JOURNAL OF BACTERIOLOGY, Mar. 1987, p. 981-989 Vol. 169, No. 30021-9193/87/030981-09$02.00/0Copyright © 1987, American Society for Microbiology

Capsule Synthesis in Escherichia coli K-12 Is Regulatedby Proteolysis

ANGEL S. TORRES-CABASSA ANtb SUSAN GOTTESMAN*Laboratory of Molecular Biology, National Cancer Institute, Bethesda, Maryland 20892

Received 29 September 1986/Accepted 24 November 1986

Ion mutants of Escherichia coli K-12 are defective in an ATP-dependent protease, are UV sensitive, andoverproduce the capsular polysaccharide colanic acid. Sit structural genes needed for capsular polysaccharidesynthesis (cps) are transcriptionally regulated by Ion as well as by three other regulatory genes, rcsA, -B, and-C (S. Gottesmnan, P. Trisler, and A. S. Torres-Cabassa, J. Bacteriol. 162:1111-1119, 1985). We have clonedrcsA, the gene for a positive regulator of capsule synthesis, onto multicopy plasmids and defined the gene byboth insertions and deletions. The product of rcsA has been identified as an unstable protein of 27 kilodaltons.RcsA has a half-life of S min in Ion' cells and one of 20 min in Ion cells. The availability of RcsA is the limitingfactor for capsule synthesis; doubling the gene ddsage of rcsA' significantly increases expression of cps genes.Our results are consistent with a model in which the presence of a lon mutation increases the synthesis ofcapsular polysaccharide via stabilization of RcsA.

Proteolytic degradation in bacteria and their viruses hasbeen implicated in such cellular events as the destruction ofabnormal polypeptides and the regulatidn of the availabilityof a number of unstable regulatory proteins (8, 10, 26). Wehave postulated that the expression of some short-lived"timing" polypeptides needed only for brief periods of timemay be regulated at the post-transcriptional level via aproteolytic degradative process (12, 26).The lon gene of Escherichia coli K-12 codes for an

ATP-dependent protease (3, 4, 9). The Lon protease hasbeen shown to affect the degradation of abnormal proteins invivo (2, 13, 24, 32) and of the bacteriophage A N protein bothin vivo (10) and in vitro (M. Maurizi, J. Biol. Chem., inpress). Mutations in lon are pleiotropic and result in in-creased sensitivity to DNA-damaging treatments (13, 14) andoverproductior of the capsular polysaccharide colanic acid(13, 21, 33). We have shown previously that the sensitivity oflon mutants to DNA damage is due to a defect in proteolysiswhich permits the accumulation of an unstable cell divisioninhibitor (SulA) by prolonging its half-life in the cell (26).

If other effects observed in lon cells are a direct result ofa defect in proteolysis, as is the case for sensitivity to DNAdamage, we would predict the existence of an unstablepositive regulator of polysaccharide synthesis which is nor-mally limiting in the cell and which would be a target of theLon protease.

Several of the enzymes involved in capsular polysaccha-ride synthesis are present in higher levels in lon mutants (18,21, 22), and at least six structural genes required for capsuleproduction (cpsA-F) are negatively regulated by lon at thetranscriptional level (33).Secondary trans-acting regulatory mutations which affect

capsule synthesis have been isolated and mapped (11). Twopositive regulatory genes, rcsA (regulator of capsuile synthe-sis) and rcsB, as well as a negative regulator, rcsC, affect thelevels of colanic acid capsular polysaccharide in cps' strainsand also change the expression of 0-galactosidase synthesisproduced in strains which carry lacZ fusions to the structuralgenes for capsule synthesis (cps::lac) (11). The existence ofthese regulatory genes suggests that the defect in regulation

* Corresponding author.

of capsule expression found in lon strains may be mediatedindirectly by the Lon protease through its action on one ofthe two positive regulators; thus, either rcsA or rcsB mightcode for a gene product which is normally limiting in the celland the stability of which is regulated by proteolysis.

In this work we have identified the product of the rcsAgene and determined its half-life in lon+ and lon cells. Theevidence presented show that RcsA protein is indeed unsta-ble and that its stability is affected by the lon protease.

MATERIALS AND METHODSBacterial strains and bacteriophage. All bacterial strains

relevant to this work and their sources or construction arelisted in Table 1.

Plasmids and bacteriophage are listed in Table 2. Isolationof A rcsA+ phage and construction of X c1857 derivativeshave been described elsewhere (11, 24).

Transductions with P1 CmclrlOO (28) were done as de-sdribed by Miller (25).

Deletions of the rcsA gene were generated by selecting fortetracycline-sensitive derivatives of strains carrying a TnlOinsert linked to rcsA by the method of Maloy and Nunn (19).Media and enzyme assays. Unless otherwise indicated,

cells were routinely grown in Luria broth (30). When re-quired, the following antibiotics were added at the finalconcentrations indicated: tetracycline (20 ,I.g/ml), ampicillin(75 p.g/ml), or kanamycin (50 p.g/ml).M56 salts [Na2HPO4, 8.2 g/liter; KH2PO4, 2.7 g/liter;

(NH4)2S04, 1.0 g/liter; FeSO4 7H20, 0.25 mg/liter;MgSO4 * 7H20, 0.1 mg/liter; CA(NO3)2 4H20, 5 mg/liter,pH 7.2) was supplemented with amino acids, thiamine,glucose (0.2%), and MgSO4 (0.01 M). M56 minimal mediumcontained 3.2% agar.MutD mutagenesis of A rcsA was carried out as described

by Silhavy et al. (30).P-Galactosidase expression in the fusion strains was mon-

itored on lactose-MacConkey indicator plates as describedpreviously (33). P-Galactosidase production was assayed asdescribed by Miller (25).

