Architecture of the outer membrane of Escherichia coli III. Protein-Lipopolysaccharide complexes in intramembraneous particles

(1)

0021-9193/78/0134-1089$02.00/0

Architecture of the Outer

Membrane of

Escherichia

coli

III. Protein-Lipopolysaccharide Complexes

in

Intramembraneous

Particles

LOEKVAN ALPHEN,'* ARIE VERKLEIJ,2 JOSE LEUNISSEN-BIJVELT,2 ANDBENLUGTENBERG' Department ofMolecular CellBiology,Microbiology' and Electron

_Microscopy

Sections,andInstitutefor

MolecularBiology,StateUniversityof Utrecht,Padualaan8, Utrecht, The Netherlands

Received forpublication21December1977

Inapreviouspaper(A. Verkleij, L.vanAlphen,J. Bijvelt, and B. Lugtenberg,

Biochim. Biophys. Acta 466:269-282, 1977) wehave hypothesized that particles ontheouterfracture face of theoutermembrane (OM), with corresponding pits

on the inner fracture face of theoutermembrane (OM), consist of

lipopolysac-charide (LPS) aggregates stabilized by divalent cations and that they might contain protein and/or phospholipid. In the present paper the roles of LPS,

cations, and proteins in these OM particlesaredescribedmoreextensively, using astrain that lacks themajoroutermembraneproteins, b,c,and d(b-c-d-), and has a reduction in the number of OM particles of 75%. To study the role of

divalent cations in the formation of OM particles, these b-c-d- cellsweregrown orincubated with

Ca,2+>Mg2+,

orputrescine. Thepresenceof

Ca2'

resulted in the

appearance ofmany OM particles and OM pits.

Mg2"

andputrescine were less

effective than Ca2+. Introduction of these particles was not accompanied by alterations in the relative amounts of LPS and cell envelope proteins.

Ca2"

treatment of a heptoseless derivative of a b- c- d- strain did not result in

morphological changes. Incubation of Ca2+-treatedcellswith ethylenediaminetet-raacetate caused the disappearance of the introduced particles as well as the

release ofmorethan 60% of the cellular LPS. These results strongly

_suppgort

the

hypothesis that LPS is involved in the formation of OM particles and OM pits. The roles of variousoutermembrane proteins in the formation of OM particles

werestudied by comparing thefreeze-fracture morphology of b-c-d- cells with

that of cells whichcontain oneof theoutermembrane proteins b,c,d, and e or

the receptor protein for bacteriophage lambda. The results showed that the

presenceofanyofthese fiveproteins inab-c-d- background resulted inalarge

increase in the number of OM particles and OM pits, indicating that these proteinsare,independent of each other, involved in the formation of OM particles

and OM pits. The simplest explanation for the results is that in wild-type cells eachparticle consists of LPS complexed withsomemolecules ofasingle protein

species, stabilized by either divalent cationsorpolyamines. It is hypothesized that

theoutermembrane of thewild-type cell containsaheterogeneous population of

particles, of which 75% consists of protein b-LPS, protein c-LPS, and protein d-LPSparticles. A function of these particlesasaqueous poresisproposed.

The outer membranes of a number of

gram-negative bacteria have been investigated by freeze-fractureelectronmicroscopy.Inwild-type cells of Escherichia coli (1, 16, 21, 30, 31),

Sal-monella typhimurium (17, 24), Pseudomonas

aeruginosa (6-8), and Acinetobacter (23), the concave or outerfracture face of the outer mem-brane (OM)iscovered withparticles4to10 nm in diameter. Inpreviouspaperswehave shown that onthecorresponding convex orinner frac-tion face (OM) ofE. coli K-12 many pits are

visible, which are probably complementary to the particles (30, 31). Particles

complementary

topitshave beeninterpreted in freeze-fracture

terms asmicelle-likestructures(32).

Studies withmutants ofE. coliandS. typhi-murium which are lacking one or more major

outermembraneproteins showed that the num-ber ofOM particlesis reduced (upto75%) (17,

21, 30, 31). Moreover, some mutants with a shortenedlipopolysaccharide(LPS)sugar chain showed the same phenomenon (24, 30, 31). In

addition, we observed in E. coli K-12 that a

reduction of the number ofOM pits coincides with a reduction in OMparticles (30).

Treatment ofwild-type cells of E. coli K-12

(2)

1090 VAN ALPHEN ET AL.

with ethylenediaminetetraacetate (EDTA)

caused (i) the release of half of the cellular LPS butnorelease of protein and (ii) areduction of

the number of OM particles and OM pits of

about 50% (30). A reduction of the number of OMparticles coinciding with release of LPSwas

also observed in P. aeruginosa (6, 7). However, in the latter species part of the cell envelope

protein is extracted together with LPS (6, 19,

24).

