Field of the Invention The present invention relates to novel compounds which stimulate the activity of immune cells in animals in a manner similar to that of lipid , methods of pre¬ paring such compounds and methods of treatment using such compounds.

Description of the Prior Art Lipid A is a component of the bacterial lipopoly¬ saccharide which is a complex amphipathic molecule which covers the outer surface membrane of Escherichia coli and other gram-negative bacteria. Lipid A is a unique hydro- phobic anchor substance which holds the lipopolysaccharide molecule in place.

The exact structure of lipid A is still unknown; however, the components include a β,l→6-linked glucosamine disaccharide, 2-3 phosphate groups, 6-7 fatty acids, an ^* aminoarabinose, and ethanolamine. The lipid is structur- ally heterogeneous because the polar aminoarabinose and phosphorylethanolamine residues occur in only a fraction of the molecules. Microheterogeneity based on the ester-linked fatty acids is also indicated by the results of a study of the purified monophosphoryl lipid. The fatty acid composition of lipid A differs from that of the conventional phospholipids by the presence of β-hydroxymyristic acid and the near absence of un- saturated fatty acids.

There is considerable interest in lipid A, as well as its precursors and its metabolites because of its biological activity. Lipid A is believed to be responsible for the endotoxic, im unostimulating, tumor cell killing, and interferon production stimulating activities of the lipopolysaccharides. A comprehensive review of the chemistry and biology of lipid A can be found in the C. Galanos et al. article which appears in the "Inter-

national Review of Biochemistry, Biochemistry of Lipids II", Volume 14, University Park Press (1977).

Substantial work has been done to determine the molecular structure of lipid A and to identify the portions of the lipid A molecule which are responsible for its biological activity.

In an article in the Journal of Biological Chemistry, Vol. 254, No. 16, pp. 7837-7844 (1979) Masahiro Nishijima and Christian R. H. Raetz noted the presence of two unidentified lipids (X and Y) which accumulated in certain Escherichia coli mutants defective in phosphatidylgiyceroi synthesis, and appeared "to be metabolites of lipopoly¬ saccharide synthesis," pp. 7837. In a later article in Journal of Bacteriology, Vol. 145, No. 1, pp. 113-121 (1981), Nishijima, Raetz and Christine E. Bulawa charac¬ terized the lipids X and Y as "precursors of lipid A bio¬ synthesis", p. 118. Still later, Nishijima and Raetz, in an article in The Journal of Biological Chemistry, Vol. 256, No. 20, pp. 10690-10696 (1981) purified milligram quantities of lipids X and Y from their Escherichia coli mutants defective in phosphatidylgiyceroi synthesis, and they speculated that the lipids X and Y were disaccharides. It has now been discovered that lipid X is an acylated cc-D- glucosamine 1-phosphate containing β-hydroxy- myristoyl groups at position 2 and 3. It is believed to be a biosynthetic precursor of lipid A and it possesses a desirable ability to stimulate the activity of immune cells in a manner similar to lipid A.

Summary of the Present Invention It is an object of the present invention to disclose the structure of lipid X and derivatives of lipid X, the preparation of such compounds, the preparation of radio- chemically labeled material, pharmaceutical compositions containing such compounds and methods of treatment employing such compositions.

Based on fast atom bombardment mass spectro etry and proton nuclear magnetic resonance studies, we have dis-

covered that lipid X is an acylated monosaccharide derived from glucosamine 1-phosphate. Lipid X has a M of 711.87 as the free acid ( ^C 34 ^H g6 ^NO i ^P ^ ^and con ains two ^β -hydroxy- myristate moieties, one attached as an amide at the 2 position and the other as an ester at the 3 position of the sugar. It has free hydroxyl groups at the 4 and 6 positions, and the anomeric configuration is alpha. The structure of lipid X closely resembles the reducing end subunit of lipid A, and it might represent a very early precursor in the biosynthesis of lipid A.

The structure of lipid X may be represented as follows:

Lipid X and its immuno stimulating derivatives may be represented by the following general formula:

OMPI WIPO frl

in which the preferred sugar stereochemistry is that of glucosamine; A and B are the same or different, and are H, or a hydrocarbon structure (as defined below) or a fatty acyl chain (as defined below) or another functional group (as defined below); R. and R ₂ are the same or different, and are H or a hydrocarbon structure or a fatty acyl chain (preferably β-hydroxymyristoyl as in the natural product (I)). When a hydroxylated substituent is present on A, B, R, and/or R~ it may be further substi- tuted with a fatty acyl chain. Substituent Z is a water- solubilizing group such as a hydroxyl, a phosphate, a succinate, a sulfate, a sugar residue, or a nucleotide.

A hydrocarbon structure may be an alkyl or a hydoxyalkyl group of 1-24 carbon atoms, or it may be an alkenyl group of 2-23 carbons. A fatty acyl chain may be an alkanoyl or a hydroxyalkanoyl chain of 2-24 carbons, or it may be an alkenoyl chain of 3-24 carbons. A functional group may be a sugar or a water solubilizing group, such as a succinoyl residue, a phosphate, a sulfate, or a nucleotide.

