Chemical Reduction of %Oxo and Unsaturated Groups in Fatty Acids of Diphosphoryl Lipid A from the Lipopolysaccharide of Rhodopseudomonas sphaeroides COMPARISON OF BIOLOGICAL PROPERTIES BEFORE AND AFTER REDUCTION*

Unlike the diphosphoryl lipid A (DPLA) derived from toxic lipopolysaccharide of Escherichia coli and Salmonella strains, the DPLA from nontoxic lipopoly- saccharide of Rhodopseudomonas sphaeroides ATCC is biologically inactive. This could be due to the presence of 3-oxotetradecanoic and A7-tetradecenoic acids. These two fatty acids in R. sphaeroides DPLA were catalytically reduced in platinum oxide/Hz to the 3-hydroxy and saturated fatty acids, respectively. The biologically active E. coli DPLA was also treated with platinum oxide/Hz, but as expected, the reduction step did not change the structure. These two preparations were then compared with the untreated

Unlike the diphosphoryl lipid A (DPLA) derived from toxic lipopolysaccharide of Escherichia coli and Salmonella strains, the DPLA from nontoxic lipopolysaccharide of Rhodopseudomonas sphaeroides ATCC 17023 is biologically inactive. This could be due to the presence of 3-oxotetradecanoic and A7-tetradecenoic acids. These two fatty acids in R. sphaeroides DPLA were catalytically reduced in platinum oxide/Hz to the 3-hydroxy and saturated fatty acids, respectively. The biologically active E. coli DPLA was also treated with platinum oxide/Hz, but as expected, the reduction step did not change the structure. These two preparations were then compared with the untreated samples for biological activity in three select in vitro assays.
Over a range of 0.01-100 ng/ml, both normal and reduced DPLA from R. sphaeroides were inactive in priming phorbol myristate acetate-stimulated superoxide anion release in human alveolar macrophages. Over a range of 10-103 ng/ml, both samples failed to induce tumor necrosis factor in the RAW 264.7 murine macrophage cell line. The reduced DPLA marginally activated 702/3 pre-B cells at concentrations of 0.1-30 pg/ml. In every case, both normal and platinum oxide/Hz-treated E. coli DPLA were biologically active.
These results indicate that the lack of biological activity of R. sphaeroides DPLA is not due to the presence of 3-oxo and unsaturated fatty acids, but rather to one or more of the following: (i) presence of only five fatty acyl groups (compared to six in active lipid A); (ii) presence of 3-hydroxydecanoic acids (rather than 3-hydroxytetradecanoic, in active lipid A); (iii) greater variation in size of the fatty acids.
The lipopolysaccharides (LPS)' of Escherichia coli and the isolate and purify the endotoxic moiety in the form of DPLA from the toxic LPS (5,6). Thus, the model toxic lipid A would be the hexaacyl DPLA obtained from LPS of E. coli and Salmonella strains. This structure has been established (6-10) (Fig.1 B ) .
The LPS of Rhodopseudomonas sphueroides has been shown to be nontoxic (11). The chemical structure of the lipid A moiety of this LPS was determined as the monophosphoryl lipid A (12). DPLA from this source (an analog of toxic DPLA) has been found to be inactive in several biological assays (13) (activation of B lymphocytes (27) and induction of cytokines in macrophages (28). Its chemical structure is shown in Fig.

lA.
We are now able to compare the structure-to-function relationship of a model toxic hexaacyl DPLA with the nontoxic pentaacyl DPLA from R. sphueroides LPS. The most interesting difference between the two structures is the presence of 3-OXO and unsaturated fatty acids in the R. sphueroides DPLA. A review article suggested that these fatty acids might contribute toward the lack of toxicity of R. sphueroides LPS (14). The question we posed was, do these two fatty acids play an important role in the essential nontoxicity of the DPLA?
We prepared the nontoxic DPLA from the LPS of R. sphaeroides, catalytically reduced this preparation in the presence of platinum oxide, and tested it in several key in vitro biological assays. We report that reduction of R. sphaeroides DPLA does not convert it to a biologically active lipid A.
Growth of Bacteria and Preparation of LPS-R. sphaeroides ATCC 17023 was grown photoheterotrophically in 2-liter Erlenmeyer flasks in medium 550 as previously described (12). Cells were grown at 26 "C for 12-14 days, concentrated using a Pellicon cassette system (Millipore Corp., Bedford, MA), and harvested by centrifugation. A yield of 850 g of cell paste was obtained from 100 liters of culture.