From the complementarityaspectof theOM

particles and OM pits and from the observed reduction in the number of particles inmutants

and inwild-type cells treated with EDTA, itwas

hypothesized that in E. coli K-12 the OM

par-ticles contain aggregates of LPS stabilized by divalent cations and possibly containing protein and/orphospholipid (30).

Inthepresentpaperthe roles ofLPS,cations,

and proteins in the OM particlesaredescribed moreextensively. Strain P6922dIwaschosen for

these experiments, because it lacks majorouter

membrane proteins b,c,and d(b-c-d- mutant),

and it hasareduction inparticle density of 75%

(9, 21, 30).Therefore, it allows the introduction of bothparticlesandpitseitherby thepresence

of divalent cationsorthepolyamine putrescine orby the "insertion"of defined proteinsin the outer membrane. The results strongly support

the hypothesis mentioned previously and show that thehypothesiscanbeextended in that the

majority of the particles contain protein-LPS

aggregates.

MATERIALS AND METHODS

Strains and growth conditions. All strainsare derivativesof E. coli K-12.Originsand relevant char-acteristics are listed in Table 1. Strain P692tut2dI,

subsequentlyabbreviatedasstrainP6922dI,ismissing majorouter membraneproteins b,c,andd,whichare

compensatedforbyincreased amounts of(wildtype)

LPS andofphospholipids (28).

Unless otherwiseindicated,bacteriaweregrownin standard tris(hydroxymethyl)aminomethane

(Tris)-based medium (22, 30), supplementedwith 0.5%

glu-cose-0.2% CasaminoAcids-20,ug of tryptophan, uracil, thymine, adenine,and guanineperml-5,ug thiamine

perml.CaCl2orMgCl2waspresent inafinal concen-tration of 0.2 mM. Insome casesthe concentration of

CaCl2 or MgCI2 wasincreased to 2 mM, or 10mM

putrescinewasaddedtothestandard medium. Cells

weregrowntothe latelogarithmic phaseat37°C. The

cellswereharvestedat4°C,washed with0.12 M

Tris-hydrochloride (pH 8.0),and used for further incuba-tion,chemicalanalyses,orfreeze-fracturing.

Incubations andanalytical procedures.To ex-amine the influenceofincubationwithdivalent cations or EDTA on the freeze-fracture morphology of the

cells, the bacteria were washed with 0.12 M

Tris-hydrochloride (pH 8.0), resuspended in1/20 volume

of the samebuffer, and further treated either for 15 minat37°Cin thepresenceof 50 mMCaCl2orMgC12 orfor 5minat37°Cwith 5 mM EDTAasdescribed

byLeiveetal. (11).Cellswereharvestedby

centrifu-gation for 15 min at4°C at 3,000 xgand used for

freeze-fracturing. Released LPS was quantitatively isolated from the supernatant solutionbytheaddition of CaCl2 and acetone (in final concentrations of 20 mM and 70%, respectively) at4°C,followed by

cen-trifugation. The pellet was washed once with 70% acetone.

Cellenvelopes wereisolated quantitatively as de-scribed before (13) except that EDTA wasomitted.

Proteinwas determined accordingto themethod of

Lowryetal. (12). Total cell proteinwasdetermined after ultrasonictreatmentof thecells.The amountof cellenvelope proteinwasexpressedper milligramof total cell protein. 3-Deoxy-D-mannooctulosonic acid wasdeterminedbythe thiobarbituric acid method of Weissbach and Hurwitz (34) asmodified byOsborn

(18) and corrected for the fact thatonlytwo ofthe three3-deoxy-D-mannooctulosonicacidmoleculesare measured(5).Theamount ofLPSwascalculatedby usingourprevious data,whichshowed that LPS con-tains 11% 3-deoxy-D-mannooctulosonic acid (wt/wt).

The protein patterns of the various fractions were

identified by sodium dodecyl sulfate-polyacrylamide gel electrophoresisasdescribed before(13).

Freeze-fracture electron microscopy. To the cellsuspension25%glycerolin0.9% NaClwasadded

asacryoprotectant. The fracture faceswerenot influ-encedbythepresenceofglycerol.Thesampleswere

quenched from4°C inamixture ofliquid and solid

nitrogenand fractured inaDenton freeze-etch

appa-TABLE 1. Strains and relevant characteristics

Strain Outermembraneprotein abnormalities Source,ref.

AB1133 Wild type P.Reeves andref.28

P692tut2dI Protein b-c-d-derivativeof AB1133 U.Henning and ref. 9 and 28

CE1097 Heptoselessderivative of P6922dI Thispaper

CE1096 Spontaneous protein b-proficientrevertant of Thispaper

P6922dI

CE1150 Spontaneous protein c-proficient revertant of Thispaper

P6922dI

CE1071 d+ exconjugantofP6922dI Ref.28

CE1145 b-c-d-derivativeofPC0479 C. Verhoef and W. van Alphen CE1146" b-c-d-e+derivative ofPC0479 C.Verhoef and W. van Alphen

" Protein eis amajor peptidoglycan-associated protein withan apparentmolecular weight of 40,000. It is probablyidentical toproteinIcdescribed by Henningetal.(10).

