JOURNAL OF BACTERIOLOGY, Apr. 2011, p. 2027–2034 Vol. 193, No. 80021-9193/11/$12.00 doi:10.1128/JB.01542-10Copyright © 2011, American Society for Microbiology. All Rights Reserved.

Tracing the Domestication of a Biofilm-Forming Bacterium�

Anna L. McLoon,1 Sarah B. Guttenplan,2 Daniel B. Kearns,2 Roberto Kolter,3 and Richard Losick1*Department of Molecular and Cellular Biology, Harvard University, Cambridge, Massachusetts 021381; Department of

Biology, Indiana University, Bloomington, Indiana 474052; and Department of Microbiology andMolecular Genetics, Harvard Medical School, Boston, Massachusetts 021153

Received 22 December 2010/Accepted 23 January 2011

Over the course of more than a century of laboratory experimentation, Bacillus subtilis has become “domes-ticated,” losing its ability to carry out many behaviors characteristic of its wild ancestors. One such charac-teristic is the ability to form architecturally complex communities, referred to as biofilms. Previous work hasshown that the laboratory strain 168 forms markedly attenuated biofilms compared with the wild strainNCIB3610 (3610), even after repair of a mutation in sfp (a gene involved in surfactin production) previouslyknown to impair biofilm formation. Here, we show that in addition to the sfp mutation, mutations in epsC, swrA,and degQ are necessary and sufficient to explain the inability of the laboratory strain to produce robustbiofilms. Finally, we show that the architecture of the biofilm is markedly influenced by a large plasmid presentin 3610 but not 168 and that the effect of the plasmid can be attributed to a gene we designate rapP. When rapPis introduced into 168 together with wild-type alleles of sfp, epsC, swrA, and degQ, the resulting repairedlaboratory strain forms biofilms that are as robust as and essentially indistinguishable in architecture fromthose of the wild strain, 3610. Thus, domestication of B. subtilis involved the accumulation of four mutationsand the loss of a plasmid-borne gene.

The soil-dwelling bacterium Bacillus subtilis has been stud-ied in the laboratory for well over a century. During this time,manipulation in the laboratory has domesticated B. subtilis,introducing mutations that have altered several of its charac-teristic behaviors. Whereas undomesticated strains have thecapacity to swarm on surfaces and form complex structuredcommunities (biofilms), many laboratory strains fail to swarmand form smooth colonies and thin pellicles at the surface ofliquids (5, 15). Here we sought to identify the full set of mu-tations common in laboratory strains that account for theirinability to form robust biofilms.

Although Christian Gottfried Ehrenberg (12) is often cred-ited with the first published description of B. subtilis, the firstauthoritative description of Bacillus subtilis was in 1877 byFerdinand Cohn, who isolated the bacterium after briefly boil-ing a culture in which hay had soaked for several hours (9).Cohn likely did not use pure cultures of this bacterium, but hedescribed many of the hallmark features of B. subtilis: thetransition between motile and filamentous cells, the develop-ment of spores, and the formation of a floating pellicle biofilmon a static liquid culture. The term “biofilm” was coined muchlater; nonetheless, Cohn clearly describes pellicle formation:“…the development after about 2 days of a delicate, iridescentfilm on the surface of the liquid. Soon thereafter the top layerbegan to become turbid and assume a slimy-flocculent or scalycharacter” (translated in reference 6).

Cohn’s work unequivocally defined the characteristics of B.subtilis. However, the ancestor of most common laboratorystrains was not isolated until around 1899 at the University of

Marburg (by Meyer and Gottheil, as described in reference10). In 1930, H. J. Conn systematically tested a dozen strainsand defined the B. subtilis type as this Marburg strain, definedby the cell and spore size, germination pattern, and the for-mation of “slightly wrinkled” colonies that stuck tightly to theagar (10). Although the culture conditions and medium dif-fered from those used today, the colony and pellicle descrip-tions from these seminal works by Cohn and Conn suggest thattheir B. subtilis strains formed biofilms and more closely re-semble the “wild” strains of NCIB3610 (3610) and ATCC 6051than the more commonly used modern laboratory strains 168,PY79, and JH642. (It is important to note that not all strainsdesignated “168” are the same. Here we tested the biofilm-forming capacities of 168 strains from the collections of R.Losick, R. Kolter, and A. L. Sonenshein. These included four168 strains as well as derivatives of 168 designated PY79,JH642, and 168 BFA. All were markedly impaired in biofilmformation, except for 168 BFA, for reasons explained herein.)

At least some of the mutations that domesticated B. subtiliswere likely introduced by Burkholder and Giles prior to theirexperimentation to irradiate the bacterium with UV light or Xrays to study amino acid synthesis and metabolism (7, 25). In1958, Spizizen demonstrated that the four surviving auxotro-phic strains mutated during the course of their work werecompetent to import and integrate genetic material from theenvironment, making genetic manipulations much easier (22).As a result of this work, strain 168 became the primary labo-ratory strain for B. subtilis research worldwide.