Combinations are excluded, in which all four of the substituents (A^R., and R,) are H, and Z is a hydroxyl, a phosphate, a succinate, or a nucleotide; or in which all four substituents (A,B,R- _/ R-) are methyl or acetyl, or combinations thereof, and Z is a hydroxyl, a phosphate, a succinate or a nucleotide.

Lipid X may be chemically synthesized or it may be isolated from bacterial sources or modified by a combin¬ ation of approaches. The isolation of lipid X from a bacterial source and verification of its structure are described in the experi¬ mental work which follows:

Experimental Procedures Bacterial Strains and Media - Temperature sensitive E. coli K12 strains MN7 (ATCC No. 39328) was grown in LB broth, which contains 10 g of NaCl, 10 g of tryptone, and 5 g of yeast extract/liter. Maximum accumulation of lipid

X occurred when a log phase culture grown at 30°C to an absorbance at 550 nm of 0.4-0.6 was shifted to 42°C for

3 h. This procedure was used whether we grew small shaker cultures or large cultures (300 liters). At the time of the harvest, the A _CC was 1.8. In a typical 300

1. fermentation, the cells were harvested by centrifuga- tion through a continuous flow centrifuge and the cell paste was stored at -80°C. The yield was about 700 g. of cell paste per 300 1. fermentation. Growth Conditions for Radiochemical

Labelling of Lipid X

Cells of E. coli strain MN7 were prepared by first growing them at 30°C in 200 ml of LB broth to an absorbance at 550 nm of 0.5. Then 1 mCi of [1- 14C]-acetate (60 mCi/mmol or higher) was added to the culture and incubation was continued at 42°C for 4 hr. The cells were harvested by centrifugation at 5,000 x g for 15 min. These cells were the source of the 14C labeled lipid X.

32

To label lipid X with P , a 50 ml culture of MN7 growing on-medium was allowed to reach A- ₅₀ = 0.8 at 30°C. The cells were collected by centrifugation and resuspended in the same volume of medium lacking phos¬ phate. Next, the cells were incubated in shaking culture at 42°C, 32P. (100 μCi/ml carrier free) was added, and the incorporation of label was allowed to take place for 3 hours. Finally, the cells were recovered by centri¬ fugation, and the lipid X was obtained by the rapid radiochemical extraction described below.

Analytical and Preparative Thin Layer Chromatography (TLC) - TLC was performed either on silica gel H or 60 using either chloroform-methanol-water-concentrated ammonium hydroxide (50:25:4:2, v/v) (solvent A) or chloroform-pyridine-formic acid (20:30:7, v/v) (solvent B). Extraction and Fractionation of Lipids - Method I. Lipopolysaccharide was prepared from 110 g of cell paste

by the method of Galanos et al. Eur. J. Biochem. 9, 245-249(1969). The yield of crude lipopolysaccharide (including lipids X and Y) was 337 mg. Next, X and Y were extracted from this crude material with 90 ml of chloroform-methanol-water (30:10:1, v/v) to yield 93 mg of a mixture predominantly consisting of lipids X and Y. This preparation (59 mg) was subjected to preparative TLC using 20 x 20 cm silica gel H (500 ^* μm) plates and solvent A at a load of 3 mg/plate. The bands were visualized with I, vapor and recovered by extracting the silica gel with chloroform-methanol-water (66:33:4, v/v). The yield of lipid X was 30.3 mg.

Method II. This method is a modification of the acidic Bligh-Dyer extraction described in Can. J. Biochem Physiol. 37, 911-918(1959), which was designed for more rapid, large scale purifications. About 35 g of cell paste was suspended in 950 ml of chloroform-methanol-water (1:2:0.8, v/v), and the mixture was shaken vigorously in a l l Erlenmeyer for 60 min. at 30°C. The cell debris was removed by centrifugation at 5,000 x g for 10 min. To the supernatant was added 250 ml each of chloroform and water to yield a two-phase system. After further addition of 10 ml. of concentrated HC1 and vigorous shaking in a 2 liter separatory funnel, the layers were allowed to separate, and the lower phase was washed once with fresh, acidic pre-equilibrated upper phase. The washed lower layer was centrifuged to break the emulsion completely, and it was concentrated by rotary evapor¬ ation. The residue was dissolved in 60 ml of chloroform- methanol-water (2:3:1, v/v) and applied to a 1.5 x 25 cm column of DEAE cellulose (acetate form) . The column was successively washed with 100 ml of the same solvent and 100 ml of chloroform-methanol-40 mM ammonium acetate, pH 7.4 (2:3:1, v/v) . Finally lipid X (along with lipid Y, phosphatidic acid and cardiolipin) was eluted from the

column with 100 ml of chloroform-methanol-100 mM ammonium acetate, pH 7.4 (2:3:1, v/v).

