By hydrophobic interaction chromatography on octyl-Sepharose, ... Hydrophobic interaction chromatography also proved an effective one-step purification ...
Eur. J. Biochem. 194, 655-661 (1990) 0FEBS 1990
Purification and fractionation of lipopolysaccharide from Gram-negative bacteria by hydrophobic interaction chromatography Werner FISCHER Institut fur Biochemie der Medizinischen Fakultat der Universitat Erlangen-Nurnberg, Federal Republic of Germany (Received June 11,1990) - EJB 90 0661
By hydrophobic interaction chromatography on octyl-Sepharose, lipopolysaccharide (LPS) of Escherichia coli Re mutant and of wild-type smooth-form (S-form) Salmonella typhimurium and Salmonella abortus equi is fractionated according to increasing amount of fatty acids. Thereby a fractionation of S-form LPS according to the length of the 0-polysaccharide chain also occurs, because with increasing of fatty acids there is a decrease in the mean length of the 0-polysaccharide chain from approximately 30 to 4 repeating units. Molecular species of Re-mutant LPS contain four 3-hydroxytetradecanoyl residues in addition to which dodecanoic, tetradecanoic and possibly hexadecanoic acid, appear in this sequence. Among the molecular species of S-form LPS, dodecanoic, tetradecanoic and hexadecanoic acids appear in the same order, but in contrast to Re-mutant LPS a significant fraction of S-form LPS contains less than four 3-hydroxytetradecanoyl residues. Hydrophobic interaction chromatography also proved an effective one-step purification procedure of LPS as was shown with a crude preparation from S-form S. typhimurium.
Lipopolysaccharide (LPS) is a prominent essential amphiphile in the outer membrane of Gram-negative bacteria. LPS from various bacteria is constructed according to a uniform principle [l]. The hydrophilic chain is composed of the 0polysaccharide, consisting of repeating oligosaccharide units, and the core oligosaccharide region which connects the 0polysaccharide chain with the hydrophobic lipid A component [2, 31. The 0-polysaccharide chain represents the major heatstable antigen, whereas lipid A is the primary agent, responsible for the endotoxicity of Gram-negative bacteria. In contrast to LPS from wild-type smooth-form (S-form) bacteria, LPS from mutant rough-form bacteria lack the 0polysaccharide chain and contain only the core or fragments of it and lipid A. On polyacrylamide gel electrophoresis, in the presence of sodium dodecyl sulfate, LPS from S-form bacteria displays a ladder-like pattern which reflects heterogeneity in the length of the 0-polysaccharide chains [4 - 71. Lipid A, a derivative of p1- 6-linked glucosamine disaccharide, also shows intrinsic heterogeneity, with molecular species differing in the number of constituent fatty acids [8 - 121. Fully acylated Escherichia-coli- and Salmonella-type LPS carry six and seven fatty acyl residues, respectively. There are four 3-hydroxytetradecanoyl residues, amide- and ester-linked to positions 2, 2’ and 3, 3’ of the disaccharide moiety. The 3hydroxytetradecanoyl residues at positions 2’ and 3‘ carry an ester-linked dodecanoyl and tetradecanoyl residue, respectively, that at position 2 may be esterified with hexadecanoic acid. In a comparative study into purification procedures of lipoteichoic acid of Gram-positive bacteria, hydrophobic interaction chromatography proved superior to gel-filtration Correspondence to W. Fischer, Institut fur Biochemie der Medizinischen Fakultat der Universitat Erlangen-Nurnberg, Fahrstrasse 17, W-8520 Erlangen, Federal Republic of Germany Abbreviations. dOclA, 3-deoxy-~-manno-octulosonicacid; LPS, lipopolysaccharide; S-form, smooth form.
