The characterization of mannan of Micrococcus lysodeikticus as an acidic lipopolysaccharide.

Ghosts of Micrococcus lysodeikticus contain a mannan that is not removed by intensive washing procedures. Purified mannan, isolated by extraction of whole cells with hot, aqueous phenol, binds to membranes in vitro. Mannan also binds to DEAE-cellulose and migrates toward the anode in neutral and sodium dodecyl sulfate disc gel electrophoresis. In aqueous solution mannan has an apparent molecular weight of 10-6, but in the presence of sodium dodecyl sulfate its apparent molecular weight is 50,000 to 100,000; removal of the detergent results in reaggregation. Purified mannan contains mannose, succinate, fatty acid, and glycerol in a ratio of 50:4.9:2.1:1.0. Treatment of mannan with mild base produces a neutral, hydrophilic polysaccharide of relatively low molecular weight that has no affinity for membranes. At least 90% of the reducing termini are blocked in a base-stable linkage. Based on these results a tentative structure is proposed for the mannan.

Mannan also binds to DEAE-cellulose and migrates toward the anode in neutral and sodium dodecyl sulfate disc gel electrophoresis. In squeous solution mannan has an apparent molecular weight of 10F, but in the presence of sodium dodecyl sulfate its apparent molecular weight is 50,000 to 100,000; removal of the detergent results in reaggregation.
Purified mannan contains mannose, succinate, fatty acid, and glycerol in a ratio of 50:4.9: 2.1: 1.0. Treatment of mannan with mild base produces a neutral, hydrophilic polysaccharide of relatively low molecular weight that has no affinity for membranes.
At least 90% of the reducing termini are blocked in a base-stable linkage.
Based on these results a tentative structure is proposed for the mannan.
However, it has not yet been possible to define the structure of the mannan in terms of a small repeating unit. *

RESULTS
Associalio~~ oj Jf artnan &tk Ghosts--To determine the affinity of mannan for the membrane, ghosts of AI. lysodeilcticus wrre prepared by lysozymc treatrnellt in hypotonic medium.
The resulting crude ghosts werr then diluted extensively, collected by centrifugation, and washed three times. This process of dilution, centrifugation, and washing was repeated a second time. Aliquots of the crude ghosts, as well as the ghost after each dilution and multiple washing stage, were analyzed for total protein and mannan content.
The latter was dctcrmined by acid hydrolysis of the membranes followed by paper chromatography, elution of the mannose, and quantitative estimation of it by the Park-Johnson method (9). The results, shown in Table 1, reveal that the mannan content of the membrane preparation remains constant throughout the ghost isolation procedure. Moreover, although there is extensive loss of protein during the first dilution and washing steps, thereafter the mannan to protein ratio of the membranes is virtually constant. Comparison OJ'" dlethods for Isolation of Jlannan~l'he firm association of the mannan with the mcmbrunc fraction of -11. Illsodeilzticus suggested that the mannnn molecule contained constituents other than mannose which could account for its affinity for  In previous studies (3), uniformly labeled ['"CJmannan was prc~parctl by growing Jf. lysodeiX-ticus in the presence of [l~C]g-lucosc. The cells were harvested and incubated at SO" in 44% phenol.
After centrifugation of the cooled suspension, the aqueous phase was incubated rvith deoxyribonuclease, ribonuclease, and pronase.
Mannan prepared in this manner was cstc~nsively characterized ant1 found to contain [14C]mannosc as the sole radioactive compound after acid hydrolysis.
To csaminc the possibility that the phenol procedure might cleave linkages of mannose residues to protein or lipid, an alternatire method not cspcctcd to rupture covalent bonds was used to isolate marnlan.
Alembrancs were suspcndcd in 20 volumes of chloroform-mrthanol (2 : 1). After the addition of 4 volumes of saline (0.9 yc Sac'1 solution) and ccntrifugation, mannan was isolated in the aqueous phase. This preparation and the mannan isolntcd by t'he phenol cstrnction procedure were compared by gel filtratioli on &+~arose 412.
'I'hc two preparations showed no apparent tlifferencc in molcculnr wright.
In addition, digestion by prona>c had no affect on the elution profile of cither sample.
130th preparations bound to I)EX&ellulose in 0.05 bl l'ris-HCl, pH 7.5, and neither pronase digestion nor incubation with bactcrial alkaline phosphatase irlfluenced the affinity of the mannan for the anion cxchangc resin. For subsequent studies, therefore, mannan was isolated by the phr~lol procedure except that the digestion with pronasc was 0mittc:tl.
Properties of .IIannan--i\lanllarl is elutrd from Fepharosr 613 near thr: exclusion volume indicating an apparent molecular v-eight on the order of IO6 (Fig. 1). 111 the presence of sodium dodccyl sulfate, however, the apparent molecular weight of mannan is reduced by about IO-fold. 1Iannan which had been preincubated in sodium dodecyl sulfate for 2 hours at 37" was still more retarded by the column, indicating that the disaggregation is time-dependent.
