Carbon Source-Induced Modifications in the Mycolic Acid Content and Cell Wall Permeability of Rhodococcus erythropolis E1

Abstract

The influence of the carbon source on cell wall properties was analyzed in an efficient alkane-degrading strain of Rhodococcus erythropolis (strain E1), with particular focus on the mycolic acid content. A clear correlation was observed between the carbon source and the mycolic acid profiles as estimated by high-performance liquid chromatography and mass spectrometry. Two types of mycolic acid patterns were observed after growth either on saturated linear alkanes or on short-chain alkanoates. One type of pattern was characterized by the lack of odd-numbered carbon chains and resulted from growth on linear alkanes with even numbers of carbon atoms. The second type of pattern was characterized by mycolic acids with both even- and odd-numbered carbon chains and resulted from growth on compounds with odd-numbered carbon chains, on branched alkanes, or on mixtures of different compounds. Cellular short-chain fatty acids were twice as abundant during growth on a branched alkane (pristane) as during growth on acetate, while equal amounts of mycolic acids were found under both conditions. More hydrocarbon-like compounds and less polysaccharide were exposed at the cell wall surface during growth on alkanes. Whatever the substrate, the cells had the same affinity for aqueous-nonaqueous solvent interfaces. By contrast, bacteria displayed completely opposite susceptibilities to hydrophilic and hydrophobic antibiotics and were found to be strongly stained by hydrophobic dyes after growth on pristane but not after growth on acetate. Taken together, these data show that the cell wall composition of R. erythropolis E1 is influenced by the nutritional regimen and that the most marked effect is a radical change in cell wall permeability.

Mycolic acids (MA) are high-molecular-weight α-alkyl, β-hydroxy fatty acids found in the cell walls of bacteria belonging to the mycolata family of actinomycetes (9, 10). This family includes the genera Rhodococcus, Gordonia, Nocardia, Corynebacterium, Tsukamurella, and Mycobacterium. MA confer resistance to chemical injury, low permeability to hydrophobic antibiotics, extremely low permeability to hydrophilic substrates, and resistance to dehydration (1). Their presence has also been correlated with peculiar adhesion properties (3).

The repeated isolation of actively dividing mycolata from biotopes contaminated with poorly available, recalcitrant pollutants like long-chain alkanes or high-molecular-weight polyaromatic hydrocarbons (PAH) suggests that members of these taxa may be physiologically favored in such pollution contexts compared to other bacteria (2, 5, 8, 11, 18, 32, 34, 39). However, neither the metabolic advantage nor the role in substrate uptake which MA might play in bacteria faced with hydrophobic contaminants has been clearly demonstrated.

MA are generally found attached to arabinogalactan, the cell wall polysaccharide of mycolata. In Rhodococcus spp., MA represent up to 40% of the cell wall skeleton and typically contain 30 to 54 carbon atoms. They can be partially free in the form of trehalose dimycolates and monomycolyl lipids (23, 28, 36). Due to their biosurfactant properties, the characteristics of these extractable forms and the growth conditions directing their synthesis have been abundantly studied in the context of biotechnological applications (15, 23, 30, 33). Arabinogalactan-bound MA, as well as free glycomycolates, are thought to be localized in the outer layer of the cell wall, where they form the basis of an outer lipid permeability barrier (13, 36). This layer is itself covered by less-characterized surface amphiphiles and capsular material, which counteract the highly hydrophobic character of the MA layer to an extent that varies from strain to strain (35, 36).

Cell wall lipids play a crucial role in the uptake of hydrophobic carbon sources. In addition to membrane abnormalities caused by specific inhibitors of lipid synthesis (1, 13, 36), structural changes in cell wall lipids occur in response to various stress conditions. Temperature shifts, starvation, low pH, and organic solvents can modify the ratios of saturated to unsaturated fatty acids or induce cis-trans isomerization and cyclopropanation (4, 12, 29, 41). These changes are interpreted as a way for bacteria to maintain membrane fluidity and impermeability. The short-chain trehalose alkyl esters produced by Rhodococcus sp. strain 51T7 have been shown to differ after feeding with different alkanes, probably as a consequence of an overlap in the enzymatic pathways for alkane catabolism and fatty acid synthesis (15). The membrane phospholipid and MA profiles used, e.g., for bacterial identification are well known to depend on the composition of the culture medium, as well as on the growth phase (7, 26, 37). In Rhodococcus sp. strain R22 (formerly Mycobacterium convolutum), changes in phospholipids were noticed in response to growth on different n-alkanes (17). In Bergey's Manual of Systematic Bacteriology, Rhodococcus spp. are described as partially positive for acid-alcohol-fast staining, meaning that the cell wall is stained by hydrophobic dyes, such as fuchsin or rhodamine, only at some stage of growth (16). Recently, the nature of the carbon source was shown to drastically affect the structure of the mycolate alkyl chains in PAH-degrading mycobacteria; hydrophobic substrates induced the synthesis of MA with longer alkyl chains compared to the chains of MA recovered when organisms were grown on water-soluble substrates (43). Growth on hydrophobic substrates is also known to increase cell wall hydrophobicity and mycobacterial adhesion to hydrophobic carriers (42). Hence, an active role of MA in these phenomena is suspected.