and anion-exchange chromatography [13]. By this procedure crude lipoteichoic acid extracted from bacteria by phenol/ water can be freed of polysaccharide, nucleic acid and protein contaminants in one step. A further advantage of hydrophobic-interaction chromatography is the partial separation of molecular species which elute from the column in the order of increasing number of fatty acids [13- 151. The present work describes hydrophobic-interaction chromatography of LPS. It proved appropriate for purification and also effected fractionation according to the number of acyl groups/molecule. Thereby it became evident that species with a low fatty acid content predominantly carry long 0-polysaccharide chains, whereas short-chain species have a higher content of fatty acids. MATERIALS AND METHODS M a terials
Octyl-Sepharose was purchased from Pharmacia LKB GmbH, Freiburg, FRG. 2-Deoxy-~-glucose,2-0x0-3-deoxyD-manno-octulosonic acid (dOclA), di(heptadecanoy1)glycerophosphocholine and pentadecanoic acid were obtained from Sigma Chemie GmbH, Deisenhofen, FRG. 2-Hydroxytetradecanoic acid was purchased from Serva Feinbiochemica GmbH & Co, Heidelberg, FRG. Lipopolysaccharides
LPS of S-form Salmonella abortus equi and LPS of the E. coli Re mutant F515 were kindly provided by C. Galanos, Max-Planck-Institut fur Immunbiologie, Freiburg, FRG, and U. Zahringer, Forschungsinstitut Borstel, FRG. S-form LPS from S. abortus equi was isolated by the phenollwater method and purified as described [16]. Re-mutant LPS from E. coli was extracted by the phenol/chloroform/light petroleum pro-
656 cedure [I 71. Both preparations were electrodialyzed and converted to the triethylamine salt form 1181. S-form LPS from Salmonella typhirnurium, prepared by phenollwater extraction [19],was purchased from Sigma Chemie GmbH, Deisenhofen, FRG.
1.0 -
w
-
3
0.8 -
m
-I
Analytical methods
Phosphorus was quantified by a modified Lowry procedure [20] and served as an internal standard in the various hydrolysates. Neutral sugars were released by hydrolysis in 2 M HCI, 100"C, 2.5 h. D-Galactose [21], D-glucose [22] and D-mannose [23] were measured enzymatically and were together with rhamnose and heptose also quantified by GLC as alditol acetates [24]. Fatty acids and glucosamine were released by hydrolysis in 2 M HCI, 100"C, 16 h. As a standard, prior to hydrolysis pentadecanoic acid was added. Its concentration was checked by separate hydrolysis with di(heptadecanoy1)glycerophosphocholine which was quantified by measurement of phosphorus. Fatty acids were extracted from the hydrolysates with light petroleum/CHCl, (4: 1,by vol.), methanolysed (1 M HCl/methanol, 80"C, 16 h), finally treated with N,O-
bis(trimethylsilyl)trifluoroacetamide/trimethylchlorosilane/ pyridine (20:5:3. by vol.) and analysed by GLC. Trimethylsilyl-3-hydroxytetradecanoicacid methyl ester was identified by GC-MS (mlz, 315; m/z, 275; mjz, 175). Thereby, a conversion of 1-2% methyl esters of all fatty acids into trimethylsilyl esters was observed. Characteristic fragments of bis(trimethylsilyl)-3-hydroxytetradecanoic acid were at m/z = 373, 257 and 233. For quantitative determination, peak areas of individual fatty acid methyl esters were divided by the respective M , ; the molar response factor of trimethylsilyl-3hydroxytetradecanoic acid was determined using the 2hydroxy analogue. Glucosamine was measured as described [25] after the hydrolysate (see above) had been dried in vucuo over KOH at 2°C. Abequose was measured after hydrolysis in 0.25 M H 2 S 0 4 (lOO"C, 30 min) according to [26], using 2-deoxy-I>glucose as a standard. dOclA was determined by a thiobarbituric acid procedure [27] after hydrolysis in 1% (by vol.) acetic acid (100 C , 20 min). Nucleic acids were measured photometrically (i= 260 nm), protein contaminants as amino acids after acid hydrolysis (5.5 M HCl, 105"C, 23 h) in an amino acid analyser. GLC was carried out on a Hewlett Packard gas chromatograph 5840 A equipped with a cold injection system (Gerstel GmbH, Miihlheim, FKG). A Durabond capillary column (DB5,30 m, internal diameter 0.25 mm, film thickness 0.25 pm) was used at 185-250°C with a temperature rise of 2'C/min for separation of alditol acetates, and at 135 - 250 "C with a temperature rise of 6 ' C/min for separation of fatty acid methyl esters. Chromatography 017 oc tyl-Sepharose
An all-glass Teflon system was used, consisting of a Gilford model 302 piston pump and Teflon tubing. Columns (0.9 cm x 35 cm, 1.5 cm x 40 cm) of octyl-Sepharose were equilibrated with buffer A or B, containing 15% (by vol.) propan-1-01, Buffer A contained 0.05 M triethylammonium acetate, pH 4.5; buffer B, 0.1 M triethylammonium acetate, pH 7.5. Details of chromatography are given under Results. For analysis propanol was removed from column fractions by evaporation at 30 C on a Vortex evaporator (Buchler
1
-
0
E 1 0.6
-
I
I3O "
20
40
60
80
100
Q
a
120
TUBE NUMBER Fig. 1. Chromatography on octyl-Sephurose ofpurified Re-mutunt LPS from E. coli. LPS (21 pmol phosphorus) in buffer A containing 15% (by vol.) propanol was applied to a column (0.9 cm x 35 cm) of octylSepharose and eluted with buffer A containing the propanol concentrations indicated. At a flow rate of 6.6 ml/h. fractions of3.3 ml were collected and analysed for phosphorus
Instruments, Fort Lilly, NJ, USA). If several fractions had to be combined, propanol and buffer were removed by dialysis.
