Characterization - Europe PMC

Vol. 35, No. 3

INFECTION AND IMMUNITY, Mar. 1982, p. 979-989

0019-9567/82/030979-11$02.00/0

Outer Membrane Proteins of Brucella abortus: Isolation and

Characterization D. R. VERSTREATE,* M. T. CREASY, N. T. CAVENEY, C. L. BALDWIN, M. W. BLAB, AND A. J. WINTER Department of Clinical Sciences, New York State College of Veterinary Medicine, Cornell University, Ithaca, New York 14853

Received 27 July 1981/Accepted 6 November 1981

Outer membrane proteins were derived from one rough and four smooth strains of Brucella abortus by sequential extraction of physically disrupted cells with Nlauroylsarcosinate and dipolar ionic detergent. Extraction of outer membrane proteins was ineffective, however, without predigestion with lysozyme. Three groups of proteins were present and could be separated in their native state by sequential anion-exchange chromatography and gel filtration. Membrane proteins contained substantial quantities of tightly adherent lipopolysaccharide which could be reduced but not eliminated by extraction of cells with trichloroacetic acid before disruption. Group 2 proteins, apparently trimers in their native state, gave rise to 43,000- and 41,000-molecular-weight bands after complete denaturation in sodium dodecyl sulfate. They were antigenically identical among all the strains, showed close resemblance in amino acid composition to each other and a general similarity to OmpF of Escherichia coli, and are proposed to be the porins of B. abortus. Group 3 proteins occurred as 30,000-molecular-weight bands on sodium dodecyl sulfate-polyacrylamide gel electrophoresis, although additional bands were frequently observed in this region. In none of the strains did group 3 proteins manifest heat-modifiable characteristics. Proteins of different strains bore a high degree of similarity to each other in amino acid composition, except in methionine, isoleucine, tyrosine, and histidine. Differences occurred consistently in amino acid composition between group 2 and 3 proteins, and some of these correspond to differences between OmpF and OmpA. Group 2 and 3 proteins were antigenically distinct from each other, but the principal group 3 antigens were shared among all the strains. Despite the lack of heat modifiability, perhaps influenced by adherent lipopolysaccharide, group 3 proteins are proposed as counterparts to OmpA. Most of the group 1 proteins, minor components, were physically associated with those of group 3 unless in sodium dodecyl sulfate. Group 1 proteins produced a major band at 94,000 and exhibited heat modifiability. No evidence was found of a low-molecular-weight lipoprotein in the outer membrane of B. abortus, but this is not taken to exclude its occurrence. The principal classes of structural proteins in the outer membrane of Escherichia coli are matrix porins (OmpC and OmpF), a heat-modifiable protein (OmpA), and murein lipoprotein (6, 45). With certain modifications, as, for example, in the absence of murein-linked lipoprotein (22, 36, 57), counterpart proteins have been detected in a wide variety of gram-negative genera (4, 6, 8, 15, 21, 25, 26, 28, 34, 37, 53) and appear to be essential constituents of gram-negative bacteria. Virtually no information is yet available on the nature of comparable proteins in Brucella abortus. The extraction by sequential treatment with N-lauroylsarcosinate and dipolar ionic detergent (Zwittergent) of a protein presumably analogous to porin was presented in a preliminary report (I. Moriyon and D. T. Berman, Abstr. Annu. Meet. 979

Am. Soc. Microbiol. 1981, K171, p. 166), and Kreutzer and Robertson (30) reported the extraction from B. abortus of a murein-associated lipoprotein, although characterization of this substance was incomplete. Our interest in proteins of the outer membrane of B. abortus stems from problems associated with eradication of brucellosis in the United States. The development of an effective subcellular vaccine devoid of lipopolysaccharide (LPS) might resolve the single greatest obstacle to eradication: detection of latently infected animals (43). Additionally, diagnostic tests not reliant on the measure of LPS antibody would be a highly desirable alternative, although progress has been made in distinguishing vaccination from infection titers by decreasing the vaccine

980

VERSTREATE ET AL.

dose (3) and by the observation that most infected cattle produce a response to the native hapten of LPS not detectable in vaccinated animals (16, 39). Although the proteins under investigation are structural constituents of the cell, and therefore unlikely to function primarily as virulence factors, they may induce immune responses useful in diagnosis and in conferring protective immunity. Outer membrane proteins have been used in other genera in serological classification (7, 8, 18, 24) and as vaccines (1, 13, 31). Because B. abortus is a facultative intracellular parasite devoid of capsule, flagella, and pili, it seemed reasonable that protective immunity would include a cell-mediated immune response to one or more protein constituents of the cell wall. The purpose of this study was to isolate and characterize outer membrane proteins of B. abortus and to compare proteins among strains used for vaccination and representative virulent field isolates. MATERIALS AND METHODS Bacterial strains and cultivation. Five strains of biotype 1 were used: two were vaccine strains used commercially in the United States (strain 19) and in other countries (strain 45/20), and three were virulent field strains of diverse geographic origin (Table 1). Strain 2308 was isolated in 1942 and has been used in experimental infection trials for over 30 years (35). The culture of 2308 that was used had been passaged twice each in cows and guinea pigs, but only four times in total in vitro (B. L. Deyoe, personal communication). Strains C-10 and Y originated from Ontario and New York State, respectively, and had been passaged in vitro no more than two times. In our laboratory, each strain was streaked over the surface of Schaedler agar plates (BBL Microbiology Systems, Cockeysville, Md.) containing 5% bovine blood and incubated for 2 days at 37°C in an atmosphere of air containing 10%O CO2. Growth from several plates was suspended in Albimi broth (Difco Laboratories, Detroit, Mich.), pooled, and frozen in portions at -70°C. For each strain, the same batch of stocks was used throughout this study. For batch cultivation, stock cultures were transferred successively onto Schaedler agar plates and Albimi agar slants. Two-day growth from slants was suspended in Albimi broth. Flasks were incubated in a shaker incubator (New Brunswick Scientific Co., New Brunswick, N.J.) at 175 rpm and continuously gassed