Insertional mutagenesis. Insertions of the transposase-deficient A16A17TnJO (ATnJO) transposon (35) in X rcsA+transducing phage were obtained by growing the phage on

981

on Decem

ber 15, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

982 TORRES-CABASSA AND GOTTESMAN

TABLE 1. Bacterial strains

Strain Relevant genotype Source or reference

MC4100 derivativesaSG20180 cps-JJ::lac-Mu dI Ion' 33SG20303 cps-11::lac-Mu dl lon-100 ton 11SG20329 ompC::TnS rcsC137 cps-li::Mu dl 11SG20581 cpsBJO::lac (imm') lon-100 33SG20582 cpsBJO::lac (imm') lon+ 33SG20587 cpsBJO::lac (imm') recA SG20582 + Pl(SG13182) +

Pl(N100)SG20595 cps-JJ::Iac-Mu dl lon-i 00" 11SG20597 cps-11::lac-Mu dl ion-100 rcsB43 zeh-S0::Tn]O0 11SG20598 cps-il::iac-Mu dl lon-100 rcsA40 zed-14::TniO" 11SG20600 cps-1l::lac-Mu dl" 11SG20643 cpsBJO::lac (imm') ompC: :TnS rcsC137 SG20581 + P1(SG20329)SG20645 cpsBJO::lac (immx) lonb rcsA3 (RcsA*3) SG20582 + Pl(SG12020)SG20681 cps-lJ::lac-Mu dI lon+ 24ATC5112 cps-ll::Mu dl ion-100 rcsA72::ATnJO SG20303 + X SY72JB3030 lon-100 rcsA+ recA cpsBlO::lac(imm`) Brill and Gottesman, in prepnJB3034 lon-100 A rcsA26 recA cpsBIO::lac (imm') Brill and Gottesman, in prepnJB3035 Ion+ A rcsA27 cpsBJO::lac (imm') Brill and Gottesman, in prepn

Other strain backgroundsC600 F- thr leu .fhuA rpsL supE NIHC strain collectionDB1255 hsdR recBC sbcB15 supF8 36N99 F- galK2 NIH strain collectionN100 F- galK2 recA NIH strain collectionRB132 F-Alac rpsL gal zxx::TnOtA4HH104(pNK217) 23SG12020 thr leuflzuA zed-650::TniO rcsA3 (RcsA*3) 11SG13182 his sulA Aleu sri::TniO rpsL 33

aAll strains contain A(lac)U169 araDflbB relA.b Contain attB bio-936 A(Sal-Xho) cI857 AH1 prophage.c NIH, National Institutes of Health.

donor strain RB132, which harbors pNK217 (his: :ATnJO), as

described previously (16, 23, 24). In this strain, transpositionfunctions are provided in trans from a chromosomal "highhopper" TnJO insertion (23). Phage carrying inserts weredetected as tetracycline-resistant lysogens. rcsA-72: :ATnJOwas isolated by the same method from X rcsA62 (RcsA*)(SY62).

Strain SG20681 (lon+ cps::lac) carries a defective heat-inducible prophage (int+ xis+ c1857 AHl) and has a Lac-phenotype on indicator plates. This strain was grown at 32°Cin TBMM,(10) with biotin (0.003%), and the cells were

shifted to 41°C for 20 min to provide int+ function. Theywere then infected with the rcsA mutagenized phage at a

multiplicity of infection of 0.1 to 0.5 and adsorbed at 32°C for1 h. Lysogens were selected on either lacto'se-MacConkey orLuria broth agar with tetracycline. Of the Tetr colonies 2 to4% were Lac-. Five independent Lac- Tetr lysogens from as

many independent mutagenized lysates were purified, andsecondary lysates were induced. Tetr phage that failed tocomplement an rcsA chromosomal mutation, and whichretained their Int- phenotype, as determined from the redplaque assay (7), were considered to carry an inactivatedrcsA gene. Phage stocks from purified plaques were used infurther analyses and in DNA isolation for restriction diges-tions.The ATnJO insertions isolated on the phage were crossed

onto the bacterial chromosome by adsorbing the A c1857rcsA+ mutagenized phage on strain SG20303 (cps: :iac Ion) at32°C. Lysogens were cured of the phage at 37°C on Luriabroth plates with tetracycline and screened for the Lacphenotype and loss of immunity to the phage. P1 bacterio-phage was grown on purified Tetr Lac- recombinants, and

the lysates were used in transductional mapping of theinsertions.

Selection of ATnJO insertions linked to rcsA+. A lysateinduced from a pool of Tetr lysogens of strain SG20681 wasused to isolate random ATnJO insertions linked to rcsA+.The lysate was adsorbed to strain SG20180 at 32°C to selectfor Tetr lysogens, which were then cured of the phage byselecting for tetracycline resistance at 37°C. These colonieswere then pooled, and a P1 lysate prepared on the cells wasused to transduce SG20581 (lon cps::lac) to Tetr. The Lac'Tetr transductants obtained carried ATnlO insertions nearrcsA+.

Isolation of AKan insertions on the plasmid. XNK1105carries a defective kanamycin-resistant TnJO derivative(AKan) and a ptac-transposase fusion outside the transposon(35). XNK1105 was adsorbed to an N99 transformant ofpATC395 at 39°C, and the resulting Kanr colonies were thenpooled. A small-scale DNA preparation was,made and usedto transform strain C600 selecting for Kanr nonmucoidcolonies. Three independent insertions were verified by theirfailure to complement two different rcsA chromosomal mu-tations (rcsA40 and rcsA72::ATnJO).Recombination of the AKan insertions from pATC395 to

the phage was done by growing SY25 or SY26 on N99derivatives harboring the plasmid to produce a lysate whichwas then used to transduce C600 to kanamycin resistance.Tetr Kanr nonmucoid colonies were purified twice and asecondary lysate was induced. Presence of the insertions onthe transducing phage was verified by their failure to com-plement an rcsA mutation and by restriction analysis ofphage DNA.AKan insertions in pATC400 were generated as for

J. BACTERIOL.

on Decem

ber 15, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

CAPSULE SYNTHESIS REGULATION 983

pATC395. Recombination into the chromosome was done as

follows. Strain DB1255 (recBC sbcB15) (36) is defective fora major nuclease activity but is able to carry out recombi-nation when transformed with linearized plasmid DNA (15).Strain DB1255 was transformed with DNA from the AKanplasmid digested with EcoRI. Kanr recombinants were se-

lected and screened for their sensitivity to ampicillin. P1lysates prepared on purified colohies of strains carrying theinsertion mutations rcsA161: :AKan, rcsA163: :Akan, andrcsAl62::AKan were used to transduce SG20581 (Ion-100cps::iac, Lac+) to Kanr. All Kanr transductants becameLac- and were complemented by A rcsA+ as expected if thercsA::AKan insei-tions had been transfetred from the plas-mid to the chromosome.