(3)

PARTICLES IN THE OUTER MEMBRANE OF E. COLI 1091 ratus as descibed before (33). Electron micrographs

weremade withaSiemensElmiskop1A. Theparticle

density was estimatedasdescribed (30). RESULTS

Influence of divalent cations and putres-cine on the density of OM particles in a mutantlacking proteins b, c, and d. In our

previouspaper we providedstrong evidence for the presence of LPS in OM particles (30). We reasonedthatmutantstrain

P6922dI,

which has astrongreductioninthe number ofOM particles

(21, 30) despiteits increasedLPS content (28), couldbeusedtoseewhetherLPS,intheabsence of threemajor

proteins,

couldbe forcedtoform OMparticles by

incubating

the cellswith

diva-lentcations orputrescine. The

complex

growth

media used in ourprevious studies (30, 31) did not allow us to study theinfluence ofdivalent

cations, since a precipitate of Ca salts was

formed. For thepresentstudycells were grown in standardTris-based medium. Growth inthis

mediumresultedinessentiallythe same

freeze-fracturemorphology asdescribedbeforein that

the OM ofthe

wild-type strain, AB1133,

is al-most

completely

covered with

particles (Fig.

1)

(30),whereasin the b- c-d- mutantP6922dIa

reduction ofabout 75% in the number of OM particlesand OM pits

(Fig.

2) (21, 30) was

ob-served.

Supplementation ofthe standard Tris-based

medium with 2 mM

CaCl2

resulted in

large

changes in the

morphology

of the outer

mem-brane of strain P6922dI. The

GM

ofthese

Ca-grown cellswas

densely

coveredwith

particles,

and numerous pits on the OM were seen (Fig.

3).Both theparticles and the pits introducedby

Ca2"

werestable in thattheyresisted washing of the cells with buffer without CaCl2. Replacing

Ca2"

in the growth medium with Mg2+ or pu-trescine resulted in a similar effect on the num-ber ofparticles andpits, althoughthe increase

was lower (from 25to 40%). Each of the three additions to the growth medium resulted in a

particle density that differed from cell to cell.

The mentioned results are averages of many

individualcells. Such aheterogeneityin the cell

populationhasbeenreported before for all mu-tants lacking protein d (30). Chemical analysis

showed that the morphological differences

caused by growth in standard Tris-based me-dium with various supplementations could be related neither tosignificant alterations in the

amount ofLPS or protein of the cell envelope permilligramof cellprotein (Table2) nor tothe

pattern of the cell envelope proteins on poly-acrylamide gels (not shown). In the wild-type

strain AB1133, no influence of the mentioned supplementations on the morphology and chem-icalcomposition ofthe outermembranescould bedetected.

The morphological effects caused by growth in the presence of 2 mM CaCl2 or MgCl2 could also beobserved afterincubationofthe mutant

cellsfor 15min in buffersupplementedwith 50 mM CaCl2 or MgCl2. These effects were also observed in the presence of chloramphenicol

(100 ,ug/ml), showing that de novoprotein syn-thesis is not required for an increase of the number of particles. Theparticlesand pits

ob-.K.;.:j4$9:..t.

...

.4

-.

I

N

4 4

'

v*

-

>

''' -

Ck

W'''J-2 Ct

'S-

.,-.

4 . -~~~~~~~~~~~'

%~~~~1

'

'~~~

-04~~~~~~>

FIG. 1. Outerfracture faces ofthe outer membrane (OM) andcytoplasmicmembrane

(OkI)

ofE. coli K- 12 strainAB1133grown in standardTris-basedmedium, quenchedfrom4°C. Bar=200nm.Arrow indicates directionofshadowing

(4)

1092 VAN ALPHEN ET AL. w.: .I . h----. .N ma,,l ..' E m'1 v*

A

*fi;,L1 C ,.'.V' N 1.~N 4V.. ..4tE ,.4 .I _.

2^

A

_~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.TsLC

FIG. 2. Fracture

_facesg9f

the outermembraneofstrainP6922dI,grown in standard Tris-based medium,

quenched from 4°C. (A)CM;(B)6Xt. Bar=200 nm.Arrowindicates directionofshadowing

served after incubation with

Ca2"

or

_Mg2e

were

lessstable, sincetheydisappeared afterwashing of thecells with buffer without divalent cations. In summary, in the outer membrane of strain P6922dIparticlesandcorresponding pits can be introduced either by growth in the presence of

orbyincubation with divalent cations without detectable alteration in the biochemical

com-positionof thecell envelope. It islikelythat in theabsence ofthe majorouter membrane pro-teinsb,c, andd,theseparticlesconsist of aggre-gates of LPSstabilized bydivalent cations. It is

likely thatthecoreregionofLPS (possiblythe

negatively charged phosphate groups) is in-volved in the aggregation process, since an

in-crease in the number of the OMparticles and

OMpits,eitherby growthorbyincubation with

Ca",

couldnotbeobserved in strainCE1097, a

heptoselessmutantof strainP6922dI.

