Bacterial Cell Wall-Induced Arthritis: Chemical Composition and Tissue Distribution of Four Lactobacillus Strains

Abstract

To study what determines the arthritogenicity of bacterial cell walls, cell wall-induced arthritis in the rat was applied, using four strains of Lactobacillus. Three of the strains used proved to induce chronic arthritis in the rat; all were Lactobacillus casei. The cell wall of Lactobacillus fermentum did not induce chronic arthritis. All arthritogenic bacterial cell walls had the same peptidoglycan structure, whereas that of L. fermentum was different. Likewise, all arthritogenic cell walls were resistant to lysozyme degradation, whereas the L. fermentum cell wall was lysozyme sensitive. Muramic acid was observed in the liver, spleen, and lymph nodes in considerably larger amounts after injection of an arthritogenic L. casei cell wall than following injection of a nonarthritogenic L. fermentum cell wall. The L. casei cell wall also persisted in the tissues longer than the L. fermentum cell wall. The present results, taken together with those published previously, underline the possibility that the chemical structure of peptidoglycan is important in determining the arthritogenicity of the bacterial cell wall.

A single intraperitoneal (i.p.) injection of bacterial cell walls isolated from gram-positive bacteria induces in the rat a chronic arthritis closely resembling human rheumatoid arthritis. Originally, a model was described in which polyarthritis was elicited by injection of Streptococcus pyogenes cell walls (5), but several other bacterial species are also arthritogenic in a similar fashion (35). They include lactobacilli (27, 30), eubacteria (26, 36–40), bifidobacteria (39), and streptococci (43).

The cell wall skeleton of all gram-positive bacteria is composed of a polymer, peptidoglycan (PG), consisting of a glycan backbone of N-acetylmuramic acid and N-acetylglucosamine, and cross-linking peptide chains containing d- and l-amino acids (34). Most variations of the PG peptide moiety do not occur in the peptide subunit but in the mode of cross-linkage and in the interpeptide bridge. Based on the anchoring point of the cross-linkage to the peptide subunit, the primary structure of PG is divided into group A (cross-linkage between positions 3 and 4) and group B (cross-linkage between positions 2 and 4), which are further classified into different subgroups and variations depending on the type or presence of the connecting interpeptide bridges and the amino acid in the third position of the PG peptide subunit (34). Apart from PG, other components of the bacterial cell wall include polysaccharide, teichoic acids, and the cell wall-associated proteins. These structures, in the cell wall complex or alone, are biologically active (24, 25). However, it has remained unclear which are the structural characteristics within this complex finally determining the arthritogenicity or nonarthritogenicity. For instance, when injected into the rat, cell wall preparations from closely related bacteria within a single genus may be either arthritogenic or nonarthritogenic; examples of such genera are Streptococcus (44), Lactobacillus (28), and Eubacterium (38, 40).

In the present work, we have attempted to elucidate some of the factors involved in the arthritogenicity of bacterial cell walls by using four different strains of Lactobacillus. To explore the question of what is decisive for arthritogenicity, we have determined the chemical composition and tissue distribution as well as the arthritogenic capacity in vivo and resistance to lysozyme in vitro of cell walls isolated from three strains of Lactobacillus casei and one strain of Lactobacillus fermentum.

MATERIALS AND METHODS

Bacterial strains.

The four bacterial strains used were purchased from the American Type Culture Collection, Manassas, Va. Two strains are L. casei B ATCC 11578 and L. casei B ATCC 25303, the cell walls of which have been reported to induce chronic arthritis in the rat (28, 30). In addition, L. casei C ATCC 25302 and L. fermentum ATCC 14931, which have not been found to be arthritogenic (28), were used. All three L. casei strains are 100% identical in a partial (273 to 304 bp) 16S ribosomal DNA (rDNA) analysis and by average are 89% (range, 86 to 91%) similar to each other, as well as to other strains of L. casei, according to the composition of cellular fatty acids; the analyses were carried out as described previously (8, 22). The major fatty acids present in these three L. casei strains are myristic acid (C_14:0), palmitoleic acid (C_16:1,cis-9), palmitic acid (C_16:0), linoleic acid (C_{18:2,cis-9,12}), oleic acid (C_18:1,cis-9), elaidic acid (C_18:1,trans-9), stearic acid (C_18:0), and nonadecanoic acid (C_19:0), the percentages of which calculated per total cellular fatty acid profile are 7.32, 5.65, 38.4, 7.76, 2.84, 14.40, 8.63, and 10.53%, respectively. In comparison with the three L. casei strains, L. fermentum shows a similarity of 93% in a 16S rDNA analysis and has, by average, a 72% (range, 70 to 77%) similar fatty acid profile. Also, it has a distinct fatty acid profile, with 1.69% of the total for myristic acid, 2.93% for palmitoleic acid, 33.4% for palmitic acid, 9.45% for linoleic acid, 3.94% for oleic acid, 9.64% for elaidic acid, 8.05% for stearic acid, and 20.96% for nonadecanoic acid. The strains of L. casei B used were originally isolated from the human oral cavity (45), L. casei C was isolated from the human intestinal flora (4), and L. fermentum was isolated from fermented beets (16); all are nonpathogenic bacteria.