The eluted material was detected by charring 5 μliter samples of each fraction, and the peak fractions were pooled (75 ml final volume). Next, enough chloro¬ form, methanol and phosphate-buffered-saline were added in Bligh-Dyer proportions to give an upper phase volume of 500 ml (approximately 1 ml of upper phase per 100 yg of lipid X) . Under these conditions lipid X partitioned into the aqueous-methanol phase, while Y and other phospholipids remained in the chloroform layer. The purified lipid X was recovered from the upper phase by adjusting the pH to 1.0 with HC1 and adding a fresh organic phase. This sample was dried by rotary evaporation, redissolved in 3 ml of chloroform-pyridine-formic acid (37:30:7, v/v), and finally purified on an 0.8 x 25 cm silicic acid column equilibrated in this solvent system. The pyridine and formic acid in samples from the final silicic acid column were removed by adding methanol and water in Bligh-Dyer proportions relative to the CHCl-. Additional concentrated HC1 was then added until the pH of the upper phase was 1. The upper phase was removed, and the lower phase was washed twice with pre- equilibrated acidic upper phase. The washed lower phase was dried under a stream of N ₂ to recover lipid X, pre¬ sumably as the free acid. Final recovery was about 30 mg and the material was stored dessicated at -80°C.

Lipid Y was isolated essentially by the same method as X. Following DEAE cellulose chromatography and partitioning at neutral pH (see above), the lower phase containing Y and some contaminating phospholipids was dried by rotary evaporation. The residue was redissolved in 4 ml of chloroform-pyridine-formic acid (60:30:7, v/v) . Final purification was achieved on a silicic acid column (0.8 x 25 cm) equilibrated with the same solvent. The pyridine and formic acid was removed as described

above for lipid X. Recovery of Y in the free acid form was about 12 mg/50 gm cell paste.

Rapid Preparation of Radiochemically Labeled Lipid X. Cells of MN7 labeled with ³² Ε _± or ¹ C acetate (as above) are extracted under Bligh-Dyer conditions but at pH 7 by use of phosphate-buffered saline as the aqueous component. In this case lipid X is recovered in the upper aqueous-methanol phase, while phospholipids and Y are in the lower phase. Relatively pure lipid X can be recovered from the upper phase by adjusting the pH to 1 with concentrated HC1 and adding fresh pre-equili- brated lower phase. In this way the X is shifted back to the lower phase, where it can be recovered. Radiochemical purity by TLC in solvents A or B is about 95%. This rapid preparation does not require column chromatography. If material of greater than 99% purity is required, this can be achieved by silicic acid chroma- tography in chloroform-pyridine-formic acid (37:30:7) as described above. Radiochemical preparation of lipid Y is not possible with the rapid technique.

Dephosphorylation of Lipid X - About 2-3 mg of sample was suspended by sonication in 2.0 ml of 0.1 N HC1, heated at 100° C for 15 min and cooled. Then 5 ml of chloroform-methanol (2:1, v/v) was added, mixed, and allowed to stand for 10 min. The upper aqueous layer was removed, and the upper aqueous layer of the blank chloro¬ form-methanol-water (10:5:6, v/v) mixture was used to wash the lower organic layer containing the dephosphory- lated product ("dephospho X") . The organic layer was filtered and dried with a stream of nitrogen. The extent of dephosphorylation was 80-90%.

Preparation of Dimethyl Derivative of Lipid X - About 2-3 mg of purified material was dissolved in 1.0 ml of chloroform-methanol (9:1, v/v) and treated with a few drops of diazomethane in diethyl ether that was sufficient to give a faint yellow color. This resulted in the

methylation of the phosphate group and the formation of a phosphate triester ("dimethyl X"). After 30 min at 25°C, the solution was dried with a stream of nitrogen and dissolved in 0.3 ml of CDC1- for NMR spectroscopy. Preparation of dephospho Y and dimethyl Y was carried out exactly as for the corresponding X derivatives.

* *

Preparation of lipid Y from lipid Y. Lipid Y was prepared from Y by hydrolysis in the presence of triethylamine (TEA). To 3 mg of lipid Y, 3 ml of water and 100 μl of TEA were added (0.24 M, pH 12.4). The mixture was heated at 100°C for 2 hours and evaporated to dryness under a stream of N,. By this procedure a con¬ trolled deacylation was achieved, yielding a partially

* deacylated lipid Y (designated Y ) and β-hydroxymyristate. The latter was removed from the residue with 2 ml of acetone. Lipid Y retains the ester-linked palmitate of lipid Y and has been analyzed further by FAB mass spectro¬ metry and NMR analysis.

