Differences in Membrane Fluidity and Fatty Acid Composition between Phenotypic Variants of Streptococcus pneumoniae

Phase variation in the colonial opacity of Streptococcus pneumoniaehas been implicated as a factor in the pathogenesis of pneumococcaldisease. This study examined the relationship between membranecharacteristics and colony morphology in a few selected opaque-transparentcouples of S. pneumoniae strains carrying different capsulartypes. Membrane fluidity was determined on the basis of intermolecularexcimerization of pyrene and fluorescence polarization of 1,6-diphenyl1,3,5-hexatriene (DPH). A significant decrease, 16 to 26% (P $≤$ 0.05), in the excimerization rate constant of the opaque variantscompared with that of the transparent variants was observed,indicating higher microviscosity of the membrane of bacterialcells in the opaque variants. Liposomes prepared from phospholipidsof the opaque phenotype showed an even greater decrease, 27to 38% (P $≤$ 0.05), in the pyrene excimerization rate constantcompared with that of liposomes prepared from phospholipidsof bacteria with the transparent phenotype. These findings agreewith the results obtained with DPH fluorescence anisotropy,which showed a 9 to 21% increase (P $≤$ 0.001) in the opaque variantscompared with the transparent variants. Membrane fatty acidcomposition, determined by gas chromatography, revealed thatthe two variants carry the same types of fatty acids but indifferent proportions. The trend of modification points to thepresence of a lower degree of unsaturated fatty acids in theopaque variants compared with their transparent counterparts.The data presented here show a distinct correlation betweenphase variation and membrane fluidity in S. pneumoniae. Thechanges in membrane fluidity most probably stem from the observeddifferences in fatty acid composition.

INTRODUCTION

Top

Introduction

Phase variation in the colonial opacity of Streptococcus pneumoniaehas been implicated as a factor in the pathogenesis of pneumococcaldisease (27). The different appearance of bacterial coloniesis assumed to result from the spontaneous and reversible phasevariation of surface components, the identity of which is notyet clear. The frequency of switching is highly variable fromisolate to isolate, ranging from 10^–3 to 10^–6 pergeneration. The significance of opacity variation in the biologyof pneumococcal infection in vivo was examined by using animalmodels of nasopharyngeal colonization and bacteremia. Transparentvariants persist in the nasopharynx in vivo and show greateradherence to human lung epithelial cells. However, experimentsperformed with an adult mouse model of sepsis showed a strongselection for organisms with the opaque morphology during invasiveinfections (28).

Genetic experiments were used to isolate a single locus ableto confer altered colony opacity at a higher frequency thanthe background rate (18). The opacity locus was found to beassociated with two genes in the presumed glycerol operon, glpFand glpD, required for glycerol metabolism in other bacteria.This finding raises the possibility that phase variation inS. pneumoniae is linked to the mechanism of synthesis of membranephospholipids where glycerol is one of the major building blocks.

The linkage between membrane characteristics and cell physiologyhas been widely covered in the literature (17, 20). Many processesassociated with cell growth and cell function are accompaniedby changes in membrane characteristics. One example is the phenomenonof "homeoviscous adaptation" in membranes of bacteria, wherechanges in growth temperature or hydrostatic pressure inducechanges in the activities of enzymes involved in fatty acidmetabolism by altering the proportion of unsaturated fatty acidsin their phospholipids (8, 26).

In this report we examine the hypothesis that pneumococci usean analogous adaptation machinery involving changes in theircytoplasmic membrane. The present work demonstrates a linkagebetween phase variation in colony morphology and the biophysicaland biochemical characteristics of the membrane bilayer in selectedopaque and transparent variants of S. pneumoniae.

MATERIALS AND METHODS

Top

Materials and Methods

Bacterial strains and growth conditions. Four opaque-transparent variants of S. pneumoniae were includedin this study (kindly provided by J. N. Weiser, University ofPennsylvania); the two variants of each strain were isolatedfrom the same ancestor colony (Table 1). The capsular type wasconfirmed by the quellung reaction (4) with antisera providedby the Statens Serum Institute of Copenhagen, Denmark. The identityin the genetic background of each pair was confirmed by pulsed-fieldgel electrophoresis (21) (Fig. 1). All strains were culturedin brain heart infusion (BHI) broth (Difco Laboratories, Becton,Dickinson and Company, Sparks, Md.) supplemented with 5% horseserum (Biological Industries, Beit Haemek, Israel). The strainswere grown at 37°C with aeration until they reached an opticaldensity at 620 nm (OD₆₂₀) of 0.2. Broth cultures were platedonto tryptic soy plates with 1% agar (Hy-Labs, Rehovot, Israel),onto which 5,000 U of catalase (Worthington Biochemical, Freehold,N.J.) was spread, and incubated at 37°C in a candle extinctionjar (27). Colony morphology was determined under magnificationand oblique transmitted illumination as described by Weiseret al. (27).

