Vol. 158, No. 1JOURNAL OF BACTERIOLOGY, Apr. 1984, p. 102-1090021-9193/84/040102-08$02.00/0Copyright C 1984, American Society for Microbiology

Isolation and Characterization of Outer Membranes of Bacteroidesthetaiotaomicron Grown on Different Carbohydrates

SUSAN F. KOTARSKI AND ABIGAIL A. SALYERS*

Department of Microbiology, University of Illinois, Urbana, Illinois 61801

Received 6 September 1983/Accepted 6 December 1983

To determine whether certain outer membrane proteins are associated with growth of Bacteroidesthetaiotaomicron on polysaccharides, we developed a procedure for separating outer membranes frominner membranes by sucrose density centrifugation. Cell extracts in 10% (wt/vol) sucrose-10 mM HEPESbuffer (N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid) (pH 7.4) were separated into two fractions ona two-step (37 and 70o [wt/vol]) sucrose gradient. These fractions were further resolved into outermembranes (p = 1.21 g/cm3) and inner membranes (p = 1.14 g/cm3) on sucrose gradients. About 20 to 26% ofthe total 3-hydroxy fatty acids from lipopolysaccharide and 2 to 3% of the total cellular succinatedehydrogenase activity were recovered in the outer membrane preparation. The inner membrane prepara-

tion contained 22 to 49% of the total succinate dehydrogenase activity and 2 to 3% of the total 3-hydroxyfatty acids from lipopolysaccharide. Outer membranes contained a lower concentration of protein (0.34mg/mg [dry weight]) than did the inner membranes (0.68 mg/mg [dry weight]). Molecular weights of innermembrane polypeptides ranged from 11,000 to 133,000. The most prominent polypeptides had molecularweights ranging from 11,000 to 26,000. In contrast, the molecular weights of outer membrane polypeptidesranged from 17,000 to 117,000. The most prominent polypeptides had molecular weights ranging from 42,000to 117,000. There were several polypeptides in the outer membranes of bacteria grown on polysaccharides(chondroitin sulfate, arabinogalactan, or polygalacturonic acid) which were not detected or were not as

prominent in outer membranes of bacteria grown on monosaccharide components of these polysaccharides.

Bacteroides thetaiotaomicron is one of the numericallypredominant species in the human colonic microflora (23).Members of this species are obligately saccharolytic andutilize a variety of monosaccharides and polysaccharides(33). Because simple sugars are virtually unavailable in thecolon, polysaccharides in the diet of the host or in hostsecretions probably provide the main sources of carbon andenergy for these bacteria. When B. thetaiotaomicron utilizespolysaccharides such as chondroitin sulfate, laminarin, orpolygalacturonic acid, it produces polysaccharidases thatare cell associated rather than extracellular (30-32). Forexample, the enzymes which degrade chondroitin sulfate aresoluble and are located either in the periplasmic space or inthe cytoplasm (30). Because these polysaccharides are toolarge to diffuse freely through pores in the outer membranewhich admit mono- and disaccharides, there may be special-ized membrane proteins that permit polysaccharides to passthrough the outer membrane. All of the polysaccharidases ofB. thetaiotaomicron that have been studied are inducible(29, 31, 32). This could also be the case for other proteins,such as outer membrane proteins, which are necessary forcatabolism of polysaccharides. To determine whether thereare any other membrane proteins that are associated with thegrowth of B. thetaiotaomicron on polysaccharides, we firsthad to develop a method for isolating outer membranes fromthis organism.There have been three previous reports of methods for

isolating Bacteroides outer membranes (5, 16, 25). Thesemethods were unsatisfactory for our purposes. First, theseprocedures employed either EDTA or Triton X-100 com-pounds which can selectively release outer membrane com-ponents (4, 8, 19, 26, 27, 35, 38). Second, the heat treatment

* Corresponding author.

used in two of these procedures may inactivate outer mem-brane proteins which bind or degrade polysaccharides. Inaddition, heat treatment inactivates respiratory enzymesthat are needed for the detection of inner membrane con-tamination of an outer membrane preparation.We have developed a procedure for isolating both outer

and inner membranes by sucrose density centrifugation.This procedure exposes membranes only to sucrose and N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid(HEPES) buffer (10 mM; pH 7.4). In this report, we describethe separation procedure, assess the suitability of variousmarkers for inner and outer membranes, and use thesemarkers to determine the percent recovery and purity of themembrane fractions obtained. Finally, we compare the pro-tein composition of outer membranes from B. thetaiotaomi-cron grown in defined medium containing either chondroitinsulfate, polygalacturonic acid, arabinogalactan, or monosac-charide components of these polysaccharides.

MATERIALS AND METHODSBacterial strains and culture conditions. B. thetaiotaomi-

cron (VPI 5482A, NCTC 10582) was obtained from theculture collection of the Anaerobe Laboratory, VirginiaPolytechnic Institute and State University, Blacksburg.Stocks were stored in chopped meat medium (11) at roomtemperature.The growth medium was the defined medium described by

Varel and Bryant (39) except that potassium phosphate (0.1M; pH 7.0) was the sole buffer, and the gas phase was amixture of 80% nitrogen and 20% carbon dioxide. Heminwas replaced by a histidine-hemin solution (12 g of hematin[Sigma Chemical Co., St. Louis, Mo.] per 10 ml of 0.2 Mhistidine [pH 8.0]) which had been stirred for 15 h at roomtemperature, sterilized by filtration through a 0.22-tim filter,

102

on April 4, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

B. THETAIOTAOMICRON OUTER MEMBRANE PROTEINS 103

and stored under a carbon dioxide atmosphere. A 0.3-mlportion of this stock solution was added to each liter ofsterile medium. Glucose, arabinogalactan, polygalacturonicacid, or chondroitin sulfate was added to the medium beforeautoclaving. Solutions (50% [wt/vol]) of N-acetylglucosa-mine, glucuronic acid, galacturonic acid, galactose, or arabi-nose were filter sterilized and added to autoclaved medium.The final concentration of carbohydrate was 0.5% (wt/vol).