Sodium dodecyl sulphatejpolyacrylamicle gel electrophoresis The gel system described by Laemmli 1281 was used. Gels were prepared as 5.7 - 35% linear gradients of acrylamide containing SDS [29]. LPS samples (2-3 nmol phosphorus) were dissolved in 0.05 M TrisIHCl, pH 6.8 (50 pl), containing 15% (mass/vol.) glycerol and 1% (massivol.) each of SDS and 2-mercaptoethanol. After heating at 100-C for 10 min, 10 p1 samples were loaded on to each well. The electrode buffer, pH 8.3, contained 25 mM Tris, 0.192 M glycine and 3.5 mM SDS. Gels were run at 12 mA and 17 mA each for 10 min then at 33 mA for approximately 2 h until the bromophenol blue reached the end of the gel. LPS was detected by silver staining [301. RESULTS LPS from the E. coli Re mutant
The preparation of Re-mutant LPS from E. coli used in this study contained glucosamine, dOclA, fatty acids, total phosphorus and inorganic phosphate in molar ratios of 2: 1.9:5.6: 3.2:0.6. The content of dodecanoic acid, tetradecanoic acid, hexadecanoic acid and 3-hydroxytetradecanoic acid were 15, 11, 2 and 72 mo1/100 mol, respectively. During chromatography on octyl-Sepharose, inorganic phosphate emerged at the void volume of the column (Fig. 1). LPS eluted as a poorly separated double peak at propanol concentrations between 26% and 42% (by vol.). Individual tubes were analysed for fatty acid composition, the results are summarized in Fig. 2. Molecular species that eluted in the first part of the double peak almost exclusively contained 3hydroxytetradecanoic and dodecanoic acids in relative abun-
657 Q
2 -.o -O
E
v-.
6-
80
82
84
86 88
90 92
94 96 98 100 102 104 106 108
TUBE NUMBER Fig. 2. Fatty acid composition of E. coli Re-mutant LPS separated on octyl-Sepharose (Fig. I ) . (B) Relative abundance of dodecanoic acid (0), tetradecanoic acid ( A ) and hexadecanoic acid (0).(A) Molar ratio of total fatty acid/lipid A (0)was calculated from the relative abundance of 3-hydroxytetradecanoic acid (see text)
TUBE NUMBER Fig. 3. Chromatography on octyl-Sepharose of crude S-form LPS from S. typhimurium. Crude LPS (150 pmol phosphorus) in buffer A containing 15% (by vol.) propanol was applied to a column (1.5 cm x 40 cm) of octyl-Sepharose and eluted with buffer A containing the propanol concentrations indicated. At a flow rate of 13.5 ml/ h, fractions of 4.5 ml were collected and analysed for phosphorus ( 0 ) and hexosyl equivalents (O), using an anthrone procedure with an equimolar mixture of galactose and mannose as a standard. The inset shows the molar ratios of hexosyl equivalents/phosphorus
o.ol, respectively. With the dance of o.83 o.l and o.l peak, the proportion Of tetrasecond part Of the decanoic acid increased up to a relative abundance of approxir Part of the Peak, there matelY 0.14. In the h ~ e descending was an increase in hexadecanoic acid approaching a relative abundance of 0.13.
