseactivity displayed by Salmonella harboring a translational fusionbetween pagM and a truncated phoA gene from Escherichia coli(27). (The subcellular location of the PagM protein could not beinvestigated because antibodies against PagM are not available,and introduction of epitope tags rendered PagM nonfunctional.)We reasoned that if PagM and/or a surface protein required for

motility in low Mg2+ were to localize to the outer leaflet of theouter membrane, protease treatment might abolish motility. Inagreement with this notion, a disk impregnated with proteinase Kabolished motility, whereas a disk with a proteinase K that hadbeen previously heat- inactivated did not impact motility (Fig. 3A).Wetting agents often promote motility on cell surfaces by

reducing the friction between a cell and its substrate (28, 29).We tested the potential role of various wetting agents by in-vestigating the motility of mutants defective in the production ofparticular surface molecules. Strains defective in the productionof capsule (yhiS), colanic acid (ugd), curli (csgB), or lipopoly-saccharide (LPS) (waaB) retained a wild-type behavior (Fig. 3B).Mutants that cannot make cellulose (bcsA) or O-antigen (rfaL)displayed slightly reduced surface migration (Fig. 3B). A wecBmutant unable to synthesize enterobacterial common antigen

Fig. 2. The PhoP-activated mgtA, mgtC, and pagM genes are required formotility on 0.3% agarose low Mg2+ media. (A) pagM mRNA levels producedby wild-type (14028s), mgtA (EG16735), and mgtC (EL4) strains grown in low(10 μM) Mg2+ media at 37 °C for 4.5 h. Expression levels were determinedas described in Materials and Methods. Shown are the mean and SD fromthree independent experiments. (B) Surface motility of pagM (SP244), mgtA(EG16735), and mgtC (EL4) strains harboring a plasmid expressing the wild-type pagM gene (pUH-pagM). Bacteria were grown in HL media for 4.5 h inthe presence of ampicillin (50 μg/mL) and IPTG (0.1 mM) at 37 °C. The vectorpUHE 21–2lacIq was used as a control. Motility was evaluated as described inthe Fig. 1A legend. (C) Surface motility of a wild-type S. enterica serovarTyphimurium (14028s) strain harboring a plasmid expressing the pagM gene(pUH-pagM). The vector pUHE 21–2lacIq was used as a control. Motility wasexamined as described in B. (D) Surface motility of the pagM motA strain(SP247) harboring a plasmid expressing the pagM gene (pUH-pagM) or thevector pUHE 21–2lacIq as a control. Bacteria were grown as described in B,and then 3 μL of a bacterial suspension (OD600 ∼1) were spotted onto the0.3% agarose low (10 μM) and high (10 mM) Mg2+. Photo image was cap-tured after bacteria were grown at 37 °C for 20 h. Shown are the repre-sentatives of three independent experiments.

Fig. 3. PagM-mediated motility requires a surface protein(s) but not certainwetting agents. (A) Proteinase K treatment inhibits PagM-mediated motility.The pagM strain (SP244) harboring a plasmid expressing the pagM gene (pUH-pagM) or the vector pUHE 21–2lacIq as a control was grown in N-minimal mediacontaining 10 μM Mg2+ (HL) in the presence of ampicillin (50 μg/mL) and IPTG(0.1 mM) at 37 °C for 4.5 h. Three microliters of a bacterial suspension (OD600

∼1) were spotted onto the same media with 0.3% agarose and paper diskcontaining 5 μL proteinase K (20 μg/μl) or heat-inactivated proteinase K. (B)Surface behavior of wild-type S. enterica serovar Typhimurium (14028s) andmutants defective in the production of cellulose (bcsA; MP133), O-antigen (rfaL;MP70), capsule (yhiS; MP105), colanic acid (ugd; EG9524), curli (csgB; MP362),ECA (wecB; MP107), or LPS (waaB; YS10) on 0.3% agarose HL. Photo image wascaptured after bacteria were grown at 37 °C for 20 h. Shown are the repre-sentatives of three independent experiments.

1852 | www.pnas.org/cgi/doi/10.1073/pnas.1422938112 Park et al.