Cell wall preparations.

Bacterial cell walls were isolated according to the method described by Lehman et al. (30), with some modifications. Harvested lactobacilli were heat treated (80°C; 30 min) to inactivate autolytic enzymes (43) and further disrupted with an MSK Cell Homogenizer (B. Braun Biotech International, Melsungen, Germany). The effectiveness of cell disruption was checked by Gram staining and light microscopy. Cell walls were collected by centrifugation at 38,700 × g, 4°C, for 20 min. To remove nucleic acids, crude cell walls were first treated with DNase and RNase A (250 μg/ml; both enzymes were purchased from Sigma, St. Louis, Mo.) and then with trypsin (250 μg/ml; Fluka Chemica-BioChemica), washing twice with phosphate-buffered saline (PBS) and once with distilled water between the cycles. To remove the cell wall-associated proteins, the preparations were treated with papain (20 μg/mg; Sigma), as described by Fox et al. (11). Purified cell walls were resuspended in PBS, and suspensions were analyzed for protein content (32) and sonicated for 120 min in an ice bath (Branson Sonifier Cell Disruptor B-15; SmithKline Co., Danbury, Conn.). To separate 10P (pellet) and 10S (supernatant) fractions (11), the cell wall fragments were centrifuged at 10,000 × g, 4°C, for 30 min. The 10S preparations were further heat treated (90°C; 30 min) and checked for sterility by bacterial culture on agar. The total carbohydrate amount in the 10S fractions was 1.12, 1.36, 10.35, and 11.20 mg per ml for L. casei B ATCC 25303, L. casei C ATCC 25302, L. casei B ATCC 11578, and L. fermentum ATCC 14931, respectively, when measured by the method described by Dubois et al. (7). These 10S fractions were used throughout the study.

Gas chromatography-mass spectrometry (GC-MS) of bacterial cell walls.

The carbohydrates and amino acids of 10S fractions were quantified with an HP 5890A gas chromatograph (Hewlett-Packard, Palo Alto, Calif.) equipped with a Noribond column (30 m by 0.25 mm [internal diameter]) (Nordion Instruments, Helsinki, Finland) coupled directly to a TRIO-1 mass spectrometer (VG Instruments, Manchester, United Kingdom). Sugars and aminosugars were analyzed as alditol acetates with mannose and N-methylglucamine as internal standards and amino acids as butyl heptafluorobutyl derivatives with norleucine, methionine, and tryptophan as internal standards, respectively (12). The interphase temperature for alditol acetates was 250°C. The column temperature started at 50°C and was programmed at a rate of 10°C/min to 270°C, where the temperature was held for 1 min. Finally, the column was heated for 5 min at 290°C. The >99% pure helium was used as the carrier gas, with a flow rate of 1 ml/min. The molecules were ionized by an electron impact method with 70 eV of energy and analyzed in the selected single-ion monitoring mode using positive ions at a mass-to-charge ratio (m/z). For butyl heptafluorobutyl derivatives, the interphase temperature was 250°C. The column temperature started at 85°C and was programmed at the rate of 10°C/min to 280°C, where the temperature was held for 1 min. Finally, the column was heated at 290°C for 5 min.

Degradation of cell walls by lysozyme.

Ninety-six milligrams of 10S cell wall fraction from each Lactobacillus strain was suspended in 0.1 M Na-acetate buffer (pH 5.0), yielding a concentration of 4 mg/ml, and incubated with 400 μg of lysozyme (Sigma) per ml for 24 h at 37°C, mixing constantly (Thermolyne Speci-Mix, Barnstead/Thermolyne, Dubuque, Iowa) (42). The 10S suspensions without lysozyme were used as controls. To calculate cell wall resistance to lysozyme, the decrease in the absorbance at 560 nm was measured and expressed as a percentage. The assays were done in triplicate.

Animals.

Pathogen-free inbred female LEW/SsNHsd rats (from the 202B colony at Harlan Sprague Dawley, Inc., Indianapolis, Ind.) weighing 110 to 150 g were used. The animals were kept in Macrolon III cages with disposable filter tops (Scanbur, Køge, Denmark); all handling was performed in a laminar-flow hood. The rats were given an autoclaved standard diet and water ad libitum. Prior to the experiments, the animals were allowed to adapt to the laboratory environment for 1 week. The animal experiments were performed in compliance with national and international laws and policies and were approved by the Institutional Committee for Animal Research.

Arthritis induction and clinical evaluation.