100
110
120
130
140
150
TUBE NUMBER Fig. 4. Carbohydrate composition of S. typhimurium S-form LPS separatedon octyl-Sepharose (Fig. 3 ) . For analysis, the column fractions 101 and 102, 103 and 104, 106 and 107, 145 and 146, 147-149, 150-152, 153-155 were combined; other fractions were analysed individually. Carbohydrate/LPS was calculated on the basis of S. typhimurium LPS containing on average 3 mol glucosamine/mol LPS [2]and 5.8 mol phosphorus/mol LPS (Table 2). Rha, rhamnose; GlcN, glucosamine; Abe, abequose
658 Table 1. Composition of crude S-form L P S (Fig. 3, peak I ) n.d., not determined Fraction
of
S. typhimurium and the material separated f r o m L P S by chromatography on octyl-Sepharose
Phosphorus __
-.
total
inorganic
nucleic acids
n. d. 52
80 65
Mannose
Galactose
Rhamnose
Abequose
Glucose Heptose dOclA
84 8
103 12
n.d. 33
n.d. 13
57 33
Glucosamine
Fatty acids
23 10
n.d. 12
pmol
Crude LPS Peak I
150 124
n.d. 1
n. d. 3
Table 2. Composition of purified L P S from S. typhimurium and S. abortus equi S. typhimurium LPS was analysed after, and S. abortus equi before column chromatography on octyl-Sepharose. Fatty acids: dodecanoic, tetradecanoic and hexadecanoic acids are given as 12:0, 14:0, 16:0, respectively; 3-OH indicates 3-hydroxy. Values are molar ratios with respect to glucosamine Source of LPS
Glucosamine
Phosphorus
dOclA
Glucose
Abequose
Rhamnose
Mannose
Galactose
Fatty acids 12:O
14:O
16:O
3-OH14:O
0.21 0.23
0.18 0.17
0.09 0.20
1.21 1.34
mol/mol S . typhimurizim S. abortus equi
2.0 I .O
1.9 2.0
0.6 0.5
1.6 0.8
5.6 7.4
The analysis of the original Re-mutant LPS suggests 4 mol 3-hydroxytetradecanoic acid/mol lipid A through all molecular species, which allows one to estimate the total amount of fatty acids from the relative abundance of 3-hydroxytetradecanoic acid. As shown in Fig. 2A, 4.8 mol fatty acids/mol lipid A were estimated for the ascending part of the peak (Fig. 1). It increased to 5.7 mol/mol and 6.7 mol/mol, respectively, when tetradecanoic and hexadecanoic acids appeared. An increase in total fatty acid/lipid A was confirmed by occasional measurement of fatty acids and glucosamine (data not shown).
4.7 5.3
4.6
5.9
5.6 6.5
phorus, which are markers of lipid A and the core polysaccharide, were present in fairly constant molar ratios, the mean molar ratios (2 gn- of glucosamine and dOclA to phosphorus being 0.47 0.05 and 0.23 k 0.04, respectively. On GLC, heptose, another core constituent, was also consistently detected. In contrast to the core and lipid A constituents, the components of the 0-polysaccharide moiety showed molar ratios to phosphorus which reflected the trend of the hexosyl/ phosphorus molar ratios of the elution profile in Fig. 3, and paralleled each other. A similar behaviour showed glucose which is present in both the 0-polysaccharide [2] and the core of S. typhimurium S-form LPS. The mean molar ratios LPS from STform S. typhimurium (X k 0,- of abequose, rhamnose and galactose to mannose The elution profile from an octyl-Sepharose column of a were 1.27 Ifr 0.1,0.85 0.06 and 1.09 f 0.06, respectively. The commercially available crude LPS preparation from S. four sugars are known to build up the tetrasaccharide repeattyphimurium is shown in Fig. 3. 83% of the total phosphorus ing unit of the 0-polysaccharide of S. typhimurium LPS [31], (Fig. 3, peak I) was eluted ahead of the propanol gradient and an additional galactosyl residue occurs in the core predominantly in the form of nucleic acid material and inor- oligosaccharide [2]. Accordingly, the values of the molar ratio ganic phosphate (Table 1). It was accompanied by protein of rhamnose/mannose approached unity when corrected for (4.3 mg) and carbohydrate which contained the sugar loss on hydrolysis, whereas the reason for the deviation components of LPS, but in atypical proportions (Table 1). from unity of the molar