(ECA) was less motile than wild-type Salmonella (Fig. 3B), andthis could be due to poor growth.

Coculture with a PagM-Expressing Strain Confers Motility upon apagM Mutant. Bacterial motility can be an individual activitywhereby a bacterium swims alone. Alternatively, motility may resultfrom the collective behavior of groups of bacteria moving in a co-ordinated fashion (30, 31). We reasoned that if PagM-mediatedmotility is a group behavior, it might be possible to promote mo-tility of a pagM mutant by coincubating it with a pagM-expressingstrain. Thus, we used a mixture of two strains to inoculate 0.3%agarose low Mg2+ N-minimal media: a pagM mutant harboringa plasmid expressing the green fluorescence protein (GFP) con-stitutively and a pagMmutant harboring the plasmid expressing thepagM gene from a heterologous promoter.The pagM-proficient strain rescued motility of the pagM mu-

tant when inoculated at a ratio of 1:4 (Fig. 4A). This rescue isindependent of flagella because motility was restored by themotA mutant harboring the plasmid vector as a helper strain(Fig. 4B). By contrast, motility requires a live pagM-expressingstrain because there was no rescue if the bacteria were previouslykilled by heating at 95 °C for 10 min (Fig. 4C).Microscopic examination of the cocultures on 0.3% agarose

low Mg2+ media revealed that the GFP-expressing pagM mutantreached the edge of the growth line together with the pagM-expressing strain (Fig. 4D). The intermixing of GFP-expressing andpagM-expressing bacteria is reminiscent of swarming in Serracialiquefaciens (32). Cumulatively, these results indicate that PagM-mediated motility is a group behavior rather than an individual one.

The PagM Allele Determines Surface Motility. We conducted aTblastn search of the complete Salmonella genomes using the

deduced amino acid sequence of the pagM gene from S. entericaserovar Typhimurium strain 14028s. We identified sequencesexhibiting 100% identity to the 14028s PagM in certain isolatesfrom subspecies I, one of the six subspecies that compriseS. enterica (33). These isolates belong to the serovars Anatum,Bovimobificans, Dublin, Heidelberg, Javiana, Paratyphi B,Schwarzegrund, Thompson, and Typhimurium. We found thatthe PagM proteins differ in length and sequence even among in-dependent isolates of a given serovar (Fig. 5A and Fig. S4). Forexample, there is a single nucleotide difference between the pagMgenes from virulent strain 14028s and certain Typhimurium strainssuch as strain LT2. This difference results in a stop codon early inthe pagM sequence (Fig. 5A and Fig. S4) that renders LT2 non-motile on 0.3% agarose low Mg2+ N-minimal media (Fig. 5B).There is a correlation between the pagM allele and surface

motility because, first, serovar Dublin encodes a PagM proteinidentical to the 14028s PagM (Fig. 5A and Fig. S4) and exhibitssurface motility similar to strain 14028s (Fig. 5B). And second,serovar Enteritidis specifies a truncated form of PagM, andserovar Pullorum encodes a PagM protein that differs in fouramino acid residues from the serovar Typhimurium 14028sPagM (Fig. 5A and Fig. S4), and neither Enteritidis nor Pullorumdisplays motility (Fig. 5B). (We verified the allelic differences bysequencing the pagM genes from the investigated strains.)If the surface motility phenotypes are due solely to the pagM

allele, it should be possible to convert a motile strain into a non-motile one and vice versa simply by swapping pagM genes. Asproposed, derivatives of serovar Typhimurium strain 14028s speci-fying a PagM protein with a lysine at position 39 or a glutamate atposition 43 did not move on 0.3% agarose low Mg2+ N-minimalmedia (Fig. 5C). These amino acid differences distinguish the PagMproteins from those in Typhimurium 14028s and Pullorum. And thenonmotile serovar Enteritidis gained motility when harboringa plasmid with the 14028s pagM gene but not with the vector control(Fig. 5D).

