THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 261, No. 23, Issue of August 15, pp. 10624-10631,1986 Printed in U. S. A.

Characterization of a Structural Series of Lipid A Obtained from the Lipopolysaccharides of Neisseria gonorrhoeae COMBINED LASER DESORPTION AND FAST ATOM BOMBARDMENT MASS SPECTRAL ANALYSIS OF HIGH PERFORMANCE LIQUID CHROMATOGRAPHY-PURIFIED DIMETHYL DERIVATIVES*

(Received for publication, February 19,1986)

Kuni TakayamaSglI, Nilofer QureshiS, Karen Hyver((, Jeff HonovichII , Robert J. Cotter11 , Paolo Mascagni**, and Herman SchneidergQ From the SMycobacteriology Research Laboratory, William S. Middleton Memorial Veterans Hospital, Madison, Wisconsin 53705, the $Department of Bacteriology, College of Agricultural and Life Sciences, University of Wisconsin, Madison, Wisconsin 53706, the I( Department of Pharmacology and Experimental Therapeutics, The Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, the **School of Pharmacy, University of London, London, United Kingdom, and the §§Department of Bacterial Diseases, Walter Reed Army Institute of Research, Washington, D. C. 20307

Monophosphoryl lipid A (MLA) obtained from the lipopolysaccharides of serum-sensitive strains of Neis- seria gonorrhoeae was fractionated on a silicic acid column to yield the hexaacyl and pentaacyl MLAs. The dimethyl derivative of the hexaacyl MLA was analyzed by proton nuclear magnetic resonance spectroscopy. The dimethyl esters of hexaacyl and pentaacyl MLAs were further purified by reverse-phase high perform- ance liquid chromatography, and all of the peaks were analyzed by laser desorption mass spectrometry. Con- siderable structural information was obtained by laser desorption mass spectrometry due to three kinds of specific fragmentations of the sugar at the reducing end. Two major fractions were also analyzed by posi- tive ion fast atom bombardment mass spectrometry. High performance liquid chromatography was able to separate the dimethyl MLA according to number, na- ture, and position of the fatty acyl groups. Since almost no structural information is available, the mass spectra of the samples were interpreted on the basis of the established structure of a model lipid A (hexaacyl MLA derived from Salmonella minnesota).

Thirteen different structures of dimethyl MLA were identified. The four prominent dimethyl MLAs found in the fractionated samples were MI (M, = 1463), Mz (Mr = 1479), M3 (Mr = 1661), and M4 (Mr = 1677). These MLAs appear to have a 1’ 4 6 linked glucosa- mine disaccharide backbone. The most prominent hexaacyl MLA was Ms. We propose that it contains hydroxylaurate at the 3- and 3’-positions in ester link- age and lauroxymyristate at the 2- and 2’-positions in amide linkage of the glucosamine disaccharide. The most abundant pentacyl MLA was Mz. We propose that it contains hydroxylaurate at the 3- and 3’-positions in ester linkage, lauroxymyristate at the %’-position in amide linkage, and hydroxymyristate at the 2-position in amide linkage of the disaccharide. The lipid A of N. gonorrhoeae appeared to differ from that of the Sal-

* This work was supported in part by the Research Service of the Veterans Administration, National Science Foundation Grants PCH- 8209954 and CHE-8410506, and National Institutes of Health Grants GM-33967 and GM-36054. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

(I To whom correspondence should be addressed William S. Mid- dleton Memorial Veterans Hospital, 2500 Overlook Terrace, Madison, WI 53705.

monella strains by the presence of shorter-chain fatty acids and by the normal fatty acid distribution in the reducing and distal subunits.

Strains of Neisseria gonorrhoeae can be defined in vitro by their susceptibility to the bactericidal action of normal human sera (1-4). Serum-sensitive strains are killed by complement- mediated immune lysis involving antibody directed against the components located on the surface of the outer membrane of the bacteria (1, 2). This action prevents the dissemination of Neisseria beyond the initial site of colonization in the mucous membrance of the host (5-10). Recent studies by Schneider and associates (4, 11-13) showed that serum re- sistance results from the absence of natural antibody directed against the LPS’ determinants that are the lytic loci on the bacterial surface. Study of the chemical structure of gonococ- cal LPS from the serum-sensitive and serum-resistant orga- nisms is important to our understanding of their role in natural human immunity and their potential as vaccines.

LPS are amphipathic macromolecules found on the outer surface of the outer membrane of Gram-negative bacteria and are composed of three regions of differing chemical and bio- logical properties (14). The polysaccharide carries the main serological specificity. It is linked to the core polysaccharide that is common to Gram-negative bacteria. The core is linked through KDO to the lipophilic lipid A component, which is responsible for most of the biological activities of LPS. The polysaccharide region of the LPS of N . gonorrhoeae differs from that of the Salmonella strains in that it is smaller ( M , 3100-7100) (11) and apparently lacking in the repeating sub- units. It appears to have multicomponents in which each component expresses different antigenic specificity (12). The structure of the polysaccharide region of N . gonorrhoeae is presently under intense investigation.’ The lipid A region of the LPS might also be important in influencing the lytic action of the host’s immune system. However, very little is known about its structure.

The complete structures of lipid A’s obtained from the LPS

The abbreviations used are: LPS, lipopolysaccharides; LD, laser desorption; FAB, fast atom bombardment; MLA, monophosphoryl lipid A , KDO, 2-keto-3-deoxyoctonate; HPLC, high performance liquid chromatography.

2R. Yamasaki, J. M. Griffiss, and H. Schneider, unpublished results.