Strain

Colony type

19 45/20 2308 C-10

Smooth Rough Smooth Smooth

Y

Smooth

INFECT. IMMUN. at a rate of 1.0 cubic foot (28.317 cubic decimeters) per

hour with a mixture of 95% air-5% CO2. Growth was terminated at 24 h (corresponding to late-logarithmic phase) by placing flasks in ice, and cells were collected by centrifugation at 4°C and washed twice in sterile phosphate-buffered saline (PBS) (pH 7.2). Because of the hazard of human infection, organisms were killed, unless otherwise noted, in PBS containing 0.5% Formalin. Killed cells were washed twice in sterile PBS, and the drained pellet was held at -20°C until extracted. Purity of growth from batch harvests was verified by examination of plates heaily streaked from the pooled flask contents. Colonial morphology was determined by standard methods (2). Extraction and purification of outer membrane proteins from physically disrupted ceUls. Cells were suspended at 1 g (wet weight) per 20 ml of 10 mM Trishydrochloride buffer (pH 7.5), and 1 mg each of DNase and RNase (Sigma Chemical Co., St. Louis, Mo.) was added per 100 ml. Disruption was accomplished by two passages through a high-pressure cell (Sorvall Ribi cell fractionator, model RF-1) at 40,000 lb/in2, with the temperature at the needle valve orifice maintained at 10 to 15°C. The disrupted suspension was centrifuged at 3,000 x g for 20 min at 4°C and the supernatant was centrifuged at 150,000 x g (average) for 60 min at 4°C to pellet the crude membranes, which were resuspended at 10 to 20 mg of protein per ml in Tris buffer. In some cases, outer membranes were separated from cytoplasmic membranes by density gradient centrifugation (49). Detergent extraction of cytoplasmic membranes was performed by using either Triton X-100 (Sigma Chemical Co.) (50) or sodium N-lauroylsarcosinate (Pfaltz & Bauer, Stamford, Conn.) (25). The resultant insoluble material was dialyzed against Tris buffer at 4°C for 72 h with repeated changes. The outer membrane-rich fraction, unless otherwise noted, was subjected to digestion overnight at 37°C with egg white lysozyme (Mann Laboratories, New York, N.Y.) (1 mg/50 mg of membrane protein). Solubilization was then performed by using Triton X100 with EDTA (50) or sodium deoxycholate (25), except that the protein concentrations used were 1 mg/ ml and the extraction was performed at 37°C for 1 h. Zwittergent 3-14 (Calbiochem, La Jolla, Calif.) (0.2%) in Tris buffer containing 0.25 M NaCl was used under the same conditions for this purpose. After extraction, the samples were centrifuged at 100,000 x g (average) for 20 min at 4°C, and the supernatants were held at 40C. Solubilized membrane fractions were concentrated by lyophilization to 10 to 20 mg/ml, equilibrated with 10 mM Tris buffer containing 0.1% Zwittergent and

TABLE 1. Derivation and properties Description Vaccine strain Vaccine strain Field strain; from aborted fetus Field strain; from supramammary lymph node Field strain; from aborted fetus

of B. abortus strains Origin U.S. Biologics Division, Dept. of Agriculture National Animal Disease Center, Ames, Iowa National Animal Disease Center, Ames, Iowa Animal Disease Research Inst., Nepean, Ontario, Canada N.Y. State Diagnostic Laboratory, Ithaca, N.Y.

VOL. 35, 1982

OUTER MEMBRANE PROTEINS OF B. ABORTUS

0.25 M NaCl and applied to a column of DEAESephacel (Pharmacia Fine Chemicals, Inc., Piscataway, N.J.) equilibrated with the same buffer. Elution was performed at room temperature with upward flow at a rate of 2 ml cm-2 h-1. After the initial wash, a gradient of NaCl (0.25 to 0.75 M) was established and collected over a period of 24 h. Protein samples from the ion-exchange column, concentrated by lyophilization, were separated under the same conditions of flow on a column of Sephacryl S-300 (Pharmacia Fine Chemicals, Inc.) equilibrated with 10 mM Tris buffer containing 0.1% Zwittergent and 0.25 M NaCl. If the proteins were not pure as judged by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE), they were either rechromatographed on Sephacryl S-300 or chromatographed with upward flow at 4 ml cm-2 h-1 on a column of hydroxylapatite (Biogel HT; Bio-Rad Laboratories, Richmond, Calif.), using a gradient of disodium phosphate (0.01 to 0.5 M). Extraction of outer membrane proteins with SDS. Living or Formalin-killed cells were extracted sequentially at 60 and 100°C with buffer containing 2% SDS (48). In one experiment, Formalin-killed cells were subjected to exhaustive extraction (17 h, with fresh extraction buffer at 1- or 2-h intervals) at 60°C in 4% SDS, until no further proteins were detectable on SDS-PAGE of a 10-fold-concentrated supernatant. The residue was subjected to lysozyme digestion and Zwittergent extraction under conditions outlined previously. In another experiment, living and Formalinkilled cells were boiled for 4 h in the presence of 4% SDS (at 1-h intervals, with fresh extraction buffer each time), after which no proteins were detectable in concentrated supernatants, and a portion of the insoluble residue, after thorough washing, was digested overnight at 37°C with lysozyme (1 mg/60 mg [dry weight]). Extraction of LPS. Acetone-dried cells were extracted with 45% hot phenol (56), and aqueous extraction of the phenol-soluble phase was performed by the method of Moreno et al. (38, 39) to obtain fraction 5, a crude preparation of biologically active LPS, which contained 43% protein, 24% carbohydrate, and 0.6% 2-keto-3-deoxyoctulosonic acid (KDO). The 50o lethal dose of this preparation in 16- to 18-g male HA(ICR) mice was 1.5 mg. Trichloroacetic acid (TCA) extractions of whole cells, or of cytoplasmic constituents derived from the 150,000 x g supernatant after pressure cell disruption, were performed by the method of Diaz et al. (16) and hydrolysis of TCA extracts by the method of Moreno et al. (39). SDS-PAGE. An equal volume of double-strength extraction buffer (48) was added to each sample, and heating, unless noted otherwise, was performed for 10 min at 100°C. In some instances, 0.1 M MgCl2 was included in the extraction buffer in an effort to visualize the low-molecular-weight lipoprotein (21). SDSPAGE was performed by the method of Laemmli (32) on 10 or 12.5% acrylamide gels. Phosphorylase b (94,000 molecular weight), bovine serum albumin (68,000), ovalbumin (43,000), carbonic anhydrase (30,000), soybean trypsin inhibitor (21,000), and lysozyme (14,300) (Bio-Rad Laboratories) were used as reference proteins. Gels were stained with Coomassie blue R-250 (Sigma Chemical Co.), and molecular

981

weight calculations were made from densitometric scans of Polaroid negatives (M. T. Creasy, submitted for publication). Chemical analysis. Amino acid analyses were performed on a Beckman model 119 CL analyzer equipped for single-column analysis. Samples were hydrolyzed in constant boiling HCl (5.7 N) containing 0.1% ethanethiol for 24 h at 110°C. Protein was determined by Peterson's modification of the method of Lowry et al. (47), with bovine serum albumin (Miles Laboratories, Inc., Elkhart, Ind.) as the standard. Total carbohydrates were estimated by the phenolsulfuric acid method (17) with glucose as a standard. The assay for 2-keto-3-deoxy sugars described by Weissbach and Hurwitz (55) was performed, using Osborn's modification (44). Results of this assay were taken as a measure of KDO in the sample, and KDO (Sigma Chemical Co.) was used as a standard. Correction for 2-deoxyaldoses was made by the method of Warren

(54).