Chondroitin sulfate (type III), larch arabinogalactan, andpolygalacturonic acid were obtained from Sigma. The com-position and molecular weight of the chondroitin sulfate andarabinogalactan used in these experiments have been de-scribed previously (28, 30). Polygalacturonic acid (Sigma)eluted in the void volume of a Bio-Gel A-1.5 column. Nolower-molecular-weight material containing uronic acid wasdetectable in this preparation.

Biochemical analyses. Protein concentrations were mea-sured by a modified Lowry procedure (22). Lipids wereextracted by the method of Bligh and Dyer as described byAmes (1), and their phosphorous content was estimated bythe method of Bartlett (2). For dry weight measurements,sucrose gradient fractions of a membrane peak were pooled,diluted with distilled water, and pelleted by centrifugation ina Beckman 70 Ti rotor (65,000 rpm; 450,000 x g; 5 h). Themembranes were then washed again, resuspended in water,and finally dried at 70°C in a tared aluminum pan.

Beta-hydroxy fatty acids were used as a measure oflipopolysaccharide (LPS) in membrane fractions. Beforeanalysis, membranes in sucrose solutions were diluted to 10ml with water, collected by centrifugation in a Beckman 50Ti rotor (50,000 rpm; 225,000 x g; 3 h; 4°C), and resus-pended in 2 ml of water. LPS in the washed membranes wasobtained by phenol extraction as described by Hofstad et al.(10). The water phases from three extractions were pooled,and traces of phenol were removed by repeated extractionwith glass-distilled diethyl ether until the water phase was nolonger cloudy. The diethyl ether phases were discarded, andmaterial in the water phase was lyophilized. Fatty acids inthe lyophilized material were hydrolyzed and methylated bytransesterification (17) at 70°C for 3 h in a solution consistingof 5 ml of methanol (spectrophotometric grade; AldrichChemical Co., Inc., Milwaukee, Wis.), 0.8 ml of benzene(spectrophotometric grade; Mallincrodt Inc., Paris, Ky.),and 1 ml of HCl (sequanal grade; Pierce Chemical Co.,Rockford, Ill.).Because 3-hydroxy tetradecanoic acid is present only in

trace amounts in B. thetaiotaomicron LPS (40), the acid(Supelco, Inc., Bellefonte, Pa.) was added to the transesteri-fication mixture as an internal control for efficiency ofesterification and recovery of 3-hydroxy fatty acids. All ofthe standard 3-hydroxy tetradecanoic acid added to thetransesterification mixture was methylated under these con-ditions. The methyl esters were extracted from the transes-terification mixture three times each with 5 ml of hexane.The extracts were dried under a stream of nitrogen, and thefatty acid methyl esters were resuspended in 50,ul of hexanefor detection by gas-liquid chromatography as described inSupelco Bulletin 767A. Samples (1 to 3 [ul) were injected intoa Hewlett-Packard 5830A gas chromatograph equipped witha flame ionization detector and a glass column (10 ft [3.048m] by 2 mm by 0.25-mm outer diameter) packed with GP 3%SP-2100 DOH on 100/120 Supelcoport (Supelco).A mixture of authentic methylated fatty acids including

saturated, straight-chain fatty acids of 11 to 20 carbons,anteisobranched pentadecanoic acid, and 3-hydroxy fattvacids of 12, 14, and 17 carbons was purchased from Supelco.

This mixture and methylated 3-hydroxy hexadecanoic acidand 3-hydroxy octadecanoic acid (Alltech Associates, Inc.,Applied Science Div., State College, Pa.) were used forcalculation of the equivalent chain lengths of the methylatedfatty acids obtained from the membrane samples. Methylat-ed 3-hydroxy hexadecanoic and methylated 3-hydroxy octa-decanoic acids were used as standards for calibration of therelative molar response of the gas chromatograph for meth-ylated 3-hydroxy fatty acids.PAGE. Membranes in HEPES buffer (30 ,ug of protein)

were diluted 1:2 with solubilization solution (which consist-ed of 10o sodium dodecyl sulfate [SDS], 0.01 M Tris [pH8.0], 0.01 M EDTA, 25% P-mercaptoethanol, 6.3% glycerol,and 0.0125% bromphenol blue), boiled for 5 min,and used immediately for electrophoresis. SDS-polyacryl-amide gel electrophoresis (PAGE) was peformed by theprocedure of Laemmli and Favre (18) with the followingchanges. Sodium chloride (0.15 M) was added to the separat-ing gel which was cast as an 8 to 20% acrylamide gradient.The SDS concentration in the running buffer was 0.15%rather than 0.1%. Gels were stained with Coomassie blue R-250. Molecular weight standards (Sigma) included: ,-galac-tosidase, 116,000; phosphorylase B, 97,400; bovine serumalbumin, 66,000; catalase, 57,000; ovalbumin, 45,000; car-bonic anhydrase 29,000; chymotrypsinogen A, 25,700; myo-globin, 18,800; lactoglobulin, 18,400; and lysozyme, 14,300.Enzyme assays. All enzyme reaction mixtures (total vol-

ume, 0.5 ml) were monitored at 37°C in a Gilford recordingspectrophotometer. Succinate dehydrogenase (SDH) activi-ty was assayed by the method described by Kasahara andAnraku (13). NAD+-independent L-(+)-lactate dehydroge-nase (LDH) activity was measured as described by Macy etal. (21), except that CaCl2 and Triton X-100 were added tothe reaction mixture at final concentrations of 4 mM and 2%(wt/vol), respectively. Malate dehydrogenase (MDH) activi-ty was assayed as described in Boehringer Mannheim'stechnical booklet Biochemica Information I. P-N-acetylglu-cosaminidase activity was assayed as the increase in absor-bance at 410 nm resulting from the hydrolysis of the p-nitrophenylglycoside (3). The measured extinctioncoefficient of p-nitrophenol in potassium phosphate buffer(20 mM; pH 6.2) was 4.4 Ml-' cm-'. NADH oxidation wasassayed by the method of Osborn et al. (24).