10624

Lipid A Obtained from LPS of N. gonorrhoeae 10625

of Salmonella typhimurium and Salmonella minnesota were recently established (15-18). In these studies, the dimethyl derivatives of the MLA were first purified by HPLC and analyzed by modern instrumental techniques; i.e. FAB-mass spectrometry and two-dimensional NMR spectroscopy. We have now purified the dimethyl MLAs obtained from LPS of N . gonorrhoeae by HPLC and have analyzed the major frac- tions by both LD- and FAB-mass spectrometry. This is the first time that mass spectrometry has been used to closely follow the extent of purification of lipid A by HPLC fraction- ation. We obtained four prominent and structurally different "free" lipid A's from the LPS of a serum-sensitive F62 strain of N. gonorrhoeae.

EXPERIMENTAL PROCEDURES

Materials-HPLC-grade chloroform, methanol, acetonitrile, and isopropyl alcohol were purchased from Burdick & Jackson Labora- tories, Inc., Muskegon, MI. Bio-Si1 HA (-325 mesh) was purchased from Bio-Rad; silica gel H thin-layer plates (250 pm) were purchased from Analabs, Inc., North Haven, CT.

Analytical Procedures-Total phosphorus and KDO contents were determined by the methods of Bartlett (19) and Osborn (20), respec- tively. Samples for glucosamine assay were hydrolyzed in 3 N HCl for 4 h a t 95 "C and analyzed by the method of Enghofer and Kress (21).

Sample (2.0 mg) for fatty acid analysis was hydrolyzed in 1.0 ml of 4 N HC1 at 100 "C for 4 h. The reaction mixture was extracted thrice with 4 ml of chloroform. The extract was dried, methylated with diazomethane, and analyzed by gas liquid chromatography. Analytical gas liquid chromatography was carried out on a Packard model 428 series gas chromatograph with a glass 3.4-m X 0.2-mm column con- taining 3% SP-2100 DOH, 100/120 mesh Supelcoport (Supelco, Inc., Bellefonte, PA). A flame ionization detector was used, and the injector temperature was 200 "C. The column was programmed at 4 "C/min from 150-230 "C, and the flow rate was 37.5 ml/min.

Anaytical TLC was performed on MLA preparations using silica gel H (250 pm) and the solvent system of chloroform/methanol/ water-concentrated ammonium hydroxide (5025:42, v/v). Standard hexaacyl and pentaacyl MLAs were prepared from LPS obtained from S. minnesota R595, as previously described (18).

Growth of Bacteria and Preparation of LPS-Two serum-sensitive strains of N. gonorrhoeae, F62 and WR213, were cultured on GC agar base (Difco) containing 1% (v/v) defined supplement (22) a t 37 "C in humidified candle extinction jars. The LPS were prepared from acetone-powdered organisms by the hot phenol-water method (23) and purified by four cycles of 100,000 X g centrifugation for 3 h.

Preparation of Lipid A-The LPS from N. gonorrhoeae (strain F62 or WR213) were suspended in 0.1 N HCl at a concentration of 2.5 mg/ml and sonicated for 10 min. The mixture was heated at 100 "C for 20 min and cooled. Five volumes of chloroform/methanol (2:1, v/ v) were added to two volumes of the hydrolysis mixture and mixed thoroughly. This mixture was centrifuged at 5000 X g for 15 min, and the clear lower organic layer plus the middle layer were recovered and evaporated to dryness on a rotary evaporator. The yield of this crude MLA was about 30-35%. A TLC analysis of such a preparation is given in Fig. 1.

Fractionation of MLA on Silicic Acid Column-About 70 mg of MLA from strain F62 (as the free acid) were applied to a 2 X 25-cm silicic acid (Bio-Si1 HA) column packed as a slurry in chloroform. The column was washed with 50 ml of chloroform. The MLA was eluted from the column with a linear gradient of 0-25% methanol in chloroform (700 ml). Six-ml fractions were collected and analyzed by spot-charring on a silica gel thin-layer plate. Char-positive peak fractions 45-57 (fraction I, 12.1 mg), 58-66 (fraction 11, 7.6 mg), 67- 79 (fraction 111, 6.1 mg), and 80-116 (23.8 mg) were pooled. Fraction I contained the hexaacyl MLA, fractions I1 and I11 contained the pentaacyl MLA, and the last fraction contained the unhydrolyzed materials. The result of TLC analysis of these fractions is shown in Fig. 1. Note that fractions I1 and I11 were contaminated with small amounts of hexaacyl MLA.

Methylation of MLA-The silicic acid-purified MLAs (fractions I- 111) were each dissolved in 1.0 ml of chloroform/methanol (41, v/v) and passed successively through a 0.8 X 3-cm Chelex 100 column (Bio-Rad) and a 0.8 X 3-cm Dowex 50 (H+) column. The columns were washed with 2.0 ml of chloroform/methanol to yield the free acid in the effluent. A slight excess of diazomethane was added to

Front -

Origin-

FIG. 1. Analytical TLC of silicic acid column purified MLA fractions. The source was LPS of F62 strain of N. gomrrhoeae. Silica gel H and a solvent system of chloroform/methanol/water- concentrated ammonium hydroxide (50:25:4:2, v/v) were used. Spots were visualized by spraying the plate with dichromate-sulfuric acid reagent and charring. Identification of samples (20-30 pg): A, crude MLA mixture before fractionation; B, silicic acid column fraction I; C, silicic acid column fraction 11; D, silicic acid column fraction 111.

each sample and allowed to stand at 22 "C for 2 min. The samples were immediately dried with a stream of nitrogen in a warm water bath; they represented the dimethyl derivatives of the hexaacyl and pentaacyl MLAs.

HPLC Fractionation-HPLC was performed with two Waters 660 solvent programmers (Waters Associates Inc., Milford, MA), a Waters U6K universal liquid chromatograph injector, a variable wavelength detector (model LC-85B, Perkin-Elmer), and a radial compression module (model RCM-100, Waters Associates, Inc.). A Radial Pak cartridge (8 mm X 10 cm) (C18-bonded 10-p silica, Waters Associates, Inc.) was used at a flow rate of 3 ml/min. For the fractionation of dimethyl MLA, a linear gradient of 20-80% isopropyl alcohol in acetonitrile was used over a period of 60 min (18). The wavelength of the detector was set at 210 nm.