View this table:
[in this window]
[in a new window]
TABLE 1. Pneumococcal strains used in this study

View larger version (107K):
[in this window]
[in a new window]
FIG. 1. Pulsed-field gel electrophoresis patterns generated by SmaI digestion of opaque (O) and transparent (T) variants of S. pneumoniae strains belonging to serotypes 23F, 6B, 6A, and 18C. ${lambda}$ , lambda ladder; R6, S. pneumoniae reference strain used as a molecular weight marker.

Measurements of membrane lateral diffusion. Lateral diffusion of the fatty acyl chains was measured by theintermolecular excimerization of pyrene. This fluorescent probe,when excited by light, can form a complex with an identicalunexcited probe molecule. Such a complex is called an excimerand is recognized by the production of a new fluorescent bandat a longer wavelength than the usual emission spectrum of themonomer. The rate of excimer formation depends on three parameters:pyrene concentration (C_Py), excimer lifetime ( ${tau}$ _Py), and the diffusionrate constant (k_a). The excimer-to-monomer ratio increases linearlywith the pyrene concentration. The probe's lateral diffusioncan be derived from the slope of the curve as follows: excimer/monomer= k_a ${tau}$ _PyC_Py (9).

Samples of bacterial cultures at steady-state growth, OD₄₅₀= 0.25, were fixed by formaldehyde (final concentration, 0.25%),washed twice with phosphate-buffered saline (PBS) containing0.25% formaldehyde (pH 7.4), and incubated with increasing concentrationsof pyrene, 0.1 to 0.4 µM (added as a 10^–4 M solutionin ethanol). The extent of incorporation of the probe was notsignificantly different for any of the strains used. An unlabeledportion of the sample, incubated under the same conditions,served as a scattering reference (less than 1% of the intensityof the labeled sample). Measurements were carried out at 37°Cwith a benchtop continuous spectrofluorometer (Ratio Master;Photon Technology International, Inc.) with excitation at 335nm, emission at 373 nm (for the monomer) and 470 nm (for theexcimer), and 4- and 4-nm slits, respectively (25).

Measurement of fluorescence anisotropy. Measurement of fluorescence anisotropy is used for studyingthe rotational diffusion of the fatty acyl chains in the membraneinterior. The sample is excited with polarized light, and theemitted light is polarized to a different extent, dependingon, among other factors, the angle between the absorption andemission dipole moments of the probe and its rotational Brownianmotion during the excited-state lifetime. The magnitude of therotational Brownian motion depends on the size and shape ofthe probe molecule and its surrounding microviscosity and temperature.

Fluorescence anisotropy is defined as the ratio of polarizedcomponents to the total intensity by the equation A = I_II –I/I_II + 2I, where I_II and I are the fluorescence intensitiesparallel and perpendicular to the direction of the excitationlight beam. Anisotropy of DPH (1,6-diphenyl 1,3,5-hexatriene)was used to monitor changes in membrane dynamics, as describedpreviously (29). Briefly, samples of bacterial cultures at steady-stategrowth, OD₄₅₀ = 0.25, were fixed by formaldehyde (final concentration,0.25%), washed twice with PBS containing 0.25% formaldehyde(pH 7.4), and then incubated for 1 h at 37°C with 5 x 10^–6M DPH (added as a 10^–4 M solution in tetrahydrofuran).Unlabeled organisms served as a scattering reference (less than3% of the intensity of the labeled sample). Steady-state fluorescenceanisotropy was measured at 37°C with a Perkin-Elmer LS50Bspectrofluorometer (Perkin-Elmer Ltd., Beaconsfield, England)with excitation at 360 nm and emission at 430 nm, 2.5- and 2.5-nmslits, respectively, and a 3-s integration time.