ratios of abequose/mannose is not Constituent fatty acids of LPS were also present, and small known. amounts of predominantly short-chain LPS could be detected Fatty acids were analysed in selected column fractions. by SDSjPAGE (not shown). The material which eluted as peak I1 had the typical composition of S-form LPS of S. The results, summarized in Fig. 5, show that molecules elute typhimurium [2, 311. To show this, aliquots (10%) of the indi- from octyl-Sepharose in the order of increasing fatty acids vidual fractions were combined and analysed. The data are content. The initial fractions contained approximately 2 mol given in Table 2 . In the purified LPS, nucleic acids and protein 3-hydroxytetradecanoic acid/mol LPS and less than 1 mol dodecanoic acid/mol LPS. While the former continously inwere no longer detectable. As shown in the inset of Fig. 3, the molar ratio of hexosyl creased to a value of four, the latter remained constant. equivalents to phosphorus of individual fractions varied con- Tetradecanoic acid and hexadecanoic acid appeared in later siderably, suggesting heterogeneity in the size of the 0- fractions in this order. Hexadecanoic acid did not separate polysaccharide moieties. This was established as shown in from LPS when the fractions were adjusted to pH 3.5 and Fig. 4. Through all fractions, glucosamine, dOclA and phos- extracted with chloroform/methanol(4:1, by vol.).
+
659
6-
1
0 1
T U B E NUMBER
Fig. 5. Fatty acid composition of S. typhimurium S-form LPS separated on octyl-Sepharose (Fig. 3). Column fractions were combined or analysed individually as indicated in Fig. 4. For calculation of fatty acid LPS, see Fig. 4. Total fatty acids ( 0 ) ;3-hydroxytetradecanoic dodecanoic acid ( 0 ) ;tetradecanoic acid (A); hexadecanoic acid (0); acid (0)
LPS from S-form S. abortus equi
The original composition of purified LPS from S-form S. abortus equi used for chromatography on octyl-Sepharose is presented in Table 2. With this LPS preparation, neither carbohydrate- nor phosphorus-containing material eluted from octyl-Sepharose ahead of the propanol gradient. On TUBE NUMBER elution with the propanol gradient S. abortus equi LPS disFig. 6. Chromatography on octyl-Sepharose of purified S-form LPS played a fractionation (Figs 6-8) similar to that of S. from S. abortus equi. LPS (135 mg, 48 pmol phosphorus) was applied typhimurium LPS. As shown in Fig. 7, the initial fractions in buffer B containing 15% propanol, to a column (1.5 cm x 35 cm) contained approximately 2 mol3-hydroxytetradecanoic acid/ of octyl-Sepharose and eluted with 0.05 M of buffer B containing the mol LPS and trace amounts of dodecanoic acid. The amount propanol concentrations indicated. At a flow rate of 15 ml/h, 3-ml of these fatty acids increased to approximately 4 mol/mol fractions were collected and analysed for phosphorus ( 0 )and hexosyl LPS and 0.8 mol/mol LPS, respectively. Tetradecanoic and equivalents (0),as described in Fig. 3. In the inset, ratio indicates hexadecanoic acids appeared in this order. Analysis of the molar ratio of hexosyl equivalents/phosphorus hydrophilic chain is summarized in Fig. 8. In all fractions, the molar ratios of glucosamine, dOclA and glucose to phosphorus were scattered around the mean values of Nucleic acids and protein separated from LPS completely. 