The ability of B. subtilis to form robust biofilms was notdescribed until 2001 by Branda et al. (4), who were workingwith strain 3610, which is closely related to the ancestor of 168(25). Strains 168 and 3610 are similar on a genomic level,suggesting that a small number of mutations are responsiblefor their phenotypic differences (11). One such mutation thatimpairs biofilm formation is in sfp, a gene required for the

* Corresponding author. Mailing address: Department of Molec-ular and Cellular Biology, Harvard University, 16 Divinity Ave.,Cambridge, MA 02138. Phone: (617) 495-4905. Fax: (617) 496-4642.E-mail: losick@mcb.harvard.edu.

� Published ahead of print on 28 January 2011.

2027

on May 20, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

production of surfactin (17, 20). We now know that surfactin isboth a surfactant required for surface motility and a signalingmolecule for biofilm formation (15, 17). However, laboratorystrains corrected for the sfp mutation remain impaired in bio-film formation, indicating the presence of additional unknowngenetic lesions that prevent the formation of architecturallycomplex communities. Here, we describe three additional mu-tations present in strain 168 but not 3610 that contribute to theloss of robust biofilm formation: (i) a previously undescribedmutation in the exopolysaccharide production gene epsC; (ii) amutation in regulatory gene swrA for the fla/che operon, whichwas not previously known to play a role in biofilm formation;and (iii) a promoter mutation for the regulatory gene degQ,which facilitates the transfer of an activating phosphate fromdegS to degU and leads to the secretion of degradative en-zymes, such as amylases and proteases (16, 24). In addition, weshow that biofilm architecture is strongly influenced by thepresence of a plasmid in 3610 that is absent in 168 and that theeffect of the plasmid can be largely attributed to a particularplasmid-borne gene, rapP. Introduction of rapP into 168 andcorrection of the four mutations restored biofilm robustnessessentially to that of the wild parent.

MATERIALS AND METHODS

Bacterial strains and culture conditions. The B. subtilis strains used in thisstudy are listed in Table 1. Escherichia coli strain DH5� was used for construc-tion and maintenance of plasmids. Strains were grown in LB medium (10 g/litertryptone, 5 g/liter yeast extract, 5 g/liter NaCl), TY medium (LB with the additionof 10 mM MgSO4 and 100 �M MnSO4), or MSgg medium (5 mM potassiumphosphate, 100 mM morpholinepropanesulfonic acid, pH 7 [MOPS], 2 mMMgCl2, 50 �M MnCl2, 50 �M FeCl3, 700 �M CaCl2, 1 �M ZnCl2, 2 �Mthiamine, 0.5% glycerol, 0.5% glutamate, 50 �g/ml threonine, tryptophan, and

phenylalanine). Solid medium contained 1.5% Bacto agar. Pellicle biofilms weregrown in 6-well microtiter plates in 10 ml liquid MSgg medium inoculated with10 �l of an LB starter culture and were incubated for 3 days at 30°C. Colonybiofilms were inoculated with 3 �l liquid starter culture, allowed to dry, andgrown at 30°C for 3 days on solid MSgg plates. Antibiotics and supplements wereincluded as appropriate at the following concentrations: ampicillin (100 �g/ml),spectinomycin (100 �g/ml), tetracycline (10 �g/ml), chloramphenicol (5 �g/ml),erythromycin (0.5 �g/ml) and lincomycin (2.5 �g/ml), kanamycin (5 �g/ml), andX-Gal (5-bromo-4-chloro-3-indolyl-�-D-galactopyranoside; 100 �g/ml).

Congression and transformation. Transformation into B. subtilis was carriedout as previously described (3). Congression transformation reactions were car-ried out as described previously (23), but cells were incubated with DNA for 2 hbefore plating. When being tested for linkage or to identify congressants withbiofilm phenotypes, cells were plated on MSgg agar containing appropriateantibiotics and, where appropriate, X-Gal (100 �g/ml) and were incubated for 2days at 37°C. Images of plates were captured with a Nikon Coolpix camera.

Markerless allele switching in epsC. We amplified a 1-kb-long stretch of DNAcentered around the C � G base pair at position 827 in epsC using primersdescribed in Table 2, cloned it into the pMAD switching plasmid, and integratedit into our recipient strain by a single crossover recombination (2). We thenscreened for derivatives of the integration strain that had spontaneously lost theplasmid by recombinational excision.

Transposon integration. To create a transposon library of PY79 and AM297,cells were transformed with the mini-Tn10-containing plasmid pIC333 and plated onsolid LB medium with macrolides-lincosamide-streptogramin B (MLS) and incu-bated for 2 days at 22°C, or until colonies appeared. Five transformants of PY79and three of AM297 were colony purified and used to inoculate separate over-night LB cultures containing spectinomycin. The overnight cultures were diluted1:100 into LB medium containing spectinomycin and grown at room temperaturefor 3 h before being shifted to 37°C for 4 h. Aliquots of each culture were platedon 2 LB-spectinomycin plates and grown overnight at 37°C. Cells were harvestedin parallel from each plate, and genomic DNA was isolated and used for mappingby transformation following standard protocols. To maximize the diversity oftransposon insertions, we conducted parallel transformation reactions with twoseparate DNA preparations from each of the 5 independent PY79 libraries andthe 3 independent AM297 libraries.

EPS precipitation. Extracellular polysaccharide (EPS) precipitation and visu-alization were carried out as described previously (13).