Localization of selected enzymes. Bacteria were grown toan optical density (650 nm) of 0.7 to 0.9 in 500-ml batchcultures and harvested by centrifugation (12,000 x g; 15 min;4°C). Cell pellets were washed in ca. 300 ml of HEPES bufferand resuspended in 20 ml of the same buffer containing 4 mgeach of pancreatic RNase type A and DNase type I (Sigma).Bacteria were disrupted by passing the cell suspension twotimes through a French pressure cell (12,000 lb/in2). Celldebris and intact cells were removed by centrifugation(17,000 x g; 5 min; 4°C). Membranes were pelleted bycentrifugation in a Beckman 50 Ti rotor (49,000 rpm; 225,000x g; 2.5 h; 4°C). The membrane pellet was washed once withHEPES buffer and then once with 1 M NaCl in HEPESbuffer. Supernatant solutions and portions of each mem-brane pellet were retained at each centrifugation step andassayed for lipid phosphorus, protein, LDH, MDH, SDH, P-N-acetylglucosaminidase, and NADH oxidation as de-scribed above.Membrane separation. Figure 1 shows a schematic dia-

gram of the entire procedure for membrane separation. Allsucrose concentrations are expressed as percentages byweight/volume in HEPES buffer (10 mm; pH 7.4). Alloperations were performed at 4°C.

VOL. 158, 1984

on April 4, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

104 KOTARSKI AND SALYERS

Bacteria were harvested in the exponential growth phase(optical density at 650 nm, 0.7 to 0.9) from batch cultures (3liters) by centrifugation (12,000 x g; 15 min). Cell pelletswere resuspended in 600 ml of HEPES buffer and harvestedby centrifugation (12,000 x g; 15 min). The washed cellpellets were drained and resuspended to a 40-ml volume with10% sucrose. After addition of 7.5 mg each of RNase A andpancreatic DNase I, the cell suspension was passed oncethrough a 23-gauge needle and then twice through a Frenchpressure cell (12,000 lb/in2). Cell debris and unbroken cellswere removed by two sequential centrifugations (17,000 x g;5 min). The resulting supernatant solution was diluted to 50ml with HEPES buffer (cell extract).

Portions (15.5 ml) of the cell extract were layered ontoeach of three two-step gradients (4 ml of 70% sucrose and 18ml of 37% sucrose) and centrifuged in a Beckman SW28rotor at 27,000 rpm (140,000 x g) for 1 h. Membranes in thesolution above the 37% sucrose solution and at the 10%/37%sucrose interface ("yellow material" in Fig. 1) were collect-ed from the tubes, pooled, and diluted to a volume of 111 mlwith HEPES buffer. Samples (37 ml) were pelleted into a37% sucrose cushion (1 ml) by centrifugation in an SW28rotor (27,000 rpm; 3 h). The material that pelleted in the 37%sucrose pad was collected and diluted to 6 ml with HEPESbuffer (mixed membrane enrichment). Membranes that re-mained in the supernatant solution after this 3-h centrifuga-tion step were diluted to a volume of 175 ml with HEPESbuffer, harvested by centrifugation in a Beckman 70 Ti rotor(55,000 rpm; 300,000 x g; 3 h), and resuspended in HEPESbuffer to a final volume of 6 ml (inner membrane enrich-ment).Membranes contained in the 37 and 70% sucrose steps of

the original two-step gradients ("white material" in Fig. 1)were pooled and diluted to 111 ml with HEPES buffer.Samples (37 ml) were pelleted into a 57% sucrose cushion (1ml) by centrifugation in an SW28 rotor (27,000 rpm; 3 h).Membranes pelleted in the sucrose cushion were pooled anddiluted to 6 ml with HEPES buffer (outer membrane enrich-ment). Samples (2 ml) of the outer membrane enrichmentwere layered onto three discontinuous gradients (3 ml of70%, 9 ml of 60%, 20 ml of 50%, and 4 ml of 40% sucrose).Samples (2 ml) of the mixed membrane enrichment werelayered onto three discontinuous gradients (3 ml of 60o, 6 mlof 50%, 20 ml of 40%, and 7 ml of 30% sucrose in HEPESbuffer). Samples (2 ml) of the inner membrane enrichmentwere layered onto three discontinuous gradients (3 ml of60%, 6 ml of 50%, 15 ml of 40%, 5 ml of 35%, and 7 ml of 30%sucrose in HEPES buffer). All gradients were placed in anSW28 rotor and centrifuged at 27,000 rpm for 14 to 16 h.

Gradients were collected in 1-ml fractions from the bot-toms of the tubes. Buoyant densities of the gradient fractionswere estimated from the measured refractive index. Mem-brane bands in the gradient were located by the 280 nmabsorbance of diluted gradient fractions. Appropriate frac-tions were pooled, diluted with an equal volume of HEPESbuffer, and harvested by centrifugation in a 70 Ti rotor(65,000 rpm; 450,000 x g; 3 h). The membrane pellets wereresuspended in 40% sucrose in HEPES buffer and stored at-200C.

