JOURNAL OF BACTERIOLOGY, Sept. 1991, p. 5677-5684 Vol. 173, No. 180021-9193/91/185677-08$02.00/0Copyright C) 1991, American Society for Microbiology

Identification, Isolation, and Structural Studies of ExtracellularPolysaccharides Produced by Caulobacter crescentus

NEIL RAVENSCROFT,1t STEPHEN G. WALKER,1 GUY G. S. DUTTON,2 AND JOHN SMITl*Departments of Microbiology' and Chemistry,2 University of British Columbia,

Vancouver, British Columbia V6T IZ3, Canada

Received 25 March 1991/Accepted 6 July 1991

Caulobacters are adherent prosthecate bacteria that are members of bacterial biofouling communities inmany environments. Investigation of the cell surface carbohydrates produced by two strains of the freshwaterCaulobacter crescentus, CB2A and CB15A, revealed a hitherto undetected extracellular polysaccharide (EPS)or capsule. Isolation and characterization of the EPS fractions showed that each strain produced a uniqueneutral EPS which could not be readily removed from the cell surface by washing. Monosaccharide analysisshowed that the main CB2A EPS contained D-glucose, D-gulose, and D-fucose in a ratio of 3:1:1, whereas theCB15A EPS fraction contained D-galactose, D-glucose, D-mannose, and D-fucose in approximately equalamounts. Methylation analysis of the main CB2A EPS showed the presence of terminal glucose and gulosegroups, 3-linked fucosyl, and two 3,4-linked glucosyl units, thus confirming the pentasaccharide repeating unitindicated by 'H nuclear magnetic resonance analysis. Similar studies of the CB15A EPS revealed atetrasaccharide repeating unit consisting of terminal galactose, 4-linked fucosyl, 3-linked glucosyl, and3,4-linked mannosyl residues. EPS was not detectable by thin-section electron microscopy techniques,including some methods designed to preserve or enhance capsules, nor was the EPS readily detected on the cellsurface by scanning electron microscopy when conventional fixation techniques were used; however, astructure consistent with EPS was revealed when samples were prepared by cryofixation and freeze-substitution methods.

Caulobacter crescentus is a gram-negative stalk-formingbacterium which is typically found in natural settings at-tached to surfaces via an adhesive holdfast organelle locatedat the distal tip of the stalk (33). The strategy for growth anddispersal of cells for this organism involves a cell differenti-ation process by which a motile dispersal (swarmer) cell isproduced. Swarmer cells express a single flagellum, pili, andthe holdfast, all of which are located at one cell pole. At aspecific time in the life cycle, all the polar features of theswarmer cell except for the holdfast are lost and a stalkdevelops at the same pole as the swarmer differentiates intoa stalked cell (33, 34, 37). Throughout the entire life cycle,the cell is completely covered with a protein surface array (Slayer) distal to the outer membrane, composed of geometri-cally arranged subunits consisting of a single protein (39, 40).We were interested in the surface molecules of caulobac-

ters on the presumption that they all participate in some wayin the persistence of caulobacters in microbial biofilm com-munities. Clearly, the adhesive holdfast has a primary role,being involved with surface attachment, and we have re-ported initial analyses for the holdfast of freshwater andmarine species (28-30). The organelle appears to be acomplex polysaccharide whose composition varies some-what between species (26, 28). A continued investigationinto the chemical nature of the holdfast has required anunderstanding of other surface polysaccharides produced bythe bacteria, since most or all are likely to be present invastly greater amounts than the minute holdfast organelleand therefore represent significant potential contaminants.Moreover, lipopolysaccharides (LPSs) and extracellular

* Corresponding author.t Present address: Department of Chemistry, University of Ca-

petown, Rondebosch, 7700, South Africa.

polysaccharides (EPSs), if present, are likely to have somerole in the competitive environment in biofilm communities,and a cataloging of their existence and composition is aprerequisite for an analysis of potential roles.Although it is presumed that there is an LPS in the outer

membrane of this bacterium and some efforts have beenundertaken to directly establish its presence (3, 8), therehave been no reports that an EPS or capsule is present on thesurface of C. crescentus strains. This is despite the fact thatthis organism has been the subject of numerous ultrastruc-tural studies (33, 36, 37, 40).EPS layers are a common cell surface component of many

bacterial species, but these highly hydrated structures oftenevade detection during ultrastructural studies because oftheir tendency to collapse or to be extracted during prepa-ration for light or electron microscopy (5, 10). Capsule layersare often revealed by electron microscopy only after thestructures are stabilized by various techniques such asantibody treatment followed by chemical dehydration andembedding in particular plastics (e.g., Lowicryl K4M), theuse of cationic ferritin, and cryofixation and freeze-substitu-tion techniques (5).

Previously, we had only slight hints that a capsule or EPSmay be present in C. crescentus strains. Smit et al. (40)suggested that a cryptic EPS layer may have been responsi-ble for the difficulties in obtaining high-quality images of theS layer on C. crescentus CB15 by freeze-etch analysis. TheIndia ink stain exclusion technique for capsules indicated athin particle-excluding layer surrounding caulobacter cells(2). However, several thin-section fixation and embeddingtechniques, including a technique recently used by Grahamet al. (17) involving cryofixation technology, revealed the Slayer on the membrane surface but little else. Finally, someunusual preservation techniques for scanning electron mi-

5677

on April 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

5678 RAVENSCROFT ET AL.

croscopy (SEM), reported here, provided an impetus for adirected search for an EPS.During an investigation of caulobacter LPS, we isolated a

high-molecular-weight carbohydrate fraction which waschemically distinct from the LPS and had physical propertiesconsistent with an EPS. In this report, we outline theisolation, purification, and chemical characterization of theEPS present on two C. crescentus strains.