TABLE 1. Strains used in this study

Strain Genotype

168 ...........................trpC23610 .........................Undomesticated strainPY79........................Prototrophic 168 derivative168 BFA..................trpC2 (gift from A. L. Sonenshein)JH642 ......................trpC2 pheA1 (gift from A. L. Sonenshein)RL3082....................3610 �swrA::tetQB5505 ...................PY79 �sigL::kanRL3090....................168 sfp� (mls linked to sfp)RL4370....................PY79 �yueE::kanAM52.......................3610 amyE::PyqxM-lacZ sacA::kanAM128.....................Congressant; AM52 DNA into RL3090AM172.....................PY79 amyE::PyqxM-lacZ sacA::kanAM271.....................168 sfp� epsC�

AM236.....................PY79 yvdP/yvdO �specAM237.....................PY79 pnbA �specAM297.....................Congressant; AM52 DNA into AM271AM305.....................3610 yvzG/yvyD �specAM311.....................168 sfp� epsC� swrA� (yvzG/yvyD �spec)AM312.....................168 sfp� epsC� swrA� degQ� (yvzG/yvyD �spec)AM373.....................168 sfp� epsC�swrA� degQ� amyE::PrapP-rapP

phrP cm (yvzG/yvyD �spec)DS991......................3610 sinR::kan tasA::Tn10 specDS2569....................3610 lacking plasmidDS5187....................3610 sinR::kan tasA::Tn10 spec epsH::tetDS5188....................168 sfp� mls sinR::kan tasA::Tn10 specDS5189....................168 sfp� mls sinR::kan tasA::Tn10 spec epsC3610

DS5190....................PY79 sfp� mls sinR::kan tasA::Tn10 specDS5191....................PY79 sfp� mls sinR::kan tasA::Tn10 spec epsC3610

DS5733....................3610 sinR::kan tasA::Tn10 spec epsC168 spec

TABLE 2. Primers and plasmids used in this study

Primer orplasmid Sequence or genotype

Primer349 rapPcompF

EcoRV..........................AGGAGGATATCTTCATCCGGAGAC

TATTTATGAACAA350 rapPcompR

BamHI..........................CTCCTGGATCCTTAGGTGGTAGCA

CCATTCTTGCAamyD for..........................CTGCTTTGCTCTCCGTTCTGGTCTamyD rev .........................CATTAATAGGGCTGATGCCGAAGAfadA for2 .........................CGTTTGGTACAGAGGTTGATGAAGfadA rev2 .........................GGAACTTGCGACATGGATTCTGCTnarG for2 .........................GGAAGGAGTGAACTCTCATGAAGAnarG rev2.........................CACGTATGGATATTTCACACGGAGykoP for............................CGAGTCAGGCTGACGAAGTACAAykoP rev ...........................TGCTGACCGAGGCATCAGTCTTTdegQ for...........................CAGGAAACGCCAAGAATCGCATATdegQ rev ..........................GACTTGTTTCCAAGTCTTTTTCACGswrA for ...........................GTTCGAATTCCTTAAAGAGATTATG

GATCATAAGTCACATswrA rev...........................GTTCGGATCCATGGCTTGGATATCC

TCAGGAGAGTepsC for............................GCTGTGTATCACACGATGTTCTTCepsC rev ...........................CCTCTTCTGGATTGTGTTCCATepsC BamHII

pMAD ..........................GTTCGGATCCGATGTTCTTCCGTCT

GTTAACCGCepsC EcoRI

pMAD ..........................GTTCGAATTCTCATTGCCGGATGTG

TTACTGTC

PlasmidpDP105.............................amyE::PrapP-rapP phrP cat amp

2028 MCLOON ET AL. J. BACTERIOL.

on May 20, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

Illumina whole-genome sequencing and analysis. Genomic DNA was isolatedfrom strains AM311 and AM312 and sheared by sonication using a Bioruptor(Diagenode). An Illumina library was prepared using the genome sample prep-aration kit (Illumina), concentrations were measured by quantitative PCR(qPCR), and the library was sequenced by 36-bp single end reads on an Illuminagenome analyzer II through the Harvard Center for Systems Biology core facility.The data were aligned to the 168 reference genome (NC_000964.2) andsorted using Bowtie and pileup on Galaxy and an in-house script. Analysiswas also done using CLC Bio Genome workbench for comparison. Possiblemutations were validated by PCR and Sanger sequencing using the primersshown in Table 2.

rapP complementation construct. To generate the PrapP-rapP phrP comple-mentation construct (pDP105), a PCR product containing the rapP phrP codingregion plus �500 bp of upstream sequence was amplified from B. subtilis 3610DNA with the primer pair 349/350 and digested with BamHI and EcoRV. Thedestination vector pDG364, containing a polylinker and chloramphenicol resis-tance cassette between two arms of the amyE gene (12a) was first digested withEcoRI and blunt ended by treatment with DNA polymerase I Klenow fragment(New England BioLabs). The linearized DNA was purified by phenol-chloro-form extraction and ethanol precipitation and digested with BamHI. The di-gested DNA fragment containing rapP phrP was then ligated into the BamHI siteon one end and a blunt end on the other of the digested pDP364 backbone togenerate pDP105.