Antisera. Male Flemish giant-chinchilla crossbred rabbits of 6 to 8 lb (-2.7 to 3.6 kg) were immunized with 2 to 3 mg of purified outer membrane protein in 1 ml of Tris buffer emulsified in an equal volume of complete Freund adjuvant. Half of the emulsion was injected intradermally in multiple sites on the back; the other half was injected intramuscularly in equal parts into each hind leg. Three weeks later, 2 mg of protein in buffer was injected intravenously. Animals were exsanguinated 1 week thereafter. Two animals were immunized with each preparation. LPS antibodies were removed by absorption with glutaraldehydefixed whole cells of strain 19, followed usually by the addition of TCA extract which contained LPS and native hapten. Absorption was considered complete when no lines developed with TCA extract in immunodiffusion reactions after 2 days of incubation at 23°C. A pool of antiserum with an agglutination titer in excess of 1:400 was obtained from four heifers 1 to 4 months after subcutaneous vaccination with strain 19. Sera from 11 cows infected with virulent field strains were obtained from J. R. Duncan (ADRI, Ontario, Canada). Immunological techniques. Immunodiffusion and immunoelectrophoresis were performed on gels composed of 1% agarose (Sea Plaque, Marine Colloids, Inc., Rockland, Maine) in 0.03 M barbital buffer (pH 8.8). In most immunodiffusion tests with bovine sera, NaCl was added to a concentration of 10o (16). Immunodiffusion tests were done with the use of templates (40). Immunoelectrophoresis was performed at 10 mA per frame in a Gelman chamber (Gelman Instrument Co., Ann Arbor, Mich.). Gels were dried and stained with 0.25% Crocein scarlet (Bio-Rad Laboratories) and 0.014% Coomassie blue R (Sigma Chemical Co.) (14). Detergent concentrations in antigen solutions were less than 0.05% to avoid artifacts which occurred at concentrations greater than 0.1%. Statistical methods. The t test (52) was used to compare the moles percent of each amino acid residue between group 2 and 3 proteins. The Spearman rank correlation method (52) was used to measure the degree of similarity between B. abortus and E. coli outer membrane proteins in amino acid composition. By this method, the rank correlation (re) can range from -1, signifying complete discordance, to +1, indicating complete concordance.

982

VERSTREATE ET AL.

RESULTS Extraction with Zwittergent and purification of outer membrane proteins. Density gradient centrifugation of crude membranes resulted in a banding pattern similar to that described by Schnaitman (49). The concentration of KDO to protein increased almost fivefold in the lower band relative to the starting material (from 0.80 to 3.85 ,ug/mg), indicating an enrichment in outer cell membranes. The sarcosinate-insoluble material from the lower band and from a portion of the same preparation not processed through gradient centrifugation was digested with lysozyme and extracted with Zwittergent. Each of the extracts contained three principal clusters of proteins on SDS-PAGE, of which the dominant bands had molecular weights of 94,000, 43,000, and 30,000 (Fig. 1). These will be referred to as groups 1, 2, and 3, respectively. The band at 14,000 was lysozyme, but no other low-molecular-weight bands were resolved on 10 or 12.5% gels, even with MgCl2 in the extraction buffer. Thus, group 1, 2, and 3 proteins were associated with the outer cell membrane, but gradient centrifugation was not a necessary condition for their extraction. In subsequent studies, this step was omitted. It was also demonstrated consistently that sarcosinate extracted cytoplasmic membrane constituents more completely than did Triton X-100 but did not remove outer membrane proteins, as has been reported for E. coli (12). Sarcosinate was therefore used in all subsequent extractions. Comparison of solubilizing efficiency on sarcosinate-insoluble outer membrane proteins of Triton X-100 + EDTA, sodium deoxycholate, and Zwittergent 3-14, with and without lyso-

INFECT. IMMUN.

TABLE 2. Solubilizing efficiencies of detergents with or without lysozyme predigestion on strain 19 sarcosinate-insoluble outer membrane proteins Protein solubilized (%) Detergent

2% Triton X-100 + 5 mM EDTA 2% Sodium deoxycholate 0.2% Zwittergent 3-14

No lysozyme Lysozyme predigestion predigestion

0.9

73.4

5.6 5.0

26.6 77.8

zyme predigestion, demonstrated that lysozyme pretreatment greatly enhanced the solubilizing efficiency of the detergents (Table 2). Of equal importance, it was noted that group 3 proteins were virtually absent from extracts prepared without lysozyme predigestion (Fig. 2c). Zwittergent was chosen for subsequent studies because of its solubilizing efficiency and convenience. Outer membrane proteins from the five strains of B. abortus contained group 1, 2, and 3 proteins (Fig. 3), and ion-exchange chromatography was highly effective in their separation.

Although there was a general pattemn of elution, variations occurred among strains and even among different extracts of the same strain. Such variations may have been owing in some measure to micelle-protein interactions. In all separations, the bulk of group 2 proteins of smooth strains (Fig. 4a and b) eluted at a conductivity at least 0.4 mS higher than those from

a b

d

e

K

.:Se

ab 94 K--

43K 30K-' 14K--

FIG. 1. SDS-PAGE of B. abortus strain 19 outer membrane proteins. (a) Crude membrane preparation subjected to density gradient centrifugation, extraction with sarcosinate, lysozyme digestion, and extraction with Zwittergent. (b) Same treatments with omission of gradient centrifugation. The band at 14,000 (14K) is lysozyme.