MATERIALS AND METHODS

Bacterial strains and growth conditions. C. crescentusCB2A and CB15A (39) were grown in a peptone-yeastextract (PYE) medium (33) at 30°C. For EPS isolation,500-ml cultures in 2-liter flasks were grown with shaking at200 rpm, harvested by centrifugation during late-log phase(optical density at 600 nm, 0.7 to 0.8), and washed with 0.1M HEPES (N-2-hydroxyethylpiperazine N'2-ethanesulfonicacid) buffer, pH 7.2, by suspension and centrifugation.EPS isolation. EPS was isolated as a by-product of an LPS

isolation strategy involving a modification of the procedureof Darveau and Hancock (11). After the nuclease digestionof cells disrupted in a French pressure cell, the cell lysatewas made up to contain 0.1 M EDTA, 2% sodium dodecylsulfate, and 10 mM Tris-HCl (pH 8.0) and then incubated at37°C for 2 h. The published procedure was followed until thecompletion of the ultracentrifugation procedure, in whichthe precipitate of an ethanol precipitation step was pelleted.The pellet was suspended in 10 mM Tris-HCl and centri-fuged at 200,000 x g for 2 h at 15°C, which pelleted the LPS.The characterization of the LPS from Caulobacter strainswill be presented elsewhere (42). A large amount of solublepolysaccharide was obtained from the supernatant of the lastLPS sedimentation step. The supernatant was dialyzedagainst distilled deionized water (DDW) and then freeze-dried. This sample was considered to be the crude EPSfraction.An initial examination of the crude EPS fraction by

steric-exclusion chromatography (SEC) on a SephacrylS-400 column (60 by 2 cm) eluted with 0.1 M pyridiniumacetate buffer (pH 7.0), followed by carbohydrate analysis ofthe major peaks, revealed contamination from undegradedRNA. In subsequent preparations, therefore, the crude EPSwas resuspended in DDW to 1/10 the original volume,treated with RNase, dialyzed against DDW, and ultracentri-fuged again at 200,000 x g for 30 h at 4°C to remove anyremaining LPS.

Colorimetric assays. Protein was determined by themethod of Markwell et al. (27), 3-deoxy-2-octulosonic acid(KDO) was determined by the method of Karkhanis et al.(24), inorganic phosphate was determined by the method ofAmes and Dubin (4), and sugars were determined by thephenol-sulfuric acid method of Dubois et al. (12).

Carbohydrate analysis. Polysaccharides were hydrolyzedwith 2 M trifluoroacetic acid (TFA) at 100°C for 12 to 18 h orwith 4 M TFA at 1200C for 1 h. The hydrolysates werereduced with NaBH4 in water at room temperature for atleast 1 h. Excess NaBH4 was destroyed with glacial aceticacid, and the sample was coevaporated with 10% acetic acidin methanol (3 times) and then with methanol (three times) toremove the borate. The sodium acetate generated was usedas the catalyst in the acetylation reaction, which wasachieved with acetic anhydride at 1000C for 1 h. In someinstances, pyridine was added to ensure acetylation. Thederived alditol acetates were extracted into chloroform andwashed sequentially with 10% sulfuric acid, water, saturated

sodium hydrogen carbonate, and water (the final wash withwater was done 5 times). Gas chromatography (GC) analysiswas performed after drying the chloroform extract withanhydrous sodium sulfate. Following the procedure recom-mended by Dudman et al. (13), methanolysis was used toprovide simultaneous determination of acidic and neutralsugars.

Methylations were carried out by the Hakomori method(18) as modified by Phillips and Fraser (32), with 2 Mpotassium methyl sulfinyl methanide, prepared by the addi-tion of dry dimethyl sulfoxide to dry KOH at 0°C, used as abase (19). Samples (5 to 20 mg) were deionized with Amber-lite IR-120 (H+) resin, freeze-dried, and then vacuum-driedprior to methylation. When complete methylation was notachieved, further methylation was carried out by the Purdiemethod (35). Methylated polysaccharide samples were re-covered by dialysis and analyzed by GC-mass spectrometry(MS) after hydrolysis (4 M TFA at 120°C for 1 h) andconversion of the aldoses into partially O-methylated alditolacetates.

Smith degradation experiments (1) were conducted onboth EPS samples. The polysaccharide sample (10 to 20 mg)was treated with 0.1 M sodium periodate (2 to 4 ml) in thedark at room temperature. After 3 days, the excess periodatewas reacted with ethane-1,2-diol and the solution was dia-lyzed for 2 days. Reduction of the oxidized polysaccharidewith an excess of NaBH4 over 2 days followed by Smithhydrolysis (1 M TFA at room temperature for 4 days) andrecovery of the polymeric material by dialysis (EPS ofCB2A) or SEC (EPS of CB15A) yielded the Smith-degradedEPS fractions.The neutral and amino sugars of the samples were deter-