Preparation of DNA. Small-scale isolation of phage DNAwas done as described by Silhavy et al. (30). For large-scalephage preparations, high-titer phage lysates were obtainedby the method of Yamamoto et al. (34) and the DNA wasisolated as described previously (30). Small- and large-scaleplasmid DNAs were prepared by the alkali method describedby Maniatis et al. (20).

TABLE 2. Bacteriophage and plasmids

Name

X derivativesSB6SB18

SB25SB26SB57SY25

SY26

SYSO

SY51

SY60

SY62

SY72

A NK1105

Other phageP1CMclrlO0

M13mpl9derivativesMAT200MAT237

PlasmidspATC352pATC395pATC400pATC401pATC402pATC408pATC410pATC45Oplon+500plonAl5OpATC119pNK217

Relevant genotype Source or reference

imm2' rcsA + 11imm21 rcsB+ Brill and Gottesman,

in prepnimm2' rcsA + This workimm2' rcsA + This workimm22 rcsA57::TnJO This workimmX cI857 rcsA + This work; immx

derivative of SB25immx c1857 rcsA+ This work; immX

derivative of SB26immX c1857 This work

rcsA50: :AKanimmX c1857 This work

rcsASI: :AKanimmx c1859 rcsA+ This work; immX

derivative of SB6immX c1857 rcsA62 This work(RcsA*62)

immx cI857 This workrcsA72::ATnJO

b522 c1857 Pam8O ninS 35

cI tslOO Cmr 28

rcsA + This workArcsA37 This work

rcsA + (Cmr Apr) pBR325 derivativercsA + (Tetr) pBR322 derivativercsA + (Ampr) pBR322 derivativercsA160::AKan (Ampr) pATC400 derivativercsAl6l::AKan (Ampr) pATC400 derivativercsA166::AKan (Ampr) pATC400 derivativercsAl68: :AKan (Amp') pATC400 derivativercsA50: :AKan (Tet') pATC395 derivative(Ampr) 24(Ampr) 24rcsA + (Ampr) pUC19 derivativeM16A17TnJO 23, 35

Restrictions, ligations, and transformations. All enzymesused for digestions were obtained from New EnglandBioLabs, and restrictions were carried out according to thesupplier's specifications. pBR322, pBR325, pUC19, andM13mpl9 DNAs were obtained from Bethesda ResearchLaboratories.

Ligation reactions and transformations were done bystandard methods (20, 30).

Generation of rcsA deletions in ml3mpl9. Construction ofrcsA deletions in M13mpl9 were done in its derivativeMAT200 by the rapid single-stranded cloning strategy forproducing overlapping clones described by Dale et al. (6).The cyclone subcloning system for M13 and pBI derivatives(system RDS) from International Biotechnologies, Inc., wasused for this purpose. The eidpoints of the deletions weredetermined by sizing the single-stranded DNAs on agarosegels and restriction enzyme analysis of the double-strandedreplicative forms.

Protein lilbeling. In vivo determination of protein pro-duced by the rcsA recombinant plasmids was done inmaxicells as described by Sancar et al. (29). The maxicellstrain JB3034 and its plasmid-containing derivatives weregrown overnight in M56 salts supplemented with 17 aminoacids, 0.2% glucose, 0.1% vitamin B1, ,MgSO4 (1 mM), andthe required antibiotic, but lacking methionine. At an opticaldensity at 600 mm of 0.6, the cultures were exposed to a UVdose of 70 J/m2 for 30 s and incubated at 37°C for 30 min,protected from light. D-Cycloserine was then added, and thecultures were grown overnight, starved for 1 h, and labeledwith [35S]methionine (50 ,Ci/ml) for 20 to 30 min (30).Labeled cells were washed, suspended in 2% sodium dode-cyl sulfate loading buffet, and boiled for 5 min; 10 RI was runin a 12% acrylamide gel, using the buffers described byLaemmli (17). Electrophoresed gels were fixed for 15 min in45% methanol-10% acetic acid and immediately soaked inEnlightning (Du Pont Co.) for 20 min, dried, and subjected toautoradiography at -70°C.The in vivo stability of the RcsA protein produced from

the recombinant plasmids was determined in growing cellscultured as indicated above. After a 1-min pulse with[35S]methionine (20 ,uCi/ml), excess nonradioactive methio-nine (2 mM) was added to stop label incorporation and0.5-ml samples were taken during a 30-min period. Thesamples were mixed with 0.125 ml of 30% trichloroaceticacid to precipitate the proteins. The pellet was washed incold acetone, suspended, and boiled in 2% sodium dodecylsulfate buffer, and 10 RI was run in a 13% acrylamide gel.After electrophoresis and autoradiography, the relativeamounts of the labeled proteins were determined bydensitometry.

RESULTSIsolation and characterization of X rcsA+ transducing

phage. We have previously identified and described muta-tions at two regulatory loci, rcsA and rcsB, which abolishtranscription from the promoters of the structural genesinvolved in capsular polysaccharide synthesis in lon cells(11). It was postulated that the lon protease may inhibitcapsule formation by degrading an unstable activator oftranscription, possibly a product of one of the two positiveregulatory genes required for cps gene expression (11, 12).

If a lon mutation results in capsule overproduction byincreasing the steady-state level of a positive regulator whichis subject to proteolytic degradation in Ion' cells, thatpositive regulator should be limiting for capsule synthesis.

VOL. 169, 1987

on Decem

ber 15, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

984 TORRES-CABASSA AND GOTTESMAN

L

C'

S

N

-I-

R H B NRNSS RS B HH RI I 1 1111 11 11

I -b- [-2Kbeft UZ* PL

ite -- - -- , -I~~~~~~~~~~~~~~~~~~~~~~~~~~~B NSS NL

ASY60 vxzv/I( rjf I 4-PL

B NLlIASY25

ASY26

S N B

3 N S N

")I N B

HRI I

RightCSiSite

-PL

B

I P.I -2bI

V1Kb-|MMA

FIG. 1. Schematic diagram showing the restriction map of immX D69 and its rcsA+ derivatives SY60 (from SB6), SY25 (from SB25), andSY26 (from SB26). The cloned fragment was inserted in the BamHI site within the int gene of A D69 (27). The hatched area represents theregion within the clone in which all insertions that disrupt the rcsA gene mapped. The restriction sites shown are as follows: NruI (N), EcoRI(R), HindIII (H), BamHI (B), and Sall (S).