Preparation of UDP-diacylglucosamine and other nucleo- tide derivatives of lipid X. To 5 mg of lipid X dissolved in anhydrous pyridine are added 6 mg of UMP-morpholidate. After an overnight incubation at 37°C in a tightly stop¬ pered tube, the product is purified by preparative thin layer chromatography in solvent B, using 500 ^* μm silica gel H plates. The nucleotide, which migrates just off the origin, is eluted and the yield is 70% of theoretical. FAB mass spectrometry of the nucleotide reveals a molecular weight of 1018, as predicted for a compound with structure V. The above method may be used to prepare any derivative of II in which Z is a nucleoside diphosphate, and may be used to make nucleotide derivatives of Y and Y .

Reaction of UDP-diacylglucosamine (V) with lipid X (I). Incubation of 0.2 mM V with 0.2 mM I in the presence of 1 mg per ml of a wild-type E. coli K12 cell-free extract in 20 mm HEPES buffer at pH 8 results in the formation of a disaccharide 1-phosphate with the structure of VI. This disaccharide can be reisolated from the

reaction mixture by the methods described above for the purification of lipid Y or by preparative TLC.

Chemical Synthesis and Modification of Lipid X and Related Compounds. Methods for selective acylation or alkylation of sugar hydroxyl groups are known. Products of partial acylation or alkylation can be separated from each other by high performance liquid chromatography or column chromatography.

Compounds of formula II can be chemically synthesized as shown in Scheme 1. Suitable known starting materials would be the allyl 2-acylamido-3- -acyl-4,6-0-benzylidene- 2-deoxy- -D-glucopyranosides.

SCHEME 1

OMPI

The allyl glycosides (1_) can be isomerized to 1-propenyl glycosides (2) by treatment with tris(triphenyl- phosphine) rhodium(I) chloride or with another suitable isomerization catalyst. Treatment of the 1-propenyl glycosides (2_) with mercuric chloride-mercuric oxide in an anhydrous medium will yield oxazolines (3_) . If water is included in the reaction mixture the products will be the 1-hydroxy compounds (4) .

The reaction of the oxazolines (3_) with precursors of Z groups, such as diesters of phosphoric acid or with alcohols or partially protected sugars in the presence of ferric chloride or other acid catalysts will result in the attachment of (protected) Z groups to position 1 of the sugar. Removal of the 4,6-0-benzylidene group, by hydrogenolysis or mild treatment with aqueous acid, and removal of any protecting groups from the Z-moiety, will give products of structure II, with A and B = H.

Other compounds of the series can be more expeditiously made by treating the 1-hydroxy compounds (4) with electro- philic reagents, designated Z ¹ in the scheme, such as acid anhydrides (e.g_. succinic anhydride), acid chlorides, or phosphorochloridates in the presence of suitable bases. Deprotection will again give II (A, B = H) .

If the products of the reactions of (3_) with ZH, and of (4) with Z', are treated so as to selectively remove the 4,6-0-benzylidene group (partial deprotection), then OH-6 and OH-4 can be alkylated or acylated, completely or selectively, by standard methods. After final depro¬ tection, the products would have the structure II, with A and B as defined above.

HPLC Fractionation - High Pressure Liquid Chroma¬ tography (HPLC) was performed with two 6000A solvent delivery systems a solvent programmer, a universal liquid chromatograph injector, a variable wavelength detector and a Radial Compression Module. A 8 mm x 10 cm Radial pak A cartridge (C _ιa -bonded silica, 10 μm) was

used. For the analysis of lipid X, a linear gradient of 0 to 100% 2-propanol-water (85:15, v/v) in acetonitrile- water (1:1, v/v) was used over a period of 60 min at a flow rate of 2 ml/min. Both solvent systems contained 5 mM tetrabutylammonium phosphate. Samples for HPLC analysis were dissolved in chloroform-methanol (4:1,v/v). The absorbance was monitored at 210 nm.

Mass Spectral analysis - FAB mass spectrometry was performed on a MS-50 mass spectrometer at ambient temper- ature, utilizing a neutral beam of xenon atoms from a saddle field discharge gun with a translational energy of 8 Kev and a discharge current of 0.4 itiA. Samples (100 μg) were dissolved in 100 μl of chloroform-methanol (1:1, v/v) and mixed with an equal volume of either triethanolamine (for negative mode) or monothioglycerol

(for positive mode). The sample (5 μliter) was deposited on the FAB probe tip and the volatile solvents were pumped away in the insertion lock chamber of the mass spectrometer. Both negative (M - H)~ and positive (M + cation) ion mass spectra were obtained by scanning at a rate of 90 atomic mass units per sec. Measurements were based on a cali¬ bration standard glycerol-potassium iodide (1:1, mol/mol) as cluster ions. Mass assignments were made with an accuracy of ±1.0 atomic mass units using the DS-55 data system.