Lifetime measurements. Changes in pyrene excimerization rate constants and DPH anisotropyare interpreted here in terms of variation in viscosity afterverifying that the DPH and pyrene lifetimes are constant underthe conditions studied. DPH and pyrene lifetime measurementsof were carried out with a single-photon counting spectrofluorometer(FLS 920; Edinburgh Instruments) under the conditions studied.

Preparation of liposomes. Bacterial phospholipids were extracted by the methanol-chloroformmethod under nonoxidizing conditions and dried (3). Dry phospholipidswere resuspended in PBS, and liposomes were prepared by gentlesonication (model VCX 750; Sonics and Materials Inc., Newtown,Conn.) three times for 30 s each time at 200 W. To determinephospholipid concentrations, phosphorus content was determinedby the method of Lanzetta et al. (12). The pyrene excimerizationrate constant was determined in solutions containing 0.5 mMphospholipids. The extent of incorporation of the probe wasnot significantly different for any of the liposome types.

Fatty acid extraction and analysis. Lipids were extracted from lyophilized biomass with chloroformand methanol as described by Bligh and Dyer (6) and separatedby two-dimensional thin-layer chromatography. Lipids were transmethylatedwith 2% H₂SO₄ in methanol at 70°C for 1 h (7). The resultingfatty acid methyl esters were analyzed by gas chromatographyon a Supelcowax 10 with a temperature gradient of 185 to 225°Cas previously described (7). A known amount of an internal standard(C_17:0) was added to each sample to allow for lipid quantitation.Fatty acid methyl esters were identified by cochromatographywith authentic standards (Sigma Chemical Co.) and by comparisonof their equivalent chain lengths (1).

Statistical analysis. The significance of differences between opaque and transparentvariants was determined by the unpaired t test.

RESULTS

Top

Results

Membrane dynamic characteristics in whole bacteria and their relation to phase variation. (i) Monitoring by pyrene excimerization. Two observations confirm that the probe was inserted into themembrane. First, the emission spectrum of pyrene in dilute solutionsdepends on solvent polarity (22). The ratio of the fluorescenceintensities of the first ( ${lambda}$ = 373 nm), third ( ${lambda}$ = 383 nm), andfourth ( ${lambda}$ = 395 nm) vibronic bands of the pyrene emission spectrawas used to sense the polarity of the environment where theprobe molecule is inserted. Figure 2 shows the emission spectraof pyrene (normalized by the intensity of the first emissionband at 373 nm) in different solvents (water, ethanol, and tetrahydrofuran),as well as in bacteria and in liposomes. The overall shape ofthe pyrene emission spectra in bacteria and in liposomes presentedwas typical of all four of the strains included in this study.The low concentrations of pyrene used in our experiments dictatedwider slits in order to get a reasonable fluorescence intensity.Therefore we had to use 4-nm slits, which provided an adequatecompromise between resolution and sensitivity. The three resolvedvibronic bands were successfully used to sense the polarityof the probe's environment. While the shift in the locationof the emission maxima is small, the ratio of the three majoremission bands of the monomer (bands I, III, and IV) changesconsiderably with solvent polarity. Table 2 shows the ratioof the three major emission bands of pyrene monomer. The intensityof the emission bands at 383 and 395 nm is higher in less polarsolvents than in polar solvents. The vibronic profile of pyrenein bacteria and liposomes resembles that of pyrene in less polarsolvents, suggesting that the probe was inserted into the hydrophobiclipid bilayer. Second, the total concentration of pyrene inthe suspension of the labeled cells is 0.2 µM, which istoo low for noticeable formation of excimers in a homogeneoussolution (25). The appearance of the excimer peak (at 470 nm)in the fluorescence spectra of bacterial cells and liposomes(Fig. 2) is thus indicative of a much higher local concentrationof the probe partitioned into the membrane.

View larger version (23K):
[in this window]
[in a new window]
FIG. 2. Fluorescence emission spectra of 0.2 µM pyrene (excitation at 335 nm), normalized by the intensity of the first emission band at 373 nm, in three different solvents (water, ethanol, and tetrahydrofuran [THF]), in bacteria, and in liposomes.