FIG. 2. SDS-PAGE of B. abortus strain 19 extracted with SDS. (a) Supernatant after extraction at 600C for 30 min. (b) Sediment from (a) heated for 10 min at 1000C with dominant band at 41,000 (41K). (c) Comparison with outer envelope proteins solubilized by deoxycholate without prior lysozyme digestion. Group 2 bands at 43,000 and 41,000 are evident, with an extremely faint band at 30,000. (d) Cells exhaustively heated in SDS at 60°C, then digested with lysozyme and extracted with Zwittergent, yielding bands at 94,000, 43,000, 41,000, and 30,000. (e) Cells exhaustively heated in SDS at 100°C, then digested with lysozyme; only the 41,000 band of the outer membrane proteins is resolved. The lowest band in (d) and (e) is

lysozyme.

VOL. 35, 1982


983

FIG. 3. Densitometric scans from Polaroid negatives of SDS-PAGE gels of Zwittergent-soluble outer membrane preparations of B. abortus strains 19 (a), 45/20 (b), 2308 (c), Y (d), and C-10 (e). In (b) and (e), 30,000 and 27,000 bands are resolved in group 3. Absorption at 580 nm is recorded on the ordinate. Numbers refer to protein groups.

rough strain 45/20 (Fig. 4c). Group 3 proteins were eluted with similar frequency before (Fig. 4b) or after (Fig. 4a) application of the gradient and sometimes in several discrete peaks (Fig. 4c). A portion of group 1 and 3 proteins was always associated in Zwittergent and could not be separated completely by sequential gel filtration and hydroxylapatite chromatography. SDS apparently disrupted this association, even at room temperature (Fig. 5a). In general, the process of extraction and purification of rough strain 45/20 was the easiest (and most complete), that of strain 19 was intermediate in difficulty, and that of the virulent strains was the most difficult. Difficulties encountered with smooth strains may have been owing in part to interactions of LPS, native hapten, or other cell wall constituents with outer membrane proteins. KDO was demonstrable in each of selected preparations tested (Table 3) and 0 antigens were detectable by immunodiffusion in all of the purified protein preparations, although very faintly in those of strain 45/20. Physical association of these proteins with LPS or native hapten was indicated by lack of demonstrable free LPS or native hapten on immunoelectrophoresis (Fig. 6a, b, c, and d) and by inability to separate LPS or native hapten from group 2 proteins through sequential ion-exchange, gel filtration, and hydroxylapatite chromatography. ExtracTABLE 3. Chemical analyses of purified outer membrane proteins of B. abortus Strain Protein group KDO (ig/mg of protein) 45/20 2 2.3 45/20 3 13.8 Y 2 2.3 Y 4.2 3

tion of cells with TCA (16) before physical disruption and detergent extraction was found to reduce substantially (Fig. 6c and d) but not to eliminate 0 antigens bound to the membrane proteins. Proteins so derived were used as antigens in the later portion of this study. In E. coli it has been shown that LPS is critical for the macromolecular association of porins (20, 45, 58), but it was not possible in this study to distinguish to what degree LPS association with outer membrane proteins occurred in the intact cell or was a consequence of the extraction process. More drastic efforts to separate 0 antigens from purified B. abortus proteins were not made because we believed that meaningful studies of antigenic relationships could be done only with proteins in their native configuration. Extraction of cells with SDS. Heating at 60°C released many proteins from the cell (Fig. 2a), including in some instances a faint band at 43,000. When the sediment after a 60°C heating was heated in SDS at 100°C for 10 min (48), the 41,000 band was the only one prominently visualized (Fig. 2b). Lysozyme digestion and Zwittergent extraction of the sediment after heating for 17 h at 60°C in 4% SDS caused release of all the principal outer envelope proteins, including bands at 94,000, 43,000, 41,000, and 30,000 (Fig. 2d). Without lysozyme digestion, even 4 h of heating at 100°C in 4% SDS failed to remove all of the outer membrane proteins from the murein. Group 3 proteins were neither detected in concentrated supernatants of 4-h boiled cells nor resolved on SDS-PAGE of the sediment (Fig. 2e). The amino acid composition of this sediment resembled a mixture of group 2 and 3 proteins. The same results were obtained with SDS extraction procedures whether performed with living or with Formalin-killed cells. Amino acid analyses. Amino acid composi-

984

VERSTREATE ET AL.

INFECT. IMMUN.

; E EE

E

C

0 cli

0

H

z

F

> m c:: 0(

D a

z

(/) m

O

0t

FRACTION NUMBER

FIG. 4. Ion-exchange chromatography on DEAESephacryl of Zwittergent-soluble outer membrane preparations of B. abortus strains 19 (a), Y (b), and 45/ 20 (c). Numbers refer to dominant protein groups in each peak; L refers to lysozyme.

tions of group 2 proteins of the five B. abortus strains were very similar. The same held true for group 3 proteins, except for large differences for methionine, isoleucine, tyrosine, and histidine (Table 4). The moles percent of tyrosine, phenylalanine, threonine, serine, glycine, and valine were significantly greater, and those of proline, leucine, and lysine were significantly lower (P < 0.05) in group 2 compared with group 3 proteins. It is notable that tyrosine and phenylalanine are proportionally more abundant and proline is less

abundant in OmpF than in OmpA of E. coli (Table 4). Rank correlations showed that although all of the B. abortus and E. coli outer membrane proteins were significantly similar (P < 0.01), a higher degree of similarity existed between B. abortus group 2 and E. coli OmpF (r, = 0.959) and between B. abortus group 3 and E. coli OmpA (rs = 0.918). The hydrophobicities (5) of group 2 and 3 proteins were 892 ± 30 calories per mol (3,732.5 ± 126 J/mol) and 980 ± 54 calories per mol (4,101.1 ± 226 J/mol), respectively, and a polarity index (9) of 42% was calculated for both protein groups, approximating values for outer membrane proteins of other gram-negative bacteria (27, 48). Heat modifiability. Purified group 2 and 3 proteins from each strain and partially purified group 1 proteins from strain 45/20 were incubated in 2% SDS for 2 h at 23 and 37°C and for 10 min at 100°C to determine effects on migration in SDS-PAGE. Group 2 proteins were at approximately 115,000 after treatment at 23 or 37°C and at 43,000 after heating at 100°C (Fig. Sd and e). After heating at 100°C, distinct bands were always present at 43,000 and 41,000 (Fig. 1 and 2c) unless gels were overloaded. Group 3 proteins displayed no heat modifiability. A band was always present at 30,000 (Fig. 1 and 2d) and in some preparations another of varying intensity at 27,000 (Fig. 4 and 5f and g). Additional minor bands sometimes observed in this region (Fig. 1 and 5) varied in occurrence among extracts of the same strain. At 23°C, group 1 protein bands ranged from approximately 84,000 to 97,000, with the heaviest band at 84,000. After being heated at 37°C, the major band was at 94,000. Heating at 100°C

K *

b c

a

.. 11 5 K L3 _'.:.:. i 94 K-_"W., .4.

d

.