RESULTSChoice of markers. In our localization experiments, all of

the SDH activity and lipid phosphorus pelleted with mem-branes, and more than 85% remained with the membranepellet even after the membranes were washed with 1 MNaCl. Since SDH activity was high (0.4 ,umol/min per mg of

CELL EXTRACT E

- 37% SUCROSE

-'70% SUCROSE

140,000 x g

hr

WHITE MA

57% SUCROSE P

140,000 x g

3 hr

37% SUCROSE N KC:140110 xg

3 hr~

3300,000xgl 3 hr

INNEROUTER MEMBRANE MIXED MEMBRANE MEMBRANE b

ENRICHMENT ENRICHMENT f ENRICHMENT l

40-70% SUCROSESTEP GRADIENT

140,000 x g16 hr

30-60% SUCROSESTEP GRADIENT

140,000 x g16 hr

-.-m

30-60% SUCROSESTEP GRADIENT

140,000 x g16 hr

FIG. 1. Summary of steps in the separation of outer from innermembranes of B. thetaiotaomicron. Membrane bands collectedfrom gradients are indicated by roman numerals I through V.

membrane protein) and was stable with freezing, we chose touse it as our inner membrane marker. NADH oxidationpartitioned similarly but was too low (0.04 pumol/min per mgof membrane protein) to be useful as a marker for the innermembrane. LDH and P-N-acetylglucosaminidase have beenreported to be membrane-bound in Bacteroides fragilis (3,21). In the case of B. thetaiotaomicron, only two-thirds ofthe total LDH and P-N-acetylglucosaminidase activitiessedimented with membrane pellets. Because the activitiesthat remained in the supernatant fluid were not pelleted athigher centrifugal forces (400,000 x g; 3 h), it is not clearwhether these enzymes are integral membrane proteins of B.thetaiotaomicron. Thus, we considered them unsuitable asmarker enzymes. MDH was chosen as a marker for solubleprotein. Only 10% of the MDH activity detected in the cellextract was recovered in the first membrane pellet, and thisactivity was released when the membranes were resus-pended in buffer. MDH activity was high (6.8 ,umol/min permg of cell protein) and was stable with freezing.Because Bacteriodes LPS does not contain 2-keto-3-deox-

yoctonoate or heptose (9, 14), we decided to use 3-hydroxyfatty acids as markers for the outer membrane. However, 3-hydroxy fatty acids in the membranes of B. thetaiotaomi-cron are found not only in LPS but also in sphingolipids (37).Therefore, we measured 3-hydroxy fatty acids in the waterphase of phenol extracts of membranes (LPS enrichments)rather than in untreated membrane fractions. The gas chrom-tograms of methylated fatty acids of the LPS enrichmentscontained 13 peaks. Three of these peaks had retention timescorresponding exactly to those of authentic standards of

J. BACTERIOL.

on April 4, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

B. THETAIOTAOMICRON OUTER MEMBRANE PROTEINS 105

methylated 3-hydroxy hexadecanoic, 3-hydroxy pentade-canoic, and anteisobranched pentadecanoic acids. Thesemethylated fatty acids constituted 17, 20, and 33%, respec-tively, of the total methylated fatty acids detected in all LPSenrichments. Wollenweber et al. (40) reported similaramounts of these three methylated fatty acids in LPS fromthe same strain of B. thetaiotaomicron. Because 3-hydroxypentadecanoic and 3-hydroxy hexadecanoic acids were themajor 3-hydroxy fatty acid species and were well separatedfrom other peaks, we used the sum (in nanomoles) of thesetwo fatty acids, detected in the water phase of phenolextracts of membrane samples, as a relative measure of LPS.The method we used to quantitate the 3-hydroxy fatty acid

content was reproducible and sensitive at levels of 0.1nmol/,ul of sample injected. Success in quantiation wasdependent upon the removal of sucrose from the membranesample by centrifugation of membranes and the removal ofphenol from the water phase of phenol extracts of mem-branes with diethyl ether. If these compounds were notremoved from the samples before transesterification, theamount of detected methylated 3-hydroxy fatty acids waslowered significantly compared with methylated nonhy-droxy fatty acids, and there was considerable variation in therelative amounts of detected methylated 3-hydroxy fattyacids from duplicate samples.

Separation of membranes. Repeated attempts to isolateouter and inner membranes of B. thetaiotaomicron on asingle sucrose density gradient by methods developed forother gram-negative bacteria (6, 7, 12, 24, 34, 36) wereunsuccessful. Most of the membrane fractions obtained bythese methods were mixtures of inner and outer membranes(buoyant densities intermediate between 1.21 and 1.14g/cm3) and contained high levels of both SDH and LPS-associated 3-hydroxy fatty acids. For example, when mem-branes from disrupted bacteria were collected by centrifuga-tion onto a 70% sucrose cushion and then placed on a 30 to70% sucrose gradient for separation by isopycnic centrifuga-tion, the pure and mixed membrane bands overlapped eachother in the resulting gradient. Similarly, membrane mix-tures predominated when membranes were harvested fromspheroplasts which had been ruptured in a French pressurecell or by osmotic lysis. To remedy these problems, wedisrupted the bacteria in a hypertonic sucrose solution andperformed the entire separtion procedure at 4°C, minimizingthe formation of membrane mixtures. We also layered thecell extract onto a two-step sucrose gradient and thusobtained enrichments of outer and inner membranes beforeisopycnic centrifugation.The highest-density membrane band in the sucrose gradi-