LD-Mass Spectrometry-LD-mass spectra were obtained on a CVC-2000 (Rochester, NY) time-of-flight mass spectrometer equipped with a Tachisto (Needham, MA) model 215G pulsed carbon dioxide laser (24). The laser wavelength was 10.6 pm. Pulse width was 40 ns, and the power density of each pulse was approximately lo6 W/cm2. The repetition rate was approximately 1 Hz. Twenty to 30 laser shots were used for each mass spectrum. The ions produced from each laser shot were gated and accelerated to 3 KeV into a 1-m drift region. They were then postaccelerated to 12 KeV and detected using dual Galileo channel plates (Sturbridge, MA). The analogue signal was recorded as 8 K channels on a LeCroy (Spring Valley, NY) 3500SA signal averaging system with a 100-MHz analogue-to-digital waveform recorder. Mass assignments were made by determining the time-of-flight (channel) centroid of each peak and comparing them with the centroids of Na+ and K+ ions that appeared in each spectrum. Mass accuracy was approximately f l atomic mass units a t 1000 atomic mass units. KC1 was added to each sample to promote pro- duction of MK' molecular ions.

Positive FAB-Mass Spectrometry-Positive ion FAB-mass spectra were acquired using a KRATOS MS-50 (Manchester, UK) mass spectrometer equipped with a 23-KG magnet and postacceleration detector. The instrument was operated at an accelerating voltage of 8 keV and 3000 resolution. A KRATOS DS-55 data system interfaced to the mass spectrometer WBS used to acquire data and to average spectra. Mass calibration was achieved using cesium iodide. Samples

10626 Lipid A Obtained f rom LPS of N. gonorrhoeae

were dissolved in chloroform/methanol (4:1, v/v) and placed on a copper probe tip. Monothioglycerol was used as the sample matrix. The samples were bombarded with an 8-kV xenon atom beam gen- erated by a saddle field source (Ion Tech, Middlesex, UK).

Proton NMR Analysis-TLC-purified dimethyl hexaacyl MLA (methylated fraction I, 5.9 mg) was dissolved in 0.5 ml of benzene-&- dimethyl sulfoxide-& (9:1, v/v) and analyzed on an XL-300 MHz Varian spectrometer. Spectra were recorded at an operating temper- ature of 33 +- 0.5 "C.

RESULTS

Preliminary Studies of LPS and Lipid A Obtained from N. gonorrhoeae-The LPS obtained from strains F62 and WR213 of N. gonorrhoeae were hydrolyzed in 0.1 N HCl a t 100 "C for 20 min to yield the MLA. Analytical TLC of such a sample from the two strains gave identical results. A rep- resentative TLC showed that two major bands co-migrated with authentic hexaacyl MLA and pentaacyl MLA obtained from the LPS of S. minnesota (18) (Fig. 1). The following bands were identified: hexaacyl MLA, R F = 0.32; pentaacyl MLA, RF = 0.26; tetraacyl MLA, R F = 0.23; diphosphoryl lipid A, R F = 0.12; unhydrolyzed LPS, origin. Only the hexaacyl and pentaacyl MLA were further studied.

Chemical analysis of the LPS showed the presence of KDO, glucosamine, and phosphate, whereas the KDO content in the crude MLA preparation was reduced to 0.01 lmol/mg to give a phosphate-to-KDO ratio of 1.OO:O.Ol. The crude MLA con- tained lauric acid, hydroxylauric acid, and hydroxymyristic acid in a ratio of 1.00:0.65:0.85. Traces of palmitic acid were also detected. Based on chemical analysis and analytical TLC, the MLAs prepared from LPS of the F62 and WR213 strains were nearly identical.

Purification of Dimethyl MU-MLA was prepared from the LPS obtained from strain F62 and fractionated on a silicic acid column by the method of Qureshi et al. (18). The column fractions (1-111) of MLA were methylated with diazomethane and subjected to preparative HPLC using a reverse-phase C18- bonded silica cartridge, as previously described (18). Subse- quent mass spectral analysis revealed major components des- ignated as MI = 1463 atomic mass units, MS = 1479 atomic mass units (pentaacyl MLA), M, = 1661 atomic mass units, and M, = 1677 atomic mass units (hexaacyl MLA), as well as several minor components.

Mass Spectral Analysis-Fraction I was separated by HPLC into four peaks, all of which were subjected to mass spectral analysis using LD (Fig. 2). The two major HPLC peaks (1-2 and 1-3) gave molecular ions (M + K+) at m / z 1716 and 1700, respectively (Fig. 2, c and d ) , corresponding to the two major hexaacyl components with molecular weights of 1677 (M,) and 1661 (M3). The mass spectrum of HPLC peak 1-3 also revealed a small amount of the heavier hexaacyl component (m/z = 1716) due to incomplete chromatographic separation.

The molecular weight of HPLC peak 1-3 was confirmed by FAB-mass spectrometry (Fig. 3), which showed an MH' ion at m / z = 1662, MH'-H,O at 1644, and MNa' at 1684. The cluster of peaks around 1678 and 1680 in the FAB-mass spectrum may be a mixture of M + NH,' and the MH' of the heavier component with molecular weight 1677. Fragmenta- tion resulting in loss of hydroxylauric acid (OHC,,) and lauric acid (nCJ gave peaks a t 1446 and 1463 atomic mass units, respectively. The base peak of the spectrum at m / z = 876 corresponded to cleavage of the bond between the distal and reducing sugars. This suggested that the component with molecular weight 1661 has the structure shown in Fig. 4. The ester-linked hydroxylauric acids attached to the 3- and 3'- positions were easily cleaved and following H-transfer pro- duced the loss of 216 atomic mass units observed in the spectrum. The acyl link between lauric acid and hydroxymy-

-3 I

w 1 4 8 4

W

1-4 1 4 9 8

1 4 8 4 1 e .