43K---

>e

&i

..

e

t

9

s:

::

.t

...

xE, ,

t

--

.....

..

...

..

-

i mm

::;-

.

FIG. 5. SDS-PAGE of group 1 (lanes a, b, and c), 2 (lanes d and e) and group 3 (lanes f and g) proteins of B. abortus strain 45/20 after different temperature treatments. Proteins were incubated in 2% SDS for 2 h at 23'C (lane a), for 2 h at 37'C (lanes b, d, and f), and for 10 min at 100°C (lanes c, e, and g). Group 1 and 3 preparations contain small quantities of the opposite protein species.

group

VOL. 35, 1982


caused further intensification of the 94,000 bands, disappearance of bands above 94,000 and generation of several lower bands (Fig. 5a, b, and c). The faint bands in each strain present between the 94,000 and 43,000 peaks (Fig. 3) were probably constituents of group 1 proteins. Antigenic comparisons. It was essential to distinguish antigens of the outer membrane proteins from the closely associated 0 antigens. The identification of antigens as LPS and native hapten was based on possession of essential properties matching those previously described (16, 39), including their presence in extracts of hot phenol and TCA (Fig. 7a), the presence of hapten in cytoplasm and cell wall (Fig. 7b), hydrolysis of LPS into hapten (Fig. 7c), the presence of antibodies for hapten in sera of field strain-infected but not strain 19-vaccinated cattle (Fig. 7d), and enhanced development of immunoprecipitates of hapten in cattle sera with 10% salt agar (data not shown). Immunodiffusion reactions were performed on purified proteins of the five strains with absorbed antisera specific for group 2 proteins of strains 19, 45/20, 2308, and Y and for group 3 proteins of strains 45/20 and Y. Each system was arranged to provide complete information on cross-reactions of heterologous with homologous antigens (Fig. 8). Group 2 proteins developed two distinct lines which in each of the reciprocal test systems were shared in an identical way among all the strains (Fig. 8a and b). Before absorption, 0 determinants were always associated with the line closer to the serum well. Group 3 proteins produced a heavy broad band close to the serum well which could be resolved into two lines when antigens were diluted. Reactions of identity among the five strains were noted with antisera specific for strains Y (Fig. 8c and d) and 45/20. In both antigen systems, one or more additional faint lines, not uniformly shared among strains, were observed with some antisera. Group 2 and 3 antigens were immunologically unrelated on the basis of immunodiffusion and immunoelectrophoresis (Fig. 6a and b). Sera from 11 cattle infected with field strains of B. abortus, after absorption of 0 antibodies, failed to react with group 2 or 3 proteins in immunodif-

fusion. DISCUSSION The weight of evidence favors the hypothesis that group 2 and 3 proteins of B. abortus are analogous to matrix porins and OmpA, respectively. Group 2 proteins had a molecular weight in their denatured form comparable to most porins, and their amino acid composition bore considerable resemblance to OmpF of E. coli (Table 4) and to porins of other gram-negative bacteria (27, 28, 45), although this in itself would

985

not ensure identity (45). Differential migration of these proteins after heating suggests their existence as trimers in the native state, in accord with observations on porins from other genera (19, 42, 45, 46, 59). Group 3 proteins were more

a_ _

c

40

CX

d FIG. 6. Immunoelectrophoresis of B. abortus antigens with anode to right. (a) Upper and lower wells, strain 19 TCA extract; middle well, strain Y-purified group 2 proteins. Upper trough, rabbit antiserum specific for strain Y group 2; lower trough, same antiserum before absorption of 0 antibodies. (b) Upper and lower wells, strain 19 TCA extract; middle well, strain Y-purified group 3 proteins. Upper trough, rabbit antiserum specific for strain Y group 3; lower trough, same antiserum before absorption of 0 antibodies. In (a) and (b), unabsorbed sera form precipitates with 0 antigens, but absorbed sera do not. The precipitate in the upper sector of (b) formed as a result of migration of unabsorbed serum from (a), lower trough, which was adjacent. Group 2 and 3 antigens produce no precipitates comparable to the TCA extract, indicating lack of unbound LPS, but precipitates of the proteins with absorbed sera are lighter, suggesting physical association of 0 antigens with group 2 and 3 proteins. The anodal streaks with protein antigens were consistently observed with rabbit antisera and are considered artifacts. (c) Upper well, strain 19 TCA extract; middle well, strain Y-purified group 2 proteins; lower well, strain 19-purified group 2 proteins from cells extracted with TCA before disruption. Troughs, antiserum from a cow with brucellosis. (d) Upper well, strain 19 TCA extract: middle well, strain 2308-purified group 3 proteins; lower well, strain 2308purified group 3 proteins from cells extracted with TCA before disruption. Troughs, same serum as in (d). Absorption experiments indicated that immunoprecipitates produced by bovine sera with group 2 and 3 proteins resulted from 0 determinants bound to the proteins (see text).

986

VERSTREATE ET AL.

INFECT. IMMUN.

TABLE 4. Amino acid compositions of group 2 and 3 proteins from B. abortus and OmpF and OmpA of

E. coli Moles percent E. colP OmpF B. abortush group 2 E. colt OmpA B. Asxc 17.4 11.7 ± 0.4 12.6 Thr 6.2 7.7 ± 0.3 6.5 Ser 4.7 5.9 ± 0.2 4.9 Glxc 7.1 9.4 ± 0.3 8.9 Pro 1.2 2.4 ± 0.4 5.8 Gly 13.5 15.4 ± 0.5 11.4 Ala 8.5 10.8 ± 0.2 8.9 Cys 0 ND' 0.6 Val 7.1 8.1 ± 0.6 7.7 Met 0.9 0.6 ± 0.3 1.5 Ile 3.5 4.0 ± 0.6 4.3 Leu 6.2 5.3 ± 0.5 6.8 Tyr 8.5 6.1 ± 0.1 5.2 Phe 5.6 4.5 ± 0.1 2.5 His 0.3 1.2 ± 0.2 1.5 Lys 5.3 4.0 ± 0.3 5.2 Arg 3.2 3.8 ± 0.5 4.0 Trp 0.9 ND 1.5 a Derived from data in references 10 and 11. b Data are means ± SD for the five strains tested. c Asx, Aspartic acid and asparagine; Glx, glutamic acid and glutamine. d *Significantly different (P < 0.05) from value of corresponding residue in group 2 proteins.