ent from the outer membrane enrichment (I in Fig. 1 andfractions 8 to 14 in Fig. 2) was white and opaque. Itcontained little SDH activity and had a high concentration of3-hydroxy fatty acids compared with the lighter-densityband (II in Fig. 1 and fractions 24 to 30 in Fig. 2). Under thecentrifugation conditions used (140,000 x g; 1 h), the 37%sucrose concentration in the two-step gradient was criticalfor maximum yield of membrane band I. Higher concentra-tions of sucrose yielded lower amounts of both membranebands I and II. Lower concentrations of sucrose decreasedthe relative yields of band I and increased the yields of bandII. Membrane bands III and IV (Fig. 1) contained both LPS-associated 3-hydroxy fatty acids and SDH (Fig. 3). A secondisopycnic density centrifugation of membrane bands II, III,or IV did not separate them into pure outer or innermembranes. Membrane band V (Fig. 1) was transluscent andbrown-yellow in color. It contained relatively high levels of

protein, lipid phosphorus, and SDH activity but little 3-hydroxy fatty acid (fractions 20 to 30 in Fig. 4). Since itcontained relatively little MDH, it was not soluble protein.Another protein peak (fractions 30 to 35 in Fig. 4), whichcontained no detectable LPS-associated 3-hydroxy fattyacids, lipid phosphorus, or SDH activity, had high levels ofMDH activity. This peak was probably soluble proteinwhich was trapped in the membrane pellet during centrifuga-tion. This peak also contained LDH and ,B-N-acetylglucosa-minidase activity (data not shown).Membrane bands from a sucrose gradient (see Fig. 2 to 4)

were harvested by centrifugation into a 40% sucrose cush-ion. If membrane bands II through V were pelleted andstored in HEPES buffer or potassium phosphate buffer (10mM; pH 7.4) without added sucrose, the membranes aggre-gated and would not stay in solution. Although homogeniza-tion or sonication resuspended the membranes, visible ag-gregation occurred in less than an hour. No apparentaggregation was observed when membrane band I wascollected by centrifugation and resuspended in HEPESbuffer without added sucrose.Recovery and composition of membranes. Over 20% of the

LPS-associated 3-hydroxy fatty acids, less than 3% of theSDH activity, and none of the MDH activity were recoveredin membrane band I (Table 1). In contrast, membrane bandV contained over 20% of the SDH activity, less than 3% ofthe LPS-associated 3-hydroxy fatty acids, and less than 1%of the MDH activity. Membrane bands I and V differedsignificantly in protein content (Table 2). Because of thedifference in protein content of the various membrane bands,enzyme activities and LPS-associated 3-hydroxy fatty acidlevels in Table 2 were normalized to total dry weight ratherthan to protein. The levels of LPS-associated 3-hydroxyfatty acids in membrane band I were 20-fold higher than

30 -A 2.0 20q0

~ 25

;2 0 X > }~~~~~~~~~~ 12^g

150

Fraction Number

FIG. 2. Analysis of fractions from the 40 to 70% sucrose stepgradient which resolved the membranes in the outer membraneenrichment (Fig. 1). Fractions 8 through 14 (corresponding to band Iin Fig. 1) and fractions 24 through 30 (corresponding to band II inFig. 1) were pooled separately for further analysis.

VOL. 158, 1984

on April 4, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

106 KOTARSKI AND SALYERS

those in membrane band V. SDH activity (normalized to dryweight) was ca. 10-fold higher in membrane band V than inmembrane band I. Based on the distribution of membranemarkers and on bouyant densities (Tables 1 and 2), weidentified membrane band I as outer membrane, membraneband V as inner membrane, and membrane bands II to IV asmixtures of inner and outer membrane.Membrane protein profiles. The SDS-PAGE profiles of

each membrane band are shown in Fig. 5. The polypeptideprofile of membrane band I was complex and containedapproximately 30 polypeptides ranging in molecular weightfrom 17,000 to 117,000 (Fig. 5). Among these, four polypep-tides predominated. These polypeptides had molecularweights which were estimated to be 41,000, 56,000, 62,000,and 102,000. They were evident in membrane bands II andIII and, to a lesser extent, in membrane band IV. Theappearance of these polypeptides in membrane bands II toIV was consistent with the levels of LPS-associated 3-hydroxy fatty acids detected in membrane bands II, III, andIV (Fig. 2 and 3). By contrast, the polypeptides of membraneband V ranged in molecular weight from 11,000 to 133,000.The polypeptides which ranged in molecular weight from11,000 to 26,000 were evident only in membrane bands IVand V. The same polypeptide profiles were obtained ifphenylmethylsulfonyl fluoride (0.05 mM) was includedthroughout the preparation procedure.

Effect of carbohydrate source. SDS-PAGE profiles of poly-peptides of outer membrane preparations from B. thetaio-taomicron grown on different carbohydrates are shown inFig. 6. All outer membrane preparations had similarly lowlevels of SDH activity (less than 100 nmol/min per mgequivalent [dry weight]). When bacteria were grown on N-acetylglucosamine, glucuronic acid, or galacturonic acid, theprofiles of outer membrane polypeptides were virtually

U)T

C.)1~

2.0-

g. 1.0

5 10 15 20 25

Fraction Number

15.0 t-1rl3.0

30 35

112 ,

108

FIG. 3. Analysis of fractions from the 30 to 60% sucrose stepgradient which resolved the membranes in the mixed membraneenrichment (Fig. 1). Fractions 2 through 13 (corresponding to bandIII in Fig. 1) and fractions 25 through 30 (corresponding to band IVin Fig. 1) were pooled separately for further analysis.