RESULTS AND DISCUSSION

A mutation that contributes to impaired biofilm formation.Laboratory strains of B. subtilis, such as 168, are often mark-edly attenuated in their ability to form robust biofilms com-pared to wild strains, such as 3610 (5). A known mutationthat contributes to impaired biofilm formation by 168 is inthe sfp gene. sfp encodes a broad-substrate-specificity phos-phopantetheinyl transferase involved in the production of nu-merous B. subtilis secondary metabolites (18). Among these issurfactin, a lipopeptide signaling molecule known to help trig-ger the expression of genes involved in biofilm formation (17,20). Accordingly, the starting point for the present investiga-tion was a derivative of 168 (RL3090) that was corrected forthe sfp mutation. As shown in Fig. 1, strain 168 itself formedflat, featureless colonies and a flat, thin pellicle, and correctionof the sfp mutation (RL3090) yielded minimal changes in mor-phology (Fig. 1; sfp�). In comparison, 3610 formed an archi-tecturally complex colony on solid biofilm-inducing medium(MSgg) and a thick and structurally complex pellicle at theair-liquid interface of a standing culture grown in MSgg.

To identify additional mutations in RL3090 responsiblefor defective biofilm formation, we used congression tomove random segments of DNA from a strain of 3610 re-sistant to kanamycin and harboring a lacZ fusion (AM52) intocompetent cells of RL3090. Colonies from congressants show-ing an enhanced wrinkly morphology were visible after 2 daysof incubation at 37°C (pink arrows in Fig. 2A) and at a fre-quency of about 3 to 5%. As a control, no such colonies wereobserved among congressants generated with donor DNAfrom a domesticated strain harboring the kanamycin resistancegene and a lacZ fusion (Fig. 2B). The more wrinkly coloniesappeared at roughly the same frequency as blue colonies aris-ing from uptake of the lacZ reporter, which appeared at afrequency of 1 to 5%. This observation is consistent with theidea that the enhanced wrinkly phenotype of each congressantresulted from correction of a single mutation or several closelylinked mutations. Congressants that produced wrinkly coloniesretained other characteristics of the recipient RL3090 strain;they were sensitive to chloramphenicol, LacZ�, and auxotro-

phic for tryptophan. The colonies and pellicles produced by thecongressants exhibited a morphology that was intermediatebetween that of the recipient strain and that of the donor 3610strain. That is, the congressants were only partially repaired intheir ability to form biofilms, an observation that suggestedthat mutations at different loci were responsible for the defec-tive biofilm phenotype of the laboratory strain.

A mutation impairing biofilm formation is located in epsC.To identify the mutation that had been corrected in generatingthe wrinkly congressants, one such congressant (AM128) wascolony purified and used to create competent cells. The com-petent cells were transformed with genomic DNA from a li-brary of transposon Tn10 insertions into the genome of strainPY79 (a derivative of 168). Next, we screened the transfor-mants for those exhibiting the impaired biofilm formation phe-notype of the donor strain. We then prepared chromosomalDNA from 56 such transformants and in each case verifiedlinkage to the mutation causing impaired biofilm formation tothe transposon by transformation. In 8 of the 56 cases, we

FIG. 1. Colony and pellicle phenotypes of strains harboring mu-tations that impair biofilm formation. Colonies were grown onMSgg agar medium for 3 days at 30°C. The scale bar represents 2mm. Rough spots often appear in the 168 sfp� and 168 sfp� epsC�

colonies, which are thought to represent spontaneous mutations inthe gene for the SinR repressor for matrix operons. The strains are168, 168 sfp� (RL3090), 168 sfp� epsC� (AM271), 168 sfp� epsC�

swrA� (AM311), 168 sfp� epsC� swrA� degQ� (AM312), 3610 withoutplasmid (DS2569), and NCIB3610. Pellicles were grown in 6-well mi-crotiter plates for 3 days at 30°C. The scale bar represents 1 cm.

VOL. 193, 2011 DOMESTICATION OF A BIOFILM-FORMING BACTERIUM 2029

on May 20, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

observed measurable linkage, with cotransformation frequen-cies ranging from 6 to 30%. Finally, the location of each of thelinked transposons was determined by sequencing DNA flank-ing chromosomal DNA. As can be seen in Fig. 3A, the trans-posons were clustered on either side of the epsA to -O (epsA-O)operon (triangles in Fig. 3A), suggesting that the mutation(s)was located within the operon.

That the mutation was in or near the operon was verified intransformation experiments using chromosomal DNA fromstrains of 3610 harboring antibiotic resistance cassettes thathad been inserted into genes pnbA and sigL located near theeps operon (black bars in Fig. 3A). Both markers showedapproximately 35% linkage to the mutation responsible for therough phenotype. Since the markers flank the epsA-O operon,this again suggested that the mutation was located within theepsA-O operon.

Sequence analysis of the epsA-O operon from RL3090 and

from the AM128 congressant that had been partially repairedfor biofilm formation revealed a single base pair difference.This was a C � G-to-T � A transition at bp 827 of epsC, whichchanged valine codon 276 (GCG) present in RL3090 to analanine codon (GTG). Further sequence analysis confirmedthat the original donor strain, 3610, had the alanine codon andthat two other laboratory strains, JH642 and PY79 (which arederivatives of 168), had the valine codon. Thus, some labora-tory strains harbor a mutation in epsC [epsC(A276V)], whichwe will refer to as epsC168.