Amino acid

e

abortusb group 3 11.4 ± 0.7 5.5 ± 0.4*d 4.6 ± 0.4* 9.8 ± 0.7 4.2 ± 0.2* 14.5 ± 0.4* 10.2 ± 0.8 ND 6.7 ± 0.6* 0.8 ± 0.7 6.7 ± 3.1 7.2 ± 0.2* 3.2 ± 2.3* 3.4 ± 0.3* 2.2 ± 1.6 6.7 ± 0.6* 3.4 ± 0.3 ND

ND, Not determined.

closely related in amino acid composition to OmpA than OmpF and differed from group 2 proteins at some of the same residues as those at which OmpA differs from OmpF (10, 11). Group 3 proteins were more refractory than were those of group 2 to detergent extraction. If this were owing to their stronger association with murein, it would compare with OmpA, now believed to interact more extensively than matrix porins with murein (45). The migration of group 3 proteins in SDS-PAGE after 100°C heating was comparable to OmpA, but unlike homologs of OmpA in other genera (4), heat-modifiable behavior of group 3 proteins did not occur. Heat modifiability of E. coli OmpA (51) and of several proteins of Pseudomonas aeruginosa exclusive of porins (21) was obviated by addition of LPS, and the LPS bound to each of our group 3 proteins may have had a similar effect. In contrast, the minor group 1 proteins, most of which were physically associated with group 3 unless in SDS, exhibited both heat modifiability and apparent disaggregation at higher temperatures. Effective solubilization of B. abortus outer membrane proteins, in particular, those of group 3, was achieved only after digestion with lysozyme. The mechanism of lysozyme enhancement is uncertain. Cleavage of murein may have weakened its association with protein or allowed better access of detergents. The requirement for lysozyme digestion was probably owing, at least in part, to the use of Formalin-killed cells.

Zwittergent extraction of outer envelopes from living and Formalin-killed B. abortus could not be compared owing to the risk of infection in preparing envelopes from living B. abortus. A comparison of this type was conducted with E. coli B and it was found that lysozyme treatment did not enhance Zwittergent extraction of outer membranes prepared from living cells but increased fourfold the percentage of protein extracted from outer membranes derived from Formalin-killed cells (D. R. Verstreate and A. J. Winter, unpublished data). It is uncertain whether multiple banding patterns within each of the protein groups on SDSPAGE reflect the existence of distinct protein species. It is possible, for example, that the two principal precipitin lines which developed with group 2 proteins were produced by antigenically distinct 43,000 and 41,000 proteins, but no direct evidence for such an association is yet available. Multiple banding patterns have been observed in the porins of a variety of gram-negative bacteria (15, 21, 34, 41, 45), but in only a few genera has it been established that these represent separate gene products (45). Multiple bands from a single protein species have been ascribed to several factors (28, 33, 48), and the variable number of bands observed, for example, in group 3 proteins, among and within strains might be explained by varying quantities of adherent murein or LPS. This question in regard to the B. abortus proteins remains open.

VOL. 35, 1982

987


In none of our preparations was it possible to visualize a low-molecular-weight protein comparable to the lipoprotein abundant in the outer membrane of many gram-negative bacteria (6, 45). In contrast to P. aeruginosa (21, 22), lysozyme predigestion did not facilitate solubilization of lipoprotein and the addition of Mg2+ to solubilized samples did not allow visualization of a low-molecular-weight band. The presence of lipoprotein covalently linked to murein could not be determined by amino acid analysis owing to the presence of other proteins on the murein even after exhaustive boiling in SDS of either Formalin-killed (Fig. 2e) or living cells. Such negative data are of course not sufficient to exclude the occurrence of a lipoprotein in the B. abortus wall. Group 2 and 3 proteins were antigenically distinct, as are the porins and heat-modifiable proteins of E. coli (23) and Neisseria gonorrhoeae (25). In contrast to Neisseria spp. (7, 18, 24) group 2 and 3 antigens were shared among five random strains, but more strains must be tested to ensure that these are species-wide antigens. Production of antibodies to these pro-

FIG. 7. Immunodiffusion reactions of B. abortus 0 antigens developed with sera from cows with brucellosis. Agar contains 1%o NaCl. (a) Wells 1 and 2, antiserum; well 3, fraction 5, derived from phenol extraction of strain 19; well 4, acid-soluble phase from TCA extract of strain 19; well 5, anodally migrating fraction derived from hydroxylapatite fractionation of strain 19 outer membrane proteins. Lines are LPS (closer to antigen well) and native hapten (29). (b) TCA extracts of whole cells (well 1) and cytoplasm (wells 5 and 6) of strain 19 and whole cells (well 2) and cytoplasm (wells 3 and 4) of strain 2308. Serum in center well. (c) TCA extract of strain 19 untreated (well 1) and hydrolyzed at 100°C in 1% acetic acid for 5, 15, 30, 60, and 120 min (wells 2 through 6, respectively). Serum in center well. (d) Wells 1, 2, and 3, sera from individual cows with brucellosis; wells 4, 5, and 6, pooled sera from heifers vaccinated with strain 19. Center well, TCA extract of strain 19. The inner LPS line is only faintly visible in the serum pool of vaccinated animals.

a

b -, M.M."INVOW "I.:,

ee

:, i . . . . .