0.5:0.24:0.34 ( D ~ 1/% 5 f O.l), whereas the molar ratios of Polysaccharides, if present, will also separate because they the 0-polysaccharide chain components, abequose, rham- elute together with nucleic acids under the conditions used nose, galactose and mannose [16] to phosphorus roughly fol- ~ 3 1 . As seen earlier with lipoteichoic acid, LPS is eluted by a lowed the hexosyl/phosphate molar ratios of the elution propropanol gradient in order of increasing number of fatty acids, file in Fig. 6. The molar ratios of abequose, rhamnose and i.e. in order of decreasing critical micellar concentration [32]. galactose to mannose were constant at values (2 D ~ 1-) of 1.3 & 0.1, 0.87 f 0.09 and 1.24 & 0.2, respectively. Fig. 9 In the case of E. coli Re-mutant LPS, the initially eluting molecular species contained 3-hydroxytetradecanoic and shows SDSjPAGE of selected fractions. In agreement with the results of Fig. 8, predominantly dodecanoic acid. In later fractions tetradecanoic and hexalong-chain species containing 18 - 40 repeating units were pre- decanoic acids appeared in this order (Fig. 2). The molar ratio sent in the ascending part of the peak (Fig. 9, lanes 1 and 2). of 3-hydroxytetradecanoic acid/glucosamine of approxiThe numbers of repeating units were determined under the mately four in the original LPS suggests that all molecular assumption that each band of the ladder-like pattern differs species contain four 3-hydroxytetradecanoyl residues. By confrom the next by one repeating unit. In the descending part, trast, the initially eluting molecular species of S-form LPS long-chain species were joined first by short-chain species from S. typhimurium and S. abortus equi contained only two containing 0 - 8 repeating units (not shown), then medium- 3-hydroxytetradecanoyl residues/molecule. While 3-hydroxychain species appeared (Fig. 9, lanes 3 and 4). Later on, the tetradecanoic acid increased to 4 mol/mol LPS, dodecanoic acid, tetradecanoic acid, and hexadecanoic acid appeared in short-chain species predominated (Fig. 9, lanes 5 and 6). this order. Hexadecanoic acid is only a trace component in the hydrolysates of lipid A from E. coli [33, 341 and S. typhimurium LPS [8]. Whether the small amount observed in DISCUSSION the later fractions of E. coli Re-mutant LPS and S. The results of Table 1 and Fig. 3 show that crude LPS can typhimurium S-form LPS (Figs 2 and 5) is actually linked to be purified as efficiently as lipoteichoic acid [13] by hydro- lipid A, requires release and analysis of lipid A from these phobic-interaction chromatography using octyl-Sepharose. molecular species.
660
E
l-
$
l
0 150
160
170
180
190
200
210
TUBE NUMBER Fig. 7. Fur?? acid composition of S. abortus equi S-form LPS sepurated on octyl-Sephurose (Fig.6 ) . For analysis, the column fractions 143149, 150-152, 153 and 154, 155 and 156, 196 and 197, 198-200, 201-205, 206-213 were combined; other.fractions were analysed individually. Fatty acid/LPS was calculated on the basis of S. ubortus equi LPS containing on average 3 mol glucosamine/mol LPS [I61 and 6 mol phosphorus/mol LPS (Table 2). Symbols for fatty acids are as in Fig. 5
0 Abe AGal 0 Man 0 Rha VGLc 4 GlcN OdOcLA
Fig. 9. SDSjPAGE of S. abortus equi Sforin L P S separated on octylSepharose. The numbers are tube numbers from the column in Fig. 6. For chemical analysis, see Fig. 8. According to [7] the fastest-moving band (lane 3 - 6, bottom) is considered to be rough-form LPS devoid of 0-polysaccharide chain components. The structural difference responsible for double bands in the short-chain fraction remains to be studied
2
01
1
150
,
I
160
I
I
170
I
I
180
I
I
190
,
I
200
,
2
TUBE NUMBER Fig. 8.Curbohpirate composition qf S. abortus equi Sform LPSsepuruted on octyl-Sephurosr (Fig.6 ) . Column fractions were combined or studied individually as indicated in Fig. 7. For calculation of carbohydrate/LPS see the legend to Fig. 7
As shown in Figs 4 and 8, the fractions from octylSepharose of both S-form LPS preparations differed considerably in the mean value of the 0-polysaccharide repeating units. Longer chains were predominantly attached to lipid A