We were surprised to discover that the published genomesequence for 168 has the wild-type T � A base pair, which weconfirmed by resequencing this region of the chromosomefrom the strain that had been used in generating the Subtilistdatabase, which we will designate as strain 168 BFA (Fig. 3A).However, when grown under biofilm conditions, 168 BFA pro-duced more robust pellicles and more wrinkly colonies thandid the other laboratory strains used in this study (data notshown). These findings confirm that the presence of a T � A atbp 827 is closely correlated with an increase in biofilm robust-ness. We acquired several other 168 strains, and only this 168BFA strain contained the T � A at bp 827. Thus, we suggestthat 168 strain variants that contain a repaired epsC allele shouldbe so designated, similar to standard discussion of strains inwhich the tryptophan auxotrophy has been repaired. To ruleout the possibility that the effect on biofilm robustness wasactually due to an unknown, unlinked mutation and to producea marker-free version of this strain, we switched the T � A basepair to a C � G base pair to repair epsC in RL3090, our originaldonor strain (see Materials and Methods). This newly builtstrain, AM271 (Fig. 1), had the same biofilm phenotype as thatobserved in our congressant strain.

The epsC168 mutation causes decreased exopolysaccharideproduction. The epsC gene is thought to play a role in exo-polysaccharide synthesis, but its precise function is unknown.Sequence analysis using the SMART protein domain databasesuggests that EpsC contains two transmembrane segments anda large catalytic domain that resembles an epimerase or dehy-drogenase. The alanine-to-valine substitution identified in cer-tain laboratory strains lies near the N-terminal end of thecatalytic domain (Fig. 3B). The region containing the substi-tution is highly conserved, with an alanine or a glycine beingfound at the homologous position in orthologs from otherspecies, such as Pseudomonas aeruginosa and Staphylococcusaureus (Fig. 3B). It is likely that the bulky side chain of valineis responsible for impairing or altering the function of EpsC,and we hypothesized that the loss of this catalytic activityprevented proper exopolysaccharide synthesis.

To investigate this possibility, we created otherwise isogenicstrains with the wild-type or mutant alleles of epsC in twodifferent domesticated backgrounds (those of 168 and PY79)and in the wild 3610 background. We introduced mutations insinR and tasA to increase overall exopolysaccharide productionand to allow the release of that exopolysaccharide into themedium (4, 8). Exopolysaccharide was first detected by theformation of aggregates following ethanol precipitation of cellsupernatants. For each strain with the epsC168 mutant allele, noexopolysaccharide aggregates were detected, whereas strains withthe wild-type allele of epsC (epsC3610) were able to producesuch aggregates (Fig. 4). Additionally, when the precipitate

FIG. 2. Identification of congressants that were partially repairedfor biofilm formation. Competent cells of a derivative of the labo-ratory strain 168 sfp� (RL3090) were transformed with genomicDNA from LacZ� Kanr-containing derivatives of 3610 (AM52)(A) or 168 (AM172) (B) DNA under conditions (DNA excess) favor-ing congression. Whereas transformation with either DNA led to sim-ilar frequencies of lacZ� congressants (blue colonies, some of whichare labeled with black arrowheads), only transformation with 3610DNA yielded congressants producing rougher colonies than the recip-ient (A; pink arrows). Colonies were grown on solid MSgg containingkanamycin and X-Gal at 37°C for 2 days.

2030 MCLOON ET AL. J. BACTERIOL.

on May 20, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

was resolved by SDS-PAGE, only strains with the wild-typeallele of epsC showed an exopolysaccharide band (Fig. 4). Forcomparison, an epsH mutant was also found to be defective inexopolysaccharide production (Fig. 4) (5, 13).

An additional mutation contributing to impaired biofilmformation. Although the derivative of RL3090 harboring thewild-type allele of epsC formed more robust biofilms than didthe original recipient strain, the biofilms formed by this 168sfp� epsC� strain were still attenuated in comparison with

those of 3610. To identify other mutations responsible for theloss of biofilm robustness in domesticated strains, we usedAM271 (168 sfp� epsC�) as a recipient strain for a secondround of congression with donor DNA from 3610. Once again,we were successful in identifying congressants that formedmore robust biofilms than did the AM271 recipient. We intro-duced random insertions of Tn10 into one such congressantand obtained a Tn10 insertion that showed 60% cotransfor-mation with a locus that caused a significantly rougher colonyphenotype.

