Effect of Variations in Lipopolysaccharide on the Fluidity of the Outer Membrane of Escherichia

The lipid hydrocarbon chains in the outer membrane of gram-negative bacteria appear from previous experiments to be less mobile than in the cytoplasmic membrane. To determine whether lipopolysaccharide, a unique outer membrane component, is a cause of this restricted mobility, outer membranes differing in the amount of lipopolysaccharide, and the length of the polysaccharide side chain, were prepared from Escherichia coli 55. Cytoplasmic membranes were prepared for comparison. The probes, 5- and 12-dox-ylstearate, were introduced into these membranes, electron spin resonance spectra were analyzed, and the order parameter (S) and empirical motion parameter (ro) were calculated. Outer membrane preparations containing long chain lipopolysaccharide were much less fluid by these criteria than were preparations containing short chain lipopolysaccharide. Removing about 40% the lipopolysaccharide from the former preparations greatly increased their fluidity. The lipid in the cytoplasmic membrane preparations was more fluid

The lipid hydrocarbon chains in the outer membrane of gram-negative bacteria appear from previous experiments to be less mobile than in the cytoplasmic membrane.
To determine whether lipopolysaccharide, a unique outer membrane component, is a cause of this restricted mobility, outer membranes differing in the amount of lipopolysaccharide, and the length of the polysaccharide side chain, were prepared from Escherichia coli 55. Cytoplasmic membranes were prepared for comparison.
The probes, 5-and 12-doxylstearate, were introduced into these membranes, electron spin resonance spectra were analyzed, and the order parameter (S) and empirical motion parameter (ro) were calculated. Outer membrane preparations containing long chain lipopolysaccharide were much less fluid by these criteria than were preparations containing short chain lipopolysaccharide.
Removing about 40% of the lipopolysaccharide from the former preparations greatly increased their fluidity. The lipid in the cytoplasmic membrane preparations was more fluid than in the outer membrane and cytoplasmic membranes were similar to each other regardless of the composition of the outer membrane.
These results indicate that lipopolysaccharide, and especially the polysaccharide portion, directly or indirectly causes the restricted mobility of the lipid hydrocarbon chains observed in the outer membrane.
The outer membrane of gram-negative cells differs in many physiological and biochemical properties from the cytoplasmic membrane.
Although their lipid content is similar, their proteins are totally different, and only the outer membrane contains lipopolysaccharide (1,2). While the cytoplasmic membrane has permeability properties common to many biological membranes, being impermeable to most ionized and organic molecules, the outer membrane is a molecular sieve, rather freely permeable to molecules of less than M, = 700, although larger ones are excluded (3,4).
Several recent studies have suggested that some or all of the lipid of the outer membrane is less mobile than is the lipid of the inner membrane.
Cheng et al. (5) found, by using fluorescence polarization, that the outer membrane of Escherichia coli showed more microviscosity than the inner membrane. for the synthesis of lipopolysaccharide. It grows normally without galactose but makes an incomplete lipopolysaccharide lacking most of its carbohydrate (8). The growth medium was made by mixing equal volumes of proteose peptone-beef extract medium (1) and basal salts medium with the glucose omitted (9). The proteose peptone No. 3 was routinely tested as previously described to insure that it lacked galactose (10). n-[l-'4Clgalactose (55 mCi/mmol) and [2-"HIglycerol (200 mCi/mmol) were obtained from the New England Nuclear Corp.
Growth and Labeling ofBacteria -For cells grown in the absence of galactose, 20 ml of an overnight culture grown in the above medium containing 1 rnM glycerol was used to inoculate 1 liter of the same medium containing 1 mM ["HIglycerol, either 0.05 or 0.1 mCi/ mmol. Cells grown with galactose were grown in the same manner except that the overnight culture also contained 2.5 rnM galactose, with 2 rnM fucose as well, to insure complete induction; and the liter culture contained 2.5 rnM [14Clgalactose, 40 $Xmmol.
The cells were grown at 37" with vigorous aeration and harvested at a density of 4 X 10X/ml. EDTA Treatment -The cells were sedimented by centrifuging at 12,000 x g for 5 min at 4". Half of the cells were used for membrane preparation without further manipulation (see below). The other half was resuspended in 50 ml of 0.12 M Tris-HCl, pH 8.0, at room temperature and warmed to 37" with gentle shaking. EDTA (1 rnM final concentration) was added and shaking was continued for 2 min. The culture was sedimented at 12,000 x g for 3 min at room temperature. The cell pellet was used for membrane preparation as described below.