The cell paste (700 g) was extracted for 1 h, with stirring, in 4 liters of ethanolln-butyl alcohol (3:1, v/v). This was followed by three 1-h extractions with 4 liters of ethanol and three 1-h extractions with 4 liters of acetone. Finally, the cells were suspended in 2 liters of diethyl ether and filtered, yielding 70.4 g of dried, pigment-depleted cells. LPS was extracted from these cells by the phenol/chloroform/ petroleum ether method (15) and precipitated as described by Qureshi et al. (7), to yield 640 mg of LPS.
Preparation of DPLA from R. sphaeroides LPS-LPS (640 mg) obtained from R. sphaeroides was suspended in 0.02 M sodium acetate, pH adjusted to 2.5 with acetic acid, at a concentration of 3 mg/ml. This was incubated for 70 min at 100 "C. The DPLA was recovered as previously described (7).
Peak C represented the nontoxic, highly purified pentaacyl DPLA.
Preparation of DPLA from E. coli LPS-The hexaacyl DPLA was prepared as previously described (6). Briefly, ReLPS from E. coli D31m4 was treated with 20 mM sodium acetate, pH 4.5, at 100 "C for 90 min. The crude DPLA was applied to a DEAE-cellulose column (acetate form). The column was eluted with a linear gradient of 0.02-0.08 M ammonium acetate in chloroform/methanol/water (2:3:1, v/ v). Hexaacyl DPLA was recovered as previously described (6). Laser desorption mass spectrometry of the methylated derivative confirmed the structure of the hexaacyl DPLA as shown in Fig. 1B (6).
Catalytic Reduction of DPLA-DPLA (20 mg) derived from the LPS of R. sphaeroides or E. coli was dissolved in 5.0 ml of n-propyl alcohol, and 30 mg of platinum(1V) oxide hydrate (Aldrich Chemical Co.) was added. Reduction was allowed to take place overnight at 22 "C with stirring in a hydrogen atmosphere. Reduced DPLA was recovered from the reaction mixture by filtration. HPLC Fractionation-Peak C was converted to the free acid by passage through a Chelex 100 (Na+) and Dowex 50 (H+) two-layered column in chloroform/methanol (4:1, v/v), methylated with diazomethane (17) to yield the tetramethyl DPLA, and fractionated by reverse-phase HPLC (18). An 8-mm X 10-cm Nova-Pak cartridge (5 pm particle size, CIR-bonded silica, Waters Associates, Inc., Milford,  MA) was used at flow rate of 2 ml/min. A linear gradient of 20-80% isopropyl alcohol in acetonitrile was used over a period of 60 min. Gas-Liquid Chromatography-Analytical gas-liquid chromatography of methyl esters of fatty acids was carried out using a Packard model 428 gas chromatograph with a glass column (0.2-mm X 3.4-m) containing 3% SP-2100 DOH, 100/200 mesh Supelcoport (12). A flame ionization detector was used and the column was programmed at 4' C/min from 150 to 230 "C. ated DPLA were obtained on a BIO-ION Nordic (Uppsala, Sweden) Mass Spectrometry-Plasma desorption mass spectra of methyl-BIN-1OK plasma desorption time-of-flight mass spectrometer equipped with a PDP 11/73-based data system. Purified tetramethyl DPLA was dissolved in chloroform/methanol (4:1, v/v) solution and electrosprayed onto a mylar-backed aluminum foil. Positive ion mass spectra were recorded with an accelerating potential of 16 kV for 3 to 9 million primary events, with resolution of 1 ns/channel. H' and Na+ were used for calibration.