The third mutation contributing to impaired biofilm forma-tion is located in swrA. Sequence analysis revealed the trans-poson was inserted in yvjA (Fig. 5A). Independent confirma-tion that the mutation was located in the yvjA region camefrom transformation experiments using a marker (spec) thathad been placed between the nearby genes yvyD and yvzG. Themarker showed�50% linkage to the additional biofilm-impair-ing mutation, and transformation of AM271 with genomicDNA from a 3610 strain containing this marker resulted inmany colonies forming more robust biofilms. One such strain,AM311, is shown in Fig. 1. This region of the genome (Fig. 5A)contains swrA, which is known to be mutated in many labora-tory strains (14). The wild-type coding sequence contains an8-bp track of A � T base pairs; laboratory strains typically con-tain an insertion of 1 bp, resulting in a frameshift mutation anda truncated, inactive protein. The wild-type SwrA protein stim-ulates transcription of the fla/che operon and is known to beneeded for swarming motility and for poly--polyglutamic acidsynthesis (14, 23). It therefore seemed attractive to suppose

FIG. 3. Domesticated strains contain a mutation in epsC that impairs biofilm formation. (A) DNA-mediated transformation was carried outusing the indicated transposon insertions (triangles) and drug resistance markers (rectangles), revealing linkage between the mutation responsiblefor the rough biofilm phenotype in strain AM128 and the epsA-O operon. The operon was sequenced, and a C � G-to-T � A missense mutation(epsC168) was found in epsC at bp 827. (B) Alignment of the amino acid sequence of a region of EpsC from B. subtilis containing the predictedA276V substitution and corresponding regions of orthologs from the following related species (with GenBank accession no. in parentheses):Bacillus amyloliquefaciens (ABS75496), Bacillus licheniformis (AAU25142.1), Bacillus pumilus (ABV63735.1), Bacillus cereus (NP_981687.1),Bacillus halodurans (BAB07437.1), Staphylococcus aureus (ZP_04016170), Streptococcus pneumoniae (ZP_02713626.1), Clostridium botulinum(YP_001392387.1), and Pseudomonas aeruginosa (ABJ12373.1). WT, wild type; dom., domesticated.

FIG. 4. The epsC168 mutation impairs exopolysaccharide produc-tion. The top row contains images of the chambers of a 12-wellmicrotiter dish containing ethanol-precipitated supernatant fromthe indicated strains. The bottom row contains ethanol-precipitatedsupernatant from the indicated strains resolved in the stacking gelof an SDS-PAGE gel stained with Stains-All. All strains containsinR::kan and tasA::Tn10 spec mutations to increase expression ofthe eps operon and to liberate the EPS from the cell surface,respectively. The indicated wild-type and mutant strains are asfollows: 3610 epsC3610 (DS991), 3610 epsH (DS5187), 3610 epsC168

(DS5733), 168 epsC168 (DS5188), 168 epsC3610 (DS5189), PY79epsC168 (DS5190), and PY79 epsC3610 (DS5191).

VOL. 193, 2011 DOMESTICATION OF A BIOFILM-FORMING BACTERIUM 2031

on May 20, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

that the swrA mutation in laboratory strains contributes toimpaired biofilm formation and that AM311 had been cor-rected for the swrA mutation. Sequence analysis confirmedthat transformants that formed more robust biofilms containedthe wild-type (8 A � T base pairs) allele of swrA and the lessrobust biofilm formers contained the mutant allele of swrA(Fig. 5A). Sequence analysis of 7 kb of DNA in the regionrevealed no other mutations. Moreover, introduction of aninsertion/deletion allele of swrA impaired the biofilm pheno-type of AM311. We conclude that a strain carrying wild-typealleles of sfp, epsC, and swrA is more robust in biofilm forma-tion than a corresponding strain that is mutant for swrA andhence that swrA also contributes to robust biofilm formation.

Despite its known role in swarming and poly--polyglutamicacid production and colony mucoidy (14, 23), why was swrA notimplicated in biofilm formation previously? We found that theintroduction of a swrA mutation into 3610 caused only a subtle,easily overlooked effect on biofilm architecture. We thereforeregard AM271 (sfp� epsC�) as a sensitized strain in which thecontribution of swrA was more readily apparent than in a fullyrobust, biofilm-forming strain.

A fourth mutation impairing biofilm formation is located inthe promoter for degQ. Since our triply corrected laboratorystrain AM311 (sfp� epsC� swrA�) was still not as robust inbiofilm formation as is the wild strain 3610, we sought toidentify further biofilm-attenuating mutations in the domesti-cated background. Another strain isolated during the samecongression and transformation experiment as AM311 dem-

onstrated more robust biofilm formation (Fig. 1), and we an-ticipated that this strain, AM312, contained the wild, repairedallele of an additional biofilm-impairing mutation in additionto being swrA�. Colonies with this phenotype appeared at afrequency of �5% during the transformation reactions fromwhich AM311 were isolated. Efforts to map the mutation bylinkage to a Tn10 insertion were unsuccessful. We thereforeturned to whole-genome sequencing to identify differences be-tween AM311 and AM312 (see Materials and Methods). Weidentified seven point mutations that differed between thestrains (Table 3). We then directly sequenced the regions con-taining these changes in the parental 3610 and 168 strains andalso from 5 independently generated congressant strains show-ing the same biofilm morphologies as AM311 and 5 indepen-

FIG. 5. Domesticated strains contain mutations in swrA and degQ that impair biofilm formation. DNA-mediated transformation was carried outusing the indicated transposon insertions (triangles) and drug resistance markers (rectangles), revealing linkage to the mutation responsible forthe rough biofilm phenotype in strains AM311 (A) and AM312 (B). (A) Laboratory strains are known to contain an A � T insertion mutation inthe swrA gene, as indicated, and AM271 (168 sfp� epsC�) also contains this insertion, whereas the more robust biofilm-forming AM311 strain doesnot. Numbered base pairs refer to 3610 sequence. (B) Illumina sequencing reveals a mutation in the degQ promoter.