Besides their usefulness in taxonomy, lipid patterns are characteristic of the physiological status. Whether specific growth substrates can induce changes in cell wall lipids has been scarcely studied except in Escherichia coli. Whether such changes can modify the physiology of environmental bacteria with respect to adhesion capacity, substrate uptake selectivity, or susceptibility to bactericidal compounds is also poorly documented. The purpose of this study was to accurately examine the effects of various carbon sources on the primary structure of MA, as well as on the cell wall physicochemical properties, in an efficient alkane-degrading strain of Rhodococcus erythropolis.

MATERIALS AND METHODS

Bacterial strain and growth conditions.

R. erythropolis E1 (= LMG 21994) was isolated from a hydrocarbon-contaminated soil in Belgium and was identified by 16S RNA sequencing (40). It was maintained on agar plates by using minimal medium MM284 (24) with sodium acetate at a concentration of 2 g liter⁻¹ as the sole carbon source. Phosphate buffer (0.05 M) instead of Tris was used to adjust the pH to 7. Cells were pregrown on agar plates at 30°C, harvested, and suspended in liquid medium MM284. The suspension was used to inoculate liquid minimal medium containing sodium alkanoates, glucose, or alkanes as carbon sources. Unless specified otherwise, the initial concentration of glucose was 2 g liter⁻¹ and the initial concentration of alkanoates and hydrocarbons was 1 g liter⁻¹. Cells were cultivated in conical flasks with rotary shaking (180 rpm) at 20°C. In order to avoid growth-phase-dependent modifications of the MA composition, all experiments were carried out with bacteria harvested in the early stationary phase. Bacterial growth was monitored by determining the optical density at 600 nm (OD₆₀₀) and by cell counting. Total consumption of the carbon source was verified by gas chromatography.

MA preparation and derivatization.

Bacteria were harvested by centrifugation (11,000 × g, 10 min, 4°C) and washed three times with medium MM284. Bacterial pellets were suspended in an appropriate amount of the same medium to obtain an OD₆₀₀ of 2, which corresponded to 3 × 10¹⁰ cells ml⁻¹. Four milliliters of this suspension was autoclaved for 1 h at 121°C with an equal volume of a 25% (wt/wt) KOH solution prepared in 50% ethanol. The pH was then adjusted to 2 with 6 M HCl, and the aqueous phase was extracted three times with 5 ml of CH₂Cl₂. After drying with anhydrous Na₂SO₄, the dichloromethane was evaporated by using a nitrogen flux.

p-Bromophenacyl esters of lipid fatty acids were prepared by adding 0.1 ml of a 0.2 M NaHCO₃ solution to dried dichloromethane extracts and evaporated to dryness. One milliliter of CH₂Cl₂ and 50 μl of p-bromophenacyl-8 reagent (Pierce Chemical Co., Rockford, Ill.) were added successively. The reaction mixtures were sealed tightly in glass tubes and heated for 25 min at 85°C. The samples were then cooled on ice, acidified by adding 1 ml of a 12 M HCl-methanol-water mixture (1:2:1), and vigorously shaken. The organic layer containing the derivatized fatty acids was carefully recovered and evaporated to dryness. Samples were solubilized in 300 μl of CH₂Cl₂ and stored indefinitely at 4°C.

Methyl esters of fatty acids were prepared by adding 10 ml of a methanol-benzene-H₂SO₄ mixture (20:10:1) to 100 mg of a dried dichloromethane extract and incubating the preparation in sealed tubes for 16 h at 70°C. The resulting fatty acid methyl esters were extracted with n-hexane, washed once with water, and evaporated to dryness with a nitrogen flux. Trimethylsilyl (TMS) ether derivatives of methyl esters were prepared by adding 0.1 ml of pyridine and 0.2 ml of bistrimethylsilyl trifluoroacetamide for each 10 mg of fatty acid methyl esters (21). The mixture was then incubated in a sealed tube at 70°C for 20 min. The solvent and the reaction by-products were coevaporated with benzene, and the resulting TMS derivatives were dissolved in a small volume of n-hexane before injection into a gas chromatography (GC)-mass spectrometry (MS) system.