View this table:
[in this window]
[in a new window]
TABLE 2. Ratios of the three major pyrene monomer emission bands in different solvents, in bacteria, and in liposomes

Membrane microviscosity in four opaque-transparent pairs, eachpair carrying a different capsular type, was examined by measuringthe lateral diffusion of pyrene (Table 3). A significant decreasein the pyrene excimerization rate constant, 16 to 26% (P $≤$ 0.05),was measured in the opaque variants compared with the transparentvariants. Pyrene excimer lifetime measurements were also carriedout in order to show that the decrease in the pyrene excimerizationrate constant in the opaque variants compared with that in thetransparent variants was due to reduced lateral diffusion ofthe probe molecules rather than to a shorter excimer fluorescencelifetime, according to the equation excimer/monomer = k_a ${tau}$ _PyC_Py.The conserved pyrene lifetime in each of the two opaque-transparentvariants (Table 3) verifies that the decrease in the excimerizationrate constants in the opaque variants is due, at least in part,to a lower rate of lateral diffusion of the probe, indicatingincreased membrane microviscosity in the opaque variant.

View this table:
[in this window]
[in a new window]
TABLE 3. Pyrene excimerization rate constants in four opaque-transparent pairs of S. pneumoniae^a

(ii) Monitoring by DPH anisotropy. Membrane microviscosity was also estimated by measuring therotational diffusion of the fatty acyl chains in the membraneinterior. This was done by the method of steady-state fluorescenceanisotropy with DPH as a probe (Table 4). The results show a9 to 21% increase in the anisotropy value (P $≤$ 0.001) in theopaque variants compared with the transparent variants, indicatingincreased microviscosity of the lipid core associated with theopaque phenotype. DPH lifetimes measured by single-photon correlationwere practically identical for each opaque-transparent pair,suggesting that the different anisotropy values between theopaque and transparent variants result from changes in membraneviscosity.

View this table:
[in this window]
[in a new window]
TABLE 4. DPH anisotropy in three opaque-transparent pairs of S. pneumoniae^a

Membrane dynamic characteristics in liposomes. To verify that the difference in membrane fluidity originatesfrom the lipids, pyrene excimerization was examined in liposomes,which were reconstructed from purified lipids of the opaqueand transparent phenotypes. As summarized in Table 5, liposomesreconstructed from phospholipids of the opaque variant showeda significant decrease in the pyrene excimerization rate constantcompared to liposomes that were created from phospholipids ofthe transparent variant, 27 to 38% (P $≤$ 0.05). Pyrene lifetimesin liposomes were very similar for each opaque-transparent pair(Table 5). The difference between the opaque and transparentvariants of each pair was even more prominent when measuredin liposomes compared to the results in whole bacteria (Table3), suggesting that the increased membrane microviscosity inthe opaque variants is due, at least in part, to differencesin their lipid compositions.

View this table:
[in this window]
[in a new window]
TABLE 5. Pyrene excimerization rate constant in liposomes that were reconstructed from purified phospholipids of the opaque and transparent phenotypes^a

Fatty acid analysis. Gas chromatographic analysis of fatty acid methyl ester compositionwas performed on four opaque-transparent pairs. In Table 6,the relative percentages of the cell fatty acids of two S. pneumoniaeopaque-transparent pairs are reported. The two variants carrythe same types of fatty acyl residues, mainly saturated andunsaturated straight C₁₆ and C₁₈ acids. However, the proportionsof the various fatty acids varied. For each serotype and phasevariant we have calculated the desaturation index as the weightedaverage of the number of double bonds per fatty acid. The mostnoticeable difference between the two variants was the degreeof unsaturation, which was higher in the transparent variantsthan in the opaque variants. For serotype 6A, the desaturationindex was 0.41 in the transparent variant and 0.32 in the opaquevariant (28% difference); for serotype 6B, it was 0.51 in thetransparent variant and 0.40 in the opaque variant (28% difference);for serotype 18C, it was 0.41 in the transparent variant and0.31 in the opaque variant (32% difference); and for serotype23F, it was 0.46 in the transparent variant and 0.38 in theopaque variant (21% difference).