Fig. 4. Coculture with a PagM-expressing strain confers motility upon apagM mutant. (A) Surface motility of a pagM mutant (SP244) harboring aplasmid expressing the gfp gene in coculture with the pagM mutant har-boring a plasmid expressing the pagM gene. Bacteria were grown inN-minimal media containing 10 μM Mg2+ in the presence of ampicillin (50μg/mL) and IPTG (0.2 mM) at 37 °C for 4.5 h. Two bacterial suspensions(OD600 ∼1) were mixed at the indicated ratios, and 3 μL of the mixture werespotted on 0.3% agarose HL containing ampicillin and IPTG. (B) Surfacemotility of coculture of the motA strain (MP7) with the plasmid vectorpFPV25 carrying a promoterless gfp gene and the pagM strain with a plas-mid (pFPV25.1) expressing the gfp gene from a constitutive promoter. Mo-tility was examined as described in A. (C) Surface motility of coculture of thepagM strain (SP244) with a plasmid expressing the pagM gene, which hadbeen previously killed by heat treatment at 95 °C for 10 min, and the pagMstrain (SP244) with a plasmid expressing the gfp gene. Motility was exam-ined as described in A. Photo images (A–C) were captured after 20 h ofgrowth at 37 °C under a transilluminator for GFP detection. (D) Phase-con-trast (Left) and fluorescent micrograph (Right) images of edge of cocultureat a ratio of 1:4 as described in A.

Fig. 5. The PagM allele determines surface motility. (A) Alignment of the de-duced amino acid sequence of pagM homologs from S. enterica serovar Typhi-murium 14828s (Typhimurium 14828s), serovar Dublin (Dublin), serovar Pullorum(Pullorum), serovar Typhimurium LT2 (Typhimurium LT2), and serovar Enteritidis(Enteritidis). (B) Surface motility of Typhimurium 14028s, Dublin, Enteritidis,Typhimurium LT2, and Pullorum. Bacteria were grown in N-minimal media with10 μM Mg2+ and 0.3% casamino acids at 37 °C for 5.5 h. Three microliters ofa bacterial suspension (OD600 ∼1) were spotted onto the same media containing0.3% agarose. (C) Surface motility of the S. enterica serovar Typhimurium 14828spagMmutant (SP244) harboring a plasmid expressing the wild-type pagM (pUH-pagM) or pagM variants (pUH-pagM-A39K and pUH-pagM-K43E). Thevector pUHE 21–2lacIq was used as a control. (D) Surface motility of the non-motile serovar Enteritidis harboring a plasmid expressing the 14028s pagM geneor the vector control. Bacteria were grown in N-minimal media containing 10 μMMg2+ in the presence of ampicillin (50 μg/mL) and IPTG (0.1 mM) at 37 °C for4.5 h. Motility was examined as described in B. Photo image was captured afterbacteria were grown at 37 °C for 20 h. Shown are the representatives of threeindependent experiments.

Park et al. PNAS | February 10, 2015 | vol. 112 | no. 6 | 1853

MICRO

BIOLO

GY

The genus Salmonella consists of two species: S. enterica andSalmonella bongori. Sequences sharing 39–40% amino acididentity were found in the deduced amino acid sequences ofthree different S. bongori genomes (Fig. S5). A search for PagMhomologs among other members of the family Enterobacteriaceaerevealed the presence of related sequences in the genomes ofSodalis sp. HS1 (43% amino acid identity), Candidatus Sodalispierantonius str. SOPE (43% amino acid identity), and Citro-bacter rodentium ICC168 (36% amino acid identity) (Fig. S5).However, other strains of the latter species lack pagM-relatedsequences. This analysis suggests that S. enterica acquired pagMby horizontal gene transfer after it split from the common an-cestor that gave rise to S. bongori.

PagM Aids Replication on Low Mg2+ Semisolid Media. We reasonedthat PagM-mediated motility might aid bacterial replication on0.3% agarose low Mg2+ media by enabling exploration of sitesaway from the site of inoculation. Such sites would still havenutrients and/or lack potentially toxic metabolic products. Toexplore this possibility, we determined the bacterial numbers ofpagM-expressing and pagM mutant strains following incubationin low Mg2+ liquid versus semisolid media. There were similarnumbers of pagM-expressing and pagM mutants when bacteriawere grown in low Mg2+ liquid media. By contrast, the number ofpagM-expressing bacteria was 30-fold higher than that of the pagMmutant after a 20-h incubation on 0.3% agarose low Mg2+ media(Fig. 6). We conclude that the pagM gene provides a growth ad-vantage to Salmonella experiencing low Mg2+ semisolid conditions.