The roles ofLPS and divalent cations in the

OM

particles

introduced by divalent cations

were further studiedby treating the cellswith

EDTA, whichis knowntoresultin the release of about half of thecellularLPS (11, 30). After

J. BACTERIOL. .rr AM. .I. I I,s b I

(5)

'.4 4.'

# .I.

~~~~~~~~'~~ ~ ~ ~ ~ A

_ ,, - - , .. . 4 ,_, ,

FIG. 3. FracturefacesofstrainP6922dIgrown in

Ca-supplemented

medium,

quenched from

4°C.

Bar = 200nm.Arrowindicates directionof

shadowing.

(6)

TABLE 2. Influenceofgrowth medium and EDTA treatment on the relative amounts of LPS and cell envelopeprotein"

Cell envelope componentper mg ofcell

StrainStrain Addition

_~medium

togrowth Outer_{protein defects}membrane protein LPS

_{EDTAf treatment)}

released

after LPS (nmol) Protein' (mg) (% oftotal LPS)

AB1133 None None 6.4 0.25 35

AB1133 CaCl2 None 6.8 0.27(0.26)" 43

P6922dI None b- c-d- 10.8-12.0 0.19 63 P6922dI CaCl2 b- c- d- 10.6-12.8 0.19(0.20)" 63 P6922dI MgCl2 b-c-d- 13.3 0.20 67 P6922dI Putrescine b-c-d- 10.7 0.24 ND" P6922dI Maltose b-c-d- 7.8 0.22 ND CE1096 None b+c-d- 8.1 0.20-0.25 ND CE1150 None b-c+d- 8.1 0.19 ND

"Cellsweregrown inTris-based mediumsupplemented,ifindicated, with2mMCaCl2,2mMMgCl2,10mM putrescine, or0.5% maltoseand washed with0.12 MTris-hydrochloride, pH8.0.Insome casesthe cellswere treated with5 mM EDTAfor5minat37°C. Cellenvelopes wereisolated,and theamountsofLPS and cell envelopeproteinweredetermined.

'Theamountsof LPSweredeterminedinthe untreated cell envelopes,inthereleasedmaterial, andinthe cells afterEDTAtreatment. Thesumofthe lattertwovalueswasalways equaltothefirst.

' Theamountof cell envelope proteinpermilligramoftotal cellprotein isquitehigh,comparedwith the

valuesobtained before (29), since EDTAwasomittedduringtheisolation of thecellenvelopes.Inthiscasealso

somecytoplasmicandperiplasmicproteinisobtainedinthe membrane fraction. d The numbersinparenthesesrepresent thevalues after EDTAtreatmentof thecells.

'ND, Notdetermined.

growthorincubation with divalentcations,more

than 60% of the cellularLPS of strain P6922dI wasreleased bythe EDTAtreatment(Table 2).

The amount of cell envelope protein did not

change (Table 2), and the

protein

patterns of cell envelopes isolated before and after EDTA

treatment were identical. EDTA treatment

caused the disappearance ofthe OM particles and the OM pits which were introduced by

divalent cations (Fig. 4). The same

freeze-frac-ture morphology was observed after EDTA

treatmentofcellsgrown in standardTris-based

medium (not shown). This treatment did not

result inasignificant reductionin the number of particles. In the latter case EDTA treatment

resulted in the release of the same fraction of thecellularLPSas wasreleasedfrom cells grown inthe presenceof divalent cations (Table 2)and also in the releaseof 3% of thecellular protein.

Thegelpatterns of theprotein releaseddidnot

resemble those of cell envelopes. Since most

bands on the gel alsooccurin the cytoplasmic fraction,thereleasedproteinwasduetolysisof

asmall fraction ofcells,mostlikely.Theresults described so far with strain P6922dI are sche-matically summarized in Fig. 5.

Role ofproteins in the formation of OM

particles.Reportsfrom severallaboratories(21, 24,30) suggest that the outer membrane proteins ofEnterobacteriaceaeplayarole in the forma-tion of the OM particles, although the role of the proteinswas not understoodin detail. The role of different outer membrane proteins in

particleformationcanbeillustrated by studying the effect of inserting a

single

protein species

into the outermembrane of the b- c-d- strain

P6922dI, sincethis strain hasareductionof75%

in particle density. Compared with strain

P6922dI, each of the

following

fivestrains

con-tainedone extramajoroutermembrane protein: (i) the spontaneousprotein b-proficient partial

revertantof strain P6922dI (strain

CE1096);

(ii)

the spontaneousprotein c-proficient partial

re-vertant of strain P6922dI (strain CE1150); (iii) the protein d-proficient exconjugant CE1071; (iv)strainP6922dI,in which the receptorprotein