Membrane Preparation -All steps were at O-4" unless otherwise indicated. All solutions were prepared in distilled water that was of E. coli also deionized by passage over Dowex 50-H+. Both the EDTA-treated and the control cell pellets were resuspended in sucrose-Tris, exposed to lysozyme, and thereafter slowly diluted with EDTA, exactly as described by Osborn et al. (1). The resultant spheroplasts were lysed osmotically as described (11 and the membranes were sedimented for 1.5 h at 142,000 x g. The pellet contained 80 to 95% of the membranes as measured by content of [3Hlglycerol precipitable by trichloroacetic acid. The membranes were resuspended in 3 ml of a sucrose solution containing 5 rnM EDTA (25% sucrose for membranes from cells grown without galactose and 30% sucrose for membranes from cells grown with galactose). The suspension was layered on gradients containing 6 ml each of 30, 35,40, 45, and 50% sucrose (for membranes from cells grown without galactosel or 35, 40, 45, 50, and 55% sucrose (for membranes from cells grown with galactosel. All solutions were made in 5 rnM EDTA. The gradients were centrifuged 41 to 50 h at 51,000 x g and samples were collected from the bottom with a Perpex pump. Portions of each sample (usually 50 ~1) were counted in a scintillation counter.
brane preparation was assessed from the order parameter (191 and from TV, an empirical motion parameter (201. The order parameters (S) are related to the mean angular deviation of the labeled fatty acid chain from its average orientation in the membrane. Low values of order parameter are associated with higher freedom of motion of membrane lipids. Since the observed values of T',, differs from the real value by a factor related to (T',, -T'rl, order parameters were calculated according to the following equation (19): Fractions comprising outer membrane or inner membrane peaks were pooled and diluted dropwise with 5 rnM EDTA to one-third or less of the original sucrose concentration.
They were then sedimented by centrifuging for 5 to 15 h at 140,000 x g. In some cases they were repurified on a second sucrose gradient prior to sedimentation. The membranes were resuspended in a small amount of Tris-Cl, pH 7.5, and stored at -150".
Enzyme Assays -n-Lactic dehydrogenase was assayed as previously described (11) and succinic dehydrogenase was assayed in an identical manner except that succinate was substituted for n-lactate. Measurements oflipopolysaccharide, Lipid, and Protein -The relative content of lipopolysaccharide in membranes from cells grown in the presence of ['YZlgalactose was determined by measuring the trichloroacetic acid-precipitable [14Clgalactose.
It has previously been shown for this strain that all (>99%) of the acid-precipitable galactose in this strain is in lipopolysaccharide (12). The relative content of lipopolysaccharide in membranes from cells grown without galactose was measured in the following manner. Cells, membranes, or the material released from EDTA-treated cells, or all three, were extracted with phenol to isolate the lipopolysaccharide (131, the dialyzed aqueous extract was treated with RNAse, dialyzed, and concentrated (14). The content of 2-keto-3-deoxyoctulonate, a component present only in lipopolysaccharide, was estimated by the thiobarbituric acid assay (15). A typical preparation yielded the following results: lipopolysaccharide from 500 ml of control cells, 23.0 pg of 2-keto&deoxyoctulonate; lipopolysaccharide from 500 ml of EDTA-treated cells, 12.6 pg; lipopolysaccharide extracted from material released by EDTA treatment, 11.0 pg. Since the stoichiometry of recovery of released lipopolysaccharide correlated well with that remaining in the cells, in many experiments the per cent loss of lipopolysaccharide due to EDTA treatment was estimated by measuring only the lipopolysaccharide recovered from the supernatant rather than assaying the lipopolysaccharide that remained in the cells or membranes.
The relative content of lipid in membranes was assayed by measuring the trichloroacetic acid-precipitable [3Hlglycerol. This compound was proven to be only in lipid by the following criteria. (a) 85 to 90% of the ["HIglycerol co-purified with the membrane fraction; (bl85 to 100% of the [3Hlglycerol was extractable in the lipid fraction by the method of Bligh and Dyer (16).
Protein was assayed by the method of Lowry et al. (17). Sodium dodecyl sulfate gel chromatography was performed as described by Ames (2).
Membranes suspended in 0.12 M Tris-HCl, pH 7.5 (0.35 ml), were incubated for 30 min at 4" with 1 ml of 2.5 rnM of the above probes dissolved in 5% aqueous solution bovine serum albumin Fraction V. Four milliliters of 0.25 M Tris-HCl, pH 7.5, solution was then added, and the membranes were sedimented by centrifugation at 140,000 x g for 5 h. The pellet was washed once and resuspended in 0.05 to 0.15 ml of 0.12 M Tris-HCl, pH 7.5, solution. The samples were then transferred to a disposable pipette, sealed at one end, and electron paramagnetic resonance spectra were obtained using a Varian E-3 spectrometer equipped with a temperature control accessory (Varian Associates, Palo Alto, Calif.).