Fast atom bombardment mass spectra were obtained on a Kratos (Manchester, England) MS-50 high-resolution, double-focusing mass spectrometer equipped with an Ion Tech (Teddington, United Kingdom) saddle-field atom gun. Samples were desorbed from the monothioglycerol matrix by a beam of 8 keV Xe atoms. Positive ion spectra were recorded with an accelerating potential at 8 kV over the mass range of 200-350 at a rate of 30 s/decade. NMR Spectroscopy-Methylated DPLA samples were dissolved in benzene-&/Me,SO-& (9:l) with internal MelSi as a chemical shift reference. 'H spectra were recorded on a GN-500 spectrometer with a spectral width of 4800 Hz at 500 MHz. One-dimensional spectra were recorded with 8,192 data points and two-dimensional 'H-'H correlation spectra were obtained by a three-pulse sequence 90"-t,-90"-90"-Acq, which (with proper phase cycling) allowed for coherence transfer through a double quantum filtered-COSY. Two sets of 384 X 2048 data points were acquired in adjacent blocks of memory, and the data were processed to obtain pure absorptive spectra. Prior to the Fourier transformation, 30' shifted sine bell functions were used in both dimensions and the tl free induction decay was zero-filled to obtain a final data matrix of 2,048 X 2,048 real points.
Superoxide Anion Release-Human bronchoalveolar cells (90% macrophages, >95% viable) were obtained from healthy nonsmoking subjects by bronchoalveolar lavage. These cells were cultured in sterile 96-well plastic microtiter plates. Cells were suspended in supplemented RPMI 1640 with 2% autologous serum, plated (1 X lo5 cells/ well), and allowed to adhere at 37 "C in a humidified 5% CO, atmosphere for 1 h. The supernatant medium and nonadherent cells were aspirated, cells were washed once in medium, and the adherent cells were conditioned overnight in supplemented RPMI 1640 with 10% autologous serum. The cells were then cultured for 48 h at 37 "C in a 5% CO, atmosphere in supplemented RPMI 1640 with 2% autologous serum and various concentrations of DPLA. Stock solutions of DPLA in 0.125% triethylamine were diluted and used for the experiment. All experiments were performed under low endotoxin conditions. PMA-stimulated superoxide anion production by alveolar macrophages was determined as superoxide dismutase-inhibitable reduction of ferricytochrome c by the method of Calhoun et al. (19). A 200-p1 final volume contained 0.1% gelatin, 1.2 mg/ml cytochrome c, and PMA (20 ng/ml). Paired reactions were run with superoxide dismutase. Absorbance was measured at 550 nm over a 2-h period. Plates were incubated at 37 "C. Kinetic analysis showed that peripheral blood monocyte and alveolar macrophage superoxide anion release increased linearly for a t least 2 h. Triethylamine in the culture medium did not affect superoxide anion generation by cultured cells.
702/3 Cell Assay-The 70Z/3 cell line is a mouse pre-B cell originally described by Paige et al. (20). The cells constitutively synthesize p chain mRNA and respond to LPS by initiating the synthesis of K chain mRNA, which leads to the expression of surface IgM. Cells (2.5 X 105/well) were added to a 48-well plate. They were cultured in RPMI 1640 medium supplemented with 5% fetal calf serum, 10 mM glutamine, 12.5 mM HEPES, 5 X M 2-mercaptoethanol, and 50 pg/ml gentamicin for 40 h at 37 "C. Cells were harvested and stained with fluoresceinated F(ab'), goat antimouse IgM (Tago, Burlingame, CA). Samples containing 5,000 cells were analyzed in a cytofluorograph (Ortho, model 50-H). Data are expressed as percent fluorescent cells above the background control (no DPLA).
TNF Assay-The induction of T N F in murine macrophage cell line RAW 264.7 by lipid A was measured with the indicator cell line L929 as previously described (13).

Preparation and Preliminary
Characterization of DPLA from LPS of R. sphaeroides-The LPS of R. sphueroides was acid hydrolyzed to yield crude DPLA. This was purified by DEAE-cellulose column chromatography to yield several peak fractions (Fig. 2). Peak A represented the monophosphoryl lipid A. Peaks B and C were two forms of DPLA that could be differentiated by TLC using silica gel H and chloroform/ pyridine/%% formic acid/water (1012:3:1, v/v) but not by the chloroform/methanol/water/concentrated ammonium hydroxide (50:25:4:2, v/v) system. Analytical TLC of peak C in several systems revealed a single char-positive band. Peak C was methylated with diazomethane and subjected to reverse-phase HPLC using a CI8-bonded silica cartridge. A representative profile of such a fractionation is shown in Fig.  3A. Two major and about seven minor peaks were observed. The set of very small peaks appearing at elution times of 10-15 min were identified as the tetraacyl series of DPLA. The pentaacyl DPLA eluted from the cartridge between 15 and 23 min. The minor shoulder fractions associated with peaks 1 and 2 and those appearing between peaks 2 and 3 were all different from the neighboring major peaks by 14 amu (one CH, group). These are thought to be products of the methylation step. Fatty acid analysis of HPLC peak 1 by gas-liquid chromatography of the acid-hydrolyzed and methylated samples (12) revealed that a single molecule of pentaacyl DPLA has two OHClo, one A7-C14:1, one OH& and one 3KC14 groups.