Proton NMR Analysis - Spectra were recorded at 200 or 270 MHz, on Nicolet NT-200 and Bruker WH-270 Fourier transform, superconducting spectrometers interfaced with Nicolet 1280 and 1180 computers, respectively. In decoupling experiments the parent spectrum was first determined with the decoupler power set "off resonance"; then the decoupled spectrum was determined. Difference spectra were generated by subtracting the parent spectrum from the decoupled spectra. Results

Analysis of Purity of Lipid X - Purified lipid X was examined by analytical TLC using solvents A and B and

found to give a single major band which could be visual¬ ized with I ^ vapor, by charring, or by spraying with an organic phosphate-specific molybdenum reagent. Lipid X was readily separable from the incomplete lipid A which migrated very slowly in the thin layer system with solvent A.

Chemical analysis of a sample of lipid X purified by TLC, and containing some silica, showed the following: total phosphorus, 0.96 μmol/mg; glucosamine, 1.02 μmol/mg ^" ; KDO, 0.04 μmol/mg. The phosphorus-glucosamine- DO molar ratio was 1.00:1.06:0.04. The presence of glucosamine as the sole amino sugar was confirmed by the use of the amino acid analyzer on an acid hydrolyzed sample. These results confirmed the results of a previous chemical analysis of lipid X.

Purified lipid X was analyzed by reverse-phase HPLC before and after preparative TLC fractionation (Method I). This analysis showed a single peak of lipid X eluting at 30 min with an estimated purity by peak height analysis of at least 95%. There was no evidence for the presence of a homologous series or further microhetero- geneity of lipid X. Under these conditions of HPLC, a purified monophosphoryl lipid A designated TLC-3 from S . typhimurium and containing 6 fatty acid residues came off the column at the end of the gradient.

Nature of the Phosphate Group in Lipid X - When [ ¹⁴ C]lipid X was treated with 0.1 N HC1 at 100°C for 15 min, a new product was formed with a R- value of 0.53 on TLC in solvent A (unhydrolyzed lipid X had R _f 0.12). The radioactivity pattern of the TLC separation showed that almost all of the label in [ ¹⁴ C]lipid X was shifted to the new product after hydrolysis. The time-course of

14 acid-catalyzed hydrolysis of [ C]lipid X was recorded.

Over a period of 15 min, there is a rapid decrease of

14 [ C]lipid X and a corresponding rise in the level of new product.

The distribution of the phosphorus content in the aqueous and organic phases of lipid X before and after 15 min hydrolysis is shown in Table I.

Table I

Distribution of phosphate in the aqueous and organic phase of control and acid-hydrolyzed lipid X. The sample was prepared as described. Then 1/3.25 volume of the aqueous phases and 1/3.75 volume of the organic phases were analyzed for phosphate content.

nmol phosphate

Two-phase system Control Acid hydrolyzed

Aqueous layer ^a 48.4 495.3 Organic layer 564.8 97.5

•a

Inorganic phosphate assay (digestion was omitted).

The phosphate content is shifted from the organic phase before hydrolysis to the aqueous phase after hydrolysis. By utilizing the data in Table I and establishing the partition coefficient of lipid X in the chloroform-methanol-water (20:10:9, v/v) system, the extent of conversion after 15 min hydrolysis was calcu- lated to be 78%. Quantitative kinetic analysis of the hydrolysis reaction is complicated by the tendency of lipid X to aggregate in aqueous solutions, and by the precipitation of the product (dephospho X). A logarithmic plot of the time-course data reveals a modest decrease of the hydrolysis rate constant with time. The evidence supports the conclusion that lipid X contains a single type of acid-labile phosphate, attached to the 1-position of the glucosamine.

The results of negative and positive mass spectrometry established the precise value of M for lipid X to be 711.87 as the free acid (C ₃₄ H ₆ gN0 ₁₂ P) . A molecule of

lipid X contains one glucosamine, two hydroxymyristate residues, and one phosphate, the latter linked to position 1 of the sugar. Since one of the hydroxy- myristic linkages is stable to mild alkali and the other is cleaved by mild alkaline treatment, one hydroxymyristate must be amide-linked (at position 2 of the glucosamine) and the other ester-linked.

Proton NMR Analysis of Lipid X - Four possible sites for the ester-linked hydroxymyristate had to be considered, namely positions 3,4, and 6 of the glucosamine residue, and the β-position of the amide-linked hydroxymyristoyl group. The principal purpose of the NMR spectroscopic analysis was to distinguish between these alternatives. Initially, spectra were taken of the free acid form of lipid X dissolved in CDC1-, but these were poorly resolved, presumably because of the aggregation of the amphipathic lipid in solution. To obtain satisfactory resolution it was necessary to esterify the phosphate moiety with diazomethane, or remove it, and dissolve the resulting derivatives (dimethyl X and dephospho X, respectively) in CdCl ₃ with 1-10 percent dimethyl sulfoxide-d _β .

The spectrum of dephospho X (not shown) resembled that of dimethyl X. Data from decoupling experiments on dephospho X substantiated the assignments made from experiments with dimethyl X.