Furthermore, if capsule overproduction in lon mutants re-sults from a decrease in proteolysis, introduction of a secondcopy of the gene which codes for the activator protein maybe sufficient to mimic the lon mucoid phenotype in lon+strains. Conversely, increasing the amount of the lon prote-ase in the cell should abolish capsule production even if theactivator protein is overproduced.To determine if one of the two positive regulatory genes,

rcsA or rcsB, is normally limiting in the cell, we studied theeffect of changing gene dosage on transcription of thecps::IacZ operon fusions by using a A transducing phagecarrying either rcsA+ or rcsB+ (J. Brill and S. Gottesman,manuscript in preparation).The rcsA gene was cloned from a Sau3A partial E. coli

digest into the BamHI site of D69 (27). Several rcsA+transducing phage were isolated by their ability to form Lac'plaques on a lawn of a lon cps::lac rcsA strain on lactose-tetrazolium plates (11). Three independent A rcsA+ clones,SB6 (previously described [11]), SB25, and SB26, wereisolated; the immX derivatives of these phage, SY60, SY25,and SY26, were used to determine the restriction enzymepattern (Fig. 1).Both phage SY25 and SY60 contained a 6.3-kilobase (kb)

BamHI insert. Results from BamHI-SaiI, BamHI-NruI, andSaiI-NruI double digests indicated that the 6.3-kb insert waspresent in opposite orientations in these two clones. PhageSY26 carried the 6.3-kb fragment in the same orientation asSB6 plus an additional 1.9-kb BamHI piece. All three phagecomplemented rcsA mutations either in plaques or aslysogens. rcsA missense mutations of two sorts were iso-lated after MutD mutagenesis of the phage: recessive rcsA(RcsA-) and dominant rcsA (RcsA*) (11).

Insertional mutagenesis of the rcsA + phage. We observedthat increasing the number of copies of rcsA+ in strainSG20681 (lon+ cps::lac) resulted in a Lac' phenotype,indicating that RcsA is normally limiting (see below). Wesupposed that lysogens of the transducing phage carrying aninactivated rcsA gene would remain Lac.SB6 was mutagenized with a defective ATnJO transposon

(see Materials and Methods), and Tetr lysogens were

screened for their Lac phenotype. About 2% of the coloniesremained Lac- and therefore were considered to have beenlysogenized by phage carrying an inactivated rcsA gene. Thephage lysates induced from these lysogens are Lac- on alawn of a lon rcsA strain (11), which confirmed that theinserts inactivated the rcsA gene.

Restriction analysis of four independent rcsA phage iso-lated in this manner demonstrated that the insertionsmapped within a 2.0-kb NruI fragment in the 6.3-kb BamHIclone (cross-hatched area in Fig. 1). The ATnlO insertionswere crossed into the chromosome by homologous recom-bination. They abolished capsule overproduction in Ionhosts and were fully complemented by the X rcsA+ transduc-ing phage.

rcsA is limiting for capsule synthesis. Lysates of compara-ble titers of A rcsA+, A rcsA::ATnJO, or X rcsB+ were used inspot tests on lawns of both lon+ and lon cps::lac strains.Expression of ,-galactosidase from the fusion was moni-tored on MacConkey-lactose indicator plates (Table 3).Introduction of additional copies of the rcsA+ gene wassufficient to increase transcription from the fusion in bothbackgrounds. In contrast, no apparent effect on cps::iacexpression was seen with X rcsB+ (Table 3).

TABLE 3. Gene dosage effect of rcsA+ and rcsB+a

Lac phenotypeb

Strain genotype No X X Xphage rcsA+ rcsA: :TnlO rcsB+(SB6) (SB57) (SB18)

Ion' - + +lon-100 + +++ + +lon-100 rcsA40 - +lon-100 rcsB43 - - - +

a The strains used were SG20600 (Ion+), SG20595 (lon-100), SG20597(ion-i00 rcsB43), and SG20598 (lon-100 rcsA40).

b Equal volumes (10 p1) from phage lysates of comparable titers were placedover dried spots of the strains on lactose-MacConkey agar and incubated at32°C for 10 to 20 h. All phage were imm21.

c + + + indicates more rapid appearance of the Lac+ phenotype.

r -

J. BACTERIOL.

on Decem

ber 15, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

CAPSULE SYNTHESIS REGULATION 985

All three rcsA+ transducing phage (SY25, SY26, andSY60) also complemented flaP and flaR mutants. As previ-ously shown, episome F'1334 which contains thefla operonsconfers mucoidy on lon+ hosts (11, 31). Clegg and Koshland(5) also observed mucoidy in cells harboring pCK202, whichcontains theflaARQP operon. Comparison of the restrictionmap of this plasmid and that of the rcsA+ phage confirmsthat the 6.3-kb BamHI fragment overlaps a Sall-HindIIIfragment in pCK202 (see below and Fig. 6). Furthermore, adeletion which eliminated essentially all of the 6.3-kb BamHIfragment abolished the plasmid's ability to confer themucoid phenotype (5). These observations are consistentwith the mapping of rcsA. We have shown elsewhere thatrcsA is not involved in the synthesis of flagella (11). There-fore, rcsA exhibits the genetic properties expected of alimiting positive regulator of the cps genes, while moderateincreases in rcsB gene dosage apparently have little effect oncps::iac expression.

Overproduction of lon protease suppresses capsule synthe-sis. We have shown previously that a Ion mutation increasesexpression of the cps operons (33). Two types of regulatorymutations which increase capsule synthesis in Ion' hostshave been described: the dominant mutations in rcsA such asrcsA3 or rcsA62 (RcsA*), or a recessive mutation in rcsC(11). When a multicopy lon+ plasmid (24) was used totransform either a Ion' rcsA3 (RcsA*) host (Table 4) or alon+ rcsC host (Table 4), capsule synthesis was suppressed.In transformants of the same strains with plasmid thatcarries the lon-S10 deletion, cps::iac expression remainedhigh (Table 4). Therefore, while rcsA (RcsA*) and rcsCincrease capsule synthesis, this increased synthesis remainssensitive to high levels of the Lon protease.