As NaOH as indicated in a total volume of 40 ~1. After incubation at 37" for 20 min, the reaction mixtures were placed on ice and 2 pmol of Tris-HCl, pH 7.5, 20 pmol of HCl, and sufficient NaOH to adjust the total to 20 rmol were added. After dilution with H,O to a total volume of 1.0 ml, each sample was applied to a column (0.6 X 3 cm) of DEAE-cellulose equilibrated with 0.05 M Tris-HCl, pH 7.5, and eluted sequentially with three l-ml portions of each of the three eluants. largely reversed when the detergent is removed by extensive dialysis. Initial studies indicated that mild base hydrolysis of mannan resulted in a marked alteration of its properties.
Intact mannan and mannan hydrolyzed by mild base were compared by gel filtration on Sepharose 613. The results, shown in Fig. 2, indicate that mild base treatment causes a reduction in apparent molecular weight of more than lo-fold.
Untreated mannan binds to DEAE-cellulose, but after mild base hydrolysis no binding is observed (Table II).
It should be noted that only 60% of the untreated mannan can be eluted with buffer containing I M NaCl; complete elution of the mannan was achieved only by the addition of the nonionic detergent Triton X-100 to the eluting buffer. This finding suggested that nonionic as well as ionic factors might play a role in the affinity of the mannan for DEAE-cellulose.
The ionic properties of the mannan were also examined by disc gel electrophoresis.
As shown in Fig. 3A, intact mannan was found to migrate toward the anode in a standard, neutral gel system. After mild base hydrolysis, however, mannan appeared to were prepared as sample gels for disc gel electrophoresis using the neutral system of Davis (21). Electrophoresis was carried out at 4.5 ma per gel for 3 hours (the tracking dye ran off the end of the gel in l>i hours). For each experiment, the sample gel, the stacking gel, and 0.25.cm slices of the resolving gel were counted to determine radioactivity after treating with NCS solubilizer (22). B, sodium dodecyl sulfate gel electrophoresis was carried out by the method reported previously (22)  be neutral since radioactivity was detected only in the sample and stacking gels. When analyzed by disc gel electrophoresis in the presence of sodium dodecyl sulfate (Fig. 3B), intact mannan migrated toward the anode but, as in the neutral gel system, basehydrolyzed mannan was detected only at the top of the gel. These results suggested that intact mannan is acidic and binds sodium dodecyl sulfate and that both of these properties are lost after mild hydrolysis.
The infrared spectrum of intact mannan indicated the presence of a carbonyl moiety (data not shown).
Analysis of mannan before and after base hydrolysis by nuclear magnetic resonance spectroscopy revealed that resonances characteristic of aliphatic groups ( < 3 ppm) were present in intact mannan but absent after base hydrolysis (Fig. 4). Th e resonances of the sugar ring proteins (3. Thin layer chromatography was carried out on Silica Gel H prepared in 2% EDTA, pH 7.2, and developed with Solvent System D. The samples were: 1, acetyl hydroxamate; 2, stearoyl hydroxamate; 8, decyl hydroxamate; 4, ether extract of the acidified reaction mixture after treatment of mannan with hydroxylamine; 5, complete reaction mixture after treatment of mannan with hydroxylamine; 6, succinyl monohydroxamate; 7, succinyl dihydroxamate; 8, n-hydroxysuccinimide. Sodium malonate (120 nmol) was added to parallel incubation mixtures ()-w, A---A).
The reaction was initiated by the addition of enzyme, and the optical density at 600 nm determined as a function of time.
mannose in a base hydrolysate of mannan.
Prior extraction of the mannan with 20 volumes of chloroform-methanol (2 : 1) did change this value. Mannan appears to contain little or no amide-linked fatty acid since the fatty acid content of a strong acid hydrolysate of mannan was not greater than the fatty acid content of a mild base hydrolysate.
Molecular Weight and E'nd Group Analyses-The quantity of reducing end group estimated by the method of Park and Johnson (9) indicated that mannan contained 0.050 pmol of reducing The sample preparation and conditions are described under "Experimental Procedure." Reaction times indicated for authentic standards were determined in parallel runs.
Since the procedure involves boiling the samples for 15 min at 100' in 0.05 M Na2C03, this value would reflect the sum of reducing sugar, if any, in intact mannan as well as reducing sugar released by base hydrolysis.
As an alternative method to determine reducing end groups, [Wlmannan was treated with base in the presence of [3H]NaBH4. The [14C]mannan was reisolated by gel filtration on Sepharose 6B. The pooled [14C]mannose-containing fractions had a mole ratio of 3H :14C of 0.046 :50, confirming the value of about 0.05 reducing terminus per 50 mannose residues. The molecular weight of base-treated [14C]mannan was estimated to be no more than 15,000 by gel filtration (see below).