ofphagelambda was inducedby replacingthe

glucoseofthegrowth mediumwithmaltose; (v) the b- c- d- e+ strain CE1146 (as protein e is

notpresent in the AB1133 background, the

re-sultsofstrain CE1146werecompared with the

b-c-d- parent strainCE1145).

Acomparisonofthefreeze-fracture morphol-ogy of strain P6922dI with those ofthe five other strains isschematicallyshown inFig.6.Starting

froma b- c-d- mutant,wefind that in all five

casesthe insertion of each single protein species is accompaniedbyastro_gincrease of the num-ber ofparticles on the OM (from 25% to 50 to

80%) and of the number of pits on the OM. For

instance,the effect ofthe insertion of the phage lambda receptor protein in strainP6922dI can be observedbycomparison of Fig.2and 7. The insertion of single proteins, all of which were present in highamounts (Fig. 8), resulted in a reduction of about 20% in the amount of LPS

(7)

PARTICLES IN THE OUTER MEMBRANE OF E. COLI 1095

FIG. 4. Fracture faces of strainP6922dI grown in Ca-supplemented medium, after treatment of the cells with EDTA. (A)Of; (B)OM.Bar=200 nm. Arrow indicates direction of shadowing.

permilligramof cell protein (Table 2) (28). The

resultsclearly show that each of the tested outer

membrane protein species is involved in the

formationof OM particles and OM pits.

DISCUSSION

Involvement of LPS and proteins in OM particles. The resultsobtainedaftergrowingor

incubating cells of the b- c- d- mutant strain P6922dI in the presence ofdivalentcationsand after treating these cells subsequently with

EDTA, summarized in Fig. 5, strongly suggest that the introducedOM particlesconsistofLPS aggregatesstabilized

by

Ca2+.

The observation

that

Ca2+

introduces both particles and pits in b- c- d- cells strongly suggests that LPS is

responsible for the pits in theOM, althoughit

cannot be excluded that otherproteinsmay be marshalled intoparticles.Thisobservation sup-ports the hypothesis (30) that in non-Ca

2-treatedcellspitscomplementarytoparticlesare

caused by micelle-like structures of LPS. The OM particles introduced

by

Ca2+ probably

do

not occur in substantial amounts in wild-type cells, since in these cells most of the LPS is

expectedtobe

complexed

withprotein.

Theresultsobtained after the insertion ofonly

oneof theprotein speciesb,c,d,ande orof the receptor protein ofbacteriophage lambda in a

b- c- d- background show that this insertion results in an outer membrane which contains many more OM particles with corresponding

OM pits (Fig. 6 and 7). The

simplest

interpre-tation of these results is that these additional

particlesrepresent protein-LPS aggregates and that the

protein

component of the

complex

is

only one protein

species.

With respect to the roles of the individualproteins, this

interpreta-tion extendsour

hypothesis

(30).

Natureof OMparticlesinwild-typecells.

According to our

hypothesis,

the outer

mem-brane of

wild-type

cells containsa

heterogeneous

populationofparticles,

consisting

of about 75% ofamixture of

b-LPS,

c-LPS,

and d-LPS

parti-cles. Previous freeze-fracture studies with

mu-tants deficient in one, two, or all three of the 134,

(8)

WILD TYPE compensated for byincreased amountsofoneor

twoof the

remaining proteins,

but such a

com-d

\

_\pensation

/

_d

wasnotobservedwhen all

three,

pro-EDTA teins

b,

c, and

d,

are

lacking

_{(3, 9, 20, 21, 29).}

* w With our

present

hypothesis,

the mentioned

compensation

effects

_{by proteins}

are

interpreted

D in terms of

compensation

by

OM

particles.

Tak-ing ab- c- strain as anexample, it isknown that

Mutation

such a straincontains increased amounts of

pro-tein d (9, 20, 21,29), whereas itsouter membrane

b,cd- less mutant contains anormal number of OM particles (21,

B 30).

According

to our

hypothesis,

the latter

re-sult was obtained because the missing b-LPS

and c-LPS particles are compensated for by increasedamounts of

_{(morphologically similar)}

.0