mined by GC of the alditol acetate derivatives on a Hewlett-Packard 5890A gas chromatograph fitted with dual flameionization detectors and a 3392A recording integrator. Sep-arations were performed on a fused silica DB17 capillarycolumn (12 m by 0.25-mm inner diameter; J & W Scientific,Rancho Cordova, Calif.) by using program A (180°C for 2min and then a heating rate of 5°C/min to bring the temper-ature to 240°C), B (160°C for 2 min and then a heating rate of1.5°C/min to 240°C), or C (170°C for 10 min and then aheating rate of 15°C/min to 240°C). Where possible, identifi-cation and quantitation were made by comparing the sugarsto authentic standards, with inositol as the internal standard.Alditol acetate derivatives of the glycosyl residues were alsoidentified by GC-MS by using the DB17 capillary column ona Varian Vista 600 series GC coupled directly to a SelsiNermag R10-10C quadrupole MS. The peaks eluted wereionized by electron impact and chemical ionization (usingammonia) MS in order to obtain both the detailed fragmen-tation pattern (23) and the molecular weight (21) of thecompounds, respectively.The absolute configurations of the sugars were assigned by

comparing the retention times of their trimethylsilylated(-)-2-butyl glycosides with those of the standards (15). Thepolysaccharide samples (2 mg) were solubilized in metha-nolic 4 M HCl for 24 h at 85°C. HCl was removed by theaddition of tert-butyl alcohol followed by evaporation undera stream of nitrogen. The methyl glycosides were vacuum-dried overnight and then treated with (-)-2-butanolic 1 MHCl (1 ml) for 8 h at 80°C. After the removal of the HCI asbefore, the residues were vacuum-dried overnight and thentreated with hexamethyldisilazane-chlorotrimethylsilane-pyridine (0.4 ml, 1:1:5) for 30 min at room temperature.Standards of the trimethylsilylated (+)- and (-)-2-butylglycosides of L-fucose, D-mannose, D-glucose, D-galactose,

J. BACTERIOL.

on April 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CAULOBACTER EXTRACELLULAR POLYSACCHARIDES 5679

and D-gulose were prepared as described above, except thatthe methanolysis step was not required. GC with the DB-17column and heating program C was performed. (+)-Alkylglycosides have the same elution profiles as their (-)-alkylglycoside enantiomers on non-chiral stationary phases; thus,both the enantiomers of each sugar were accounted for.400 MHz 'H nuclear magnetic resonance (NMR) spectra

were recorded on a Bruker WH-400 spectrometer at 80 to90°C. 'H chemical shifts were measured with reference tointernal acetone at 8 2.23. Samples were prepared for NMRby dissolving them in 99.7% D20 after freeze-drying themthree times from D20 solutions.

Electron microscopy techniques. (i) Thin-section transmis-sion electron microscopy (TEM). Strains CB15A and CB2Awere fixed and processed for thin-section electron micros-copy by conventional methods (including treatment withruthenium red) essentially as previously described (40). Wealso prepared cells by the cryofixation and freeze-substitu-tion methods of Graham and Beveridge (16). Thin sectionswere prepared, mounted on Parlodion-coated, carbon-stabi-lized grids, and stained with aqueous uranyl acetate and leadcitrate. Grids were examined in a Siemens lOlA transmis-sion electron microscope, operated at 60 kV.

(ii) SEM. PYE (25 ml) was inoculated with 25 ,ul of a

stationary-phase culture of CB2A. Pieces of fused aluminumoxide particles prepared from a ceramic foam product pro-vided by the Selee Corporation (Hendersonville, N.C.) were

also added, and incubation continued overnight. The culturewas decanted from the ceramic pieces, and the pieces were

washed twice by the addition and decantation of 1 mMMgCl2-1 mM CaCl2 (Mg-Cl). The washed ceramic was thensuspended in 100 ml of Mg-Cl and incubated at room

temperature without shaking for 3 h.The ceramic pieces with attached bacteria were then

processed for SEM in two ways. Conventional processingwas done by placing the ceramic pieces in 2.5% glutaralde-hyde in buffer (0.1 M sodium cacodylate [pH 7.5], 5 mMCaCl2, 5 mM MgCl2) and incubating them for 18 h at 4°C.The samples were washed three times with buffer and placedin 1% osmium tetroxide in buffer for 2 h at 4°C. The sampleswere then washed three times with buffer, dehydrated bysequential transfer to increasing concentrations of acetoneup to anhydrous acetone, critical-point dried from C02, andsputter coated with gold.By adapting the method of Graham and Beveridge (16) for

SEM, a cryofixation and freeze-substitution method was

also used. Pieces of the ceramic with bound cells were

removed from the Mg-Cl buffer and immediately plungedinto liquid propane held at - 196°C by liquid nitrogen.The frozen ceramic was transferred to glass vials held at- 196°C containing a substitution cocktail (2% osmiumtetroxide, 1% uranyl acetate, and 5% acidified 2,2'-di-methoxypropane [0.5 ,ul of concentrated HCI per ml ofdimethoxypropane] in anhydrous acetone). The dimethoxy-propane was incorporated just before use, since it has some

reactivity with osmium tetroxide. The vials were placed in a

-70°C freezer for 72 h, transferred to -20°C for 1 h, warmedto room temperature, washed four times with anhydrousacetone, and then critical-point dried and sputter coated as

above.Samples were examined with an ETEC Autoscan scanning

electron microscope operated at 20 kV and photographedwith T-Max 400 film (Eastman Kodak, Rochester, N.Y.).