MA analysis.

p-Bromophenacyl esters were analyzed by high-performance liquid chromatography (HPLC) (Waters) by using an RP-C18 column (Novapak; 3.9 by 300 mm; 4 μm; 60 Å; Waters). The mobile phase consisted of a linear gradient of CH₂Cl₂ and methanol (from 0 to 13 min, 0 to 10%; from 13 to 17 min, 10 to 25%; from 17 to 34 min, 25 to 75%; from 34 to 41 min, 30 to 70%; and from 41 to 45 min, 100 to 0%) at a flow rate of 1 ml/min. The separated esters were detected by UV absorption at 254 nm. HPLC quantitative analysis of lipids was performed by using an internal standard, pentacosanoic acid p-bromophenacyl ester (C₂₅). This molecule was synthesized in our laboratory from the corresponding methyl ester (Fluka) by using the protocol described above. Relative ester proportions were determined on the basis of the integrated surfaces calculated for peaks with retention times between either 4 and 13 min (phospholipid fatty acids) or 17 and 32 min (MA). HPLC-MS analysis of p-bromophenacyl esters was performed with an LCQ mass spectrometer (Finnigan Mat) by using the atmospheric pressure chemical ionization method in positive ion mode under the following conditions: vaporizer temperature, 520°C; sheath gas pressure, 60 lb/in²; auxiliary gas pressure, 30 lb/in²; discharge current, 5 μA; capillary temperature, 190°C; capillary voltage, 3 V; and tube lens offset, −20 V.

TMS derivatives of methyl esters were separated by GC (series II 5890; Hewlett-Packard) by using a fused silica capillary column coated with methyl silicone (30 m by 0.25 mm; SPB-1; Supelco Inc., Bellefonte, Pa.). The oven temperature was programmed to increase from 230 to 350°C at a rate of 3°C per min. The final temperature was maintained for 5 min. The injector port temperature was maintained at 350°C. Helium at a flow rate of 1 ml/min was used as the mobile phase. The gas chromatograph was coupled with a TSQ 7000 Finnigan MAT mass spectrometer combined with an electron impact ion source at 70 eV. The interface was heated at 350°C. For rough estimation of the individual MA species proportions, peak heights displayed in the GC chromatograms were measured and ratios were calculated.

XPS.

R. erythropolis E1 cells harvested in the early stationary growth phase were centrifuged for 10 min at 11,000 × g and 4°C and washed three times in distilled water. The pellets were suspended in 2 ml of distilled water, frozen in liquid nitrogen, and stored at −20°C. Cells were lyophilized prior to X-ray photoelectron spectroscopy (XPS) analysis, which was performed as described by Dufrêne et al. (14).

MATH.

The cell surface hydrophobicity of bacteria grown on acetate or pristane (2,6,10,14-tetramethylpentadecane) was measured by the assay for microbial adhesion to hydrocarbons (MATH) (38) by using n-hexadecane, n-decane, ethyl acetate, and chloroform as solvents. Bacteria were first washed three times in distilled water, and the pellets were suspended in enough phosphate buffer (10 mM, pH 7) so that the OD₆₀₀ was 0.5. Three milliliters of each bacterial suspension was mixed with 0.15 ml of the organic solvent in a glass tube and vigorously shaken for 10 s. After the preparations rested for 10 min, the OD₆₀₀ values of the aqueous phase were determined, and an additional 0.15 ml of fresh solvent was added to each mixture. The process was repeated 10 times. The affinity of bacteria for the different solvents was evaluated by computing log(OD_600-i × 100/OD_600-0), where OD_600-i is the OD₆₀₀ determined after addition of solvent aliquot i and resting of the mixture and OD_600-0 is the initial OD₆₀₀ of the bacterial suspension.

Acid-fast staining.

The acid-fast staining test is a classic bacteriological test aimed at identifying MA-containing bacteria. This test involves using hydrophobic dyes, carbol fuchsin and auramine-rhodamine, which form complexes with MA. The complexes are resistant to acid-alcohol washing (19).

(i) Ziehl-Neelsen staining.