19

4520

2308

C10

4520

Y

d

C 19

Y

4520

CO

V

2308

FIG. 8. Immunodiffusion reactions of purified outer membrane proteins with specific rabbit antisera freed of 0 antibodies by absorption. (a) and (b), bottom wells, group 2 proteins (2 mg/ml) of designated strains; top wells, antiserum specific for group 2 proteins of strain 45/20. Split in strain Y precipitin line is an artifact owing to an edge effect. (c) and (d), bottom wells, group 3 proteins (250 sLg/ml) of designated strains; top wells, antiserum specific for group 3 proteins of strain Y. One band with strain C-10 developed very faintly but could be visualized when antigens were more concentrated.

teins in rabbits proved difficult owing to the dominant response to LPS. It is likely that the proteins functioned in some measure as carriers for the tightly associated LPS determinants. Recent experiments have demonstrated that proteins can be almost completely separated from LPS by brief sonication in Zwittergent followed by TCA precipitation (D. R. Verstreate, unpublished data). The antibody response in naturally infected cattle is known to be directed predominantly to LPS and native hapten (16, 39), and in single samples from a few animals, we found no precipitins to group 2 and 3 proteins. Responses to these proteins in vaccinated and infected cattle are being evaluated currently by enzyme-linked immunosorbent assay and blastogenesis assays (C. L. Baldwin, unpublished data). Preliminary data indicate that the antigenic relatedness of these proteins between vaccine strains and field strains does not limit their usefulness in differentiating vaccinal responses from clinical disease, and their role in conferring protective immunity is currently under investigation. ACKNOWLEDGMENTS We thank C. Fulimer for performing the amino acid analyses, G. Barker for photography, and J. Reyna for preparing the manuscript. This research was supported in part by U.S. Dept. of Agriculture grant 59-2361.-02-080-0. LITERATURE CITED 1. Adamus, G., M. Mulczyk, D. Wikowska, and E. Romanowska. 1980. Protection against keratoconjunctivitis shigellosa induced by immunization with outer membrane proteins of Shigella spp. Infect. Immun. 30:321-324.

988

VERSTREATE ET AL.

2. Alton, G. G., L. M. Jones, and D. E. Pietz. 1975. Laboratory techniques in brucellosis. World Health Organization monograph series no. 55. World Health Organization, Geneva, Switzerland. 3. Barton, C. E., and J. R. Lomme. 1980. Reduced-dose whole herd vaccination against brucellosis: a review of recent experience. J. Am. Vet. Med. Assoc. 177:12181220. 4. Beher, M. G., C. A. Schnaitman, and A. P. Pugsley. 1980. Major heat-modifiable outer membrane protein in gramnegative bacteria: comparison with the OmpA protein of Escherichia coli. J. Bacteriol. 143:906-913. 5. Bigelow, C. C., and M. Channon. 1976. Hydrophobicities of amino acids and proteins, p. 209-243. In G. D. Fasman (ed.), Handbook of biochemistry and molecular biology: nucleic acids section, 3rd ed. CRC Press, West Palm Beach, Fla. 6. Braun, V., and K. Hantke. 1974. Biochemistry of bacterial cell envelopes. Annu. Rev. Biochem. 43:89-121. 7. Buchanan, T. M., and J. F. Hildebrandt. 1981. Antigenspecific serotyping of Neisseria gonorrhoeae: characterization based upon principal outer membrane protein. Infect. Immun. 32:985-994. 8. Caldwell, H. D., J. Kromhout, and J. Schachter. 1981. Purification and partial characterization of the major outer membrane protein of Chlamydia trachomatis. Infect. Immun. 31:1161-1176. 9. Capaldi, R. A., and G. Vanderkooi. 1972. The low polarity of many membrane proteins. Proc. NatI. Acad. Sci. U.S.A. 69:930-932. 10. Chen, R., C. Kramer, W. Schmidmayr, and U. Henning. 1979. Primary structure of major outer membrane protein I of Escherichia coli B/r. Proc. Natl. Acad. Sci. U.S.A. 76:5014-5017. 11. Chen, R., W. Schmidmayr, C. Kramer, V. ChenSchmeisser, and U. Henning. 1980. Primary structure of major outer membrane protein II* (Omp A protein) of Escherichia coli K-12. Proc. Natl. Acad. Sci. U.S.A. 77:4592-4596. 12. Chopra, I., and S. W. Shales. 1980. Comparison of the polypeptide composition of Escherichia coli outer membranes prepared by two methods. J. Bacteriol. 144:425427. 13. Craven, D. E., and C. E. Frasch. 1979. Protection against group B meningococcal disease: evaluation of serotype 2 protein vaccines in a mouse bacteremia model. Infect. Immun. 26:110-117. 14. Crowle, A. J., and L. J. Cline. 1977. An improved stain for immunodiffusion tests. J. Immunol. Methods 17:379-381. 15. Darveau, R. P., W. T. Charnetsky, and R. E. Hurlbert. 1980. Outer membrane protein composition of Yersinia pestis at different growth stages and incubation temperatures. J. Bacteriol. 143:942-949. 16. Diaz, R., P. Garatea, L. M. Jones, and I. Moriyon. 1979. Radial immunodiffusion test with a Brucella polysaccharide antigen for differentiating infected from vaccinated cattle. J. Clin. Microbiol. 10:37-41. 17. Dubois, M., K. A. Giles, J. K. Hamilton, P. A. Rebers, and F. Smith. 1956. Colorimetric method for determination of sugars and related substances. Anal. Chem. 28:350-356. 18. Frasch, C. E., and E. C. Gotschlich. 1974. An outer membrane protein of Neisseria meningitidis responsible for serotype specificity. J. Exp. Med. 140:87-104. 19. Frasch, C. E., and L. F. Mocca. 1978. Heat-modifiable outer membrane proteins of Neisseria meningitidis and their organization within the membrane. J. Bacteriol. 136:1127-1134. 20. Gmeiner, J., and S. Schlecht. 1980. Molecular composition of the outer membrane of Escherichia coli and the importance of protein-lipopolysaccharide interactions. Arch. Microbiol. 127:81-86. 21. Hancock, R. E. W., and A. M. Carey. 1979. Outer membrane of Pseudomonas aeruginosa: heat- and 2mercaptoethanol-modifiable proteins. J. Bacteriol. 140:902-910.