View this table:
[in this window]
[in a new window]
TABLE 6. Fatty acid compositions of opaque and transparent phase variants of S. pneumoniae^a

DISCUSSION

Top

Discussion

Variability and adaptability are crucial characteristics oforganisms possessing the ability to survive and prosper undervarious environmental conditions (2). S. pneumoniae has theability to thrive in a number of different host environments,including the bloodstream and the mucosal surfaces of the upperrespiratory tract. The pneumococcus, when grown under differentconditions, undergoes spontaneous and reversible phase variation,which is apparent as differences in colony opacity on transparentagar surfaces. Phase variation involves changes in the amountof phosphorylcholine and the amounts of several important surfaceproteins (28). The nature of the mechanism controlling phasevariation is still under investigation.

A universal conserved adaptation response observed among bacteriais adjustment of the membrane lipid composition to various growthconditions. It has been shown (10) that the general responsemechanism of certain thermotolerant strains or species to superoptimaltemperatures, as well as to oxidative stress, is associatedwith an increased degree of fatty acid unsaturation or decreasedfatty acid chain length. The changes in lipid composition enablethe microorganisms to maintain membrane biophysical characteristicsand biochemical functions in the face of environmental fluctuations.

The cell membrane of the pneumococcus is a typical bilayer composedmainly of various phospholipids, glycolipids, and proteins (23).Changes in lipid composition, like the degree of unsaturationof fatty acyl chains and their length and degree of branching,result in alterations in the biophysical characteristics ofthe membrane. The dynamic characteristics are expressed by theso-called "membrane microviscosity" parameter, which relatesto the physical state of the lipid acyl chains. One can describethree distinctive modes of motion of the acyl chains as (i)lateral diffusion of individual molecules within the face ofthe bilayer, (ii) rotational diffusion of the fatty acyl chainsin the bilayer interior, and (iii) transbilayer diffusion ("flipflop"), which is quite rare. In the course of the present study,membrane characteristics were documented in phase variants ofS. pneumoniae. We used two different fluorescence assay methodsthat provide information on the two diffusion modes.

To avoid any changes caused by enzymatic activities during labelingand measurements, formaldehyde fixation was used during samplepreparation. As reported before (5) and checked by us for eachstrain used in this study, fixation itself had no effect oneither pyrene lateral diffusion or the fluorescence anisotropyof DPH (data not shown).

The rate of lateral diffusion of membrane phospholipids wasevaluated by determining the intermolecular excimerization ofpyrene. The monomer of this highly hydrophobic molecule exhibitsfive well-resolved major vibronic bands between 370 and 400nm generally labeled I to V in progressive order. The ratioof the emission intensities of vibronic bands I, III, and IVdepends on solvent polarity and is therefore commonly used tocharacterize the polarity of structured and unisotropic medialike micelles and biological membranes (22). Although the probemolecules may be distributed in the membrane at different sitesand depths, having a distinct type of polarity, the spectrarepresent an average of the emission characteristics of theenvironment sensed by the individual probe molecules. The emissionspectra of pyrene and the intensity ratios (listed in Table2) in bacteria and in liposomes clearly show that pyrene wasimmersed in a nonpolar environment, namely, the membrane.

Differences in the lateral diffusion of pyrene were found forthe two colony morphologies when measured in four opaque-transparentpairs of different serotypes. The excimerization rate constantsin the opaque variants were significantly lower than those intheir transparent counterparts (16 to 26%, P $≤$ 0.05), indicatinghigher membrane microviscosity in the opaque variants than inthe transparent variants. The pyrene lifetime, measured underthe conditions studied, showed no change, thus supporting ourinterpretation that the difference in the pyrene excimerizationrate constants was linked to variation in membrane microviscosity.

It is well known that polypeptides, proteins, and other nonlipidmembrane constituents may significantly reduce the lateral diffusionof lipids in biological membranes (16). In addition, pyrenediffusion and excimer formation may be hampered by the inhomogeneityof the membrane. In order to verify that the difference in membranedynamic characteristics originated from the phospholipids andnot other, nonlipid, membrane constituents, we repeated thepyrene experiment with liposomes, which were reconstructed frompurified phospholipids from cell membranes of opaque and transparentvariants. Differences in lateral diffusion were most notablein liposomes, as pyrene excimerization rate constants were lowerby 27 to 38% (P $≤$ 0.05) in the opaque variants than in the transparentvariants, suggesting that the increased membrane microviscosityin the opaque variants was due, at least in part, to a differentlipid composition.