Smears of sample bacteria were flame fixed and covered with a hot carbol fuchsin solution (2.5 g of basic fuchsin dissolved in 25 ml of 100% ethanol, 12.5 ml of liquid phenol, and 250 ml of distilled water) for 30 min. The slides were washed with running tap water and bleached for 100 s with ethanolic HCl (0.4% HCl in 70% ethanol) until no more red color left the preparation. The slides were rinsed with tap water once more and counterstained for 5 min with a solution containing 1.4 g of methylene blue per liter. The slides were washed with water and dried in air before microscopic examination.

(ii) Auramine-rhodamine staining.

Smears were covered with an auramine-rhodamine solution (Difco, Detroit, Mich.) for 15 min. The stained samples were washed with running tap water and bleached for 5 min with ethanolic HCl (0.5% HCl in 70% ethanol). The slides were then washed with tap water and dried in air prior to microscopic observation. Fluorescence was analyzed by using a Leica DMR (Wetzlar, Germany) microscope. The filter band passes were as follows: 525 to 550 nm (emission) and 440 to 470 nm (excitation).

Surface tension measurements.

The surface tensions of total cultures and derived preparations were determined by using a Du Noüy interfacial tensiometer. To prepare cell suspensions, cells were harvested by centrifugation (11,000 × g, 10 min, 4°C), washed three times with 0.9% NaCl when necessary, and diluted in fresh culture medium to obtain the initial cell concentration. Cell-free supernatants were obtained by filtration through 0.2-μm-pore-size nitrate-cellulose membranes.

Antibiotic resistance test.

The sensitivity of R. erythropolis E1 to antibiotics was roughly evaluated with antibiotic disks (Becton Dickinson, Franklin Lakes, N.J.). The tetracycline and rifampin MICs were determined in multiwell plates (water-soluble substrates) or in sealed glass tubes (alkanes). Serial dilutions of the antibiotics were made in minimal medium 284, and the concentrations ranged from 10 to 0.04 mg liter⁻¹ for tetracycline and from 100 to 0.4 mg liter⁻¹ for rifampin. Carbon sources were added to a final concentration of 2 g liter⁻¹ (water-soluble substrates) or 20 g liter⁻¹ (alkanes). Samples were inoculated with R. erythropolis E1 pregrown on acetate at an initial cell density of 1.5 × 10⁷ CFU ml⁻¹ and were grown at 30°C for 6 days (water-soluble substrates) or 7 days (alkanes).

RESULTS

MA composition after growth on different carbon sources.

MA profiles for R. erythropolis E1 were established by HPLC by using UV detection after derivatization into p-bromophenacyl esters. MA profiles were compared after growth either on n-alkanoic acid salts with lengths ranging from C₂ to C₇ or on saturated n-alkanes with lengths ranging from C₉ to C₁₅. In order to avoid growth-phase-dependent modifications of the MA composition, all experiments were carried out with bacteria harvested in the early stationary phase (see Materials and Methods). MA eluted at retention times between 17 and 32 min. Basically, two types of profiles were observed when pure carbon sources were used. The type I profiles resulted from growth on molecules containing an even number of carbon atoms (acetate, butyrate, decane, and dodecane) and were characterized by a series of five major signal clusters with mean retention times of 21, 24, 26, 27.5, and 28.5 min (Fig. 1, panels C2, C4, C10, and C12) and in certain cases by a signal near 17.5 min (Fig. 1, panels C10 and C12) or 29.5 min (Fig. 1, panels C2 and C4). The type II profiles resulted from growth on molecules with an odd number of carbon atoms (propionate, valerate, nonane, and undecane) and were characterized by the same peaks plus peaks at intermediate retention times (Fig. 1, panels C3, C5, C9, and C11). Similarly, use of n-hexanoate and n-heptanoate as carbon sources resulted in type I and type II MA profiles, respectively, while glucose-grown bacteria produced a type I profile (data not shown). Growth on a branched alkane, such as pristane (2,6,10,14-tetramethylpentadecane), on mixtures of different n-alkanes or n-alkanoates, or on a complex aliphatic hydrocarbon mixture, such as diesel fuel, resulted in MA profiles that resembled the type II profile but had increased complexity and a Gaussian-like distribution of peak intensities (Fig. 1). MS analysis was conducted on line for HPLC-separated MA esters and showed that the sizes of the major peaks observed in type I profiles differed by 28 mass units (Δm/z = 28), which is consistent with a difference of two methylene residues in the alkyl chains. The intermediate peaks observed only in type II profiles were shown to differ by one methylene residue (Δm/z = 14) from the adjacent major peaks. Since acetate and pristane yielded MA chromatograms with the most marked differences, these two carbon sources were used in subsequent experiments to better characterize the cell wall modifications.

FIG. 1.