DiscussionWe have now identified a form of motility that S. enterica serovarTyphimurium uses to move on surfaces when experiencing lowMg2+ semisolid conditions. This form of motility is independentof flagella (Fig. 1B) and still takes place in mutants defective ineach of 13 fimbriae operons (Fig. 1C). By contrast, it requires thePhoP/PhoQ regulatory system to promote expression of themgtA, mgtC, and pagM genes (Fig. 1A). The pagM gene appearsto be directly involved in the uncovered motility because, first,pagM expression from a heterologous promoter conferred mo-tility even when Salmonella experienced high Mg2+, which isa noninducing condition for the PhoP/PhoQ system (Fig. 2D).Second, pagM expression from a heterologous promoter restoredmotility to mgtA and mgtC mutants (Fig. 2B), but mgtA- andmgtC-expressing plasmids failed to rescue motility in a pagMmutant (Fig. S3B). Third, mgtA and mgtC mutants displayed lowpagM mRNA levels (Fig. 2A), but inactivation of pagM had noeffect on the mRNA levels of mgtA or mgtC (Fig. S3A). Fourth,nonmotile Salmonella isolates have pagM alleles different fromthat present in motile Typhimurium strain 14028s (Fig. 5 A andB). And fifth, motility was eliminated or conferred simply byaltering the pagM allele (Fig. 5 C and D).PagM-mediated motility appears to constitute a group be-

havior because a pagM-expressing strain rescued motility of apagM mutant when incubated together with it (Fig. 4). PagM-dependent motility appears to involve a surface protein(s) be-cause proteinase K and heat treatment abolished motility (Figs.3A and 4C). The surface protein could be PagM itself, given thatit has a predicted signal sequence (Fig. S4).PagM may promote sliding motility by which bacterial move-

ment results from expansive forces in a growing colony combinedwith surface properties that reduce the friction between the celland its substrate (4). In support of this notion, Salmonella dis-played faster surface migration in high than low Mg2+ when thepagM gene was expressed constitutively (Fig. 2D), which corre-lates with better bacterial replication at high than low Mg2+ (26).Sliding motility typically requires surfactants to lower surfacefriction—for example, serrawettin in Serratia marcescens (28, 34),acetylated glycopeptidolipids in Mycobacteria smegmatis (35, 36),

and surfactin in Bacillus subtilis (37). In S. enterica, however, lowMg2+-promoted motility was not affected when the genes re-sponsible for the production of cellulose, capsule, colanic acid,LPS, ECA, or O-antigen were inactivated (Fig. 3B), and theidentity of a potential wetting agent reducing surface friction inSalmonella remains unknown.Vibrio cholerae and E. coli can move on low-melting-temper-

ature agarose minimal media in a flagella-independent manner(38). However, the motility of these two species appears to beunrelated to that promoted by PagM in Salmonella becauseV. cholerae requires LPS for motility (38), whereas Salmonella doesnot (Fig. 3B), and also because V. cholerae moves on a surfaceindependently of growth (38), whereas pagM-mediated motilityis stimulated by growth (Fig. 2D). In addition, V. cholerae andE. coli lack homologs of PagM, which exhibits a fairly limitedphylogenetic distribution.The opposite regulation of flagella and pagM by the Salmo-

nella PhoP/PhoQ system is reminiscent of the control exerted byproteins that repress flagella but promote expression of othersurface molecules. For example, the fimZ and pefI genes specifyactivators of distinct fimbriae operons and repress flagella ex-pression when transcribed from heterologous promoters (39, 40).However, PagM promotes motility (Figs. 1–5), whereas fimbriaefurther surface adhesion.PagM may enable Salmonella to move and scavenge Mg2+, and

perhaps other nutrients, when experiencing low Mg2+ for an ex-tended period. This is because, first, low Mg2+ activates the PhoP/PhoQ system, which hinders flagella-mediated motility by repres-sing transcription of flagellin (9) and by decreasing membranepotential, which hinders flagellar rotation (10). And second, pagMtranscription requires the Mg2+ transporter MgtA and the F1FoATPase inhibitor MgtC. This makes PagM-mediated motility