""""I

FIG. 2. A, HPLC of MLA fraction I, and LD-mass spectra of: B, peak 1-1; C, peak 1-2; D, peak 1-3; and E, peak 1-4.

ristic acid was also easily cleaved with H-transfer, showing loss of 200 atomic mass units in the spectrum. The amide- linked hydroxymyristic acid was not cleaved.

The structure and arrangement of fatty acids in the di- methyl MLA were partially derived from FAB-mass spectrom- etry, which gave molecular ions, losses of 200 and 216 atomic mass units, and formation of oxonium ion at m / z 876. This structure was further supported by LD-mass spectrometry, which produced a far different fragmentation pattern. The LD-mass spectrum of the same HPLC peak (Fig. 2 d ) showed the same losses of hydroxylauric and lauric acids (at m / z = 1484 and 1500) from the MK' ion at m / z = 1700. The oxonium ion was not observed, however. Instead, as reported previously by Coates and Wilkens (25), LD produced prominent frag- mentation within the sugar rings, and this fragmentation provided direct information on the identity and positions of the fatty acid chains.

We propose three major fragmentation pathways which are summarized in Scheme 1. Simultaneous cleavage of the C1-0 bond and the C2-C3 bond of the reducing sugar results in the loss of a portion of the molecule which includes the amide- linked fatty acid. If the N-linked fatty acid (R4) is lauroxy- myristate (C1,OCI4), then the loss of 467 atomic mass units produces the peak a t m / z = 1233 observed in the spectrum (Fig. 2d). Analogously, hydroxylauroxymyristate or hydroxy- myristate at that position would have produced losses of 483 and 285, atomic mass units, respectively. A second two-bond cleavage at C1-C, and C3-C, includes the loss of both the amide-linked (R4) and ester-linked (R3) fatty acids from po- sitions 2 and 3 and can be used to distinguish between normal and hydroxylauric acid at the 3 position once R4 is known. The loss of 665 atomic mass units in the mass spectrum of HPLC peak 1-3 to give the peak at m / z = 1035 established R3 as hydroxylauric acid. A third fragmentation involves simultaneous cleavage of the C4-Cs and C1-0 bonds. The ion that results (see below) can be used to establish the identity of R1 and R,.

(distal subunit - 0 - CH, - CH = 0) + K'

When Rl was hydroxylauric acid and R, was lauroxymyristic acid, an ion at m / z 974 resulted. A peak at 990 atomic mass units indicated the presence of an additional OH group at the distal subunit. This fragment was also indicative of the pres-

Lipid A Obtained from LPS of N. gonorrhoeae 10627

L > IO01

HPLC peak 1-3 from MLA fraction FIG. 3. FAB-mass spectrum of

I. L ? 40

R3 R4 FIG. 4. Proposed structure of hexaacyl MLA (M3) of M, =

1661. We have provisionally assigned the phosphate group to the 4"position according to the model compound, the hexaacyl MLA derived from Salmonella minnesota.

-0CHz

( ~ X H - 483 R 4 = OHC120C14 HO

- 467 R4 = C120C14

1 - AH I

- 285 R 4 = OHC14

-0CHz

(LbH - 665 R 3 = O H C 1 2

HO R4 = c120c14

NH - 467 R 3 = H

I R4 = c120c14 R4

HO $ g k H

NH I R 4

974 R 1 = O H C 1 2

R 2 = c120c14 990 R 1 = OHC12

R 2 = OHC120C14

SCHEME 1

ence of a 1' + 6 glycosidic linkage in the dimethyl MLA disaccharide. If the linkage was 1' + 4, one might expect a fragment ion "60 + K+. This was not observed.

Using this fragmentation scheme, we assigned structures to

r M,t Ht -H20 1644

x5 ~ M,+H+

1662 M4+ H'

I.

x+ 950 1450

M I 2

all of the other peaks from the HPLC separated fractions. The other major component of fraction I (see Fig. 2c) showed a loss of 216 atomic mass units corresponding to hydroxylauric acid, but the loss of 200 atomic mass units was greatly reduced, suggesting that an additional hydroxylauric acid replaced one of the lauric acids. The peak at m/z = 1249 was a loss of 467 atomic mass units from the molecular ion, indicating that the extra hydroxylauric acid is not on the reducing end. The shift of 16 atomic mass units of the peaks at 990, 1050, and 1249 indicated that the distal lauric acid has been replaced by hydroxylauric acid. The two other peaks from the HPLC separation of fraction I were both mixtures and somewhat difficult to analyze. However, they were both minor compo- nents of fraction I. The first peak (Fig. 2b) had traces of the component in 1-2, but also showed several lower mass com- ponents. A molecular ion at 1688 atomic mass units, 28 mass units below 1716, suggested that one c14 fatty acid has been replaced by a C,, fatty acid. Other molecular ion peaks sug- gested differences in the number of hydroxy fatty acids. The multiplicity of fragment ion peaks suggested that this substi- tution is random. The final peak (1-4) appeared to be a mixture of (1-3) along with methyl and dimethyl derivatives. The peak at 1498 was the loss of OHC,,, while the peak at 1484 included the loss of methylated hydroxylauric acid (OMeC12) from 1714 and the loss of OHC1, from m/z = 1700. The peaks at 974, 988, and 1002 indicated that the distal portion of the molecule may be unmethylated, monomethy- lated, or dimethylated. The second methyl group must be on the sugar. The peaks at 1232 and 1246 suggested that addi- tional methylation may occur in the reducing subunit. There- fore, we concluded that HPLC peak 1-4 contains a number of species with randomly methylated hydroxy fatty acids and/or sugar OH groups. It is not possible at this time to determine whether these methoxy fatty acids were naturally occurring or were generated by the diazomethane treatment. The struc- tures are summarized in Table I and indicate, in general, that the number of hydroxylated fatty acids decreased with reten- tion time.