HPLC profiles of p-bromophenacyl-derivatized MA as a function of the growth substrate. The relative 254-nm UV signal (in arbitrary units) is expressed as a function of the retention time (in minutes). The following carbon sources were used: acetate (C2), propionate (C3), butyrate (C4), valerate (C5), n-nonane (C9), n-decane (C10), n-undecane (C11), and n-dodecane (C12).

The primary structures of MA could not be determined from the p-bromophenacyl ester derivatives by HPLC-MS. GC-MS analysis of TMS ethers of MA methyl esters was performed instead. The mass spectrum of each detectable peak was analyzed by adopting the mass fragmentation pattern described previously (27, 28). The results are shown in Table 1. A total of 80 different MA molecules were detected; 22 of these molecules specifically originated from growth on acetate, 58 molecules originated from growth on pristane, and 8 molecules were found in both profiles. The total number of carbon atoms ranged from 30 to 40. The proportions of the individual MA species resulting from growth on either carbon source varied from 0.1 to 15.9%. Only species that accounted for at least 2% of the compounds in at least one profile are listed in Table 1. Fully saturated MA containing either 30 or 40 carbons represented less than 1% of the total MA content in both types of profiles. The most abundant MA species were C_34:0, C_36:0, C_38:1, and C_38:0 in profiles resulting from growth on acetate and C_32:0, C_34:0, C_35:0, and C_36:0 in profiles resulting from growth on pristane. The most notable observation was the lack of detectable MA with odd numbers of carbon atoms after growth on carbon sources with even numbers of carbon atoms. Indeed, no MA with a β-chain structure 20:0, 22:0, 22:1, 24:0, or 24:1 was detected in those profiles. Unsaturated MA species were present at a level of approximately 36.7% in acetate-grown bacteria; the proportion was about 13.9% in pristane-grown bacteria.

TABLE 1.

GC-MS analysis of MA composition after growth on either pristane or acetate

Molecular mass of cationa	MA	Retention time (min)	Cβb	Cαb	% of total MA after growth on:c
					Acetate	Pristane
567	C32:0	28	19:0	10:0	6.3	1.5
			20:0	9:0		10.6
581	C33:0	29.72	20:0	10:0		5.3
593	C34:1	30.9-31.2	18:0	13:1	2.0
			21:1	10:0	2.6	0.3
595	C34:0	31.5	15:0	16:0		2.0
			18:0	13:0	2.0
			19:0	12:0	6.4	0.6
			20:0	11:0		15.9
			21:0	10:0	9.9
			22:0	9:0		3.9
609	C35:0	33.2	20:0	12:0		2.4
			21:0	11:0		4.2
			22:0	10:0		3.3
621	C36:1	34.2-34.6	21:1	12:0	7.0
			22:1	11:0		2.1
			23:1	10:0	8.3
			24:1	9:0		2.5
623	C36:0	34.8	19:0	14:0	3.3
			20:0	13:0		3.1
			21:0	12:0	15.0
			22:0	11:0		9.8
			23:0	10:0	4.4	1.3
649	C38:1	37.7-37.9	23:1	12:0	9.5
			25:1	10:0	5.4
651	C38:0	38	21:0	14:0	5.2
			22:0	13:0		3.1
			23:0	12:0	9.7	0.4
			24:0	11:0		1.6

^aIn the molecular masses of cations 15 equals one CH₃ group.

^bNumber of C atoms in MA.

^cThe percentages of the individual MA species were determined on the basis of peak heights displayed in the GC chromatograms. Only species that accounted for at least 2% of the total MA are listed.

Quantification of MA.

In order to compare the MA contents of R. erythropolis E1 cultures grown on pristane and on acetate, quantitative HPLC analysis of MA and short-chain fatty acids resulting from phospholipid hydrolysis was performed. Table 2 shows that samples derived from pristane-grown bacteria contained 1.4 ± 0.1 times more derivatized fatty acids than samples derived from acetate-grown bacteria contained. Fatty acids originating from phospholipids (retention times, 4 to 13 min) were twofold more abundant after growth on pristane than after growth on acetate; they accounted for 55% of the total fatty acids in pristane-grown bacteria and 38% of the total fatty acids in acetate-grown bacteria (Table 2). Identical amounts of MA were detected in pristine- and acetate-grown bacteria. Control GC analysis ensured that no residual growth substrate was present in the lipids analyzed (data not shown).

TABLE 2.