An additional mode of motion, the rotational diffusion of thefatty acyl chains in the bilayer interior, was evaluated byfluorescence anisotropy, with DPH as a probe. It is a rodlike,highly hydrophobic molecule that intercalates into the membranelipid core. DPH is oriented parallel to the axis of the lipidacyl chain, and its mode of motion is assumed to resemble therotational diffusion of the lipid chains. Anisotropy changesare interpreted here in terms of microviscosity variation afterverifying that the DPH lifetime did not change under the conditionsstudied. It is important to note that small changes (of about10%) in fluorescence anisotropy may reflect pronounced changes(of about 25%) in membrane microviscosity (19). The resultsobtained with DPH show a 9 to 21% increase in the anisotropyvalues (P $≤$ 0.001) in the opaque variants compared with the transparentvariants, indicating an increase in membrane microviscosityof about 25%. These findings agree with the results obtainedwith pyrene, where lateral diffusion in the membrane of theopaque variant was lower than in the transparent variant, signifyinghigher microviscosity in bacteria with the opaque phenotype.

It was shown in Escherichia coli that changes in cell surfacephysical properties, such as phase transition temperature andmembrane microviscosity, show a positive linear correlationwith the proportion of unsaturated fatty acids in the bacteriallipids (14). In this study, the lipid acyl chain profile offour opaque-transparent S. pneumoniae strains shows that thetwo variants carry the same types of fatty acyl residues, mainlysaturated and unsaturated straight C₁₆ and C₁₈ acids, as reportedbefore for other strains of S. pneumoniae (24). However, theratio of unsaturated to saturated fatty acids was increasedin the transparent variants compared with the opaque variants.These results can explain the lower microviscosity observedin bacteria with the less saturated transparent phenotype. Preliminarydata indicate that the enhancement of the desaturation indexin the transparent variant is likely to result from an acrossthe board increase in the desaturation level of the fatty acids,regardless of their lipid host. However, we are aware that changesin the ratio of membrane lipids may also have an effect on theproperties of the different phenotypes. We are currently comparingthe lipid distribution in the opaque variants with that in thetransparent variants.

In general, introduction of double bonds into acyl chains canbe achieved either anaerobically during fatty acid synthesis(13) or aerobically by modification of existing fatty acidsthrough fatty acid desaturases (26). The cellular fatty acidcomposition is a result of a sum of complex phenomena maintainingoptimal viability of the cell under various conditions. Therefore,it is difficult to understand the adjustment mechanisms linkingfatty acid composition to various oxygen pressure or other stressfactors.

The transition from benign pneumococcal colonization in thenasopharynx to invasive disease has been shown to be affected,both in vivo and in vitro, by the reversible phase variationin colony morphology from transparent to opaque. A molecularexplanation for these divergent capabilities has begun to emergefrom the understanding of the biochemical differences betweenthe phenotypic variants. In this study we showed that the twovariants differ in their membrane fluidity characteristics andfatty acid compositions. The function of the bacterial membraneresponse to environmental alterations, like oxygen availability,temperature, etc., can be rationalized on the basis of the advantagesthat it offers to the pneumococcus whose survival depends onvariable environmental conditions. In general, tolerance tooxygen is known to involve enzyme-based detoxification and freeradical scavenging mechanisms that have been described for manydifferent organisms; these mechanisms are strongly induced bya mild oxidative burst (15). The decreased number of doublebonds may confer an advantage on the opaque variants by decreasingtheir susceptibility to lipid peroxidation during an oxidativeburst.

Further experiments will be carried out in our laboratory withgenetically defined mutants that are essentially deficient inH₂O₂ production to show the effect of peroxide production onthe biophysical characteristics and fatty acid composition ofthe membrane under various oxygen pressures. These experimentswill provide more data on the adaptation processes endured bythe pneumococcus under various environmental conditions.

ACKNOWLEDGMENTS

We thank Sofia Kolusheva and Ronit Trefler for expert technicalassistance.

This work was partially supported by grant 5589 from the Ministryof Health, Israel.

FOOTNOTES

* Corresponding author. Mailing address: Pediatric Infectious Disease Unit, Soroka University Medical Center, P.O. Box 151, Beer Sheva 84101, Israel. Phone: 972-8-6400839. Fax: 972-8-6232334. E-mail: npurat{at}bgumail.bgu.ac.il.

REFERENCES

Top