0

0)at0

4

04osL

0)c]0

m

0 ' 5 1 0 1 5 20 T 25

dextran fraction No. GlucoseFIG. 1. Fractionation of the EPS of CB2A (E) and CB15A (*) on

Sephacryl S-400. Carbohydrate monitored by the phenol-sulfuricacid assay (optical density at 490 mm).

RESULTS

Isolation and chemical characterization of EPS from CB2Aand CB15A. Fractionation of the crude EPS by SEC (Fig. 1)yielded similar profiles for both strains. Carbohydrate anal-ysis (Table 1) showed that the void volume peak contained aheteropolysaccharide, whereas the second peak containedonly ribose. Additionally, the second peak contained signif-icant amounts of phosphate, and the peak was thereforeattributed to undegraded RNA. The CB2A crude EPS alsocontained an additional minor peak (B) between the het-eropolysaccharide (A) and the RNA (C) peaks. On the basisof exclusion limits reported for Sephacryl S-400, the appear-ance of the EPS peaks in the void volume suggests aminimum molecular mass of one to two million daltons.The results displayed in Table 1 show that the EPS of

CB2A and CB15A differed in monosaccharide composition.The EPS produced by CB2A contained glucose, fucose, andan initially unknown component which eluted early in theGC analysis of the derived alditol acetates (Fig. 2). Thiscompound was identified by GC-MS analysis as a 1,6-anhydroaldopyranose derivative. Chemical-ionization MSusing ammonia gave a molecular weight of 288, typical of atri-O-acetylated anhydrohexose derivative (21), while moredetailed fragmentation by electron impact MS (20) showed

TABLE 1. Carbohydrate composition of Caulobacter EPS

Sugara CB2A CB15AI II III IV (V)

Anhydrohexose 12 9 15 5 NDCRibose 28 21 ND 4 NDFucose 12 23 14 trd 17Mannose 5 4 ND 42 20Glucose 43 43 71 41 31Galactose tr ND ND 8 32

a Analysis of derived alditol acetate derivatives by GC-MS, determined ona DB17 capillary column by using heating program A.

b I, crude EPS fraction of CB2A after treatment with 2 M TFA at 100°C for12 h; II, crude EPS fraction of CB2A after methanolysis; III, peak A of CB2A(Fig. 1) after treatment with 2 M TFA at 100°C for 12 h; IV, peak B of CB2A(Fig. 1) after treatment with 2 M TFA at 100°C for 12 h; V, purified EPSfraction of CB15A after treatment with 2 M TFA at 100°C for 18 h.

' ND, not detected.d tr, trace.

VOL. 173, 1991

on April 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

5680 RAVENSCROFT ET AL.

FIG. 2. GC trace of alditol acetates from crude EPS of CB2A.Alditol acetate derivatives of gulose (1), ribose (2), fucose (3),inositol standard (4), mannose (5), and glucose (6) are shown. Thesample was chromatographed on a DB17 column using beatingprogram A.

that the anhydro ring was 1,6 linked (Fig. 3). Hydrolysis ofthe methyl glycosides of gulose followed by reduction andacetylation yielded the anhydro sugar (89%) and gulitol(11%) derivatives as determined by GC analysis (gulitolhexaacetate has the same retention time as its enantiomer,the glucitol derivative) (7). Consequently, the early peakdetected by GC was attributed to the 1,6-anhydrogulopyra-nose. The EPS of CB2A therefore contained glucose, fu-cose, and gulose in the approximate ratio 3:1:1, whereas theEPS produced by CB15A contains galactose, glucose, man-nose, and fucose in approximately equal amounts as deter-mined by GC-MS analysis. The CB2A EPS fraction alsocontained a minor low-molecular-weight component com-posed of equimolar amounts of mannose and glucose (Fig. 1,peak B). Because of the difficulty in obtaining adequatequantities of this material and because of the possibility thatit was not an EPS judging from its small size, it was notexamined further.Methanolysis of EPS samples followed by reduction with

NaBD4 prior to conversion to the alditol acetal derivatives

A. Chemical Ionization

(M+NH4)+ = 306

(M+Ht)+ = 289

(M+H-AcOH)+ = 229

M.Wt. = 288Anhydrohexose

HaC-0AO

ACO OAC

2882p

220 2.0 to03e 0 3

mle

B. Electron impact

I3

15 av

Il 112r5

Ii

7 ~~~H2C-o

VO

22 ACO*o 21 S

Wle

1,6 - anhydrohexopyranoseFIG. 3. MS analysis of gulose derived from the EPS of CB2A.

and GC-MS analysis would allow for the detection of uronicacids; none were found. The absolute configurations of thesugars of both EPS samples were determined by GC analysisand were found to be D for all the hexoses and L for fucose.'H NMR analysis of the EPS fractions confirmed both the

repeating nature of these polymers and that they differ incomposition. The spectrum recorded for the EPS of CB2Ashowed a fucosyl methyl signal at 8 1.20 and five discernibleanomeric signals. These corresponded to three ot linkages (55.41, 5.34, and 5.29) and two 3 linkages (5 4.83 and 4.77).The spectrum obtained for the EPS of CB15A showedsimilar features, namely, a fucosyl methyl signal at high field(5 1.21) and four major signals in the anomeric region. Thesecorresponded to three a. linkages (5 5.27, 5.22, and 5.07) andI linkage at 8 4.71. Borderline signals at 8 4.57, 4.46, and4.39 were attributed to ring protons (H-2 of mannose and H-5of fucose are typically found at these 8 values) (41). The 'HNMR spectrum of the Smith-degraded CB15 EPS revealedtwo anomeric signals at 8 5.01 and 4.82 which were assignedto ox-linked mannose and n-linked glucose on the basis ofcoupling constant analysis.