Fractions I1 and I11 from the silicic acid column produced mainly pentaacyl MLA components (with some hexaacyl com- ponents at long retention times). The structure of one possible pentaacyl MLA with molecular weight 1479 is shown in Fig. 5.

Fraction I1 produced six peaks by HPLC separation (Fig. 6). Peak 2-1 had a molecular ion peak (M + K+) at 1534, corresponding to a molecular weight of 1495, or 16 atomic mass units more than the structure shown in Fig. 5. The loss of 216 atomic mass units produced the peak at 1318, indicative of hydroxylauric acid, but the absence of a loss of 200 atomic mass units suggested that there are no lauric acids present. The peak at 1249 was a loss of 285 atomic mass units from the molecular ion, indicating that the 2-position had hydrox- ymyristic acid (OHCI4) rather than C120Cl,. The peak at 989 (990) indicated that hydroxylauric acid is in the distal sugar as hydroxylauroxymyristate. This fully hydroxylated compo-

10628 Lipid A Obtained f rom LPS of N. gonorrhoeae TABLE I

Fatty acid distribution of HPLC-fractionated dimethyl MLA, fraction I

HPLC Retention peak“ time

mirz 1-1 32.6

1-2 34.2

1-3 34.8

1-4 35.8

Fatty acid distribution’

RP R3 R4

HPLC. a Silicic acid column fraction I was methylated and subjected to

*Refer to Fig. 4 for structure. Abbreviations: OHCI2 = hydroxy- laurate; OMeCI2 = methoxylaurate; C120C14 = lauroxymyristate; Cl20Cl2 = lauroxylaurate; OHCI20Cl4 = hydroxylauroxymyristate; OHC120-C12 = hydroxylauroxylaurate.

FIG. 1479.

H3C0, 0 +,P

H3C0

R3 R4

5. Proposed structure of pentaacyl MLA (M,) of M, =

nent had the lowest retention time of all the fractions exam- ined.

Peak 2-2 (Fig. 6c) coincided with the structure shown in Fig. 5 and was confirmed by FAB-mass spectrometry (Fig. 7). The spectrum showed only a small loss of 200 atomic mass units to suggest the presence of lauric acid. However, peaks a t 974 and 1034 indicated that lauric acid is indeed present at the distal subunit as lauroxymyristate. Conversely, the peak a t 1233 (285 atomic mass units below the molecular ion) indicated that it is not on the reducing subunit, since this corresponded to a ring cleavage involving hydroxymyristate.

Peak 2-3 (Fig. 6d) was a mixture and contained some of the components in 2-2. The molecular ion at 1518 corresponded to a molecular weight of 1479. The peaks at 989 and 1048 suggested that the distal subunits of some of the molecules carry an additional hydroxy group, most probably as hydrox- ylauroxymyristate. The small peak at 1233 (285 atomic mass units below the molecular ion) indicated that the reducing end of some of the molecules contains hydroxymyristate. Since attachment of lauric acid to the 3-position of the sugar is unlikely (see “Discussion” below), species with hydroxy- myristate on the reducing end probably corresponded to the poorly separated 2-2 component. Since the peaks a t 1034 and 1050 atomic mass units were also losses of 483 and 467 from the molecular ion, then both lauroxymyristate and hydroxy- lauroxymyristate may be present on the reducing subunit.

I 2-51 a

w > I e .

I , 1 0 3 6

0 1 2 2 4 3 6

MINUTES

I / 1 4 8 4 1 7 0 0

MASS

FIG. 6. A, HPLC of MLA fraction 11, and LD-mass spectra of: B, peak 2-1; C, peak 2-2; D, peak 2-3; E, peak 2-4; F, peak 2-5; and G, peak 2-6.

Peak 2-4 (Fig. 6e) is a methylated derivative of a component with molecular weight 1479, since it showed losses of both hydroxylauric and methoxylauric acid. The peak at 1065 indicated that the lauric acid is attached to myristic acid on the reducing end.

Peak 2-5 (Fig. Sf, had a molecular ion (MK’ at 1502) that corresponded to a molecular weight of 1463 and contained one less hydroxyl group than that shown in Fig. 5. The predominance of the loss of 200 atomic mass units suggested that there are fewer hydroxylauric acids. The peak at 974 indicated that this did not occur on the distal subunit. The loss of 467 atomic mass units at 1034 rather than 285 atomic mass units suggested that the reducing Cl20Cl4 is intact.

Peak 2-6 was a hexaacyl MLA identical to the major com- ponent observed on peak 1-3 and appeared at a comparably long retention time. The structures of the components from fraction I1 are summarized in Table 11.

The final fraction I11 was separated into four peaks and analyzed by LD-mass spectrometry (data not shown). The analysis showed that the components found in these peaks were virtually identical to those in the previous two fractions.