~

/ \ d-LPS

particles.

The involvement both of LPS and of the proteins b, c, d, and eand the lambda receptor +Ca++ n_oc\I-ca+++ca++_{} D}3 EDTA protein in particle formation predicts that these_{proteins interact} with LPS. Interactions ofthe

4-

proteins

b,

c

_(4;

unpublished

results cited in

FIG. 5. Schematic illustrationof the effects of

di-valent cations and EDTAon the freeze-fracture mor- //WILD,TYPE\

phology of the outer membrane. dM particles with / * \

corresponding

6-M

pits are indicated by black /

spheres.

The

O6f

_ofparent

strain AB1133 is covered _// _*\ with particles(A).The numberof Ok particles in the

b- c- d- strainP6922dI is reduced to 25% (B). Sup- @0

plementationofthegrowthmediumofstrainP6922dI

withCa" results in an increasein theparticle density

_b-/d-

* * to 75 to80%o (C).Incubation with Ca"+ has the same

effect. The particles introducedby growth in the pres- @00

enceof

Ca2"

orbyincubation with

Ca2"

areremoved

/@

by EDTA treatment, resulting in the morphology

observed for strain P6922dIgrown in standard Tris- b,c,d+ btc-,d-based medium (B). EDTAtreatment of the latter cells

hardly influences the morphology (not shown). In \

contrast, EDTA treatmentofwild-type cells results in /

the disappearanceof about half of theOiIparticles

b-,c-,d-and

6(5

pits (D). EDTA treatment releases more *

thanhalf of the cellular LPS (see Table2).

proteins b, c, and d showed that the number of e+b-,c-d-

?Arec+b-,c-,d-particles instrainsmissing a single protein or in FIG. 6. Schematic representation of the effect of b- c- mutants was almost normal, whereas the thepresence ofproteins b, c, d, ande and thephage number of particles in a b- c- d- strain was lambda receptorprotein on thefreeze-fracture

mor-strongly decreased (21, 30, 31). Although the phology ofthe outer membrane. Symbols are as in latter resultsuggesteda roleoftheseoutermem- Fig. 5. The b-

c-

d- strain P6922dI, a derivative of brane proteins, the results obtained with the wild-typestrain

AB1133,

has an extreme reduction in

single-protein mutants and with, c strains the

density of

Of

particles

and

corresponding

6-I

did..nogtrhatfoeltlolidtsootahcinoi,,s

stidpe(33tlaertrtntpits.

Thisfigureillustratesthat,starting fromtheb,

didnot allow solidconclusions with respect to

_6c

d-strain, theacquisition ofprotein b, c, d, or e or

the roles of theindividual proteins (21, 30, 31). the lambda receptorprotein results in a strong to

However, it hasbeen reported that the lack of extensMve increase in the numbers of OkI particles

one or two of the proteins b, c, and d is often and OMpits.

(9)

PARTICLES IN THE OUTER MEMBRANE OF E. COLI 1097

uo-wW.

O M. --e.gr_K:'""~

7

-~Om

s

FIG. 7. OM of strain P6922dI after growth in Tris-based medium,supplementedwith maltosetoinducethe

phage A receptor. Bar=200 nm.Arrow indicates direction of shadowing.

;

t_

~~-Xrec

a

b

cY

-0

C

CD

CD cu

-dr) (N 0) U) Cq

0)

0 -m -o

_o-

a-m _(D LuLU w uj

(De

E

<

.L u W XL CS 0

FIG. 8. Polyacrylamide gel pattern of the cell en-velope proteins of themutants in Fig. 6. Only the relevant part of the gel is shown. Strain AB1133 is wild type with respecttoouter membrane

composi-tion;strain P6922dI lacks proteins b,c,andd.Strain CE1096 has thephenotype b+c-d-, strain CE1150 is b- c+ dT,strainCE1071 is b-c- d, strain CE1146 is b-c- d- e+, and in strain P6922dI grownonmaltose the receptorprotein for phage lambda has been

in-duced.

reference 35), and d (4, 27) with LPS in vitro haveindeedbeenreported.

Inwild-type cellsabout half of the OM parti-cles are stabilized by divalent cations (30). The

abilityofputrescineto introduce OMparticles

in strain P6922dI suggests thatputrescine and

possiblyotherpolyaminescould be involved in the stabilization of the other half of the OM

particles.

From studies involving the extraction of P.

aeruejnosa

withEDTA it was concluded that theOMparticlesinthisspecies arealso

protein-LPS complexes (6). This conclusion was based