Fig. 6. PagM aids replication on low Mg2+ semisolid media. Fold change incolony forming units (cfu) of the pagM mutant (SP244) harboring either thevector control (pUHE 21–2lacIq) or the pagM-expressing plasmid (pUH-pagM)following incubation on 0.3% agarose media with 10 μMMg2+ at 37 °C for 20 h.Bacteria were grown in N-minimal media containing 10 μM Mg2+ (HL) in thepresence of ampicillin (50 μg/mL) and IPTG (0.1 mM) at 37 °C for 4.5 h. Threemicroliters of a bacterial suspension (OD600 ∼0.6) were spotted onto 0.3% HLwith ampicillin and IPTG (0.5 mM) and grown at 37 °C for 20 h. To measureviable bacterial numbers, bacteria were collected, serially diluted in phosphatesaline buffer, and then plated on Luria–Bertani agar plates containing ampicillin.The number of cfus was determined following incubation at 37 °C overnight.

1854 | www.pnas.org/cgi/doi/10.1073/pnas.1422938112 Park et al.

contingent on the cytosolic conditions that promote transcriptionelongation into the mgtA and mgtC coding regions: a decrease inMg2+ (18) and an increase in ATP (25), respectively. These cy-tosolic conditions may be indicative of a need to explore otherlocales. Indeed, PagM helps Salmonella replication (Fig. 6),possibly by facilitating access to nutrients and escape from toxicmetabolic byproducts. Finally, unlike flagella-mediated motility,which has been implicated in virulence (41–43), deletion of thepagM gene did not alter Salmonella pathogenicity (21), suggest-ing PagM-mediated motility aids Salmonella survival outside amammalian host.

Materials and MethodsBacterial Strains, Plasmids, and Growth Conditions. Bacterial strains andplasmids used in this study are listed in Table S1. Details of strains andplasmid constructions are described in SI Materials and Methods. Primers arelisted in Tables S2 and S3. Bacteria were grown at 37 °C in Luria–Bertanibroth or in N-minimal media, pH 7.7 (44) supplemented with 0.1% casaminoacids, 38 mM glycerol, and the indicated concentrations of MgCl2. For ge-netic manipulations, such as transformation and transduction, ampicillin wasused at a final concentration of 50 μg/mL and chloramphenicol at 20 μg/mL.To induce the LacI-repressed lac promoter derivative in plasmid pUHE21-2lacIq, we used isopropyl-thio-β-galactoside (IPTG) at 0.1–0.5 mM.

Determination of Transcript Levels. Total RNA was extracted using RNeasyMini Kit (Qiagen). cDNA was synthesized using High Capacity RNA to cDNA

Master Mix (Applied Biosystems) following the manufacturer’s instructions.Quantification of transcripts was performed by real-time PCR using FastSYBR Green Master Mix (Applied Biosystems) in an ABI 7500 Sequence De-tection System (Applied Biosystems). Data were normalized to the levels ofthe 16S ribosomal RNA rrs gene. A list of primers used for quantitative RT-PCR is presented in Table S2.

Surface Motility Assay. Bacteria were grown overnight in N-minimal media,pH 7.7, containing 10 mM MgCl2. They were then washed with N-minimalmedia, inoculated into 2 mL of N-minimal media containing 10 μM MgCl2(1:50 dilution), and grown at 37 °C for 4.5–5 h with aeration. The harvestedbacteria were adjusted to OD600 ∼1 with N-minimal media. Three microlitersof cell suspension were spotted on the 0.3% agarose N-minimal media with10 μM or 10 mM MgCl2 and grown at 37 °C for 20 h. To test the motility ofstrains with plasmid harboring the pagM gene, bacteria were grown in thepresence of ampicillin (50 μg/mL) and IPTG (0.1 mM) and spotted on thesame media with 0.3% agarose and IPTG (0.1–0.5 mM). To test swimmingmotility, bacteria were spotted onto 0.1% agarose N-minimal media with10 μM MgCl2. The data are representative of three independent experi-ments, which gave similar results.