The four major components of MLA of N. gonorrhoeae with molecular weights of 1463, 1479 (pentaacyl MLA), 1661, and 1677 (hexaacyl MLA) are summarized in Table 111. Additional peaks in the HPLC fractions were observed either as meth- ylated derivatives of the major components or as several minor components. There was a small pentaacyl component in which the distal Cl20Cl4 was replaced by an OHC120C14 and had a molecular weight of 1495 (peak 2-1). There were also several minor hexaacyl components in which one of the N-linked hydroxymyristic acids was replaced by hydroxylauric acid (peak 1-1). For both the major and minor components, one can derive several generalizations. For the pentaacyl MLA,

Lipid A Obtained from LPS of N. gonorrhoeae 10629

, , , , , , , , , , ,; Mz+Hf -H20

1462

I! M z + H *

I I -0HC12 I 4 r Mz +No'

FIG. 7. FAB-mass spectrum of ' 6o I - G z 1502

HPLC peak 2-2 from MLA fraction E I--ICH,OIZ POZH I- 40 11. a 1445

ii 20

$ 0

M z + K' 1518

1264

a 1281 1304 1320 1355

1 ~ ~ " 1 ~ " ' l ' ' ~ ~ I ' ~ ' 1 ' ' l I 't,, ,, 11 11 Ill ,

900 950 1250 I300 1350 1400 1450 1500

TABLE I1 Fatty acid distribution of HPLC-fractionated dimethyl M U , fraction

II HPLC Retention peak" time

Fatty acid distribution*

R. R, R, R.

min 2-1 25.0 OHCIz

2-2 26.4 OHClZ

2-3 27.5 OHClZ OHGz OHCu

2-4 29.4 OMeClz

2-5 30.6 OHC12

2-6 34.8 OHC12

HPLC. Silicic acid column fraction I1 was methylated and subjected to

* Refer to Figs. 4 and 5 for structures and Table I for abbreviations.

TABLE I11 Fatty acid distribution of the major MLA components obtained from

the LPS of Neisseria gonorrhoeae strain F62 Mono-

Dimethyl isotopic Fatty acid distribution"

MLA mass R1 Rf R, R4

MI 1463.025 OHC12 c12oc14 H c12oc14 M2 1479.020 OHClz c12oc14 OHC12 OHC14 M3 1661.186 OHClz c12oc14 OHClz c12oc14 Ma 1677.181 OHC12 OHC1zOC14 OHC12 c12oc14

a Refer to Figs. 4 and 5 for structures and Table I for abbreviations.

the absence of lauric acid as lauroxymyristate generally oc- curred on the reducing end. The absence of hydroxylauric acid attached directly to the sugar also occurred, generally in the reducing subunit. For both hexaacyl and pentaacyl MLA, hydroxylauroxymyristate was generally observed on the distal subunit. In general, the distal subunit carried more hydroxy fatty acids than the reducing subunit. Mass spectral analysis revealed that the MLA obtained from the LPS of strains F62 and WR213 were similar.

Proton N M R Spectroscopy-The silicic acid column puri- fied hexaacyl MLA (fraction I) was methylated with diazo- methane and analyzed by proton NMR spectroscopy by a previously described method (17, 18). Two doublets appeared at 8.1 and 7.4 ppm, which represented the amide protons of the distal (NH') and the reducing (NH) end subunits, respec- tively (data not presented). Downfield shifts of the resonances were observed for protons at the 3- and 3'-positions of the glucosamine disaccharide from the normal 3.5-4.1 ppm to new values a t 5.67 and 5.63 ppm, respectively, indicating that these two positions were esterified. Because this sample was not sufficiently pure, further structural information could not be obtained.

M / Z

DISCUSSION

The LPS of N. gonorrhoeae appears to be directly involved in natural human immunity and susceptibility to dissemi- nated gonococcal infections. Since lipid A is the most biolog- ically active part of the LPS molecule, it is conceivable that the structure of lipid A might influence the nature and mag- nitude of responses made by the host to infection or immu- nization. For example, the absence of a reducing-end phos- phate group makes the lipid A (MLA) relatively nontoxic by several criteria (26, 27). The absence of a normal fatty acid in lipid A (as in precursor lipid A) also reduces the toxicity (26) and allows LPS nonresponder mice to become responder (28). Incorrect placement of the fatty acids into the glucosa- mine disaccharide in the organic synthesis of lipid A results in products with low biological activities (29,30). These recent studies showed that subtle differences in the structure of lipid A can result in significant changes in their biological proper- ties. For the gonococcus in particular, recent work indicates that lipid A is crucial to epitope expression by some compo- nents in a strain's LPS3 The removal of acyl-linked fatty acids ablates or reduces the reactivity of such components with a monoclonal antibody. Covalent coupling of the LPS- derived oligosaccharide with long-chain aliphatic acids en- hanced the antigenicity of the oligosaccharide. Knowledge of the fine structure of the lipid A moiety of gonococcal LPS is of primary importance in defining the role of LPS in the pathogenesis of and immunity to gonorrhea.

The LPS of N. gonorrhoeae is much less cytotoxic to macrophage than the LPS of other Gram-negative organisms. Peavy et al. (31) showed that 50 Fg of LPS from N. gonor- rhoeae reduced cell viability to 14%, whereas LPS from Esch- erichia coli Olll:B4, Salmonella enteritidis, and Serratia mar- cescens produced a similar degree of cytotoxicity at one- hundredth the concentration. This cytotoxicity is directly related to lipid A.

Knowledge of the structure of lipid A obtained from the LPS of N. gonorrhoeae is meager. Jennings et al. (32) deter- mined the chemical composition of the LPS from a related organism, Neisseria meningitidis. Perry et al. (33), Stead et al. (9), Gregg et al. (34), and Schneider et al. (4) all determined the chemical composition of LPS and/or lipid A obtained from the LPS bf N . gonorrhoeae. The results of these early studies showed the presence of KDO in their LPS prepara- tions. The lipid A preparations were shown to contain phos- phate, glucosamine, laurate, hydroxylaurate, and hydroxy- myristate.

Since almost no structural information is available, an assumption has been made that the basic structure of the MLA obtained from the LPS of N. gonorrhoeae is similar to that of MLAs from other Gram-negative bacteria. This as- sumption is supported by the literature (35). The model lipid A chosen for comparison in the present study was the hexaacyl MLA derived from S. minnesota (18). Ester-linked fatty acyl

Yamasaki, R., Program of the Annual Meeting of the American Chemical Society, New York, April 1986 (CARB 32).