Methylation analysis of the CB2A polysaccharide gavefour major peaks representing terminal glucose, 3-linkedfucose, 3,4-linked glucose, and a terminal hexose (ascribedto gulose) in a ratio of 1:1:1:2, which is consistent with apentasaccharide repeating unit as indicated by NMR analy-sis. The assignments were made on the basis of retentiontimes (coinjection with standards) and GC-MS analysis (7).Methylation analysis of the methyl glycosides of guloseyielded the 2,3,4,6-tetra-O-methyl acetylated derivative witha retention time of 0.96 relative to terminal glucose (heatingprogram B), thus corroborating the assignment made above.Periodate oxidation followed by methylation analysis gavepeaks representing 3-linked fucose and 3-linked glucose (3),thus confirming that both the terminal gulose and glucosegroups removed during the Smith degradation must beattached to 0-4 of the 3,4-linked glucose units in the originalpolysaccharide. Figure 4A shows the two possible structuresof the CB2A repeating unit based on the above data.

Methylation analyses of the CB15A EPS were compli-cated by undermethylation despite several Purdie treatments(35) following methylation by the Hakomori procedure (18).Methylation analysis of the CB15A polysaccharide gave fourmajor peaks representing terminal galactose, 4-linked fu-cose, 3-linked glucose, and 3,4-linked mannose, which isconsistent with a tetrasaccharide repeating unit as indicatedby NMR analysis. Periodate oxidation of the CB15A EPSfollowed by reduction with NaBH4 yielded a polyalcoholcontaining equal amounts of mannose and glucose, which isconsistent with the results of methylation analysis presentedabove. Mild hydrolysis of the polyalcohol yielded the Smith-degraded product which was isolated by SEC. Methylationanalysis of Smith-degraded CB1SA showed the presence ofterminal glucose and 4-linked mannose. This result suggeststhat the terminal galactose must be linked to 0-4 of the3,4-linked mannose in the native polysaccharide and that thepolymer backbone consists of --4-Fuc-3-Glc-4-Man--*. Fig-ure 4B illustrates the proposed structure of the CB1SArepeating unit based on the above data. Note that, for bothCB2A and CB15A EPS, the number of at and P linkages wasdetermined, but specific assignments cannot be made.

Colorimetric analysis of purified CB1SA and CB2A EPSshowed they were free of detectable amounts of protein butdid contain trace amounts of phosphorus and KDO. By wayof comparison, the second ribose peak (presumed to beRNA) contained significant amounts of phosphorus.

J. BACTERIOL.

_-t

on April 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CAIULOBACTER EXTRACELLULAR POLYSACCHARIDES 5681

A.

-_ 3) Glc (1 3) Fuc (1 -_ 3) Glc (14

Glc

4

1

Gul

or

-_ 3) Glc (1 3) Glc (1 -_ 3) - Fuc - (1i4

Gic

4

1

Gul

B.

4) - Fuc- (1 3) - GIc - (1-. 4) - Man - (1-_3

Gal

FIG. 4. Proposed structures for the Caulobacter exopolysaccha-rides. (A) The two possible structures for the CB2A EPS repeatingunit. (B) The structure of the CB15A EPS repeating unit. Note thatno assignment has been made for the linkages (a or ,B) between thesugar units.

Assessing the degree of EPS cell association. In one repeti-tion of the isolation procedure, CB15A and CB2A cells were

washed five times by centrifugation with 0.1 M HEPESbuffer before the extraction procedure to determine whetherthe EPS could be readily washed from the cells was begun.The yield of crude EPS from this experiment was notsignificantly different than that of cells which had beenwashed only once or not at all (data not shown). Correspond-ingly, the culture supernatants from 500-ml batch culturesfor CB2A and CB15A were freeze-dried, dialyzed againstwater, freeze-dried again, and analyzed to determine percentcarbohydrate by weight. No significant amount of carbohy-drate was detected. We conclude that the EPS in both CB2Aand CB15A was significantly adherent to the cell.

Electron microscopy analysis. Conventional fixation proce-

dures for SEM analysis of C. crescentus CB2A cells at-tached to an aluminum oxide surface demonstrated theexpected image; cells were attached to the surface via the tipof the stalk, the cell surface was smooth, and there was littleto indicate that a significant amount ofEPS or capsule mightbe present (Fig. 5A). In contrast, the cells prepared by thecryofixation and freeze-substitution methods (Fig. 5B and C)showed numerous fibrous networks, more prevalent in areasof highest cell density. This is typical of SEM images ofhighly hydrated polysaccharide polymers that have aggre-gated during the removal of the liquid phase during critical-point drying (25).Such results encouraged us to reinvestigate thin-section

TEM analysis in an attempt to localize the EPS identified bychemical extraction. While the inner membrane, peptidogly-can layer, outer membrane, and S layer were well preservedby both preparation techniques, neither conventional fixa-tion techniques nor the cryofixation substitution methodol-ogy revealed the presence of a stainable layer.