Mass Spectral Analysis of Tetramethyl DPLA-Since the structure of the R. sphaeroides DPLA was never strictly established, we decided to examine its structure by mass spectrometry. The HPLC-purified tetramethyl DPLA peaks (Fig. 3A) were analyzed by plasma desorption mass spectrometry (Fig. 4). The molecular ions MNa' for the major peaks 1 and 2 were observed at m/z 1,578 and 1,580, respectively. The loss of PO:+2,H5 (-108) accompanied by proton transfer gave rise to ions at m/z 1,470 and 1,472, respectively. Additional loss of OHClo yielded MNa' minus OHClo and P03C,Hb ions at m/z 1,281 and 1,283, respectively. These results showed that the M , for peaks 1 and 2 (Fig. 3A)  for components 1 and 2 (Table I).
These results are consistent with the sizes of dimethyl monophosphoryl lipid A established by Qureshi et al. (12), which were 1,447 and 1,449, suggesting the presence of an additional phosphate group (P03CzHs, 108 amu) in the DPLA series. There was no evidence of a pyrophosphate group. A peak at m/z 124 in the negative ion mass spectrum (data not shown) corresponds to the dimethylated phosphate anion P03(CH3); and most likely results from the cleavage of the C-0 bond of the sugar 1-phosphate. An analogous peak corresponding to the anion of a pyrophosphate group at this position was not observed.
Fast atom bombardment and plasma desorption mass spec-  Fig. 1 for chemical structures. bThese two are the major components of peak C and represent This represents HPLC peak 4 (Fig. 3B) of the catalytically dEcDPLA is the model toxic hexaacyl DPLA derived from the e The properties of this synthetic product (pentaacyl DPLA) were HPLC peaks 1 and 2 (Fig. 3A).

TABLE I1
'H NMR chemical shifts (in ppm from internal Me4Si) and vicinal coupling constants (J, Hz) for the disaccharide of HPLC-purified tetramethyl DPLA from LPS of R. sphaeroides H-1 to N-H represent protons of the reducing-end sugar, whereas H-1' to N'-H represent protons of the nonreducing-end sugar. The sample analyzed was HPLC peak 1 of Fig. 3A. Accuracy of chemical shifts is kO.01 ppm.  Fig. 3A), summarized in Table 11, was made from the two-dimensional DQF-COSY spectrum using the anomeric and the amide 'H signals as the starting point. Our results are similar to those reported by Strain et al. (21) for the disaccharide precursors IVA and IVB isolated from Salmonella typhimurium. In particular, the coupling constant between H-1' and H-2' of the nonreducing terminal residue of GlcNAc (8.2 Hz) implies that both protons are trans (axial), as expected for the @-anomer. The one-dimensional spectrum recorded at a resolution of 0.5 Hz shows that the signal assigned to H-1 of the reducing terminal is split, with coupling constants of 2.8 and 6.6 Hz. The Hl-H2 cross-peak in the DQF-COSY spectrum shows that the smaller coupling is active (antiphase), while the larger coupling is passive, as expected for the coupling to the 31P of the anomeric phosphate. The small coupling (2.8 Hz) between H-1 and H-2 shows that H-1 is equatorial and the reducing-end glucosamine is in the a-anomeric configuration (22). Similar results were obtained with HPLC peak 2 (Fig. 3A).

Catalytic Reduction of E. coli and R. sphaeroides DPLA-E.
coli DPLA and peak C were reduced with H2 in the presence of platinum oxide and analyzed for completeness of reaction. Fatty acid analysis of the reduced peak C by gas-liquid chromatography revealed the complete absence of 3KCI4 and A7-Clkl (data not presented). Reverse-phase HPLC of the tetramethyl ester showed a shift in the position of the two major peaks from 15 min for peak 1 and 18 min for peak 2 to a single major peak 4 at 24 min (Fig. 3, A and B). Peak 4 was analyzed by plasma desorption mass spectrometry (Fig. 4C). It showed a molecular ion MNa' at m / z 1,582. Loss of P03CzH5 gave rise to ions at m/z 1,474. Additional loss of OHClo yielded MNa' minus PO3CZH5 and OHClo ion at m / z 1,285. This showed that the M , of peak 4 is 1,559 (Fig.  3B, Table I), indicating that the 3KC14 and A7-C14:1 were converted to OHC14 and c 1 4 : 0 , respectively. Catalytic reduction of 3KC14 should yield a racemic acid containing equivalent amounts of D-and L-enantiomers.