10630 Lipid A Obtained from LPS of N . gonorrhoeae

groups are cleaved by FAB-mass spectrometry whereas amide- linked fatty acyl groups are not (16, 18). The normal fatty acids are generally involved in acyloxyacyl linkages (15-18). This information, along with the results obtained from the present study, was used to characterize the dimethyl MLAs derived from N. gonorrhoeae.

We applied the fractionation procedures that had been used to successfully determine the complete structures of the MLAs derived from the Salmonella strains (16-18). The MLAs were first fractionated on a silicic acid column to yield the crude hexaacyl and pentaacyl MLA, each fraction was then meth- ylated, and reverse-phase HPLC was performed to yield a series of peak fractions. These peaks were analyzed by LD- mass spectrometry. FAB was used on two major fractions. A total of 13 different structures of dimethyl MLA were identi- fied. From the results of these analyses, it became clear that the backbone structure of the MLA from N. gonorrhoeae is a glucosamine disaccharide. We discovered that positive ion FAB and LD caused similar cleavage of the fatty acyl residues. Only the hydroxylauric and lauric acids were cleaved from the dimethyl MLA. There was no indication of the release of the two possible acyloxyacyl groups (lauroxylaurate or lauroxy- myristate), demonstrating that the only ester-linked fatty acids were hydroxylauric and lauric acids. The hydroxymyris- tic acid must then be nitrogen linked. NMR spectrum of dimethyl hexaacyl MLA indicated that fatty acyl groups are esterified to the 3- and 3’-positions of the disaccharide, dem- onstrating that these are the sites of attachment of the hy- droxylaurate.

The common oxonium ion at 876 seen in the FAB-mass spectrum of a hexaacyl (M3, peak 1-3) and pentaacyl (M2, peak 2-2) suggests that the distal subunit of Mz and M3 contain one each of glucosamine, phosphate, laurate, hydrox- ylaurate, and hydroxymyristate. The peak at 974 in the LD- mass spectrum of M1, MP, and M, suggest that all three of these components have a similar composition in the distal subunit, while the peak a t 990 in the mass spectra of M4 (peak 1-2) suggests an additional hydroxyl group in the distal sub- unit.

This study shows that LD-mass spectrometry complements the FAB-mass spectrometry in the structural analysis of dimethyl MLA. LD-mass spectrometry causes specific frag- mentations of the reducing end sugar (Scheme 1) that allows one to establish the nature and location of the fatty acids at the reducing end.

Since there are only two available positions for the three fatty acyl groups in the distal subunit, one of these two positions must be occupied by an acyloxyacyl group. A similar situation appears to exist with the fatty acyl groups in the reducing-end subunit of M,. Our results support a structure of M3 (Fig. 4) where R1 = RB = hydroxylaurate, and Rz = R4 = lauroxymyristate. The suggested structure of MP (Fig. 5) indicates R1 = R, = hydroxymyristate, R2 = lauroxymyristate, and R4 = hydroxymyristate.

There were four major structures of free lipid A generated by the acid hydrolysis of LPS from N. gonorrhoeae. This can be related to the structural variations to be found in the lipid A moiety of LPS. The sizes of the lipid A(s) from this source were generally 84 mass units (or 6 CH2 groups) smaller than the corresponding lipid A(s) obtained from the Salmonella strains. Moreover, the lauric acid (normal fatty acid) was singularly present in both reducing and distal subunits of the hexaacyl MLA. This compares with the presence of two normal fatty acids at the distal subunit of the hexaacyl MLA from the Salmonella strain (16-18). The influence of this

difference in the fatty acid distribution on biological activity is not known.

We examined the structural relationship among the four prominent components of the MLAs derived from the serum- sensitive strains of N. gonorrhoeae (Table 111). The pentaacyl MLAs (M1 and Mq) appeared to differ by the presence of either a laurate or a hydroxylaurate group. The additional laurate in MI is linked to hydroxymyristate to give an acylox- yacyl group.

The hexaacyl MLAs (M3 and M4) differed by the presence of an additional oxygen atom in M4, and we found that lauroxymyristate is replaced by a hydroxylauroxymyristate in the distal subunit of M,. The presence of a hydroxyacyloxy- acyl group in the lipid A of Gram-negative bacteria has been reported (36). This higher polarity of M4 over M, could have led to the observed retention time that was slightly lower for M, than for M,.

This study was undertaken to demonstrate that one can obtain much useful structural information on the lipid A by utilizing a combination of HPLC and mass spectrometry. We were able to obtain precise structural information on the components of MLAs found in N. gonorrhoeae that might be useful for biological studies on the LPS. In this regard, it would now be interesting to compare the fine structure of MLAs obtained from a panel of LPS of the serum-sensitive and serum-resistant strains of N. gonorrhoeae. Such a study is planned.

Although it is unlikely that further study of the MLAs of N . gonorrhoeae will reveal new structural features, it is needed to unequivocally establish their complete structures. Although LD-mass spectrometry suggests 1’ + 6 glycosidic linkage of the glucosamine disaccharide, methylation analysis should be done to confirm this (37). Two-dimensional NMR spectros- copy should be done on HPLC-purified dimethyl MLA to establish the position of the phosphate group in the distal subunit and the anomeric configurations of the distal and reducing end sugars (18). These studies are presently in progress.

Acknowledgments-Mass spectral determinations were carried out at the Middle Atlantic Mass Spectrometry Laboratory, a National Science Foundation Shared Instrumentation Facility, The Johns Hopkins University School of Medicine, Baltimore, MD 21205.

REFERENCES

1. Glynn, A. A., and Ward, M. E. (1970) Infect. Immun. 2, 162-168 2. Tramont, E. C., Sadoff, J. C., and Wilson, C. (1977) J. Immunol.