These results suggest, therefore, that a maximum of 10% of the mannose polymers released by base hydrolysis terminate in a reducing sugar; in at least 90% of the polymannose chains the reducing end is blocked by a base-stable linkage.
In considering possible compounds linked to the reducing terminus of most of the polymannose chains, acidic groups such as phosphate, glycerophosphate, or glyceric acid were considered unlikely as the mannan has no acidic character after base hydrolysis.
On the other hand, a linkage of the mannose polymer to glycerol would be consistent with the chemical and physical properties of mannan and, moreover, this type of linkage is known to occur in dimannosyl diglyceride found in large quantities in the membrane of M. lysodeikticus.
Mannan was analyzed for glycerol by the formation of [""PIglycerophosphate in the presence of glycerokinase and [Y-~~P]-ATP.
Acid hydrolysis of intact mannan or of an unfractionated base hydrolysate of mannan released 1.0 pmol of glycerol per 50 pm01 of mannose.
Extraction of an aqueous solution of mannan with 20 volumes of chloroform-methanol (2 : 1) did not change the ratio of glycerol to mannose.
Control assays containing samples of untreated mannan or a base hydrolysate of mannan contained no free glycerol prior to acid hydrolysis.
Base-treated mannan was subjected to gel filtration on Sepharose 6B, and analyses of acid hydrolysates of the effluent established that a single peak of glycerol eluted just after the peak of radioactivity (Fig. 8). The elution profile would be consistent with the release by base hydrolysis of a pofydisperse population of mannose-containing polymers each covalently bound to a single glycerol and ranging in size from as few as 10 to more than 100 mannose residues per polymer.
The discrepancy in the elution positions of the peak of radioactivity and the peak of glycerol would arise because the maximal glycerol would reflect a number average molecular weight (approximately 9,000) whereas the higher specific radioactivity of the longer polymers would shift t,he radioactive profile toward higher apparent molecular weight (approximately 15,000). The absence of a second peak of glycerol at the inclusion volume as well as the failure of chloroform-methanol extraction to reduce the glycerol content argues against significant contamination of the mannan with glyccrolipids.
InJZuence of Base Treatment on Binding of Xannan to Jlembranes-After incubation of purified [Klmannan with 111. lysodeikticus membran&, mannan was found to sediment with the particulate fraction ( Fig. 9). At the highest concentration of purified mannan tested, the amount of bound mannan represents a 17% increase over the amount of endogenous marman A---A), the samples were incubated at 30" for 30 min in a total volume of 1.0 ml, centrifuged for 30 min at 100,000 X g, 4", and the radioactivity in the pellet was determined.
(0.54 pmol of mannose per mg of protein (3)). To a lesser extent, mannan also sedimented with membranes from B. subtilis. In contrast, base-treated mannan has no affinity for the membranes from either organism. DISCUSSION The recent discovery that several bacterial polysaccharides are covalently linked to protein or lipid indicates a new dimension in both the structural complexity and potential physiological significance of these macromolecules. Braun and Kehn (25) have demonstrated lhal the peptidoglycan of Escherichia coli is co-valent& linked to a lipoprotein,'and the structure of the lipoprotein has been fully characterized (26). The lipopolysaccharide of the outer membrane of E. coli has also been shown to be covalently bound to protein (27). Lipoteichoic acids have recently been isolated from a number of gram-positive bacteria (28-30).
Synthesis of the cell wall teiehoic xcid of Stapkylo-COCCUS aureus II, polyribitol phosphate, is dependent on an acceptor which contains fatty acids as well as glycerol, phosphate, and glucose. This acceptor is indistinguishable in its composition and properties from the membrane-bound lipoteichoic acid of the same organism (31). Thus, all three of these major polysaccharides of the bacterial cell envelope have been shown to have a lipophilic region.
In addition, a novel class of O-methylglucose-containing lipopolysaccharidcs has been found in Jfycobacteriunl phlei. These arc of particular interest because of the finding that they stimulate fatty acid biosynthesis (32). The structural studies of Gray and Ballou (33) have established that the backbone of the lipol")lysacchsricle is composed of 7 glucose and 11 0-methylglucose residues linked at the reducing terminus to a glyceric acid moiety.
The lipopolysaccharidc molecule contains six short chain fatty acids and zero to three succinyl groups in ester linkage.
The results reported in this communication establish that the mannan of N. lysodeikticus is an additional example of a membrane-bound polysaccharide containing covalently linked lipid and permit its chemical composition to be correlated with its physical properties.