Q2

0l0.1-1

5 10 15 20 25 30 35

Fraction Number

U,0nf-: Z1

.16

12

108:

FIG. 4. Analysis of fractions from the 30 to 60% sucrose stepgradient which resolved the membranes in the inner membraneenrichment (Fig. 1). Fractions 20 through 31 (corresponding to bandV in Fig. 1) were pooled separately for further analysis.

identical. When bacteria were grown on neutral sugars, suchas glucose, galactose, or arabinose, the pattern of outermembrane polypeptides was similar, but not identical, to thepolypeptide banding pattern associated with growth on theother monosaccharides. In particular, bacteria grown onglucose had a prominent polypeptide with a molecule weightof ca. 100,000.There were significant differences among the polypeptide

profiles of outer membrane preparations of bacteria grownon different polysaccharides. In outer membrane prepara-tions of bacteria grown on chondroitin sulfate, there wereseveral outer membrane polypeptides (Mr, ca. 24,000,30,000, 63,000, 69,000, 93,000, and 104,000) which were notas prominent in outer membranes of bacteria grown onglucuronic acid or N-acetylglucosamine. N-acetylgalactos-amine and glucuronic acid are the monosaccharide compo-

TABLE 1. Recovery of various markers in pooled fractions fromthe sucrose gradients (Fig. 2 to 4)

% of total marker detected in cell extractsaMembrane

band SDH 3-Hydroxy MDHfatty acids

I 2, 3b 21, 26 0, 0II 14, 5 15, 8 0, 0III 22, 25 25, 32 0, 0IV 27, 16 5, 2 <0.1, <0.1V 22, 49 3, 2 <0.1, <0.1

a Percentage of total = (amount of enzyme [or LPS-associated 3-hydroxy fatty acids] in membrane band/total amount of enzyme [orfatty acids] in cell extract) x 100.

b Each value is the mean of at least duplicate determinations. Twovalues, from different membrane separation experiments, areshown.

J. BACTERIOL.

on April 4, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

B. THETAIOTAOMICRON OUTER MEMBRANE PROTEINS 107

TABLE 2. Comparison of pooled fractions from the sucrosegradients (Fig. 2 to 4)

Protein Lipid 3-Hydroxy SDH

band (g/cm3)a D(mg/mgb (g/mg (nmol/mg (nmol/min perband (g/cm)[dry wt] [dry wtJ) [dry wtJ)C mg [dr wtI)c

I 1.21 0.34 4.5, 4.1 21, 22 51, 95II 1.19 0.42 5.4, 4.8 14, 12 411, 270II 1.18 0.47 6.1, 4.2 16, 12 460, 221IV 1.15 0.57 6.2, 4.7 5, 3 730, 616V 1.14 0.68 4.2, 4.6 1, 3 564, 694

a Density was estimated from sucrose gradients as described inthe text.

b Values are the mean of duplicate determinations. Variation fromone membrane separation experiment to another was less than 10%.

Each value is the mean of at least duplicate determinations. Twovalues, from different membrane separation experiments, areshown.

nents of chondroitin sulfate. We used N-acetylglucosaminefor the present comparison because N-acetylgalactosamineis too expensive for use in large-scale culturing of bacteria.When bacteria were grown on polygalacturonic acid, therewere two outer membrane polypeptides (Mr, 63,000 and104,000) which were more prominent than those in outermembrane preparations of bacteria grown on galacturonicacid, the main component of polygalacturonic acid. In outermembrane preparations of bacteria grown on arabinogalac-tan, three polypeptides (Mr, 84,000, 67,000, and 37,000) werenot as prominent in the outer membrane preparations ofbacteria grown on arabinose or galactose, the components ofarabinogalactan.

DISCUSSIONUsing the procedure outlined in Fig. 1, we separated outer

from inner membranes of B. thetaiotaomicron without ex-posing the membranes to EDTA, detergent, or heat. Yieldsof purified outer and inner membranes were at least 20%.The inner membrane preparation contained less than 5%contamination by outer membranes, as estimated by theamount of LPS-associated 3-hydroxy fatty acids. To ourknowledge, this is the first report of isolation of Bacteroidesinner membranes. The outer membrane preparation con-tained ca. 10% contamination by inner membranes, as

I 11 III IV V

116K-_ 97K--§ = a. w ~~~~66K5jS7K

|= w -~~~~~~45K

-26K

5q -14K

FIG. 5. Comparison of SDS-PAGE profiles of polypeptides inpooled gradient fractions corresponding to bands I through V in Fig.1. Mobilities of the molecular weight standard are indicated byarrows.

GicNAc GIcA CS GalA PG GIc AG Gal Ara

_e4 _~116K97K

w_ 9 q _ 1 =-! i 5 ~~-66K_5~~~~~~~~~~~~~7

==;- > _~~-45 K

-, -26K

-18K

-14K

FIG. 6. Comparison of SDS-PAGE profiles of polypeptide inouter membranes obtained from B. thetaiotaomicron which hadbeen grown on N-acetylglucosamine (GlcNAc), glucuronic acid(GlcA), chondroitin sulfate (CS), galacturonic acid (GalA), polyga-lacturonic acid (PG), glucose (Glc), arabinogalactan (AG), galactose(GAL), or arabinose (Ara). Mobilities of the molecular weightstandards are indicated by arrows.