The platinum oxide/Hz-treated E. coli DPLA was similarly analyzed by HPLC and found to be identical to the untreated sample (data not presented). As expected, treatment of E. coli DPLA with platinum oxide/Hz had no effect on the structure as determined by plasma desorption mass spectrometry. These preparations served as the positive controls in the biological assays. (23) first showed that when peritoneal macrophages are exposed to LPS, the cells are primed to respond with increasing oxygen anion (0;) production after stimulation with PMA. This secretion of oxygen intermediate was suggested by Nathan (24) to be a biochemical marker of macrophage activation. Since this appears to be an excellent and sensitive model system for LPS activation, we examined the ability of both normal and reduced R. sphaeroides DPLA to prime the PMA-stimulated 0; release in human alveolar macrophages. When these two DPLAs were tested over a concentration range of 0.01-100 ng/ml, they showed virtually no priming activity (Fig. 5). Thus, the conversion of the %oxo group to 3-hydroxy and saturation of the double bond in the fatty acids of R. sphaeroides DPLA did not change its lack of priming activity.

Effects of R. sphaeroides DPLA on the Priming of PMAstimulated Superoxide Anion Release by Human Alveolar Macrophages-Pabst and Johnston
Control samples of normal and platinum oxide/Hz-treated E. coli DPLA were very effective in priming macrophages (Fig. 5). Activity was observed even at 0.01 ng/ml, and reached a maximum a t about 10 ng/ml. This showed that treatment of E. coli DPLA with platinum oxide/H2 does not affect its ability to prime the macrophages.
Effect of Reduced R. sphneroides DPLA on Induction of T N F i n Macrophages-We have already shown that R. sphaeroides DPLA is unable to induce the formation of T N F by RAW 264.7 cells (13). Table I11 again shows this to be the case. Reduced R. sphaeroides DPLA was also inactive in the induction of TNF. The value of 367 T N F units a t lo5 ng/ml is only 1.8% of the units that the toxic E. coli DPLA showed at the same concentration. The E. coli DPLA and the platinum oxide/H,-treated E. coli DPLA were equally effective in inducing T N F by the RAW 264.7 cells. These results showed that the reduction of R. sphaeroides DPLA does not cause it to become active in this assay.
Activation of 702/3 Cells by Reduced R. sphaeroides DPLA-We have shown that the pentaacyl DPLA from LPS of R. sphaeroides is not able to activate 70Z/3 cells. Tested over a  range of 0.1-30 pg/ml, the R. sphaeroides DPLA is indeed not active (Fig. 6). When the reduced DPLA was tested over the same concentration range, it showed a gradual but small increase in the activation of 70Z/3 cells from about 8.5% at 0.1 pg/ml to 17.5% at 1.0 pg/ml. With increase in concentration beyond this point, there was a gradual decrease to about 10% at 30 pg/ml. We consistently observed this pattern. Thus the R. sphaeroides DPLA appears to give a small increase in the activation of 70Z/3 cells after reduction.
These results were compared with those for active lipid A. The toxic E. coli DPLA showed a near-linear rise in activity from 9.5% at 0.1 pg/ml to 62% at 30 pg/ml. The platinum oxide/H2-treated E. coli DPLA appeared to be slightly more active over the same range, but this was not statistically significant. These results suggest that the conversion of R. sphaeroides DPLA to the reduced DPLA causes a small increase in its ability to activate B lymphocytes. Because of the nonlinearity of this rise, it is difficult to interpret the results. However, it is clear that the reduction of R. sphaeroides DPLA does not convert it to an impressively active compound.

DISCUSSION
We have developed a simplified two-step procedure for preparing pentaacyl DPLA from the LPS of R. sphaeroides.