INFECT. IMMUN. 22. Hancock, R. E. W., R. T. Irvin, J. W. Costerton, and A. M. Carey. 1981. Pseudomonas aeruginosa outer membrane: peptidoglycan-associated proteins. J. Bacteriol. 145:628-631. 23. Hofstra, H., and J. Dankert. 1980. Preparation and quantitative determination of antibodies against major outer membrane proteins of Escherichia coli 026 K60. J. Gen. Microbiol. 117:437-447. 24. Johnston, K. H. 1977. Surface antigens: an outer membrane protein responsible for imparting serological specificity to Neisseria gonorrhoeae, p. 274-283. In R. R. Roberts (ed.), The gonococcus. John Wiley & Sons, Inc., New York. 25. Johnston, K. H. 1978. Antigenic profile of an outer membrane complex of Neisseria gonorrhoeae for serotypic specificity, p. 121-129. In G. F. Brooks, E. C. Gotschlich, K. K. Holmes, W. D. Sawyer, and F. E. Young (ed.), Immunology of Neisseria gonorrhoeae. American Society for Microbiology, Washington, D.C. 26. Johnston K. H., and E. C. Gotschlich. 1974. Isolation and characterization of the outer membrane of Neisseria gonorrhoeae. J. Bacteriol. 119:250-257. 27. Jones, R. B., P. A. Jemison, W. J. Newhall V, and R. A. Haak. 1980. Resolution of basic gonococcal outer membrane proteins by nonequilibrium pH gradient electrophoresis. Infect. Immun. 30:773-780. 28. Kabir, S. 1980. Composition and immunochemical propertites of outer membrane proteins of Vibrio cholerae. J. Bacteriol. 144:382-389. 29. Knight, E., Jr. 1975. Heterogeneity of purified mouse interferons. J. Biol. Chem. 250:4139-4144. 30. Kreutzer, D. L., and D. C. Robertson. 1979. Surface macromolecules and virulence in intracellular parasitism: comparison of cell envelope components of smooth and rough strains of Brucella abortus. Infect. Immun. 23:819828. 31. Kuusi, N., M. Murminen, H. Saxen, M. Valtoneu, and P. H. Makela. 1979. Immunization with major outer membrane proteins in experimental salmonellosis of mice. Infect. Immun. 25:857-862. 32. Laemmli, U. K. 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature (London) 227:680-685. 33. Lester, E. P., J. B. Miller, and S. Yachnin. 1978. Human alpha-fetoprotein: immunosuppressive activity and microheterogeneity. Immunol. Commun. 7:137-161. 34. Loeb, M. R., and D. H. Smith. 1980. Outer membrane protein composition in disease isolates of Haemophilus influenzae: pathogenic and epidemiological implications. Infect. Immun. 30:709-717. 35. Manthei, C. A. 1950. Brucellosis in cattle, p. 172-187. In C. L. Larson and M. H. Soule (ed.), Brucellosis. American Association for the Advancement of Science, Washington, D.C. 36. Martin, H. H. 1966. Biochemistry of bacterial cell walls. Annu. Rev. Biochem. 35:457-484. 37. McCoy, E. C., H. A. Wiltberger, and A. J. Winter. 1976. Major outer membrane protein of Campylobacter fetus: physical and immunological characterization. Infect. Immun. 13:1258-1265. 38. Moreno, E., M. W. Pitt, L. M. Jones, G. G. Schurig, and D. T. Berman. 1979. Purification and characterization of smooth and rough lipopolysaccharides from Brucella abortus. J. Bacteriol. 138:361-369. 39. Moreno, E., S. L. Speth, L. M. Jones, and D. T. Berman. 1981. Immunochemical characterization of Brucella lipopolysaccharides and polysaccharides. Infect. Immun. 31:214-222. 40. Munoz, J. 1971. Double diffusion in plates. Methods Immunol. Immunochem. 3:143-156. 41. Nakae, T., and J. Ishii. 1978. Transmembrane permeability channels in vesicles reconstituted from single species of porins from Salmonella typhimurium. J. Bacteriol. 133:1412-1418. 42. Newhall, W. J., W. D. Sawyer, and R. A. Haak. 1980.

VOL. 35, 1982

43. 44.

45.

46. 47.

48. 49. 50.

51.


Cross-linking analysis of the outer membrane proteins of Neisseria gonorrhoeae. Infect. Immun. 28:785-791. Nicoletti, P. 1980. The epidemiology of bovine bruceliosis. Adv. Vet. Sci. Comp. Med. 24:70-98. Osborn, M. J. 1963. Studies on the gram-negative cell wall. I. Evidence for the role of 2-keto-3-deoxyoctonate in the lipopolysaccharide of Salmonella typhimurium. Proc. Natl. Acad. Sci. U.S.A. 50:499-506. Osborn, M. J., and H. C. P. Wu. 1980. Proteins of the outer membrane of the gram-negative bacteria. Annu. Rev. Microbiol. 34:369-422. Palva, E. T., and L. L. Randall. 1978. Arrangement of protein I in Escherichia coli outer membrane: crosslinking study. J. Bacteriol. 133:279-286. Peterson, G. L. 1977. A simplification of the protein assay method of Lowry et al. which is more generally applicable. Anal. Biochem. 83:346-356. Rosenbusch, J. P. 1974. Characterization of the major envelope protein from Escherichia coli. J. Biol. Chem. 249:8019-8029. Schnaltman, C. A. 1970. Protein composition of the cell wall and cytoplasmic membrane of Escherichia coli. J. Bacteriol. 104:890-901. Schnaitman, C. A. 1971. Effect of ethylenediaminetetraacetic acid, Triton X-100, and lysozyme on the morphology and chemical composition of isolated cell walls of Escherichia coli. J. Bacteriol. 108:553-563. Schweizer, M., D. Hindennach, U. Garten, and U. Henning. 1978. Major proteins of the Escherichia coli outer

52. 53.

54. 55. 56.

57.

989

membrane. Interaction of protein II* with lipopolysaccharide. Eur. J. Biochem. 82:211-217. Snedecor, G. W. 1956. Statistical methods, 5th ed. Iowa State University, Ames, Iowa. Tsai, C., and C. E. Frasch. 1980. Chemical analysis of major outer membrane proteins of Neisseria meningitidis: comparison of serotypes 2 and 11. J. Bacteriol. 141:169176. Warren, L. 1959. The thiobarbituric acid assay of sialic acids. J. Biol. Chem. 234:1971-1975. Weissbach, A., and J. Hurwitz. 1959. The formation of 2keto-3-deoxyheptonic acid in extracts of Escherichia coli B. I. Idenification. J. Biol. Chem. 234:705-709. Wesphal, O., and K. Jann. 1965. Bacterial lipopolysaccharides. Extraction with phenol-water and further applications of the procedure. Methods Carbohydr. Chem. 5:8396. Winter, A. J., W. Katz, and H. H. Martin. 1971. Murein (peptidoglycan) structure of Vibrio fetus. Comparison of a venereal and an intestinal strain. Biochim. Biophys. Acta

244:58-64. 58. Yamada, H., and S. Mizushima. 1980. Interaction between major outer membrane protein (0-8) and lipopolysaccharide in Escherichia coli K12. Eur. J. Biochem. 103:209218. 59. Yu, F., S. Ichihara, and S. Mizushima. 1979. A major outer membrane protein (0.8) of Escherichia coli K-12 exists as a trimer in sodium dodecyl sulfate solution. FEBS Lett. 100:71-74.