Cellular fatty acid quantification

Carbon source	No. of treated CFU	Total amt of fatty acids (nmol/cell)	Amt of phospholipid fatty acids (nmol/cell)a	Amt of MA (nmol/cell)a	MA/phospholipid fatty acid ratio
Acetate	5 × 1010	4.2 × 10−8	1.6 × 10−8	2.6 × 10−8	62:38
Pristane	5 × 1010	5.8 × 10−8	3.2 × 10−8	2.6 × 10−8	45:55

^aBased on HPLC UV signal quantification after hydrolysis and esterification. An internal standard (pentacosanoic acid p-bromophenacyl ester) was used to calibrate the analysis.

Cell surface hydrophobicity and culture surface tension.

To monitor cell surface hydrophobicity, the affinities of acetate- and pristane-grown R. erythropolis E1, harvested in the stationary growth phase, were determined for different water-organic solvent interfaces by using n-hexadecane, n-decane, ethyl acetate, and chloroform as representative solvents.

There was no significant withdrawal of cells from the aqueous phase (data not shown), indicating that the cells were relatively hydrophilic, following growth on either acetate or pristane. Surface tensions of different culture fractions (whole culture, culture supernatant, suspensions of washed and unwashed bacteria) of R. erythropolis E1 were determined. The results are presented in Table 3. An important decrease in the surface tension of the pristane culture was observed in the early exponential phase. Compared to the surface tension of the whole culture, the surface tension of the cell-free supernatant measured at that time was nearly double, indicating that the biosurfactant activity was not released but rather was associated with bacteria (Table 3). No significant differences between acetate- and pristane-grown bacteria were seen after the carbon sources were completely exhausted.

TABLE 3.

Surface tensions of R. erythropolis El cultures and culture fractions

Prepn	Surface tension (mN m−1) fora:
	Pristane-grown culture	Acetate-grown culture
Whole culture medium	67.3	71.5
Whole culture medium + inoculum	59.2	68.7
Whole culture (early exponential phase)	27.5	61.2
Supernatant (early exponential phase)b	46.5	69.7
Whole culture (late exponential phase)	39.8	50.6
Supernatant (late exponential phase)b	50.6	59.2
Whole culture (stationary phase)	49.4	60.1
Supernatant (stationary phase)b	56.1	57.1
Suspension of unwashed bacteria (stationary phase)	58.9	65.0
Suspension of washed bacteria (stationary phase)c	65.3	65.1

^aMean values for three independent cultures.

^bPreparation obtained by filtration through 0.2-μm-pore-size membranes.

^cPreparation washed three times in 0.9% NaCl and then diluted in fresh culture medium to obtain the initial cell concentration.

Cell surface chemical composition.

XPS showed that the surface of R. erythropolis E1 is mainly composed of carbon, oxygen, and nitrogen; hydrogen is not analyzed by this technique. The surface of pristane-grown bacteria was composed of 70.2% carbon, 24.4% oxygen, and 5.4% nitrogen. In the case of acetate-grown bacteria, the cell surface was composed of 66.2% carbon, 30.6% oxygen, and 3.2%nitrogen. The levels of phosphorus and potassium were below the detection limit, while traces of sodium were detected only at the surface of acetate-grown bacteria. The carbon and oxygen peaks were decomposed as described previously (14). Nitrogen was only in the unprotonated form, as it is in amine or amide functions. Table 4 shows the concentrations of carbon and oxygen present in different functions with respect to the total carbon concentration.The data show that bacteria grown on pristane contained about 10% more carbon in the CH₂ or CH₃ form [C—(C,H)] than bacteria grown on acetate contained. On the other hand, acetate-grown bacteria contained about 10% more oxygenated carbon [C—(O,N) or C=O].

TABLE 4.

XPS analysis of R. erythropolis El cell surface composition

Growth substrate	Molar ratios (%)a
	C—(H,C)/C	C—(O,N)/C	C=O/C	O—C/C	O=C/C	O/C	N	Na/C
Acetate	49.5, 47.6	35.7, 36.6	15.7, 14.8	36.0, 34.1	9.8, 12.9	45.8, 46.9	4.5, 5.2	0.5
Pristane	59.0, 56.6	26.9, 28.1	14.1, 15.3	22.7, 22.0	11.7, 13.2	34.4, 35.3	6.8, 8.5	NDb

^aThe values are 100× molar ratios with respect to total carbon. In most cases the values for two independent cultures are given.

^bND, not determined.

The elemental composition (C, N, O) was used to divide the bacterial surface molecules into three classes of typical compounds (i.e., hydrocarbon-like compounds, polypeptides, and polysaccharides), as described previously (19). Based on these calculations, the percentages of carbon in hydrocarbon-like compounds, polypeptides, and polysaccharides were 41.4, 27.4, and 31.1%, respectively, for pristane-grown bacteria and 33.8, 14.4, and 48.9%, respectively, for acetate-grown bacteria. The standard deviations of these values did not exceed 4%.