The first step is the acid hydrolysis of LPS with 0.02 M sodium acetate, pH 2.5 (instead of the usual pH 4.5 used with E. coli and Salmonella LPS (5,6)), to yield the crude DPLA.
The second step is the DEAE-cellulose column fractionation of the hydrolyzed LPS, which effectively separated the unhydrolyzed LPS, monophosphoryl lipid A, and several forms of DPLA. This new procedure was used to prepare the DPLA for this study.
R. sphaeroides DPLA was characterized by combined reverse-phase HPLC and mass spectrometry and 'H NMR spectroscopy. Plasma desorption and fast atom bombardment mass spectrometry were utilized to establish and confirm the structure of the two major HPLC fractions (components 1 and 2) as the tetramethyl DPLA instead of the dimethyl monophosphoryl lipid A, as was done previously (12). The aanomeric configuration of the reducing-end phosphate was established by 'H NMR spectroscopy.
This purified nontoxic pentaacyl DPLA (peak C) was catalytically reduced with platinum oxide. The completeness of reduction was confirmed by gas-liquid chromatography of the fatty acids as well as reverse-phase HPLC and plasma desorption mass spectrometry of the reduced DPLA.
The reduced DPLA was then tested for its ability to prime the PMA-stimulated superoxide anion release in human alveolar macrophages, induce TNF production in the RAW 264.7 macrophage cell line, and activate 70213 pre-B cells. We found that reduction of R. sphaeroides DPLA did not convert it to an active form in any of these assays. The slight activation of 70213 cells by the reduced R. sphaeroides DPLA is not considered significant in this study. Both the R. sphae- Control was 2.18 f 0.81 nmol. These results showed that the %oxo group and unsaturation of the fatty acids of R. sphaeroides DPLA are not the factors that made the compound inactive and nontoxic.
When one compares the structures of the model toxic hexaacyl DPLA (Fig. 1B) with the nontoxic pentaacyl DPLA (Fig. lA), the following differences are noted: (i) the toxic DPLA contains six fatty acids, whereas the nontoxic DPLA has only five; (ii) the nontoxic DPLA has much shorter fatty acids (OHClo) at R1 and R3 than the toxic DPLA; (iii) the nontoxic DPLA has greater variation in the sizes of fatty acids than the toxic DPLA; (iv) the nontoxic DPLA has an unsaturated fatty acid in acyloxyacyl linkage at Rz; (v) the nontoxic DPLA has a 3KCl4 at R4. Any of these factors could account for the difference in their biological activities. However, the results of this study strongly suggest that the lack of biological activities of R. sphaeroides DPLA is not due to the last two factors.
However, very similar activities were observed for the hexaacyl and pentaacyl DPLA in the B cell mitogenic assay. It is interesting to note that this Ib-isomer has structural features in common with our reduced R. sphaeroides DPLA (compare reduced RBDPLA with synthetic DPLA (Ib-isomer) in Table  I). However, two major differences between the reduced R. sphaeroides DPLA and Ib-isomer exist in the composition of R1 and R3. The R1 and R3 of reduced R. sphaeroides DPLA are OHClo, wheras those of Ib-isomer are OHClr. Similar differences exist when we compare DPLA from R. sphaeroides and E. coli, a8 shown by the molecular models in Fig. 7. These structural representations emphasize the importance of the size of the hydroxy fatty acids at R1 and Rs of the DPLA in determining biological activity. We propose that when R1 and R3 have hydroxy fatty acids equal to or less than Clo and the total fatty acid content is five, the DPLA will be nontoxic and biologically inactive. When R1 and RI are OHClr and the total fatty acid content is six, the DPLA should be toxic and biologically active. This raises an interesting question: what is the biological property of a DPLA when R1 and R3 are both OHC12? For such a study, we plan to isolate and purify the DPLA from LPS of Neisseria gonorrhoecae (26) and compare its biological activities with the model toxic DPLA from E. coli LPS and the reduced pentaacyl DPLA from R. sphaeroides LPS.
If the %oxo group in the lipid A moiety is not making the LPS nontoxic, one might wonder what its physiological role might be. It is established that the hexaacyl lipid A is more active than the pentaacyl lipid A (25). The 3-oxo group (but not the 3-hydroxy group) at R4 would prevent the acylation of that fatty acid (to form the acyloxyacyl group) to increase the fatty acid content to six and thus yield a more active and toxic LPS.