DISCUSSION

During growth in broth culture, both CB2A and CB15Aproduced sufficient quantities of an EPS to be isolated in anaqueous phase as a by-product of a general-purpose LPSisolation procedure (11). The inability to wash the EPS offthe surface by repeated centrifugations and suspensionsindicated that the polymers were not loosely associatedslime layers (31). NMR and methylation analysis confirmedthat the polymers consist of repeating units containing botha- and 1-linked sugar residues, including a fucosyl sugarunit. The repeating sugar units indicated that the isolatedpolymers had the general features of a bacterial EPS orcapsule. The significant chemical differences in the EPSs ofthe two strains illustrate that they have evolved indepen-dently so as to present different chemical motifs to theexternal environment. Like strains of other bacterial spe-cies, C. crescentus strains may produce many different EPSchemotypes.The apparent firm attachment to the surface is a property

shared with LPS, and it might be argued that in both caseswe have in fact isolated an LPS. That is not likely to be thecase, at least for the typical LPS of caulobacters. In the LPSextraction procedure, we differentiated between the LPSand EPS first on the criterion that the LPS was pelleted byultracentrifugation at 200,000 x g for 30 h in aqueoussolution (a standard criterion for LPS) whereas the EPSremained in the supernatant. Subsequent chemical analysishas shown that the LPS fraction contained a single "rough"species of LPS, had a completely different chemical compo-sition from the EPS in both CB2A and CB15, and had KDOlevels typical of LPS (42).

It is possible, however, that the EPS fraction is technicallya large species of "smooth" LPS. That is, the long four- orfive-sugar repeat-structure oligosaccharide might be an-chored to the outer membrane by attachment to a singlerough LPS molecule (consisting of lipid A and core oligosac-charide moieties). The size of such an "LPS" moleculewould be considerable; the SEC findings suggest a minimumof 1,000 to 3,000 repeat units of four or five sugars. Such ananchoring arrangement has been suggested for the group Icapsular polysaccharide antigens of Escherichia coli (22).The presence of trace amounts of phosphate and KDO in theEPS preparations might support that notion. It is difficult,however, to rule out the possibility (or likelihood) that slightcontamination of the EPS with LPS would lead to the sameresult. Anchoring of the EPS to the surface might also bemediated by other lipids, as has been shown for group IIcapsular polysaccharides of E. coli, in which the EPS islinked to the cell surface by phosphatidic acid (22). At thispoint, the means of apparent surface adherence for theCaulobacter EPSs is unresolved.

Since the extraction procedure demonstrated that at leastmost of the EPS remained on the cell during typical centrif-ugation procedures, it might be expected to be visible bythin-section TEM methods. We were unable to observe anyindication of an EPS layer on unattached cells prepared forthin-section TEM by standard or cryofixation/freeze-substi-tution methods, even when dyes commonly used to reveal

VOL. 173, 1991

on April 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

5682 RAVENSCROFT ET AL.

FIG. 5. SEM of adherent C. cresc'entits C12A prepared by (A) conventional and (B and C) cryopreservation methods. Bars. 5 p.m (A andB.) and 1 A.m (C).

J. BACTERIOL.

on April 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CAULOBACTER EXTRACELLULAR POLYSACCHARIDES 5683

polysaccharides (e.g., ruthenium red) were incorporated intothe procedures. Recently, Graham et al. (17), as part of alarger study on the use of freeze-substitution methods,achieved similar results with strain CB15A. Yet the cryofix-ation/freeze-substitution technique has been used to suc-cessfully preserve the EPS layer on Leptothrix discophora(6) and E. coli K30 (43). The EPS of both Caulobacterstrains, however, contained only neutral monosaccharides.It is perhaps not surprising that the cationic dyes often usedto stain surface polysaccharides do not stain a neutralpolymer. We surmise, then, that the capsule layer is presentbut, so far, is insufficiently stained by any of the heavymetals used during the thin-section procedures to be de-tected.We are, in particular, interested in determining the loca-

tion of the EPS with respect to the S layer of caulobacters,a dominant feature of Caulobacter cell surface design (39,40). We have learned, for example, that another oligosac-charide of the Caulobacter surface, termed SAO, involvedwith attachment of the S layer to the surface, is hidden fromspecific antisera when the S layer is present (14). Othermethods of capsular stabilization, such as pretreatment withantibody directed against the EPS or chemical dehydrationand Lowicryl embedding (5), may be required to visualizethe layer by thin-section analysis. Moreover, the composi-tion of the polymers of the two strains differed enough that itshould be possible to detect and differentiate between thesetwo polymers by immunological means. In this context, it isof interest to note that CB2A no longer produces an S layer(39), presumably as a consequence of prolonged laboratorysubculturing, but does produce and correctly assembles theS-layer protein from CB15A when the gene is introducedinto CB2A on a plasmid (14). Apparently, S-layer assemblywas not affected by differing EPS molecules.We also note that we have not in the past had difficulty in

labeling the S-layer protein with specific antibodies andcolloidal gold labels for electron microscopy (30, 38). It maybe that highly hydrated neutral polysaccharides, incapable ofion bridging or charge repulsion effects, have little ability tolimit access of antibody molecules to the cell surface.