Cell wall permeability.

In order to evaluate the effects of different growth substrates on cell surface permeability, acid-fast staining and antibiotic susceptibility were determined. Our results show that R. erythropolis E1 grown on acetate released both carbol fuchsin (data not shown) and auramine-rhodamine dyes (Fig. 2) upon washing with acid-alcohol, while pristane-grown bacteria gave positive results, meaning that they retained the dyes after acid-alcohol washing. This differential staining was observed at any time during cultivation but was most marked during the stationary phase. The same test performed with bacteria grown on propionate, on butyrate, and on pure n-alkanes resulted in intermediate, rather negative results (data not shown).

FIG. 2.

Epifluorescence microscopy of R. erythropolis E1 cells grown on acetate (A) and on pristane (B) after staining with auramine-rhodamine (acid-fast staining).

In order to reveal possible differences in cell wall permeability during growth on different carbon sources, the resistance of R. erythropolis E1 to antibiotics was examined. Rifampin and tetracycline chloride were chosen as representative hydrophobic and hydrophilic antibiotics, respectively. Figure 3 shows that bacteria grown on n-alkanoates were more sensitive to tetracycline, while bacteria grown on different n-alkanes were more sensitive to rifampin. In addition, some correlation was observed between the even-odd nature of the alkanoates supporting growth and the susceptibility to both antibiotics: growth on C₂ or C₄ compounds resulted in greater antibiotic resistance than growth on C₃ or C₅ compounds. In contrast, a correlation between MICs and carbon chain length was not clearly seen during growth on alkanes. n-Decane-grown bacteria were found to be more resistant to tetracycline (MIC, 21.2 mg liter⁻¹) than bacteria cultivated on other alkanes. The medium used for precultivation had no influence on the antibiotic susceptibility (data not shown).

FIG. 3.

Susceptibility of R. erythropolis E1 to antibiotics as a function of the growth substrate. Tetracycline (open bars) and rifampin (solid bars) MIC were determined during growth on different hydrophilic (A) or hydrophobic (B) carbon sources. C2, acetate; C3, propionate; C4, butyrate; C5, valerate; C10, n-decane; C11, n-undecane; C12, n-dodecane; C13, n-tridecane; Mix, mixtures of the n-alkanoates (A) or n-alkanes (B).

DISCUSSION

There have been few previous studies in which the effects of carbon sources on the cell wall lipids of mycolata have been systematically analyzed (15, 17, 43). It has been demonstrated, however, that hydrocarbons are completely incorporated into the membrane phospholipids of hydrocarbon-degrading rhodococci (31). In the present work, different types of carbon sources were used to support growth of R. erythropolis E1, an efficient hydrocarbon degrader. The MA composition and a few selected cell wall properties were analyzed. In contrast to observations made for PAH-degrading mycobacteria (43), no major differences in MA length were found for R. erythropolis E1 growing on water-soluble or insoluble substrates. On the contrary, a strong correlation was observed between the MA profile and the even-odd nature of the carbon chain serving as the substrate. Virtually no MA with odd-numbered carbon chains were synthesized by R. erythropolis E1 grown on even-numbered, linear carbon sources that were either hydrophilic or hydrophobic, while both even- and odd-numbered MA chains were found in the cell wall when odd-numbered molecules served as carbon sources.

According to the currently available information for MA biosynthesis, elongation of the carbon chains may result from sequential addition of acetyl coenzyme A and from addition of propionyl coenzyme A (20), which result in chain length increases of two and three carbon atoms, respectively. A third mechanism involving intramolecular rearrangement around double bonds has been described (25). The results presented here support the hypothesis that there are two different pathways for synthesis of MA in R. erythropolis and indicate the importance of the available carbon sources for establishment of MA profiles. Nishiuchi et al. (28) noticed that most of the MA identified in Rhodococcus, Nocardia, and Gordonia have even-numbered carbon chains. These observations were made after growth on a rich medium containing glucose, yeast extract, and peptone. The use of a mineral medium supplemented with single carbon sources as reported here turned out to be a convenient way to study the biosynthesis and the regulatory mechanisms governing MA biosynthesis in environmental bacteria belonging to the mycolata group.