Construction and characterization of rcsA recombinantplasmids. The 6.3-kb fragment containing the rcsA gene wasinitially subcloned into the BamHI site of pBR325 to gener-ate plasmid pATC352 (Fig. 2). C600 (Ion') transformants ofthis plasmid were extremely mucoid. When pATC352 wasintroduced into a Ion cps::lac strain in the presence orabsence of an rcsA chromosomal mutation, the levels of,-galactosidase increased dramatically (data not shown).The plasmid was found to be very unstable and was lost at ahigh frequency in media without ampicillin.A 4.5-kb PstI fragment containing 2.4 kb of the original

insert was subcloned into pBR322 to generate pATC395(Fig. 2). This 2.4-kb BamHI-PstI piece included the region ofrcsA::ATnJO insertions. The plasmid was found to be signif-icantly more stable than pATC352 in the absence of antibi-otic selection and fully complemented rcsA chromosomalmutations. While pATC352 had complementedflaP andflaRmutants, pATC395 failed to do so. This confirmed that rcsAwas entirely within the BamHI-PstI fragment and was sep-

TABLE 4. Effect of a lon+ plasmid on capsule synthesisa

Strain ,B-Galactosidase unitsbgenotype plon+ plonA510

lon-100 0.3 201lon+' 0.01 2.1lon+ rcsA3 (RcsA*3) 0.2 456lon+ rcsCJ37 4 475

a The cps-1: :lac fusion strains were SG20581 (lon-100), SG20582 (lon+),SG20645 [rcsA3(RcsA*3)], and SG20643 (rcsCJ37).

b Cells were grown in glucose-M56 medium supplemented with CasaminoAcids and ampicillin (50 ,ug/ml) at 32°C.

B BI I

Bam HI-'

pATC 352 N(10.5Kb)

BN P

Pstl

H

pATC 395(8.9Kb)

BH

pBR 325

H

Ip B

pBR 322

FIG. 2. Plasmid constructions. The 6.2-kb BamHI bacterialDNA insert from X SB6 was inserted into pBR325 to make pATC352(rcsA+). A 2.4-kb BamHI-PstI fragment (cross-hatched area) wassubcloned by restricting pBR322 and pATC352 with PstI followedby ligation with T4 DNA ligase to produce pATC395, which wasrcsA+ as well. Restriction sites are as follows: BamHI (B), Hindlll(H), NruI (N), Sall (S), PstI (P), and ClaI (C).

arate from the fla operons. Apparently, DNA segmentswhich cause rapid plasmid loss are also separate from rcsA.

Smaller plasmids containing the rcsA+ gene were gener-ated as shown in Fig. 3. A 2.4-kb BamHI-PstI fragment inthe M13mpl9 derivative MAT200 complemented rcsA mu-tations to increase expression of cps::iac and of capsularpolysaccharide. Deletions of the rcsA+ clone were generatedin this phage by using T4 DNA polymerase (6). Deletionsfrom the EcoRI site that went beyond the EcoRV siteinactivated the gene (Fig. 4). The largest deletion that stillcomplemented an rcsA chromosomal mutation, A37, left a0.86-kb fragment. This piece was subcloned in two orienta-tions into pUC18 and pUC19 to generate pATC118 andpATC119, respectively. The 2.4-kb piece was also subclonedfrom MAT200 into pBR322 to produce pATC400. All ofthese plasmids confer mucoidy to Ion' hosts.

Insertional mutagenesis of plasmids. Insertions linked toand in the rcsA gene cloned in pATC395 were isolated byusing the AKan transposon as indicated in Materials andMethods. rcsA::AKan insertions were identified by the plas-mid's loss of its ability to complement chromosomal rcsAmutations. In plasmids with linked AKan inserts, rcsAremained fully active. Similar AKan insertion mutationswere isolated in pATC400 (Fig. 4).pATC395 and its derivative pATC450, which carries the

rcsA-50::AKan insertion, were used to transform an isogenic

VOL. 169, 1987

on Decem

ber 15, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

986 TORRES-CABASSA AND GOTTESMAN

R B

Am

TMO p

wS(6.4Kb)D

-N /11Bam HI+ Pstl

R

E Rp F

pBR32

EcoRI + Sphl

R

X-NMAT200

R5b (9.7Kb)

Sp

Deletion

m R 37I H C&~~(37) R

pUC19 R lac

RaR M AT27

\C (886 1(b

Ain R5 EcoRI + Hind IIl

HP

pATC119(3.8Kb)

FIG. 3. Construction of pATC400 and pATC119. Plasmid deriv-atives are shown on the left; M13 derivatives are shown on the right.The 2.4-kb BamHI-PstI fragment was subcloned into the polylinkersite of M13mpl9 to produce phage MAT200. The RcsA deletionderivative A37 was generated by the cyclone method in MAT200 (6).A37 was subcloned from phage MAT237, using EcoRI and HindlIl,into pUC19 to generate pATC119. Restriction sites are as follows:BamHI (B), EcoR5 (R5), HindIII (H), Sall (S), NruI (N), ClaI (C),SphI (Sp), PstI (P), EcoRI (R), ScaI (Sc), and XmnI (X).