on the observations (i) that EDTA treatment

causes areduction inthe numberofOM parti-cles and(ii)that thereleased materialconsisted of acomplex containing60%protein (incontrast to E. coli K-12), 30% LPS, and 10% loosely bound lipid. The protein fraction consisted mainly of two major outer membrane proteins (6, 19, 25).

Proteinsb, c, d, and eand thephage lambda receptorprotein, which areinvolvedinparticle formation (Fig.6), arereportedtobeinvolvedin thefunctioning of hydrophilicpores with differ-ent

specificities

(2, 14, 15, 26, 29a, 29b).

There-fore, it is tempting to speculate that the OM

particlesaremorphological reflections of

hydro-philic pores throughthe outer membrane ofE.

coliin whichthespecificityof the pore function isdeterminedbythe

protein.

Thepitsobserved on thesurface ofE.coli cells(1) could represent

the entrances of these pores.

ACKNOWLEDGMENTS

Wethank C. Verhoef andW. vanAlphenfortheirgiftof strains and P. G. de HaanandP. H. J. T.Ververgaert for criticalreadingof themanuscript.

LITERATURE CITED

1. Bayer,M.E., andL.Leive.1977. Effect of ethylenedia-minetetraacetateupon thesurface ofEscherichiacoli. J.Bacteriol. 130:1364-1381.

2. Braun, V.,and H. J.Krieger-Brauer.1977. Interrela-tionshipofthephageA receptorproteinand maltose transport in mutants ofEscherichiacoli K-12. Biochim. Biophys.Acta469:89-98.

VOL. 134, 1978

(10)

3. Chai, T.-J., andJ.Foulds.1977.Escherichia coli K-12 tolF mutants: alterations in protein composition of the outermembrane. J. Bacteriol. 130:781-786.

4. Datta, D.B., B. Arden, and U. Henning. 1977.Major proteins of the Escherichia coli outer cell envelope membrane as bacteriophage receptors. J. Bacteriol. 131:821-829.

5. Droge, W., V. Lehman, 0. Luderitz, and 0. Westphal. 1970.Structural investigationsonthe 2-keto-3-deoxy-octonateregion of lipopolysaccharides. Eur. J. Biochem. 14:175-184.

6. Gilleland, H. E., Jr.1977.Ultrastructuralalterationof the outer membrane of Pseudomonasaeruginosa as-sociatedwith resistance topolymixinB and toEDTA, p. 145-150. In D. Schlessinger (ed.), Micro-biology-1977. American Society for Microbiology, Washington, D.C.

7. Gilleland,H.E., Jr.,and R. G. E.Murray. 1976. Ultra-structural studyofpolymyxin-resistantisolates of Pseu-domonasaeruginosa. J.Bacteriol. 125:267-281. 8. Gilleland,H.E., Jr.,J.D. Stinnett, and R. G.Eagon.

1974.Ultrastructural and chemical alteration of the cell envelope of Pseudomonas aeruginosa,asssociated with resistance to ethylenediaminetetraacetate resulting fromgrowthin aMg2"-deficientmedium.J.Bacteriol. 117:302-311.

9. Henning, U., and I.Haller. 1975. Mutants of Esche-richiacoli K-12 lacking all 'major' proteins of theouter cellenvelope membrane. FEBS Lett.55:161-164. 10.Henning, U., W.Schmidmayr, and I. Hindennach.

1977.Major proteins of theouter cellenvelope mem-brane ofEscherichia coli K-12: multiple species of protein I. Mol. Gen. Genet.152:293-298.

11.Leive, L.,V. K.Shovlin,and S. E.Mergelhagen.1968. Physical, chemical and immunological propertiesof li-popolysaccharide released fron Escherichia coli by ethylenediaminetetraacetate. J. Biol. Chem. 243: 6384-6391.

12. Lowry, 0.H.,N. J.Rosebrough,A. L.Farr,and R. J. Randall. 1951. Protein measurement with the Folin phenol reagent. J. Biol.Chem. 193:265-275.

13. Lugtenberg, B.,J.Meijers,P.vanderHoek,R. Pe-ters, and L.vanAlphen.1975.Electrophoretic reso-lution of the'majoroutermembraneprotein'of Esch-erichia coli K-12 into four bands. FEBS Lett. 58:254-258.

14. Lutkenhaus,J. F.1977. Role of amajoroutermembrane proteininEscherichia coli. J.Bacteriol. 131:631-637. 15.Manning,P.A.,A.P.Pugsley,andP.Reeves.1977. Defectivegrowth functions inmutantsof Escherichia coli K-12lackingamajoroutermembraneprotein. J. Mol. Biol.116:285-300.

16.Nanninga, N.1970.Ultrastructureof thecell envelope of Escherichia coli B afterfreeze-etching. J. Bacteriol. 101:297-303.

17. Nurminen, M., K. Lounatmaa, M. Sarvas, P. H. Miikela,and T. Nakae.1976.Bacteriophage-resistant mutants ofSalmonella typhimurium deficient in two major outer membrane proteins. J. Bacteriol. 127:941-955.