Standard procedures for preparing an adherent populationof C. crescentus for observation by SEM also showed noindication of an EPS layer, which in part explains why anEPS has not been previously noticed in caulobacters. It istypical for capsular material to collapse onto the surfaceafter the removal of water during acetone dehydration andcritical-point drying (25). It is quite possible that glutaralde-hyde and osmium tetroxide in the conventional procedurehave no chemical cross-linking activity toward this neutralpolysaccharide and therefore could not prevent the collapse.On the other hand, the technique involving rapid freezingand exchange of water for acetone at very low temperaturemay have reduced the degree of this collapse. Subsequently,SEM, which does not depend on the binding of heavy metalions to the polysaccharide for visibility, might well reveal thepolysaccharides.Given their apparent invisibility by most techniques, one

must address the question as to whether the chemicallycharacterized polymers are EPSs; that is, do they reside onthe cell surface? Recently, we have been attempting toisolate the outer membrane fraction of Caulobacter enve-lopes and have noted that washing intact cells with anNaCl-EDTA solution releases LPS, SAO, and S-layer pro-tein, as well as three to five other specific proteins andabundant amounts of EPS. The treatment leaves cells intactby phase-contrast and thin-section TEM, and the EPS frac-

tion is free ofRNA contamination (data not shown), suggest-ing again that the cells remained intact. Also, there is noprecedent for the presence of such large polymers in eitherthe cytoplasm or the periplasm of gram-negative bacteria;we think it unlikely that the polymers are located in interiorcell regions.

Finally, our growth conditions for the samples in the SEMstudy may have promoted production of additional EPS,especially during the incubation period in MgCl after thecells became attached to the ceramic matrix. It has beensuggested that, for some bacteria, adherence to a surfacemay promote the production of EPS above the level pro-duced by free-ranging cells (9). We have not yet attempted toisolate the EPS produced by adherent C. crescentus popu-lations in order to determine whether more of the polysac-charide is produced per cell or whether it has the samechemical composition as that produced on nonadherentcells. Such information would help further define the role ofcaulobacters as members of natural biofilm communities.

ACKNOWLEDGMENTS

We acknowledge the assistance of Steve Smith in thin-sectionTEM analysis and thank Michael Armstrong of the Selee Corpora-tion for providing the ceramic foam. We also thank the MassSpectrometry and NMR units of the Department of Chemistry at theUniversity of British Columbia for spectra recorded.

This work was supported by grants to J.S. from the U.S. Office ofNaval Research (N00014-89-J-1749) and the Natural Sciences andEngineering Research Council of Canada (OGP 0036574).

REFERENCES1. Abdel-Akhar, M., J. K. Hamilton, R. Montgomery, and F.

Smith. 1952. A new procedure for the determination of the finestructure of polysaccharides. J. Am. Chem. Soc. 74:4970-4971.

2. Agabian, N., and J. Smit. Unpublished data.3. Agabian, N., and B. Unger. 1978. Caulobacter crescentus cell

envelope: effect of growth conditions on murein and outermembrane protein composition. J. Bacteriol. 133:987-994.

4. Ames, B. N., and D. T. Dubin. 1960. The role of polyamines inthe neutralization of bacteriophage deoxyribonucleic acid. J.Biol. Chem. 235:769-775.

5. Bayer, M. E. 1990. Visualization of the bacterial polysaccharidecapsule. Curr. Top. Microbiol. Immunol. 150:129-175.

6. Beveridge, T. J. 1988. The bacterial surface: general consider-ations towards design and function. Can. J. Microbiol. 34:363-372.

7. Bock, K., and H. Th0gersen. 1982. Nuclear magnetic resonancespectroscopy in the study of mono- and oligosaccharides. Annu.Rep. NMR Spectrosc. 13:1-57.

8. Button, D., and R. D. Bevill. 1970. Polysaccharides in the cellwall of Caulobacter crescentus. Bacteriol. Proc., p. 57.

9. Costerton, J. W., K.-J. Cheng, G. G. Geesey, T. I. Ladd, J. C.Nickel, M. Dasgupta, and T. J. Marrie. 1987. Bacterial films innature and disease. Annu. Rev. Microbiol. 41:435-464.

10. Costerton, J. W., R. T. Irvin, and K.-J. Cheng. 1981. Thebacterial glycocalyx in nature and disease. Annu. Rev. Micro-biol. 35:299-324.

11. Darveau, R. P., and R. E. W. Hancock. 1983. Procedure forisolation of bacterial lipopolysaccharides from both smooth andrough Pseudomonas aeruginosa and Salmonella typhimuriumstrains. J. Bacteriol. 155:831-834.

12. Dubois, M., K. A. Gilles, J. K. Hamilton, P. A. Rebers, and F.Smith. 1956. Colorimetric method for determination of sugarsand related substances. Anal. Chem. 28:350-356.

13. Dudman, W. F., L.-E. Franzen, J. E. Darvill, M. McNeil, A. G.Darvill, and P. Albersherin. 1983. The structure of the acidicpolysaccharide secreted by Rhizobium phaseoli strain 127 K36.Carbohydr. Res. 117:141-156.

14. Edwards, P., S. G. Walker, N. Ravenscroft, and J. Smit.Unpublished data.

VOL. 173, 1991

on April 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

5684 RAVENSCROFT ET AL.

15. Gerwig, G. J., J. P. Kamerling, and J. F. G. Vliengenthart. 1978.Determination of the D and L configurations of neutral monosac-charides by high-resolution capillary g.l.c. Carbohydr. Res.62:349-357.

16. Graham, L. L., and T. J. Beveridge. 1990. Effect of chemicalfixatives on accurate preservation of Escherichia coli and Ba-cillus subtilis structure in cells prepared by freeze-substitution.J. Bacteriol. 172:2150-2159.

17. Graham, L. L., R. Harris, W. Villiger, and T. J. Beveridge.1991. Freeze-substitution of gram-negative eubacteria: generalcell morphology and envelope profiles. J. Bacteriol. 173:1623-1633.