containing smaller amounts of fatty acids, whereas shortchain species in general contained more fully acylated lipid A. SDSjPAGE revealed that the continous decrease of the mean chain length in the descending part of the peak in Fig. 8 is actually caused by mixtures in which medium- and short-chain species increasingly replace long-chain species. LPS species from S-form bacteria that differ in both the number of fatty acids and the length of the 0-polysaccharide chain have not been observed until recently. Using solvent partition, Galanos et al. obtained three fractions of S. abortus equi LPS, designated long-chain, short-chain and R fractions [7]. All fractions contained lipid A and the core constituents but, as shown by analysis and SDS/PAGE, the R fraction was devoid of 0-polysaccharide constituents, whereas the longchain and short-chain fractions contained 20 - 50 and 0 - 6 repeating units, respectively. The amount of fatty acids increased with decreasing length of the hydrophilic chain, the mean number of acyl groups/mol LPS of the long-chain, shortchain and R fractions, being 4.3, 5.8 and 6.4, respectively. In parallel, the number of 3-hydroxytetradecanoyl residues
661 increased from 2.4 through 3.6 to 3.9, and analysis of lipid A released from these fractions revealed that the underacylation with this fatty acid concerns the residues at all four positions [12]. SDSjPAGE suggests that on chromatography on octylSepharose, in contrast to phase partition [7], short-chain and R fractions did not separate, and fractions with overlapping chain lengths were observed (Fig. 9). Separation by phase partition appears, therefore, to be primarily based on the hydrophilicity of LPS, determined by the length of the 0chains, whereas separation on octyl-Sepharose seems to be essentially governed by hydrophobicity, dependent on the amount of fatty acids. Biosynthetic studies using certain mutant strains of E. coli and S . typhimurium [35-381 or drug inhibition of dOclA incorporation [39] suggest a lipid A precursor that already carries four 3-hydroxytetradecanoyl residues, to which, at later stages, presumably after the addition of phosphate and dOclA residues, the non-hydroxylated fatty acids are added. One is tempted to speculate that the molecular species of Re-mutant LPS separated on octyl-Sepharose (Fig. 2) might reflect the so-far-unknown order in which dodecanoic, tetradecanoic and possibly hexadecanoic acid are incorporated. Until now, no metabolic route is known which leads to the underacylated LPS species discovered in this and previous studies [12]. Enzymatic deacylation during synthesis of the 0polysaccharide chain has been suggested [12] although the physiological role of such process is still obscure. Studies into the physicochemical properties and biological activities of the diverse molecular species which can now be separated would be of future interest. The skilful technical assistance of Edeltraud Ebnet and Birgitta Reiter is gratefully acknowledged. We wish to thank Professor H. J. Horstmann and Mrs R. Weber for help with electrophoresis, Mr G. Distler for performance of GC-MS analyses. This work was supported by a grant from the Deutsche Forschungsgemeinschaft (Fi 21 814-7).
REFERENCES 1. Westphal, 0. & Luderitz, 0. (1954) Angew. Chem. 66,407-417. 2. Luderitz, O., Freudenberg, M. A,, Galanos, C., Lehmann, V., Rietschel, E. Th. & Shaw, D. H. (1982) Curr. Top. Membr. Transp. 17,79 - 151. 3. Brade, H., Brade, L. & Rietschel, E. Th. (1988) Zentralbl. Bakteriol. Mikrobiol. Hyg. Ser. A 268, 151- 179. 4. Palva, E. T. & Makela, P. H. (1980) Eur. J. Biochem. 107, 137143. 5 . Goldman, R. C. & Leive, L. (1980) Eur. J . Biochem. 107, 145153. 6. Peterson, A. A. & McGroarty, E. J. (1985) J . Bacteriol. 162,738145. 7. Galanos, C., Jiao, B., Komuro, T., Freudenberg, M. A . & Luderitz, 0. (1988) J . Chromatogr. 440, 397-404. 8. Qureshi, N., Takayama, K. & Ribi, E. (1982) J . Biol. Chem. 257, 11808-11815. 9. Seydel, U., Lindner, B., Wollenweber, H.-W. & Rietschel, E. T. (1984) Eur. J. Biochem. 145, 505-509. 10. Qureshi, N., Mascagni, P., Ribi, E. & Takayama, K. (1985) J . Biol. Chem. 260, 5271 -5278.