The normal range of the resonances for H-3 and H-4, and the downshifting of these resonances on acylation at the respective positions, is well established by obser¬ vations on numerous glucosamine derivatives, including synthetic lipid A analogs. Thus, the downfield position of the H-3 signal in the present case clearly delineates 0-3 of the glucosamine as the site of attachment of the ester-linked hydroxymyristate in lipid X.

Discussion The complete structure of the glycolipid, lipid X, which accumulates in certain phosphatidylglycerol- deficient mutants of E. coli (particularly at nonper-

missive temperature) has now been established. Lipid X is 2-deoxy-2- [ (R)-3-hydroxytetradecanamido]-3-0-[(R)-3- hydroxytetradecanoyl]-α-D-glucopyranose 1-phosphate and it bears a striking resemblance to the reducing end subunit of lipid A. The ^' structure of lipid Y, which is essentially the same as X but contains an additional esterified palmitoyl residue on the βOH of the N-linked hydroxy¬ myristate, has also been completed using techniques essentially identical to those described above for X. Lipid X might be a very early precursor of lipid A biosynthesis. Thus radiolabeled lipid X could be useful as a substrate to study the pathway for the enzymatic synthesis of lipid A in a cell-free system.

Lipid X might serve as a good model compound to study the relationship between the structure of lipid A and its numerous biological activities. The attractive feature of lipid X is the simplicity of its structure and the ease with which the structure can be modified by controlled degradation. Preliminary experiments show that lipid X is relatively nontoxic in the chick embryo lethality test and that it is a B-cell mitogen. There appears to be an interaction between phosphatidylgiyceroi metabolism and lipid A bio¬ synthesis. Mutants like 11-2 or MN7 (pgsA444 pgsBl) are isolated by a two stage mutagenesis. They are temp¬ erature-sensitive for growth and phosphatidylglycerol- deficient only when both genetic lesions are present. Based on subcloning of pgsA and detailed enzymological studies, it is certain that pgsA represents the structural gene for phosphatidylglycerolphosphate synthase. The pgsB may code for an enzyme in the lipid A pathway, for instance one that converts lipid X to the next inter¬ mediate. In this regard, it is relevant that strains with the genotype pgsA pgsBl regain normal phosphatidyl- glycerol levels and are not temperature-sensitive, but they continue to have more than 100 times the lipid X present in wild-type cells (pgsA pgsB ). Perhaps, the

OMPI ,W IPO .^J

accumulation of lipid X caused by the pgsBl mutation interferes with phosphatidylglycerophosphate production when the synthase specified by the pgsA444 allele is present. Furthermore, phosphatidylgiyceroi itself might be required for lipid X utilization, since all components of phospholipid metabolism and lipid A biosynthesis are membrane-bound.

The Biological Activity of Lipid X The outer membrane of gram-negative bacteria such as Escherichia coli and Salmonella typhimurium consists of proteins, phospholipids and lipopolysaccharide. The latter substance is localized almost exclusively on the outer surface of the outer membrane. It accounts for many of the immunological and endotoxic properties of gram-negative organisms. Although the complete structure of lipopolysaccharide is unknown, it consists of three domains. The outer sugars, which are highly variable, give rise to the O-antigenic determinants. The core sugars are relatively conserved and may be involved in cell penetration by certain bacteriophages. The lipid A molecule (which is highly conserved between species) functions as a hydrophobic anchor holding lipopolysaccharide in place. Since lipopolysaccharide represents several percent of the dry weight of gram- negative bacteria, it follows that lipid A (and not phospholipid) must account for most of the outer leaflet of the outer membrane.

Since lipid X is easy to purify and can be further manipulated by controlled chemical degradation, we have examined its ability to activate mouse lymphocytes. We have found that our lipid X preparations are almost as active as intact lipopolysaccharide by this criterion, and that the ester-linked hydroxymyristate at position 3 is crucial for this biological function. The results suggest that lipid X, like lipid A, is a B cell mitogen.

Methods for Evaluating Mitogenic Activity of Lipid X

Lipid X was isolated from strain MN7. The mild alkaline hydrolysis product of lipid X (which retains the N-linked but not the O-linked hydroxymyristate residue) was obtained under standard conditions for

* deacylation of phospholipids. Lipid Y was obtained from lipid Y by treatment with triethylamine for 3 hours at 100°C. Dephosphorylated lipid X was generated by mild acid treatment (0.1 M HC1 at 100°C for 60 min). A lipid A derivative lacking the 1-phosphate moiety at the reducing end was generated as described in the literature. Microspheres (polystyrenedivinylbenzene beads, 5.7 ± 1.5 μM diameter, were coated with lipopolysaccharide or dephospho-X by known methods.