set of lon+ and Ion cps: :lac fusion strains, and their effect on3-galactosidase expression from the cps: :lac fusion was

determined (Table 5). The level of the enzyme increasedabout 300-fold in pATC395 transformants of lon+ strains. Inlon hosts, which already express high levels of P-galactosidase, pATC395 increases expression fourfold. Thesame plasmid conferred high levels of cps: :lac expression oncells carrying a chromosomal rcsA deletion. The plasmidcomplementation data confirm that rcsA is normally limitingin the cell.While the level of enzyme obtained from lon+ rcsA or lon

rcsA transformants of pATC450 was similar to that of theuntransformed control (Table 5), a 100-fold increase insynthesis was observed in a lon+ rcsA+ transformant withthe same rcsA: :AKan plasmid (cf. columns 1 and 3 of line 1,Table 5). Similar results were observed with several otherpATC395 derivatives carrying different AKan inserts as wellas with pATC400 AKan insert plasmids (data not shown).These results suggest that the high-copy plasmid titrates anegative effector, allowing expression of the chromosomalcopy of rcsA+ and, consequently, escape synthesis ofcps::lac. It is unclear whether such escape synthesis is seenin lon-100 hosts, which already show relatively high levels of3-galactosidase synthesis.Identification of the rcsA gene product. Determination of

the proteins synthesized by plasmids pATC119 andpATC400 and several of its AKan insertion derivatives wasdone in the ion-100 maxicell strain JB3034 (Fig. 5). Onemajor protein of about 27 kilodaltons (kDa) was synthesizedby pATC119 (lane 1) and pATC400 (lane 2). All fourrcsA::AKan insertions in pATC400 failed to produce thispolypeptide (lanes 3 through 6). No unique protein wassynthesized by pATC401 (rcsA160::AKan) (lane 3). In cellsharboring pATC410 (rcsA168: :AKan) a band of about 16 kDawas seen (lane 4). A 23-kDa protein was synthesized bypATC408 (rcsAJ66::AKan) (lane 5), and pATC402(rcsA161: :AKan) produced a 19-kDa protein (lane 6). If thesenew bands are truncated RcsA proteins, then the direction oftranscription of the gene would be from the PstI site to theClal site (Fig. 6). The results are consistent with the identi-fication of the 27-kDa protein as the rcsA gene product.

Stability of RcsA protein. To determine if the RcsA gene

166161 168 160

R5TUrCT

* 0.45 1' 0.25 .- 0.40- *

164 165-IB t t X R5 C \PSp

IF////A I I II -

*0.5- 0.69 0 0.46 0.450o25440-.5L 0.15 / F:: --::-:::::::: A

l JA 103......:: A 9

A 15A 37

FIG. 4. Detailed restriction map of the rcsA clone in pATC400 and deletions generated in phage MAT200. The extent of the deletions areshown by thin lines. Closed circles indicate AKan inserts linked to rcsA+. Plasmids carrying these inserts confer mucoidy upon cps' hostsand a Lac' phenotype to cps::lac rcsA strains. Closed triangles indicate AKan inserts which inactivate the rcsA gene.

x

Pv RI

22

RcsA

i

I

J. BACTERIOL.

on Decem

ber 15, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

CAPSULE SYNTHESIS REGULATION 987

product is affected by lon proteolysis, we assayed its stabil-ity in growing cells, using isogenic lon+ and lon cellscarrying pATC119 transformants. This plasmid, derivedfrom pUC19, is stable even in the absence of an F'iQ episomeand expresses RcsA at high levels. Exponentially growingcells were pulse-labeled with [35S]methionine for 1 min andchased with an excess of unlabeled methionine for up to 30min. Samples were removed at various times during thechase, and the proteins were analyzed by electrophoresisand autoradiography (Fig. 7). The amount of rcsA deter-mined by densitometry of this gel and others, using bothpATC119 and other rcsA+ plasmids, indicate that whereasthe half-life of the RcsA protein is about 5 (±2) min in lon+cells, it is increased to 20 (+5) min in lon strains. Otherproteins synthesized from the same plasmids, the Kanr orthe Bla protein, are stable in both lon+ and lon cells underthese conditions (half-life much greater than 30 min). There-fore, as predicted by our model, RcsA is an unstable proteinin wild-type cells.

DISCUSSION

Mutations at the lon locus of E. coli, which codes for anATP-dependent protease, are pleiotropic and result in over-production of the capsular polysaccharide colanic acid andsensitivity to UV irradiation (3, 4, 12, 13). The UV sensitiv-ity phenotype of lon has been shown to result directly fromthe defect in proteolysis that allows for the accumulation ofSulA, an unstable inhibitor of septation in the cell (26). Weshow here that, as for UV sensitivity, capsule overproduc-tion in lon cells can be explained by the accumulation of anunstable protein, RcsA, which is sensitive to the lonproteolytic system.Two trans-acting positive regulatory loci, rcsA+ and

rcsB+, have been previously identified (11). Introduction ofan additional copy (in X rcsA+ lysogens) or multiple copies(in strains carrying rcsA plasmids) of rcsA+ in a Ion'background increases both ,-galactosidase synthesis fromcps::lacZ fusions and capsule synthesis in cps' strains.These results confirm the observations of Silverman andSimon (31) and Clegg and Koshland (5) that introduction ofthe episome F'1334 and of plasmids carrying the flaA andflaP operons resulted in a mucoid phenotype. rcsA is tightlylinked to fla, at min 43 on the E. coli map (1, 11). Weconclude from our results that rcsA is normally limiting inthe cell. Small increases in rcsB+ gene dosage, on the otherhand, did not increase cps gene expression.Our hypothesis was that the limiting positive regulator

should also be unstable. We have now demonstrated that thercsA protein is a 27-kDa polypeptide which turns over witha half-life of about 5 min in lon+ cells. The RcsA half-lifeincreases to 20 min in lon strains. In contrast, the rcsB gene

TABLE 5. Expression of P-galactosidase in transformantsa

P-Galactosidase unitsbStrain

genotype No pATC395 pATC450plasmid (rcsA+) (rcsA50::AKan)

lon+ rcsA+ 1.5 598 118lon+ ArcsA27 1.1 419 1lon-100 rcsA+ 73 296 34lon-100 ArcsA26 0.1 200 0.8

a The strains used were the recA derivatives SG20587 (lon + rcsA +), JB3035(lon' rcsA), JB3030 (lon-100), and JB3034 (lon-100 rcsA).

b Cells were grown in minimal-M56 Casamino Acids medium with tetracy-cline at 37'C. Data are an average of two samples.

1 2 3 4 5 6

Amp--Rcs A--

24--19-- .:

16-..