No remarkable difference in the cell surface hydrophobicity of bacteria harvested in the stationary growth phase was noticed when water-soluble substrates rather than insoluble alkanes were used to feed cultures of R. erythropolis E1. Adhesion (MATH) tests performed following growth under both conditions showed that the cell surface was rather hydrophilic. This observation contrasts with results obtained with PAH-degrading mycobacteria, for which a significant increase in cell surface hydrophobicity was observed upon a switch from glucose to PAH feeding (43). A decrease in the surface tension of whole pristane cultures was observed during the early exponential growth phase, a phenomenon which was also noticed previously by other authors (6, 15). Following separation of pristane-grown cells from their culture supernatant, it was noticed that the biosurfactant activity was mostly associated with the bacterial cells. Production of biosurfactant by various Rhodococcus strains in the early exponential phase is thought to be required to initiate the subsequent degradation of hydrophobic substrates (41).

XPS analysis of R. erythropolis revealed differences in cell surface composition between pristane- and acetate-grown bacteria. The major carbon constituents exposed at the surface of bacteria grown on acetate were polysaccharides (49%), whereas in pristane-grown bacteria hydrocarbon-like compounds were predominant (41%). However pristane-grown bacteria still had a surface polysaccharide concentration of 31%, which explains the global hydrophilic character. The presence of extracellular polysaccharides, which apparently masks the hydrophobic character of the cell wall, was indeed demonstrated for several Rhodococcus strains (35). As the level of phosphorus was below the detection limit, no information was obtained about the concentrations of phospholipids and MA near the cell surface. Note that XPS probes only a limited depth at the surface (about 5 nm) (19). Since the cell wall thickness of gram-positive bacteria is several times greater than this, the observed differences reflect only modifications of the external components of the surface, while changes in the deeper MA layer are not detected by this method.

Accurate HPLC quantification showed that the amounts of MA were identical in pristane- and acetate-grown bacteria. By contrast, phospholipid fatty acids were twofold more abundant in bacteria grown on pristane than in bacteria grown on acetate. Given the fixed quantity of lipids associated with the cytoplasmic membrane, additional short-chain fatty acids can only be associated with the intracellular compartment or with the external part of the cell wall. The XPS results described above are compatible with an increase in the amount of lipid-like compounds at the cell surface of pristane-grown bacteria. On the other hand, intracellular short-chain lipids are known to make up the membrane of the cytoplasmic inclusion bodies observed in hydrocarbon-degrading rhodococci grown on alkanes (22). Hence, both the cell wall surface and residual inclusion bodies are likely to account for the higher short-chain fatty acid concentrations measured by quantitative HPLC in pristane-grown bacteria.

Rhodococcus species are generally described as partially acid fast, meaning that the cell wall reacts positively with hydrophobic dyes only at some stages of growth (16). Interestingly, all cells of an R. erythropolis E1 culture were acid fast after growth on pristane, while they were not acid fast after growth on acetate. Growth on pristane or on other complex alkanes could be useful for facilitating microscopic identification of Rhodococcus species and hence could solve a recurrent problem in classical bacteriology related to the difficulty of staining Rhodococcus MA.

The relative permeability of R. erythropolis cell walls to hydrophilic and hydrophobic molecules was analyzed under physiological conditions by studying the bacterial susceptibility to tetracycline chloride, a water-soluble antibiotic, and rifampin, which is rather hydrophobic. The MIC of tetracycline chloride was found to be about four times higher for R. erythropolis grown on n-decane than for R. erythropolis grown on acetate. Under the same conditions, a sevenfold difference in the rifampin MIC was observed. It is thus clear that the permeability of the R. erythropolis cell wall to hydrophobic molecules is increased upon growth on a hydrophobic carbon source. The opposite is observed for hydrophilic compounds, which are less efficiently transferred to the cytoplasm of cells grown on hydrophobic substrates.

In conclusion, the permeability of the R. erythropolis cell wall is influenced by the class of carbon source used to support growth. Whether the different MA profiles reported here after growth on different carbon sources might influence the selectivity of the uptake and transport of alkanes is an interesting issue that deserves further investigation. Although the present study clearly demonstrated that the cell wall of R. erythropolis E1 becomes more permeable to at least some hydrophobic molecules upon a shift from a hydrophilic carbon source to a hydrophobic carbon source, we were unable to assign this permeability shift to a specific structural change. A hypothetical explanation involves the ratio of free MA to bound MA in the cell wall. It has been suggested previously that free MA-containing glycolipids contribute to the selective permeability of the cell wall, but this has never been demonstrated (36). Also, phospholipid fatty acids were not accurately analyzed in detail here, but their composition might influence hydrophobic substrate uptake and/or active transport as well. These questions could be addressed in future investigations.