The freedom of motion of the spin-labeled fatty acids in the mem- where T',, and T; (in gauss) are equal to one-half the separation of the outer and inner spectral extrema, respectively, and C = 1.4 G -0.053 (T',, -T;). In our hands, repeated determinations of S on a given preparation yielded a standard deviation of +O.Ol. The motion parameter (T,J was calculated according to Henry and Keith (20) from the expression: where W, is the line width of the midfield line and h,, and h-, are the heights of the mid-and high field lines on a first derivative absorption spectrum. Since the departure from isotropic tumbling is related to the proximity of the nitrogen atom to the carboxyl group of the fatty acid, it was shown that 12-doxylstearate, though far from being spherical, moves in a nearly isotropic fashion, warranting the use of Kivelson's formula (21) for determining rotational correlation time (~~1. Nevertheless, the spectra reported here are too slow for the oneline shape theory to apply (7, > lo-" s); thus, for comparative purposes, the empirical parameter T" is used (221. Only 7" values faster than lo-" s are presented, as the high field lines at values slower than lo-* s showed nonconsistent variability.

Preparation and Composition
of Membranes-Cells were grown in the presence and absence of galactose as described under "Materials and Methods." Half of each culture was then exposed to EDTA as described and membranes were prepared from all four samples. These membranes were centrifuged to equilibrium and membrane fractions isolated as previously reported (1,10).
The properties of the membranes thus isolated are shown in Table I. The identification of the peaks was based on their content of lipopolysaccharide, an outer membrane constituent, and their content of succinic and lactic dehydrogenases, which are found in the inner membrane.
Since cells treated with EDTA yielded more unseparated membrane material than control cells, in some experiments the outer membrane fractions were pooled, diluted to reduce the concentration of sucrose, and recentrifuged to equilibrium to purify them further. The contamination of outer membranes by inner membranes, when judged by the specific activity of lactic and succinic dehydrogenases, was the same whether or not the cells had been treated with EDTA.
The density of the outer membrane from cells grown with galactose was shown previously to be higher than the density of membranes grown without galactose, presumably because of the higher polysaccharide content of the former (10). In accord with those results, loss of lipopolysaccharide by EDTA treatment lowers the density of the outer membrane slightly, without affecting the density of the inner membrane (Table I). Outer membranes from EDTA-treated cells contained 50 to 70% as much lipopolysaccharide, and 75 to 100% as much lipid relative to protein as did control outer membranes consistent with previous results on the effect of EDTA treatment on whole cells (14). It was of interest that, while lipopolysaccharide was always released by EDTA, in some experiments lipid was released as well, while in some it was not (compare Table   by  ; however, the ESR results obtained, and described below, were the same for both sorts of experiments. The protein composition of the outer membrane of Escherichia coli 55 has been reported not to vary when the cells are grown with or without galactose (23). In these experiments, the composition of outer membranes from cells grown with and without galactose, with or without EDTA treatment, was the same as indicated by identical protein bands after sodium dodecyl sulfate gel electrophoresis.
The gels provided another important control: since proteins affect the fluidity of phospholipid membranes (241, it was important to determine whether there might be sufficient differential binding of lysozyme by EDTA-treated versus control cells to affect the results. Examination of the gels revealed that little or none of the protein (less than 5%) in any of the outer membrane preparations banded in the region of lysozyme. it was necessary to determine whether lipopolysaccharide loss occurs during manipulation of these particular cells grown under these conditions. Cells were grown iri the presence of galactose and the culture divided.
Spheroplasts were prepared from one portion of the culture exactly as described under "Materials and Methods," except that the spheroplasts were not lysed but sedimented by centrifugation for 10 min at 13,000 x g. Less than 10% of the lipopolysaccharide, as measured by trichloroacetic acid-precipitable ['Qgalactose, was in the supernatant after such a procedure. The remaining cells, treated with EDTA as described under "Materials and Methods," released 50% of their lipopolysaccharide to the supernatant.