1 he composition of mannan is summarized in Table 1V. The presence of esterified succinate is consistent with the observation that mannan binds to LIEAE-cellulose and migrates toward the anode in neutral disc gel electrophoresis. In addition, mannan contains long chain fatty acid esters which provide a lipophilic portion of the molecule available for interaction with the corresponding hydrophobic portion of another mannan molecule or with membrane lipids. The lipophilic  Fig. 8).
nature of mannan is indicated by the observation that mannan exists as an aggregate in aqueous solution.
In the presence of sodium dodecyl sulfate the mannan dissociates but it is not clear that the product of dissociation is the monomer. III a short report, published while this manuscript was in preparation, evidence for the presence of long chain fatty acyl groups in mannan was reported (34). It was also demonstrated t)hat the mannan was acidic, but the presence of succinyl groups was not established.
Three types of evidence indicate that succinate and long chain fatty acids are covalently linked to mannan.
1 he ratio of ester to mannose is constant through diverse purification procedures employing organic solvents, detergents, and affinity chromatography.
In addition, hydrolysis of the esters with mild base results in a concomitant alteration of the properties of mannan including loss of acidic character, loss of affinity for the membrane, and a dramatic reduction in molecular weight.
Finally, comparison of the nuclear magnetic resonance spectra of untreated and base-treated mannan provides evidence for a small proportion of acylated mannose residues, indicating that at least some of the acyl moieties may be linked directly to the polysaccharide.
Base-treated mannan contains a very low content of reducing termini (about 0.05 per 50 mannose residues), and a maximum of 10% of the mannosc polymers may terminate in a reducing sugar or in an alkali-labile glycosidic linkage to a hydroxyamino acid. Glycerol was identified as a constituent of mannan, and it is present in sufficient quantity to be the residue blocking the reducing termini.
In addition, the linkage of mannan to glycerol is stable to mild base but is cleaved by strong acid, which are properties consistent with a glycosidic ether linkage of the polysaccharide to glycerol.
However, direct proof for such a glycosidic linkage has not been obtained.
The ratio of fatty acid to glycerol in untreated mannan (2.1: 1) suggests that the long chain fatty acyl groups may be linked to glycerol.
Purified mannan preparations contain a small amount of protein, which constitutes about 3% of the dry weight.
Preliminary amino acid analyses of the proteins released by base indicated a typical spectrum of amino acids. The quantity of protein present would be equivalent to only 3 amino acids per 50 mannose residues. Thus, whereas linkage of each mannose polymer to a short peptide of constant composition is unlikely, the attachment of a small proportion of the polymers to a larger polypeptide cannot be ruled out.
The data presented arc consistent with the tentative, partial structure of mannan shown in Fig. 10. An average length of the polysaccharide chain of 60 mannose residues may be calculated from the number average molecular weight determined by gel filtration (about 9000). Th' 1s value is in reasonable agreement with an average chain length of 50 mannose residues (calculated ?*IW 8000) based on the ratio of glycerol to mannose, assuming a single glycerol per polymer.
However, it should be noted that determination of the molecular weight by gel filtration is of limited value because both the calibration standards and the base-treated mannan appear to be polydispersoid. Further structural features of the polysaccharide chain remain to be cstablished.
As noted in the introductory section, it was not possible to deduce the structure of a short oligosaccharide repeating unit in mannan from the results of the permethylation and acetolysis of mannan.
Analogous studies on purified, deacylated mannan might, however, be more fruitful.
A possible functional analogy between membrane-bound mannan of dl. lysodeikticus and the lipoteichoic acid found in most gram-positive membranes but absent in 31. lysodeikficus has been suggested (34). It is of interest) that recent preliminary reports conclude that the level of both polymers is higher in the mesosomal membrane than in the plasma membrane (35,36). Although no attempt has been made in our study to separate these two membrane fractions, the strong affinity of mannan for the membrane is established by the observation that the mannan content of the ghosts remained constant despite extensive washing steps. Additional experiments (data not shown) involving treatment of membranes with a high concentration of salt or mild detergents corroborate this conclusion; the finding that purified, untreated mannan binds to membranes in vitro, whereas base-treated mannan does not, indicates that the covalently bound acyl groups are responsible for the association of the polysaccharide with the membrane.
Finally, it should be noted that in the partial structure proposed for mannan, the lipophilic portion of the molecule is identical with the major glycolipid of Jl. lysodeikficus, mannosyl-mannosyl-diglyceride, thus suggesting a possible biosynthetic relationship between these two membrane-associated compounds.
Ackno~ledymenfs--m-c wish to thank M. Kennedy of this laboratory for B. subfi2is membranes and Dr. I>. '1?'. Cochran of this institution for carrying out the nuclear magnetic resonance studies. 111 addition, we are grateful to Dr. K. Fenselau, Dr. C. H. Robinson, and Dr. I>. A. I'owcrs of this institution for mass spectroscopy, infrared spectroscopy, and amino acid analyses, respectively.