Mitogenesis Experiments

C57B1/10 and C3H/HeJ mice were anesthetized with ether, and the spleens were excised immediately. After opening the splenic capsule, the cells were suspended in 10 ml of DMEM, washed twice with the same, and then g resuspended at 5 x 10 /ml in DMEM, supplemented with

10% fetal bovine serum, 2mM L-glutamine, 1 mM sodium pyruvate, 10 mM MEM non-essential amino acids, 10 mM

Hepes, 50 μM 2-mercaptoethanol, 100 units/ml penicillin G and 100 μg/ml streptomycin. Multiple wells of a 96 well

Costar microtiter dish were seeded with 0.1 ml of this cell suspension (5 x 10 /well). Next, an additional 0.1 ml of

DMEM supplemented with an appropriate amount of test mitogen was added. Following addition of mitogen, the dishes were incubated for 2 days at 37°C, 100% humidity and 7.5% CO,.

3 Finally, fmethyl- H]-thymidine was added at 1 μCi/well, and the cells were incubated at 37°C for an additional six hours. Cells were washed free of medium and excess thymidine with 0.9% NaCl using a Bellco microharvester. The washed cells were retained on glass fiber filter strips, and radioactivity incorporated into the cells was quantitated by liquid scintillation counting. Each mitoσen

concentration was tested in triplicate wells, and the standard deviation of the measurement was approximately ±10%. Induction of plaque forming cells by various mitogen preparations was quantitated using a monolayer of sheep red blood cells as the indicator.

Results Mitogenic effect of the chloroform-soluble fraction from mutant MN7. Membrane phospholipids were extracted under acidic conditions from a strain of E. coli which is wild-type with respect to its membrane lipids or from mutant MN7 which accumulates lipid X. The crude lipids were dried under ₂ to remove CHC1-, suspended in 1 mM EDTA adjusted to pH 6 with NaOH, and dispersed at a concentration of 1-2 mg/ml by sonic irradiation for 5 min at 25°C in bath sonicator. E. coli lipopolysaccharide was dissolved in 1 mM EDTA in a similar manner. Next, triplicate sets of mouse lymphocyte cultures were incubated for 48 ^" hours with increasing amounts of added phospholipid (0-5 μg/ml final concentration) . Following this, the cells were labeled for 6 hours with rmethyl-3

₃ H]-thymidine, and incorporation of H into DNA was deter¬ mined.

The phospholipid dispersions derived from MN7 stimu¬ lated lymphocyte proliferation to a much greater extent than similar preparations from strains of E. coli with a wild-type lipid composition. The stimulation of cells by commercial E. coli lipopolysaccharide was compared. The results suggest that there is an additional mitogenic factor in the CHC1,-soluble fraction of strain MN7 that is not present in the wild-type. The main difference between MN7 and wild-type E. coli lipids has been attributed to the presence of glycolipids X and Y in the mutant at levels that are 10,000-fold (or more) greater than normal. Mitogenic Activity of Purified Lipid X

The predominant glycolipid that accumulates in MN7 is lipid X, a diacylglucosamine 1-phosphate, substituted

with β-hydroxymyristate at positions 2 and 3. Lipid X is readily dispersed in 1 mM EDTA (pH 6) at concentrations of 1-2 mg/ml, especially with mild ultrasonic irradiation. Dispersions of lipid X (tested in the range of 0-50 μg/ml final concentration) are almost as effective as commercial lipopolysaccharide in stimulating lymphocyte proliferation. However, unlike the commercial lipoply- saccharide, this activity can now be attributed directly to a highly purified and structurally defined molecule. Phosphatidic acid, which is structurally similar to lipid X, has no mitogenic activity. Phosphatidylcholine, sphingomyelin, phosphatidylinositol, cardiolipin, phos- phatidylserine, CDP-diglyceride, lysophosphatidylcholine, lysophosphatidylethanolamine and lysophosphatidic acid are also ineffective. Polymyxin B (50 μg/ml), which inhibits lipopolysaccharide-induced mitogenesis, also abolishes lymphocyte stimulation by lipid X. This supports the notion that lipids A and X have similar mechanisms of action and that the stimulation of the B lymphocytes is involved.

To further demonstrate that lipid X exerts its effects by the same mechanism(s) as lipopolysaccharide (or lipid A), we examined splenic lymphocytes from C3H/HeJ mice. These are unresponsive to lipopoly- saccharide, presumably because they lack a membrane receptor (or enzyme) that recognizes this molecule. The C3H/HeJ lymphocytes also respond poorly (if at all) to lipid X. However, they are readily activated by the T cell mitogen concanavalin A, or by PPD-tuberculin, which function by different mechanisms. T cells do not seem to be required for lipid X mitogenesis, since splenic lymphocytes from nude athymic mice respond well to lipid X and lipopolysaccharide, but not to concanavalin A. Molecular Requirements for Mitogenicity Mild alkaline hydrolysis of lipopolysaccharide abolishes its mitogenic activity. Since the locations of the ester-linked fatty acids in lipid A are not

precisely known, it is unclear which particular ester- linkage is required. The very fact that lipid X has strong mitogenic activity implies that the β-hydroxymy- ristate esterified at the 3 position of this molecule must mediate this function. To prove this, we subjected lipid X to mild alkaline hydrolysis and isolated a monoacyl glucosamine 1-phosphate derivative, bearing only the amide-linked β-hydroxymyristate. The removal of the esterified hydroxymyristate completely abolishes the B cell proliferation. Simultaneous addition of equimolar β-hydroxymyristate and monoacyl glucosamine 1-phosphate or of glucosamine 1-phosphate plus β-hydroxymyristate did not cause significant proliferation.