FIG. 5. In vivo labeling of proteins expressed by recombinantplasmids in maxicells. Maxicell strain JB3034 containing differentplasmids was labeled as described in Materials and Methods. Theproteins were separated on a 12% sodium dodecyl sulfate-polyacrylamide gel. Lanes: 1, pATC119 (rcsA+); 2, pATC400(rcsA+); 3, pATC401 (rcsA160::AKan); 4, pATC410(rcsA168::AKan); 5, pATC408 (rcsA166::AKan); 6, pATC402(rcsA161::/Kan). Amp, Ampicillin.

product is completely stable in Ion' cells (Brill and Gottes-man, in preparation). Thus the RcsA protein is unstable andits degradation is affected by the lon protease.The lon effect on RcsA stability is similar to that seen for

XN protein, an in vivo substrate for Lon proteolysis (10). Inboth cases, the half-lives of the proteins in Ion hosts areabout 20 min. We believe this residual turnover is due todegradation of these proteins by other cellular proteases. Itseems unlikely that Lon is acting indirectly to affect turnoverof these proteins. In the case of XN protein, the in vitrodegradation of purified N protein by purified lon protease(Maurizi, in press) confirms that the in vivo effects of lonmutations are reflected by the direct degradation of N byLon in vitro.The increase in cps::lac expression caused by rcsA62

(RcsA*62), a dominant allele of rcsA, can be reversed byintroduction of a lon+ multicopy plasmid. This suggests thatthe protein made by this mutant is still sensitive to highlevels of the Lon protease. The exact nature of rcsA (RcsA*)

ABNS Sc X X R5C P NIII I I:1KWb-|

rcsA

p N B H

fla P Q R A

FIG. 6. Fine-structure map of the region of the bacterial chro-mosome carrying the rcsA gene. The map includes restrictioninformation from the X rcsA+ phage and the rcsA plasmids as well asfrom plasmid pCK202, which contains theflaARQP operon (5). TheSaII-HindIII fragment is contained in both pCK202 and X rcsA+.The deletion in plasmid pCK202 (A7) which abolishes its ability toconfer the mucoid phenotype extends from AvaI to HindIII. Thewavy arrows show the direction of transcription of thefla operon (5)and that postulated for the rcsA gene. Restriction sites are asfollows: AvaI (A), BamHI (B), NruI (N), SalI (S), ScaI (Sc), XmnI(X), EcoRV (R5), ClaI (C), PstI (P), and HindIII (H).

VOL. 169, 1987

on Decem

ber 15, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

988 TORRES-CABASSA AND GOTTESMAN

1 3345B

6 1 2 3 4 5 6 7 8a

_ _ _ _ _

RAmpI AmpA

RcsA.-4 -

.t.**k*

:3;:I.-.

FIG. 7. Pulse-chase labeling of RcsA in lon+ and lon cells. Cells were pulse-labeled with [35S]methionine for 1 min and chased with excessunlabeled methionine (2 mM). (A) lon+ (JB3035): lane 1, pUC19; lane 2, pATC119, 1-min pulse; lane 3, 2-min chase; lane 4, 4-min chase; lane5, 6-min chase; lane 6, 8-min chase. (B) lon (JB3034): lane 1, pATC119, 1-min pulse; lanes 2 to 8, chases of 2, 4, 6, 8, 10, 15, and 30 min. Amp,Ampicillin.

is not known. Its gene product could be a protein withincreased but not complete resistance to proteolysis. Evenrelatively small differences in RcsA protein half-life wouldbe expected to increase capsule synthesis significantly.Alternatively, the RcsA* phenotype may reflect higher lev-els of synthesis of the wild-type protein or a more activeRcsA protein.The regulation of capsule synthesis and possibly that of

rcsA may be part of a complex regulatory pathway. Ourresults indicate that rcsA itself may be under the negativecontrol of some effector. rcsC, a negative regulator ofcapsule synthesis (11), may act by controlling the level ofRcsA synthesis. We observed that high-level expression ofcps::lac in rcsC backgrounds is still sensitive to proteolysiswhen a multicopy lon+ plasmid is introduced in the cell(Table 4). In addition, rcsC137 rcsA hosts are defective forcps::lac expression (data not shown). These results areconsistent with models in which rcsC acts before rcsA toregulate capsule synthesis. We are currently examining therole of rcsA, -B, and -C in capsule synthesis regulation.Many bacterial species produce capsules in nature and

under laboratory growth conditions. For example, nonmu-coid mutants of the normally mucoid corn pathogen Erwiniastewartii have been isolated and characterized (D. Coplin,personal communication). Some of these mutants are defec-tive in a positive regulator of capsule synthesis with similar-ities to rcsA (D. Coplin, A. S. Torres-Cabassa, and S.Gottesman, manuscript in preparation). Capsule synthesiscould be a cellular response to some environmental stresscondition; environmental factors such as temperature, me-dium, and growing conditions influence capsule expression.It is possible that a mechanism for capsule regulation couldhave been evolutionarily conserved and that similar path-ways are involved in the synthesis of other complex polysac-charides.

ACKNOWLEDGMENTSWe thank Julie Brill and Patsy Trisler for constructing some of the

strains used in this work and David Coplin for communicating his

unpublished results to us. Sankar Adhya, Michael Maurizi, andOlivier Huisman provided helpful comments on the manuscript. Wealso acknowledge the help of Althea Gaddis in preparing themanuscript for publication.

LITERATURE CITED

1. Bachmann, B. J. 1983. Linkage map of Escherichia coli K-12,edition 7. Microbiol. Rev. 47:180-230.

2. Bukhari, A. I., and D. Zipser. 1973. Mutant of Escherichia coliwith a defect in the degradation of nonsense fragments. Nature(London) New Biol. 243:238-241.

3. Charette, M. F., G. W. Henderson, and A. Markovitz. 1981.ATP hydrolysis-dependent protease activity of the lon (capR)protein of Escherichia coli K-12. Proc. Natl. Acad. Sci. USA78:4728-4732.

4. Chung, C. H., and A. L. Goldberg. 1981. The product of the lon(capR) gene in Escherichia coli is the ATP-dependent protease,protease La. Proc. Natl. Acad. Sci. USA 78:4931-4935.

5. Clegg, D. D., and D. Koshland, Jr. 1985. Identification of abacterial sensing protein and effects of its elevated expression.J. Bacteriol. 162:398-405.

6. Dale, R. M., B. A. McClure, and J. P. Houchins. 1985. A rapidsingle-stranded cloning strategy for producing a sequential se-ries of overlapping clones for use in DNA-sequencing: applica-tion to sequencing the corn mitochondrial 18S rDNA. Plasmid13:31-41.