18.Osborn,M. J. 1963.Studies onthe gram-negative cell wall. I. Evidence forthe role of2-keto-3-deoxyoctonate inthelipopolysaccharide of Salmonella typhimurium. Proc. Natl. Acad. Sci.U.S.A. 50:499-506.

19.Rogers, S. W., H.E.Gilleland, Jr., and R. G. Eagon. 1968. Characterization of aprotein-lipopolysaccharide complex released from, cell walls of P.seudomonas aeruginosaby ethylenediaminetetraacetic acid. Can. J. Microbiol. 15:743-748.

20. Schnaitman, C.A. 1974. Outer membraneproteins of Escherichia coli. IV.Differences in outermembrane proteinsdue to strainand cultural differences. J. Bac-teriol. 118:454-464.

21. Schweizer, M., H. Schwartz, I. Sonntag, and U. Hen-ning. 1976. Mutational change of membrane architec-ture.MutantsofEscherichia coli K-12 missing major proteins of the outer cell envelope membrane. Biochim. Biophys. Acta 448:474-491.

22. Sinsheimer, R. L., B. Starman, C. Nagler, and S. Gurthrie.1962.The processofinfection with bacterio-phage 4X174.I. Evidencefor a "replicative form." J. Mol.Biol. 4:142-160.

23. Sleytr, U.B.,M. J.Thornley,and A. M.Glauert.1974. Location of the fracture faces within the cell envelope ofAcinetobacter species strain MJT/F5/5. J. Bacteriol. 118:693-707.

24. Smit, J.,Y.Kamio, and H. Nikaido.1975.Outer mem-brane ofSalmonellatyphimurium: chemical analysis andfreeze-fractu're studies withlipopolysaccharide mu-tants. J. Bacteriol. 124:942-958.

25. Stinnett, J.D., H. E.Gilleland, Jr.,andR.G.Eagon. 1973. Proteins releasedfromcellenvelopes of Pseudo-monasaeruginosaonexposureto ethylenediaminetet-raacetate: comparison with dimethylformamide-ex-tractableproteins. J. Bacteriol.114:399-407. 26. Szmelcman,S., and M.Hofnung.1975.Maltose

trans-port in Escherichia coliK-12:involvementof the bac-teriophage lambda receptor. J. Bacteriol.124:112-118. 27. vanAlphen, L., L.Havekes,andB.Lugtenberg.1977. Major outer membrane protein dofEscherichia coli K-12.Purification and in vitro activity ofbacteriophage K3 and F-pilus mediated conjugation. FEBS Lett. 75:285-290.

28. vanAlphen, L.,B.Lugtenberg,R.vanBoxtel,and K.Verhoef.1977.Architecture ofthe outermembrane ofEscherichia coli K-12. I. Action ofphospholipasesA2 and C onwild type strains andoutermembrane mu-tants. Biochim.Biophys.Acta466:257-268.

29. vanAlphen, W., B.Lugtenberg,andW.Berendsen. 1976.Heptose-deficientmutantsofEscherichia coli K-12 deficient in up to three major outer membrane proteins. Mol. Gen. Genet. 147:263-269.

29a. vanAlphen,W., R.vanBoxtel,N.vanSelm,and B. Lugtenberg. 1978. Pores in the outermembrane of Escherichia coliK-12.Involvementofproteinsb andc in the permeation of cephaloridine and ampicillin. FEMS Lett. 3:103-106.

29b. vanAlphen,W.,N.vanSelm,and B.Lugtenberg. 1978. Pores in theoutermembraneofEscherichiacoli K-12.Involvement ofproteinsb andcin thefunctioning of poresfornucleotides. Mol. Gen. Genet. 159:75-83. 30. Verkleij, A., L. van Alphen, J. Bijvelt, and B. Lug-tenberg. 1977.Architecture of the outer membrane of Escherichia coli K-12.II.Freezefracturemorphology ofwild type andmutantstrains. Biochim.Biophys.Acta 466:269-282.

31. Verkleij,A. J., E. J. J. Lugtenberg, and P. H. J. T. Ververgaert. 1976. Freeze etch morphology of outer membrane mutants of Escherichia coli K-12. Biochim. Biophys. Acta 426:581-586.

32. Verkleij,A.J.,and P. H.J.T.Ververgaert. 1975. The architecture of biological and artificial membranes as visualized by freeze etching. Annu. Rev. Phys. Chem. 26:101-122.

33. Ververgaert,P.H. J. T., P. F. Elbers, A. J. Luitingh, and H. J. van de Berg. 1972. Surface patterns in freeze-fracturedliposomes. Cytobiologie 6:86-96. 34. Weissbach, A., and J. Hurwitz. 1959. Theformation

of 2-keto-3-deoxyheptonic acid in extracts of Esche-richia coli B. I. Identification. J. Biol. Chem. 234:705-709.

35. Yu, F.,andS.Mizushima. 1977.Stimulation by lipo-polysaccharide of the binding of outer membrane pro-teins 0-8and0-9to thepeptidoglycan layer of Esche-richia coli K-12. Biochem. Biophys. Res. Commun. 74:1397-1403.