11. Qureshi, N., Cotter, R. J. & Takayama, K. (1986) J . Microbiol. Methods 5, 65 - 77. 12. Jiao, B., Freudenberg, M. & Galanos, C. (1989) Eur. J . Biochem. 180, 515-518. 13. Fischer, W., Koch, H. U. & Haas, R. (1983) Eur. J . Biochem. 133, 523 - 530. 14. Fischer, W. (1987) Eur. J . Biochem. 165, 639-646. 15. Fischer, W., Mannsfeld, T. & Hagen, G. (1990) Biochem. Cell Biol. 68, 33 -43. 16. Galanos, C., Luderitz, 0. & Westphal, 0. (1979) Zentralbl. Bakteriol. Mikrobiol. Hyg. 1 Abt Orig. A 243, 226 - 244. 37. Galanos, C., Luderitz, 0. & Westphal, 0. (1969) Eur. J . Biochem. 9,245 - 249. 18. Galanos, C. & Luderitz, 0. (1975) Eur. J. Biochem. 54,603 -610. 19. Westphal, 0. & J a m , K. (1965) Methods Carbohydr. Chem. 5, 83-91. 20. Schnitger, H., Papenberg, K., Ganse, E., Czok, R., Bucher, T. & Adam, H. (1959) Biochem. 2.332, 167-185. 21. Beutler, H. 0. (1984) in Methods of enzymatic analysis, 3rd edn (Bergmeyer, H. U., Bergmeyer, J. & Grassl, M., eds) vol. 6 , pp. 104- 112, Verlag Chemie, Weinheim, Deerfield Beach, Florida, Basel. 22. Kunst, A,, Draeger, B. & Ziegenhorn, J. (1984) in Methods of enzymatic analysis, 3rd edn (Bergmeyer, H. U., Bergmeyer, J. & Grassl, M., eds) vol. 6, pp. 163 - 172, Verlag Chemie, Weinheim, Deerfield Beach, Florida, Basel. 23. Gawehn, K. (1984) in Methods of enzymatic analysis, 3rd edn (Bergmeyer, H. U., Bergmeyer, J. & Grassl, M., eds) vol. 6, pp. 262 - 267, Verlag Chemie, Weinheim, Deerfield Beach, Florida, Basel. 24. Clamp, J. R., Bhatte, T. & Chambers, R. E. (1971) in Methods of biochemical analysis (Glick, D., ed.) pp. 229 - 244, Interscience Publishers, New York, London, Sydney, Toronto. 25. Strominger, J. L., Park, J. T. & Thompson, R. E. (1959) J . Bid. Chem. 234,3263 - 3268. 26. Cynkin, M. A. & Ashwell, G. (1960) Nature 186, 155- 156. 27. Karkhanis, Y. D., Zeltner, J. Y., Jackson, J. J. & Dennis, J. C. (1978) Anal. Biochem. 85, 595 - 601. 28. Laemmli, U. K. (1970) Nature 227, 680-685. 29. Anselstetter, V. (1979) J . Chromatogr. 172, 49-56. 30. Tsai,C. M. &Frasch,C. E. (1982) Anal. Biochem. 119,115-119. 31. Hellerqvist, C. G., Lindberg, B. & Svenson, S. (1968) Carbohydr. Res. 8, 43 - 55. 32. Tanford, C. (1980) The hydrophobic effect: formation of micelles and biological membranes, John Wiley & Sons, New York. 33. Rosner, M. R., Tang, J., Barzilay, I. & Khorana, H. G. (1979) J . Biol. Chem. 254, 5906 - 5917. 34. Imoto, M., Kusumoto, S., Shiba, T., Naoki, H., Iwashita, T., Rietschel, E. Th., Wollenweber, H.-W., Galanos, C. & Luderitz, 0. (1983) Tetrahedron Lett. 24, 4017-4020. 35. Raetz, C. R. H., Purcell, S., Meyer, M. V., Qureshi, N. & Takayama, K. (1985) J . Biol. Chem. 260,16080- 16088. 36. Strain, S. M., Armitage, I. M., Anderson, L., Takayama, K., Qureshi, N. & Raetz, C. R. H. (1985) J . Biol. Chem. 260, 16089 - 16098. 37. Anderson, M. S., Bulawa, C. E. & Raetz, C. R. H. (1985) 1.Bid. Chem. 260, 15 536 - 15541. 38. Anderson, M. S., Robertson, A. D., Macher, I. & Raetz, C. R. H. (1988) Biochemistry 27, 1908-1917. 39. Goldman, R. C., Doran, C. C. & Capobianco, J. 0. (1988) J . Bacteriol. 170, 2185-2191.