estimated by the amount of SDH in this preparation. Thedensities of the inner and outer membrane preparations of B.thetaiotaomicron (1.14 g/cm3 and 1.21 g/cm3, respectively)are similar to the densities of inner and outer membranesfrom other gram-negative bacteria. The density of our outermembranes preparation is lower than that repcvrted byKasper et al. (15) for an outer membrane complex from B.fragilis. The procedure of Kasper et al., particularly the heattreatment, could have produced aggregates which migratedat higher densities than the outer membrane preparationsdescribed here. Alternatively, the difference is densitiescould reflect differences in membrane composition betweenthese two Bacteroides species.The polypeptide pattern of our outer membrane prepara-

tion was similar, but not identical, to the polypeptide patternof the Triton X-100 insoluble material obtained by Diedrichand Martin from B. thetaiotaomicron ATCC 29148 (5).These differences may have been due to differences inmedium composition. Diedrich and Martin grew their bacte-ria in complex media rather than the defined medium used inour studies; however, they did not use inner membranemarkers to estimate the purity of the outer membranepreparations. Therefore, we could not evaluate whetherthese differences in polypeptide profiles were due to differ-ent levels of inner membrane contamination in the twopreparations.There were several polypeptides in the outer membrane

preparations of cells grown on chondroitin sulfate, polyga-lacturonic acid, or arabinogalactan which were either lessprominent or not detectable in outer membrane preparationsof bacteria grown on component monosaccharides of thesepolysaccharides. One polypeptide in particular (Mr, 104,000)was prominent only in the outer membrane preparations ofbacteria which had been grown on one of the two negativelycharged polysaccharides, chondroitin sulfate, or polygalac-turonic acid. These polypeptides were probably not the cell-associated polysaccharidases which have been found previ-ously in these organisms. The two chondroitin sulfate lyases(Mr of 104,000 and 107,000) which are produced when B.thetaiotaomicron is grown on chondroitin sulfate are solubleenzymes (20). B. thetaiotaomicron produces both solubleand membrane-bound polygalacturonases when grown on

VOL. 158, 1984

on April 4, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

108 KOTARSKI AND SALYERS

polygalacturonic acid. However, the membrane-bound ac-tivity appears to be located in the inner membrane (R.McCarthy and A. Salyers, unpublished data). Some of theseouter membrane polypeptides could be enzymes whichdegrade the polysaccharide into large fragments. Such poly-saccharidases may have been missed in earlier studies of thelocation of polysaccharide-degrading activities because ofrapid breakdown by the soluble enzymes. Alternatively,some of these outer membrane polypeptides may function asbinding proteins or as porins which may act in concert tofacilitate the passage of these polysaccharides through theouter membrane. These polypeptides may not be the onlyones which are necessary for utilization of polysaccharides.We have not ruled out the possibility that there are outermembrane polypeptides which are involved in polysaccha-ride utilization but which are produced constitutively or arenot easily detected by one-dimensional SDS-PAGE.

ACKNOWLEDGMENTSWe thank Diane Braun for excellent technical assistance.This work was supported by Public Health Service grant AI-17876

from the National Institute of Allergy and Infectious Diseases.

LITERATURE CITED

1. Ames, G. F. 1968. Lipids of Salmonella typhimurium andEscherichia coli: structure and metabolism. J. Bacteriol.95:833-843.

2. Bartlett, E. R. 1959. Phosphorous assay in column chromatogra-phy. J. Biol. Chem. 234:466-468.

3. Berg, J.-Q. 1981. Cellular localization of glycoside hydrolases inBacteroides fragilis. Curr. Microbiol. 5:13-17.

4. DePamphilis, M. L., and J. Adler. 1971. Attachment of flagellarbasal bodies to the cell envelope: specific attachment to theouter, lipopolysaccharide membrane and the cytoplasmic mem-brane. J. Bacteriol. 105:396-407.

5. Diedrich, D. L., and A. E. Martin. 1981. Outer membraneproteins of the Bacteroides fragilis group. Curr. Microbiol.6:85-88.

6. Ding, D. H., and S. Kaplan. 1976. Separation of inner and outermembranes of Rhodopseudomonas sphaeroides. Prep. Bio-chem. 6:61-79.

7. Hancock, R. E. W., and H. Nikaido. 1978. Outer membranes ofgram-negative bacteria. XIX. Isolation from Pseudomonas aer-uginosa PAO1 and use in reconstitution and definition of thepermeability barrier. J. Bacteriol. 136:381-390.

8. Hardaway, K. L., and C. S. Buller. 1979. Effect of ethylenedia-minetetraacetate on phospholipids and outer membrane func-tion in Escherichia coli. J. Bacteriol. 137:62-68.

9. Hofstad, T., and T. Kristoffersen. 1970. Chemical characteristicsof endotoxin from Bacteroides fragilis NCTC 9343. J. Gen.Microbiol. 61:15-19.

10. Hofstad, T., K. Sveen, and G. Dahlen. 1977. Chemical composi-tion, serological reactivity and endotoxicity of lipopolysaccha-rides extracted in different ways from Bacteroides fragilis,Bacteroides melaninogenicus, and Bacteroides oralis. Acta.Pathol. Microbiol. Scand. Sect. B 85:262-270.

11. Holdeman, L. V., and W. E. C. Moore (ed.). 1977. Anaerobelaboratory manual, 4th ed. Virginia Polytechnic Institute andState University, Blacksburg, Va.

12. Johnston, K. H., and E. C. Gotschlich. 1974. Isolation andcharacterization of the outer membrane of Neisseria gonor-rhoeae. J. Bacteriol. 119:250-257.

13. Kasahara, M., and Y. Anraku. 1974. Succinate dehydrogenaseof Escherichia coli membrane vesicles. J. Biochem. 76:959-966.

14. Kasper, D. L. 1976. Chemical and biological characterization ofthe lipopolysaccharide of Bacteroides fragilis subspecies fragi-lis. J. Infect. Dis. 134:59-66.

15. Kasper, D. L., A. B. Onderdonk, B. F. Polk, and J. G. Bartlett.1979. Surface antigens as virulence factors in infection with

Bacteroides fragilis. Rev. Infect. Dis. 1:278-288.16. Kasper, D. L., and M. W. Seiler. 1975. Immunochemical charac-

terization of the outer membrane complex of Bacteroides fragi-lis subspecies fragilis. J. Infect. Dis. 132:440-450.