TABLE 3. Allelic differences between AM311 and AM312identified by whole-genome sequencing

Location Mutation Change Correlation withbiofilm defect

ykoN A to T Ile1103Phe NoamyD A to T His843Leu NoBefore degQ G to A Promoter YesyusK T to C Lys363Glu NoflgM T to C Silent NSa

ggaB G to A Silent NSnarG G to A Thr533Ile No

a NS, not sequenced.

2032 MCLOON ET AL. J. BACTERIOL.

on May 20, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

dently generated hybrid strains showing the same morpholo-gies as AM312. Only one of the mutations—the one near theregulatory gene degQ—met the criteria of being present in 168,AM311, and other congressants with impaired biofilm forma-tion but absent in 3610, AM312, and independently isolatedcongressant strains with the more robust biofilm phenotype.Consistent with the assignment of the degQ mutation as beingresponsible for impaired biofilm formation in the tested strains,DNA-mediated transformation using a marker inserted in thechromosome in a gene (yueE) located near degQ revealed highlinkage (55% cotransformation) between the marker and themutation impairing biofilm formation (Fig. 5B). Gratifyingly,Kobayashi (16) had previously and independently found thatmutations in degQ impair biofilm formation.

Sequence analysis revealed that the mutation was a T � A-to-C � G transition at the upstream (5) end of the �10 regionof the promoter for degQ. The wild-type T � A base pair at thisposition is highly conserved in promoters recognized by RNApolymerase containing �A (19). Indeed, during the initial char-acterization of degQ, Yang et al. observed that a mutant bear-ing this T � A base pair produced more DegQ (then calledSacQ) activity, as judged by increased levels of secreted amy-lases, proteases, and levansucrase (1, 24). Finally, we note thata degQ null mutation has been shown to impair poly--poly-glutamic acid synthesis in certain wild strains, which causes amucoid colony phenotype (16, 23). Poly--polyglutamic acid,however, is not a component of the matrix, and its productionis not required for robust biofilm formation (4).

A plasmid-borne gene influences biofilm morphology. It isknown that 3610 contains an 80-kb plasmid. In other work, oneof us (D.B.K.) generated a derivative of 3610 that was cured ofthe plasmid (unpublished results). Interestingly, the cured strainexhibited a colony morphology phenotype that was highly similarto that of strain AM312, the strain that had been corrected forsfp, epsC, swrA, and degQ. Thus, it seemed possible that theonly remaining, relevant difference between AM312 and 3610was the absence of the plasmid in the corrected laboratorystrain.

In an earlier screen for mutations altering matrix gene ex-pression in 3610 by transposon-mediated insertional mutagen-esis, we had in fact identified an insertion in a plasmid-bornegene. We designate this gene rapP because of the similarity ofits inferred product to other members of the family of responseregulator aspartate phosphatases from B. subtilis (21). To in-vestigate whether the absence of rapP in the corrected labora-tory strain could account for the difference in morphology from3610, we introduced DNA containing rapP (and the adjacentdownstream gene phrP) at amyE in AM312. Strikingly, theresulting strain (AM373) formed colonies on biofilm-inducingmedium that were essentially indistinguishable in architecturefrom those of 3610 (Fig. 1).

In toto, the domestication of B. subtilis with respect to bio-film formation can be largely, if not entirely, attributed to fivegenetic alterations. These are point mutations in the codingsequences of sfp, epsC, and swrA; a promoter mutation in degQ;and the absence of the plasmid-borne gene rapP.

This has been in part an historical investigation in which wehave sought to identify mutations that have accumulated overtime in the domestication of a laboratory bacterium. It is dif-ficult to know precisely when, and in which order, the muta-

tions arose. For instance, did the epsC mutation arise in thelaboratory of Burkholder and Giles, or did subsequent inves-tigators inadvertently (or conceivably purposefully) select fordecreased biofilm robustness by choosing smoother, more-re-producible-looking colonies? In any event, the lesson here isthat generations of investigators have robbed laboratory strainsof some of the important biological properties of the ancestralstrains first studied by Meyer and Gottheil over a century agoand first described by Ehrenberg and Cohn a century and a halfago (10).

ACKNOWLEDGMENTS

We thank A. L. Sonenshein, P. Piggot, J. Perkins, and D. Ziegler forhelpful discussions, A. L. Sonenshein for strains, and A. Kurger forhelp with Illumina data analysis.

This work was supported by grants from BASF to R.L. and R.K. andby NIH grants GM18569 to R.L., GM093030 to D.B.K., and GM58213to R.K.

REFERENCES

1. Amory, A., F. Kunst, E. Aubert, A. Klier, and G. Rapoport. 1987. Charac-terization of the sacQ genes from Bacillus licheniformis and Bacillus subtilis.J. Bacteriol. 169:324–333.

2. Arnaud, M., A. Chastanet, and M. Debarbouille. 2004. New vector forefficient allelic replacement in naturally nontransformable, low-GC-content,gram-positive bacteria. Appl. Environ. Microbiol. 70:6887–6891.