ACKNOWLEDGMENTS. We thank Jennifer Aronson for editorial assistanceon the manuscript and Sangjin Kim for help with fluorescence microscopy.This research was supported, in part, by Grant AI49561 from the NationalInstitutes of Health (to E.A.G.). E.A.G. is an investigator of the HowardHughes Medical Institute.

1. Costerton JW (1995) Overview of microbial biofilms. J Ind Microbiol 15(3):137–140.2. Harshey RM (2003) Bacterial motility on a surface: Many ways to a common goal.

Annu Rev Microbiol 57:249–273.3. Verstraeten N, et al. (2008) Living on a surface: Swarming and biofilm formation.

Trends Microbiol 16(10):496–506.4. Henrichsen J (1972) Bacterial surface translocation: A survey and a classification.

Bacteriol Rev 36(4):478–503.5. Kearns DB (2010) A field guide to bacterial swarming motility. Nat Rev Microbiol 8(9):

634–644.6. Harshey RM, Matsuyama T (1994) Dimorphic transition in Escherichia coli and Sal-

monella typhimurium: Surface-induced differentiation into hyperflagellate swarmercells. Proc Natl Acad Sci USA 91(18):8631–8635.

7. Kutsukake K (1997) Autogenous and global control of the flagellar master operon,flhD, in Salmonella typhimurium. Mol Gen Genet 254(4):440–448.

8. Lawhon SD, et al. (2003) Global regulation by CsrA in Salmonella typhimurium. MolMicrobiol 48(6):1633–1645.

9. Adams P, et al. (2001) Proteomic detection of PhoPQ- and acid-mediated repression ofSalmonella motility. Proteomics 1(4):597–607.

10. Alteri CJ, Lindner JR, Reiss DJ, Smith SN, Mobley HL (2011) The broadly conservedregulator PhoP links pathogen virulence and membrane potential in Escherichia coli.Mol Microbiol 82(1):145–163.

11. Shin D, Groisman EA (2005) Signal-dependent binding of the response regulatorsPhoP and PmrA to their target promoters in vivo. J Biol Chem 280(6):4089–4094.

12. García Véscovi E, Soncini FC, Groisman EA (1996) Mg2+ as an extracellular signal:Environmental regulation of Salmonella virulence. Cell 84(1):165–174.

13. Prost LR, et al. (2007) Activation of the bacterial sensor kinase PhoQ by acidic pH. MolCell 26(2):165–174.

14. Bader MW, et al. (2005) Recognition of antimicrobial peptides by a bacterial sensorkinase. Cell 122(3):461–472.

15. Harari O, Park SY, Huang H, Groisman EA, Zwir I (2010) Defining the plasticity oftranscription factor binding sites by deconstructing DNA consensus sequences: ThePhoP-binding sites among gamma/enterobacteria. PLOS Comput Biol 6(7):e1000862.

16. Park SY, Groisman EA (2014) Signal-specific temporal response by the SalmonellaPhoP/PhoQ regulatory system. Mol Microbiol 91(1):135–144.

17. Cromie MJ, Groisman EA (2010) Promoter and riboswitch control of the Mg2+

transporter MgtA from Salmonella enterica. J Bacteriol 192(2):604–607.18. Cromie MJ, Shi Y, Latifi T, Groisman EA (2006) An RNA sensor for intracellular Mg(2+).

Cell 125(1):71–84.19. Gunn JS, Alpuche-Aranda CM, Loomis WP, Belden WJ, Miller SI (1995) Characteriza-

tion of the Salmonella typhimurium pagC/pagD chromosomal region. J Bacteriol177(17):5040–5047.

20. Blanc-Potard AB, Groisman EA (1997) The Salmonella selC locus contains a pathoge-nicity island mediating intramacrophage survival. EMBO J 16(17):5376–5385.

21. Gunn JS, Belden WJ, Miller SI (1998) Identification of PhoP-PhoQ activated geneswithin a duplicated region of the Salmonella typhimurium chromosome. MicrobPathog 25(2):77–90.

22. McBride MJ (2001) Bacterial gliding motility: Multiple mechanisms for cell movementover surfaces. Annu Rev Microbiol 55:49–75.