3. Rice, P. A,, McCormack, W. M., and Kasper, D. L. (1980) J.

4. Schneider, H., Griffiss, J . M., Williams, G. D., and Pier, G. B.

5. Goldschneider, I., Gotschlich, E. C., and Artenstein, M. S. (1969)

6. Schoolnik, G. K., Buchanan, T. M., and Holmes, K. K. (1976) J.

7. Nicholson, A,, and Lepow, I. H. (1979) Science 205,298-299 8. Brooks, G. F.. Israel, K. S., and Peterson, B. H. (1976) J. Infect.

118,1843-1851

Immunol. 124,2105-2109

(1982) J. Gen. Microbiol. 128, 13-22

J. Exp. Med. 129, 1307-1326

Clin. Inuest. 58, 1163-1173

Dis. i34,450-462 9. Stead. A.. Main. J. S.. Ward. M. E.. and Walt, P. J. (1975) J .

Gen. Microbiol. 88, 123-131 10. Apicella, M. A,, Breen, J. F., and Gagliardi, N. C. (1978) Infect.

Immun. 20,228-234 11. Schneider, H., Hale, T. L., Zollinger, W. D., Seid, R. C., Jr.,

Hammack, C. A., and Griffiss, J . M. (1984) Infect. Immun. 45, 544-549

12. Mandrell, R., Schneider, H., Zollinger, W., Apicella, M. A., and Griffiss, J. M. (1985) in T h e Pathogenic Neisseria. Proceedings of the Fourth International Symposium (Schoolnik, G. K., Brooks, G. F., Falkow, S., Frasch, C. E., Knapp, J. S., Mc- Cutchan, J. A., and Morse, S. A,, eds) pp. 379-384, American

Lipid A Obtained from LPS of N. gonorrhoeae 10631

Society for Microbiology, Washington, D. C. 26. Takayama, K., Qureshi, N., Ribi, E., and Cantreil, J. L. (1984)

Schneider, H. (1985) in The Pathogenic Neisseria. Proceedings 27. Takayama, K., Qureshi, N., Raetz, C. R. H., Ribi, E., Peterson, of the Fourth International Symposium (Schoolnik, G. K., J., Cantrell, J. L., Pearson, F. C., Wiggins, J., and Johnson, A. Brooks, G. F., Falkow, S., Frasch, C. E., Knapp, J. S., Mc- G. (1984) Infect. Immun. 45,350-355 Cutchan, J. A., and Morse, S. A., eds) pp. 373-378, American 28. Vogel, s. N., Madonna, G. S., Wahl, L. M., and Rick, P. D. (1984) Society for Microbiology, Washington, D. C. Rev. Infect. Dis. 6, 535-541

14. Galanos, C., Luderitz, O., Rietschel, E. T., and Westphal, 0. 29. Tanamoto, K., Zahringer, U., McKenzie, G. R., Galanos, C., (1975) Int. Rea Biochem. 14,280-288 Rietschel, E. Th., Liideritz, O., Kusumoto, S., and Shiba, T.

15. Qureshi, N., Takayama, K., and Ribi, E. (1982) J. Biol. Chem. (1984) Infect. Immun. 44,421-426

16. Qureshi, N., Takayama, K., Heller, D., and Fenselau, C. (1983) Shiba, T. (1984) Infect. Immun. 44,427-433

17. Takayama, K., Qureshi, N., and Mascagni, P. (1983) J. Biol. Immun. 21,310-319

13. Yamasaki, R., OBrien, J. P., Rice, P. A., Griffiss, J. M., and Rev. Infect. Dis. 6,439-443

257,11808-11815 30. Tanamoto, K., Galanos, C., Luderitz, O., Kusumoto, S., and

J. Biol. Chem. 258,12947-12951 31. Peavy, D. L., Baughn, R. E., and Musher, D. M. (1978) Infect.

Chem. 258, 12801-12803 32. Jennings, H. J., Hawes, G. B., and Adams, G. A. (1973) Can. J.

Biol. Chem. 260,5271-5278 Qureshi? N', Mascami, ", Ribit E', and Takayama, J* 33. Perry, M. B., Daoust, V., Diena, B. B., Ashton, F, E., and Wallace,

Biochem. 51, 1347-1354

R. (1975) Can. J. Eiochem. 53,623-629

20. Osborn, M. J. (1963) Proc. Natl. Acud. Sci. U. S. A . 50,499-506 19* G' R' (1959) " BioL 234i 466-468; 469-471 34. Gregg, C. R., Melly, M. A., Hellerqvist, C. G., Coniglio, J. G., and McGee, Z. A. (1981) J. Infect. Dis. 143,432-439

22. White, L., and Kellogg, D. S. (1965) Appl. Microbiol. 13,171-174 23. Westphal, O., and Jann, K. (1965) in Methods in Carbohydrate

Lindner, B., Seydel, U., Bradaczek, H., Barnickel, G., Labis-

C h e m ~ t ~ (Whistler, R. L., ed) Voi. 5, pp. 83-91, Academic chinski, H., and Giesbrecht, P. (1984) in ~ a ~ ~ o ~ of ~ n ~ o t o ~ i n

Press, New York (Rietschel, E. Th., ed) Vol. 1, pp. 187-220, Elsevier Science

24. Van Breemen, R. B., Snow, M., and Cotter, R. J. (1983) Int. J. 36. Wollenweber, H.-W., Seydel, U., Lindner, B., Luderitz, o., and Publishers, New York

Mass Spectrom. Ion Phys. 49, 35-49 Rietschel, E. Th. (1984) Eur. J. Eiochem. 145,265-272

21. E., and Kressv H' (1979) Carbohydr. Res. 76, 233-238 35. Rietschel, E. Th., Wollenweber, H.-W., Bra&, H., Zahringer, U,,

25. Coates, M. L., and Wilkens, C. L. (1985) Biomed. .Vass Spectrom. 37. Hase, S., and Rietschel, E. Th. (1976) Eur, J. Biochem. 63, 93- 12,424-428 99