In addition to lipid X, mutant MN7 accumulates lipid Y, especially after 3 hours at 42°C. This material is similar to X but has an additional esterified palmitate moiety on the β-hydroxyl group of the N-linked hydroxy¬ myristate.

Lipid Y is much less soluble in H,0 than lipid X, but can still be dispersed at 1 mg/ml by sonic irradiation at pH 6. The material is also mitogenic, though somewhat less under these conditions, possibly because of poor solubility. Selective removal of the

3-0-linked -hydroxymyristate from the glucosamine ring of lipid Y with triethylamine gives rise to the derivative * Y . This substance retains the esterified palmitate and therefore is very similar to lipid X in its physical

* properties. However, lipid Y is less active as a mitogen and in fact may be a useful specific inhibitor. Mild acid treatment can be used to remove the

1-phosphate moiety from lipid X, leaving the two fatty acid residues in place. The resulting substance, termed dephospho lipid X, is very insoluble in H.O and cannot be dispersed by sonic irradiation. Fine suspensions of dephospho lipid X or preparations immobilized on hydro- phobic beads do cause slight cell proliferation (not shown). The finding that dephospho lipid X retains

significant biological activity strongly suggests that the sugar 1-phosphate moiety is not obligatory for the mitogenic response. This conclusion is further supported by the observation that TLC-3, a lipid A derivative from S. typhimurium G30/C21 lacking the 1-phosphate residue, also is fully mitogenic when tested under the same conditions.

Formation of PIagu -Forming Cells We evaluated the stimulation of plaque-forming cells by I . coli lipopolysaccharide, lipid X, or other prepar¬ ations. All derivatives which stimulated radioactive thymidine incorporation also increased the incidence of antibody-producing cells significantly. These results show that lipid X, lipid Y, and perhaps also dephospho lipid X all stimulate true lymphocyte proliferation and

3 that the observed increase in fmethyl- H]thymidine incorporation is not a radiochemical artifact.

Since the complete covalent structure of lipid A is not known, it has been difficult to elucidate the molecular mechanisms by which lipid A (or lipopolysaccharide) trig¬ gers diverse physiological responses, such as B cell proliferation or endotoxic shock. The discovery of biologically active monosaccharide lipid A fragments with defined structures makes it possible for the first time to explore lipid A function at a biochemical level. The lipid X molecule retains most of the properties of intact lipopolysaccharide with respect to the induction of B lymphocyte proliferation. In this case the ester-linked -hydroxymyristate at position 3 of the glucosamine ring is critical for function, while the phosphate moiety may enhance biological activity by increasing the solubility of the acylated sugar.

Lipid X also possesses other activities normally associated with lipopolysaccharide. Recently we have found that sheep respond to intravenous injection of lipid X in a manner that resembles the pathophysiological effects of Gram-negative endotoxin including both the

characteristic pulmonary hypertension and the increased lung lymph flow. Further, the limulus lysate assay for Gram-negative endotoxin is positive with lipid X under certain conditions, and removal of the ester-linked hydroxymyristate abolishes clot-forming activity. On the other hand, lipid X is relatively nontoxic as judged by the chick embryo lethality test (CELD ₅₀ >10 μg) . It appears that many of the biological activities of lipid A are mediated by the esterified hydroxymyristatoyl group at position 3 of the sugar.

The compound lipid X and its immunostimulating derivatives may be introduced to the immune cells of an animal in place of lipid A to stimulate the immune cells when such stimulation is desired. When thus employed the compound(s) may be administered in the form of parenteral solutions containing the selected immunostimulating compound in a sterile liquid suitable for intravenous administration. The exact dosage of the active compound to be administered will vary with the size and weight of the animal and the desired immunological response.

Generally speaking, the amount administered will be less than that which will produce an unacceptable endotoxic response in the animal.

It will be readily apparent to those skilled in the art that the foregoing description has been for purposes of illustration and that a number of changes may be made without departing from the spirit and scope of the invention. For example, although specific microorganisms that produce lipid X in recoverable amounts have been described, it will be apparent that other microorganisms, including those modified by genetic engineering, may be used to prepare lipid X. It will also be apparent that lipid X and some of its useful derivatives may be chemically synthesized using conventional techniques. Therefore, it is to be understood that the invention is not to be limited except by the claims which follow:

C PI