7. Enquist, L. W., and R. A. Weisberg. 1977. A genetic analysis ofthe att-int-xis region of coliphage lambda. J. Mol. Biol. 111:97-120.

8. Goldberg, A. L. 1972. Degradation of abnormal proteins inEscherichia coli. Proc. Natl. Acad. Sci. USA 69:422-426.

9. Goldberg, A. L. 1985. Role of ATP and protease La, the longene product, in protein breakdown in Escherichia coli, p.340-345. In L. Leive (ed.), Microbiology-1985. AmericanSociety for Microbiology, Washington, D.C.

10. Gottesman, S., M. E. Gottesman, J. E. Shaw, and M. L. Pearson.1981. Protein degradation in E. coli: the lon mutation andbacteriophage lambda N and cIl protein stability. Cell 24:225-233.

11. Gottesman, S., P. Trisler, and A. S. Torres-Cabassa. 1985.Regulation of capsular polysaccharide synthesis in Escherichiacoli K-12: characterization of three regulatory genes. J. Bacte-

A

J. BACTERIOL.

on Decem

ber 15, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

CAPSULE SYNTHESIS REGULATION 989

riol. 162:1111-1119.

12. Gottesman, S., P. Trisler, A. S. Torres-Cabassa, and M. Maurizi.1985. Regulation via proteolysis: the Escherichia coli lon sys-tem, p. 350-354. In L. Leive (ed.), Microbiology-1985. Amer-ican Society for Microbiology, Washington, D.C.

13. Gottesman, S., and D. Zipser. 1978. Deg phenotype of Esche-richia coli lon mutants. J. Bacteriol. 133:844 851.

14. Howard-Flanders, P., E. Simson, and L. Theriot. 1964. A locusthat controls filament formation and sensitivity to radiation inEscherichia coli K-12. Genetics 49:237-246.

15, Kanaya, S., and R. J. Crouch. 1984. The rnh gene is essential forgrowth of Escherichia coli. Proc. Natl. Acad. Sci. USA 81.3447-3451.

16. Kleckner, N., J. Roth, and D. Botstein. 1977. Genetic engineer-ing in vivo using translocatable drug-resistance elements. Newmethods in bacterial genetics. J. Mol. Biol. 116:125-159.

17. Laemmli, U. K. 1970. Cleavage of structural proteins duringassembly of the head of bacteriophage T4. Nature (London)227:680-685.

18. Mackie, G., and D. B. Wilson. 1972. Regulation of the galoperon of Escherichia coli by the capR gene. J. Biol. Chem.247:2973-2978.

19. Maloy, S. R., and W. D. Nunn. 1981. Selection for loss oftetracycline resistance by Escherichia coli. J. Bacteriol. 145:1110-1112.

20. Maniatis, T. E., E. F. Fritsch, and J. Sambrook. 1982. Molecularcloning: a laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.

21. Markovitz, A. 1977. Genetics and regulation of bacterial capsu-

lar polysaccharide synthesis and radiation sensitivity, p.

415-462. In I. Sutherland (ed.), Surface carbohydrates ofthe prokaryotic cell. Academic Press, Inc. (London), Ltd.,London.

22. Markovitz, A., M. M. Liebermann, and N. Rosenbaum. 1967.Derepression of phosphomannose isomerase by regulator gene

mutations involved in capsular polysaccharide synthesis inEscherichia coli K-12. J. Bacteriol. 94:1497-1501.

23. Maurer, R., B. C. Osmond, E. Shekhtman, A. Wong, and D.Botstein. 1984. Functional interchangeability of DNA replica-tion genes in Salmonella typhimurium and Escherichia coli

demonstrated by a general complementation procedure. Genet-ics 108:1-23.

24. Maurizi, M., P. Trisler, and S. Gottesman. 1985. Insertionalmutagenesis of the lon gene in Escherichia coli: lon is dispens-able. J. Bacteriol. 164:1124-1135.

25. Miller, J. H. 1972. Experiments in molecular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

26. Mizusawa, S., and S. Gottesman. 1983. Protein degradation inEscherichia coli: the lon gene controls the stability of sulAprotein. Proc. Natl. Acad. Sci. USA 80:358-362.

27. Mizusawa, S., and D. Ward. 1982. A bacteriophage lambdavector for cloning with BamHI and Sau3A. Gene 20:317-322.

28. Rosner, J. L. 1972. Formnation, induction and curing ofbacteriophage P1 lysogens. Virology 48:677-689.

29. Sancar, A., A. M. Hack, and W. D. Rupp. 1979. Simple methodfor identification of plasmid-encoded proteins. J. Bacteriol. 137:692-693.

30. Silhavy, T. J., M. L. Berman, and L. W. Enquist. 1984.Experiments with gene fusions. Cold Spring Harbor Labora-tory, Cold Spring Harbor, N.Y.

31. Silverman, M., and M. Simon. 1973. Genetic analysis of flagellarmutants in Escherichia coli. J. Bacteriol. 113:105-113.

32. Simon, L. D., M. E. Gottesman, K. Tomczak, and S. Gottesman.1979. Hyperdegradation of proteins in Escherichia coli rhomutants. Proc. Natl. Acad. Sci. USA 76:1623-1627.

33. Trisler, P., and S. Gottesman. 1984. lon transcriptional regula-tion of genes necessary for capsular polysaccharide synthesis inEscherichia coli K-12. J. Bacteriol. 160:184-191.

34. Yamamoto, K. R., B. M. Alberts, R. Benzinger, L. Lawhorne,and G. Treiber. 1970. Rapid bacteriophage sedimentation in thepresence of polyethylene glycol and its application to large-scalevirus preparation. Virology 40:734-744.

35. Way, J. C., M. A. Davis, D. Morisato, D. E. Roberts, and N.Kleckner. 1984. New TnlO derivatives for transposon mutagen-esis and for construction of lacZ operon fusions. Gene 32:369-379.

36. Wyman, A. R., L. B. Wolfe, and D. Botstein. 1985. Propagationof some human DNA sequences in bacteriophage X vectorsrequires mutant Escherichia coli hosts. Proc. Natl. Acad. Sci.USA 82:2880-2884.

VOL. 169, 1987

on Decem

ber 15, 2020 by guesthttp://jb.asm

.org/D

ownloaded from