3. Bott, K. F., and G. A. Wilson. 1967. Development of competence in theBacillus subtilis transformation system. J. Bacteriol. 94:562–570.

4. Branda, S. S., F. Chu, D. B. Kearns, R. Losick, and R. Kolter. 2006. A majorprotein component of the Bacillus subtilis biofilm matrix. Mol. Microbiol.59:1229–1238.

5. Branda, S. S., J. E. Gonzalez-Pastor, S. Ben-Yehuda, R. Losick, and R.Kolter. 2001. Fruiting body formation by Bacillus subtilis. Proc. Natl. Acad.Sci. U. S. A. 98:11621–11626.

6. Brock, T. D. 1975. Milestones in microbiology. American Society for Micro-biology, Washington, DC.

7. Burkholder, P. R., and N. H. Giles, Jr. 1947. Induced biochemical mutationsin Bacillus subtilis. Am. J. Bot. 34:345–348.

8. Chu, F., D. B. Kearns, S. S. Branda, R. Kolter, and R. Losick. 2006. Targetsof the master regulator of biofilm formation in Bacillus subtilis. Mol. Mi-crobiol. 59:1216–1228.

9. Cohn, F. 1877. Untersuchungen uber Bacterien. IV. Beitrage zur Biologieder Bacillen. Beitr. Biol. Pflanz. 2:249–276.

10. Conn, H. J. 1930. The identity of Bacillus subtilis. J. Infect. Dis. 46:341–350.11. Earl, A. M., R. Losick, and R. Kolter. 2007. Bacillus subtilis genome diver-

sity. J. Bacteriol. 189:1163–1170.12. Ehrenberg, C. G. 1835. Dritter Beitrag zur Erkenntniss grosser Organization

in der Richtung des kleinsten Raumes. Physikal. Abh. Koniglich. Akad. Wiss.Berl. Jahr. 1833–1835:135–336.

12a.Guerot-Fleury, A. M., N. Frandsen, and P. Stragier. 1996. Plasmids forectopic integration in Bacillus subtilis. Gene 180:57–61.

13. Guttenplan, S. B., K. M. Blair, and D. B. Kearns. 2010. The EpsE flagellarclutch is bifunctional and synergizes with EPS biosynthesis to promote Ba-cillus subtilis biofilm formation. PLoS Genet. 6:e1001243.

14. Kearns, D. B., F. Chu, R. Rudner, and R. Losick. 2004. Genes governingswarming in Bacillus subtilis and evidence for a phase variation mechanismcontrolling surface motility. Mol. Microbiol. 52:357–369.

15. Kearns, D. B., and R. Losick. 2003. Swarming motility in undomesticatedBacillus subtilis. Mol. Microbiol. 49:581–590.

16. Kobayashi, K. 2007. Gradual activation of the response regulator DegUcontrols serial expression of genes for flagellum formation and biofilm for-mation in Bacillus subtilis. Mol. Microbiol. 66:395–409.

17. Lopez, D., M. A. Fischbach, F. Chu, R. Losick, and R. Kolter. 2009. Struc-turally diverse natural products that cause potassium leakage trigger multi-cellularity in Bacillus subtilis. Proc. Natl. Acad. Sci. U. S. A. 106:280–285.

18. Mootz, H. D., R. Finking, and M. A. Marahiel. 2001. 4-Phosphopantetheinetransfer in primary and secondary metabolism of Bacillus subtilis. J. Biol.Chem. 276:37289–37298.

19. Moran, C. P., Jr., et al. 1982. Nucleotide sequences that signal the initiationof transcription and translation in Bacillus subtilis. Mol. Gen. Genet. 186:339–346.

20. Nakano, M. M., N. Corbell, J. Besson, and P. Zuber. 1992. Isolation andcharacterization of sfp: a gene that functions in the production of the lipo-peptide biosurfactant, surfactin, in Bacillus subtilis. Mol. Gen. Genet. 232:313–321.

VOL. 193, 2011 DOMESTICATION OF A BIOFILM-FORMING BACTERIUM 2033

on May 20, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

21. Perego, M., et al. 1994. Multiple protein-aspartate phosphatases provide amechanism for the integration of diverse signals in the control of develop-ment in B. subtilis. Cell 79:1047–1055.

22. Spizizen, J. 1958. Transformation of biochemically deficient strains of Ba-cillus subtilis by deoxyribonucleate. Proc. Natl. Acad. Sci. U. S. A. 44:1072–1078.

23. Stanley, N. R., and B. A. Lazazzera. 2005. Defining the genetic differences

between wild and domestic strains of Bacillus subtilis that affect poly-gam-ma-dl-glutamic acid production and biofilm formation. Mol. Microbiol. 57:1143–1158.

24. Yang, M., E. Ferrari, E. Chen, and D. J. Henner. 1986. Identification of thepleiotropic sacQ gene of Bacillus subtilis. J. Bacteriol. 166:113–119.

25. Zeigler, D. R., et al. 2008. The origins of 168, W23, and other Bacillus subtilislegacy strains. J. Bacteriol. 190:6983–6995.

2034 MCLOON ET AL. J. BACTERIOL.

on May 20, 2020 by guest

http://jb.asm.org/

Dow

nloaded from