17. Kates, M. 1964. Simplified procedure for hydrolysis or methano-lysis of lipids. J. Lipid Res. 5:132-135.

18. Laemmli, U. K., and M. Favre. 1973. Maturation of the head ofbacteriophage T4. I. DNA packaging events. J. Mol. Biol.80:575-599.

19. Leive, L. 1965. Release of lipopolysaccharide by EDTA treat-ment of Escherichia coli. Biochem. Biophys. Res. Commun.21:290-296.

20. Linn, S., T. Chan, L. Lipeski, and A. A. Salyers. 1983. Isolationand characterization of two chondroitin lyases from Bacteroidesthetaiotaomicron. J. Bacteriol. 156:859-866.

21. Macy, J. M., L. G. Ljungdahl, and G. Gottschalk. 1978. Path-way of succinate and propionate formation in Bacteroidesfragilis. J. Bacteriol. 134:84-91.

22. Markwell, M. A. K., S. M. Haas, L. L. Bieber, and N. E.Tolbert. 1978. A modification of the Lowry procedure tosimplify protein determination in membrane and lipoproteinsamples. Anal. Biochem. 87:206-210.

23. Moore, W. E. C., and L. V. Holdeman. 1974. Human fecal flora:the normal flora of 20 Japanese-Hawaiians. Appl. Microbiol.27:961-979.

24. Osborn, M. J., J. E. Gander, E. Parisi, and J. Carson. 1972.Mechanism of assembly of the outer membrane of Salmonellatyphimurium. Isolation and characterization of cytoplasmic andouter membrane. J. Biol. Chem. 247:3962-3972.

25. Poxton, I. R., and R. Brown. 1979. Sodium dodecyl sulphate-polyacrylamide gel electrocphoresis of cell surface proteins asan aid to the identification of the Bacteroides fragilis group. J.Gen. Microbiol. 112:211-217.

26. Roberts, N. A., G. W. Gray, and S. G. Wilkinson. 1970. Thebacterial action of ethylenediaminetetra-acetic acid on Pseudo-monas aeruginosa. Microbios 2:189-208.

27. Rogers, S. W., H. E. Gilleland, Jr., and R. G. Eagon. 1969.Characterization of a protein-lipopolysaccharide complex re-leased from cell walls of Pseudomonas aeruginosa by ethylene-diaminetetraacetic acid. Can. J. Microbiol. 15:743-748.

28. Salyers, A. A., R. Arthur, and A. Kuritza. 1981. Digestion oflarch arabinogalactan by a strain of human colonic Bacteroidesgrowing in continuous culture. J. Agric. Food Chem. 29:475-480.

29. Salyers, A. A., and S. F. Kotarski. 1980. Induction of chondroi-tin sulfate lyase activity in Bacteroides thetaiotaomicron. J.Bacteriol. 143:781-788.

30. Salyers, A. A., and M. O'Brien. 1980. Cellular location ofenzymes involved in chondroitin sulfate breakdown by Bacte-roides thetaiotaomicron. J. Bacteriol. 143:772-780.

31. Salyers, A. A., J. K. Palmer, and T. D. Wilkins. 1977. Laminar-inase (,-glucanase) activity in Bacteroides from the humancolon. Appl. Environ. Microbiol. 33:1118-1124.

32. Salyers, A. A., J. K. Palmer, and T. D. Wilkins. 1978. Degrada-tion of polysaccharides by intestinal bacterial enzymes. Am. J.Clin. Nutr. 31:S128-S130.

33. Salyers, A. A., J. R. Vercellotti, S. E. H. West, and T. D.Wilkins. 1977. Fermentation of mucin and plant polysaccharidesby strains of Bacteroides from the human colon. Appl. Environ.Microbiol. 33:319-322.

34. Schnaitman, C. A. 1970. Protein composition of the cell wall andcytoplasmic membrane of Escherichia coli. J. Bacteriol.104:890-901.

35. Schnaitman, C. A. 1971. Effect of ethylenediaminetetraaceticacid, Triton X-100, and lysozyme on the morphology andchemical composition of isolated cell walls of Escherichia coli.J. Bacteriol. 108:553-563.

36. Smit, J., Y. Kamio, and H. Nikaido. 1975. Outer membrane ofSalmonella typhimurium: chemical analysis and freeze-fracturestudies with lipopolysaccharide mutants. J. Bacteriol. 124:942-958.

37. Stoffel, W., K. Dittmar, and R. Wilmes. 1975. Sphingolipidmetabolism in Bacteroideaceae. Hoppe-Seyler's Z. Physiol.

J. BACTERIOL.

on April 4, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

B. THETAIOTAOMICRON OUTER MEMBRANE PROTEINS 109

Chem. 356:715-725.38. van Alphen, L., T. Riemens, J. Poolman, and H. C. Zanen. 1983.

Characteristics of major outer membrane proteins ofHaemophi-lus influenzae. J. Bacteriol. 155:878-885.

39. Varel, V. H., and M. P. Bryant. 1974. Nutritional features ofBacteroides fragilis subsp. fragilis. Appl. Microbiol. 28:251-

257.40. Wollenweber, H.-W., E. T. Rietschel, T. Hofstad, A. Weintraub,

and A. A. Lindberg. 1980. Nature, type of linkage, quantity, andabsolute configuration of (3-hydroxy) fatty acids in lipopolysac-charides from Bacteroides fragilis NCTC 9343 and relatedstrains. J. Bacteriol. 144:898-903.

VOL. 158, 1984

on April 4, 2020 by guest

http://jb.asm.org/

Dow

nloaded from