23. Mattick JS (2002) Type IV pili and twitching motility. Annu Rev Microbiol 56:289–314.24. McClelland M, et al. (2001) Complete genome sequence of Salmonella enterica se-

rovar Typhimurium LT2. Nature 413(6858):852–856.25. Lee EJ, Pontes MH, Groisman EA (2013) A bacterial virulence protein promotes patho-

genicity by inhibiting the bacterium’s own F1Fo ATP synthase. Cell 154(1):146–156.26. Soncini FC, García Véscovi E, Solomon F, Groisman EA (1996) Molecular basis of the

magnesium deprivation response in Salmonella typhimurium: Identification of PhoP-regulated genes. J Bacteriol 178(17):5092–5099.

27. Belden WJ, Miller SI (1994) Further characterization of the PhoP regulon: Identifica-tion of new PhoP-activated virulence loci. Infect Immun 62(11):5095–5101.

28. Matsuyama T, et al. (1992) A novel extracellular cyclic lipopeptide which promotesflagellum-dependent and -independent spreading growth of Serratia marcescens.J Bacteriol 174(6):1769–1776.

29. Martínez A, Torello S, Kolter R (1999) Sliding motility in mycobacteria. J Bacteriol181(23):7331–7338.

30. Fraser GM, Hughes C (1999) Swarming motility. Curr Opin Microbiol 2(6):630–635.31. Spormann AM (1999) Gliding motility in bacteria: Insights from studies ofMyxococcus

xanthus. Microbiol Mol Biol Rev 63(3):621–641.32. Eberl L, Molin S, Givskov M (1999) Surface motility of serratia liquefaciens MG1.

J Bacteriol 181(6):1703–1712.33. Brenner FW, Villar RG, Angulo FJ, Tauxe R, Swaminathan B (2000) Salmonella no-

menclature. J Clin Microbiol 38(7):2465–2467.34. Matsuyama T, Fujiita M, Yano I (1985) Wetting agent produced by Serratia marces-

cens. FEMS Microbiol Lett 28(1):125–129.35. Recht J, Martínez A, Torello S, Kolter R (2000) Genetic analysis of sliding motility in

Mycobacterium smegmatis. J Bacteriol 182(15):4348–4351.36. Recht J, Kolter R (2001) Glycopeptidolipid acetylation affects sliding motility and bio-

film formation in Mycobacterium smegmatis. J Bacteriol 183(19):5718–5724.37. Kinsinger RF, Shirk MC, Fall R (2003) Rapid surface motility in Bacillus subtilis is de-

pendent on extracellular surfactin and potassium ion. J Bacteriol 185(18):5627–5631.38. Brown II, Häse CC (2001) Flagellum-independent surface migration of Vibrio cholerae

and Escherichia coli. J Bacteriol 183(12):3784–3790.39. Wozniak CE, Lee C, Hughes KT (2009) T-POP array identifies EcnR and PefI-SrgD as

novel regulators of flagellar gene expression. J Bacteriol 191(5):1498–1508.40. Clegg S, Hughes KT (2002) FimZ is a molecular link between sticking and swimming in

Salmonella enterica serovar Typhimurium. J Bacteriol 184(4):1209–1213.41. Carsiotis M, Weinstein DL, Karch H, Holder IA, O’Brien AD (1984) Flagella of Salmo-

nella typhimurium are a virulence factor in infected C57BL/6J mice. Infect Immun46(3):814–818.

42. Lockman HA, Curtiss R, 3rd (1990) Salmonella typhimurium mutants lacking flagellaor motility remain virulent in BALB/c mice. Infect Immun 58(1):137–143.

43. Kimbrough TG, Miller SI (2002) Assembly of the type III secretion needle complex ofSalmonella typhimurium. Microbes Infect 4(1):75–82.

44. Snavely MD, Miller CG, Maguire ME (1991) The mgtB Mg2+ transport locus of Sal-monella typhimurium encodes a P-type ATPase. J Biol Chem 266(2):815–823.

Park et al. PNAS | February 10, 2015 | vol. 112 | no. 6 | 1855

MICRO

BIOLO

GY