18. Hakomori, S.-I. 1964. A rapid permethylation of glycolipid andpolysaccharide catalysed by methyl sulphinyl carbanion indimethyl sulphoxide. J. Biochem. (Tokyo) 55:205-208.

19. Harris, P. J., R. J. Henry, A. B. Blakarey, and B. A. Stone. 1984.An improved procedure for the methylation analysis of oligosac-charides and polysaccharides. Carbohydr. Res. 127:59-73.

20. Heyns, K., and H. Scharmann. 1966. Substituenteneinfluss aufdas Fragmentierungsverhalten von 1,6-Anhydro-hexopyra-nosen; Zerfallsmechanismen von peracetylierten 1,6-Anhydro-hexopyranosen mit stickstoffhaltigen Ringen. Chem. Ber. 99:3461-3476.

21. Horton, D., J. D. Wander, and R. L. Foltz. 1974. Analysis ofsugar derivatives by chemical-ionization mass-spectrometry.Carbohydr. Res. 36:75-96.

22. Jann, B., and K. Jann. 1990. Structure and biosynthesis of thecapsular antigens of Escherichia coli. Curr. Top. Microbiol.Immunol. 150:19-41.

23. Jansson, P. E., L. Kenne, H. Liedgrer, B. Lindberg, and J.Lonngrer. 1976. A practical guide to the methylation analysis ofcarbohydrates. Chem. Commun. (Stockholm Univ.) 8:1-75.

24. Karkhanis, Y. D., J. Y. Zeltner, J. J. Jackson, and D. J. Carlo.1978. A new and improved microassay to determine 2-keto-3-deoxyoctonate in lipopolysaccharide of gram-negative bacteria.Anal. Biochem. 85:595-601.

25. Kelienberger, E. 1987. The response of biological macromol-ecules and supramolecular structures to the physics of speci-men cryopreparation, p. 35-63. In R. A. Steinbrecht and K.Zierold (ed.), Cryotechniques in biological electron microscopy.Springer-Verlag, New York.

26. MacRae, J. D., and J. Smit. 1991. Characterization of caulobac-ters isolated from wastewater treatment systems. Appl. Envi-ron. Microbiol. 57:751-758.

27. Markwell, M. A. K., S. M. Haas, L. L. Bieber, and N. E.Tolbert. 1978. A modification of the Lowry procedure to sim-

plify protein determination in membrane and lipoprotein sam-ples. Anal. Biochem. 87:206-210.

28. Merker, R. I., and J. Smit. 1988. Characterization of theadhesive holdfast of marine and freshwater caulobacters. Appl.Environ. Microbiol. 54:2078-2085.

29. Mitchell, D., and J. Smit. 1990. Identification of genes affectingproduction of the adhesion organelle of Caulobacter crescentusCB2. J. Bacteriol. 172:5425-5431.

30. Ong, C., M. L. Y. Wong, and J. Smit. 1990. Attachment of theadhesive holdfast organelle to the cellular stalk of Caulobactercrescentus. J. Bacteriol. 172:1448-1456.

31. 0rskov, F., and I. 0rskov. 1990. The serology of capsularantigens. Curr. Top. Microbiol. Immunol. 150:43-63.

32. Phillips, L. R., and B. A. Fraser. 1981. Methylation of carbohy-drates with dansyl potassium in dimethyl sulfoxide. Carbohydr.Res. 90:149-152.

33. Poindexter, J. S. 1964. Biological properties and classification ofthe Caulobacter group. Bacteriol. Rev. 28:231-298.

34. Poindexter, J. S. 1981. The caulobacters: ubiquitous unusualbacteria. Microbiol. Rev. 45:123-179.

35. Purdie, T., and J. C. Irvine. 1903. The alkylation of sugars. J.Chem. Soc. 83:1021-1037.

36. Schmidt, J. M., and R. Y. Stanier. 1966. The development ofcellular stalks in bacteria. J. Cell Biol. 28:423-436.

37. Shapiro, L. 1985. Generation of polarity during Caulobacter celldifferentiation. Annu. Rev. Cell Biol. 1:173-207.

38. Smit, J., and N. Agabian. 1982. Cell surface pattering andmorphogenesis: biogenesis of a periodic surface array duringCaulobacter development. J. Cell Biol. 95:41-49.

39. Smit, J., and N. Agabian. 1984. Cloning of the major protein ofthe Caulobacter crescentus periodic surface layer: detectionand characterization of the cloned peptide by protein expressionassays. J. Bacteriol. 160:1137-1145.

40. Smit, J., D. A. Grano, R. M. Glaeser, and N. Agabian. 1981.Periodic surface array in Caulobacter crescentus: fine structureand chemical analysis. J. Bacteriol. 146:1135-1150.

41. Vliegenhardt, J. F. G., L. Dorland, and H. van Halbeek. 1983.High-resolution 'H-nuclear magnetic resonance spectroscopyas a tool in the structural analysis of carbohydrates related toglycoproteins. Adv. Carbohydr. Chem. Biochem. 41:209-375.

42. Walker, S. G., N. R. Ravenscroft, G. S. Dutton, and J. Smit.Unpublished data.

43. Whitfield, C., G. Schoenhals, and L. Graham. 1989. Mutants ofEscherichia coli 09:K30 with altered synthesis and expressionof the K30 antigen. J. Gen. Microbiol. 135:2589-2599.

J. BACTERIOL.

on April 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from