ESR Results - Table  II shows that at 37" the order parame- The per cent change in the observed order parameter at 37" with the changing position of the label on the fatty acid was much more dramatic for the cytoplasmic membrane than for the outer membrane preparations. Thus, for galactose-grown cells, cytoplasmic membranes yielded order parameters of 0.48 and 0.32 for 5-doxylstearate and 12-doxylstearate, respectively, while outer membranes had order parameters of 0.75 and 0.67 under the same conditions. Fig. 1 shows representative ESR spectra of 12-doxylstearate in outer membrane preparations from cells grown with or without galactose. The spectra, which indicate the presence of both a membrane-bound and a freely tumbling nitroxide spin label, showed great similarity, but the large hyperfine splitting (2T,,), as well as the line height ratio (h&J, differed. Order parameters calculated from such spectra were much higher for the outer membrane preparation of E. coli cells grown with galactose than for those grown without galactose 1. Representative electron paramagnetic reionance spectra, obtained at 30", of 12-doxylstearate in outer membrane preparations from Escherichia coli cells grown with or without galactose. Inset, spectral parameters used in determining the molecular motion of the spin label in the membrane. (Table II). The results indicate that outer membranes containing long chain polysaccharide restrict the motion of the probe more than membranes containing short chain polysaccharide. The restricted molecular motion of the spin-labeled fatty acid in the outer membrane of glactose-grown cells was further demonstrated in an Arrhenius plot of the motion parameter CT,,) uersus OK-' (Fig. 2). At each temperature tested, the motion parameter was higher for outer membranes containing long chain lipopolysaccharide than for those containing short chain lipopolysaccharide.
The Arrhenius plot for both membrane preparations revealed straight lines with activation energies of 15 and 6 kcal/mol for outer membranes with long and short chain lipopolysaccharide, respectively (Fig. 2). These results substantiate and expand those reported above, indicating that the fluidity of the outer membrane is reduced when the carbohydrate moiety of the lipopolysaccharide is long.
Outer membranes from galactose-grown cells that had been treated with EDTA and thus had lost 40% of their lipopolysaccharide, showed a marked increase in the freedom of motion of the spin label relative to untreated cells. This increase was manifested as a decrease in order parameter determined at 37 (Table III) and also as a decrease in the empirical motion parameter at various temperatures (Fig. 2). In contrast, the effect of EDTA treatment was negligible for outer membrane preparations from cells grown without galactose. These results will be discussed below. chain. Exposure of the cell to EDTA removes 35 to 50% of the lipopolysaccharide, with a small amount of outer membrane protein and variable amounts (0 to 30%) of the phospholipid (14). Thus, the effect is primarily on the lipopolysaccharide content.
The current results show that varying only the length of the carbohydrate chain of lipopolysaccharide greatly affects the mobility of a lipid probe added to outer membrane preparations. Reducing the lipopolysaccharide content by EDTA treatment increases the mobility of the probes in a membrane containing a long chain lipopolysaccharide, and the results were independent of whether or not phospholipid was also lost. The important moiety of lipopolysaccharide is not the lipid A, but the polysaccharide, since outer membranes containing the same amount of lipid A but less polysaccharide (strain 55 grown without galactose) show mobility more similar to cytoplasmic membranes; furthermore, reducing the lipopolysaccharide content of these membranes has a negligible effect. Presumably, differences still observed between outer membranes containing short polysaccharides and the cytoplasmic membrane may be due to factors other than the lipopolysaccharide moiety.
An intrinsic limitation of electron spin resonance measurements is the ability of the probes to sample conditions only in the portion of the membrane in which they are contained. If the probes were located in vastly different domains in each experimental situation, interpretation would be difficult. Such a possibility appears unlikely in these experiments because approximately equal amounts of probe were absorbed by outer membranes regardless of their lipopolysaccharide chain length, or whether or not they had been exposed to EDTA. Furthermore, both removal of long chain lipopolysaccharide by EDTA and abbreviation of chain length, which should perturb the membrane in different ways, resulted in internally consistent results.
Parenthetically, in the current experiments, no melting point of lipids (or break in the Arrhenius curve, Fig. 2 lipids of wild type E. coli exhibit only a very gradual melting curve over the temperatures assayed (25). Although at first glance it may appear strange that the presence of polysaccharide side chains should affect the mobility of a probe in the lipid bilayer, there is at least one possible interpretation.
It is known that polysaccharide chains can interact with each other, as measured by x-ray analyses (26). If lipopolysaccharide molecules interact with a certain percentage of the phospholipid of the outer membrane, then interaction of polysaccharide side chains might restrict mobility both of lipopolysaccharide and of the associated lipid. This hypothesis is consistent with data from two sources. First, as mentioned above, x-ray analysis indicates that polysaccharides can interact with each other to form 2-, 3-, and 4fold helices (26