On day zero, four groups of rats were injected i.p. with sterile PBS suspensions of cell wall from L. casei B ATCC 25303, L. casei C ATCC 25302, L. casei B ATCC 11578, or L. fermentum ATCC 14931. The injected cell wall dose was based on the carbohydrate content of 5 mg/3 ml per rat. At least five animals per group were used. Control rats (n = 8) were injected i.p. with 3 ml of PBS alone. To monitor the development of arthritis, the front and hind paws were scored two to five times per week. The arthritic symptoms were graded from 0 to 4, based on the degree of erythema, edema, and functional disorder of the ankle and metatarsal joints (wrist and metacarpal joints), by two independent observers as described previously (14). Such an evaluation has been widely used, and the results are parallel to those by histological grading (14, 41, 48–50). Rats were sacrificed at different time points by cardiac puncture bleeding under Metofane (Pitman-Moore, Inc., Washington Crossing, N.J.) anesthesia.

Detection of muramic acid in tissues.

Liver, spleen, and mesenteric lymph nodes were collected from rats injected with L. casei B ATCC 11578 or L. fermentum ATCC 14931. For this purpose, three animals per group were killed 14, 28, or 63 days after cell wall injection. Organs from two control rats were also collected. The suspensions from homogenized organs were passed through steel meshes, and mononuclear cells were isolated with Lympholyte-Rat (Cedarlane Laboratories, Ltd., Ontario, Canada) gradient centrifugation according to the manufacturer's instructions. Viable mononuclear cells were counted and stored at −70°C until used. Muramic acid content was analyzed as an alditol acetate derivative by GC-MS as previously described (31). Briefly, the molecules were ionized by the electron impact method and analyzed in the single-ion monitoring mode using positive ions at m/z of 403 and 445 for the detection of muramic acid and an m/z of 327 for the detection of N-methyl-d-glucamine as the internal standard. The sensitivity of the method is ca. 10 ng of the derivatized muramic acid.

Statistics.

The differences between study groups were compared by the Mann-Whitney U test for unpaired data and the Wilcoxon matched-pairs test for paired data. A P value of <0.05 was considered significant.

RESULTS

Clinical observations.

Studies on arthritogenicity in vivo were carried out in two experiments, using L. casei B ATCC 25303 and L. casei C for the first experiment and L. casei B ATCC 11578 with L. fermentum for the second one. Injection of cell wall preparations from all L. casei strains induced chronic arthritis, whereas a cell wall preparation from L. fermentum caused only a mild acute arthritis, completely subsiding in 14 days (Table 1; Fig. 1).

TABLE 1.

Arthritogenicity and degradation by lysozyme of cell wall preparations used

Source of bacterial cell wall	Degradation by lysozymea	Incidence of chronic arthritisb
L. casei B ATCC 25303	0	4/5
L. casei C ATCC 25302	0.2	4/5
L. casei B ATCC 11578	0	18/18
L. fermentum ATCC 14931	44.0	0/18

^aExpressed as the percent decrease of OD at 560 nm. 

^bExpressed as number of arthritic rats/total. 

FIG. 1.

Development of arthritis in female LEW/SsNHsd rats injected i.p. with cell walls from L. casei B ATCC 25303 (■) or L. casei C ATCC 25302 (□) (A) and L. casei B ATCC 11578 (●) or L. fermentum ATCC 14931 (○) (B). The cell wall dose per rat was determined to correspond to 5 mg of total carbohydrate/3 ml. The arthritis score is calculated as the mean value ± standard error of the mean until day 84 for four rats, thereafter for three rats (A) or for the number of rats indicated (B).

Chemical composition of cell walls.

For the purpose of finding clues for the arthritogenicity or nonarthritogenicity, the content of carbohydrates and amino acids was determined in the four cell wall preparations used. The results are presented in Table 2. Four details are worth mentioning separately. (i) The presence or absence of rhamnose is not decisive for arthritogenicity or nonarthritogenicity of Lactobacillus cell walls, which is opposite to what has been suggested previously (28, 43); also, in Eubacterium species, rhamnose is not a major factor contributing to the cell wall arthritogenicity (37, 38). (ii) The cell wall of the nonarthritogenic L. fermentum contains a substantial amount of galactose, whereas the three arthritogenic strains do not contain this sugar. Regarding N-acetylgalactosamine, the situation is reversed. (iii) Results from the amino acid analysis suggest that all three arthritogenic cell walls have alanine, glutamine, lysine, and asparagine as PG amino acids; this is consistent with the previously reported occurrence of the variation α in the PG subgroup A4 in these L. casei strains (15, 21, 34). In contrast, the L. fermentum strain using the variation β in the PG subgroup A4 has already been reported (34), and our finding of alanine, glutamine, asparagine, and ornithine as PG amino acids is in agreement with this. (iv) From the non-PG amino acids, valine was observed to be present in the three arthritogenic cell walls and absent in the nonarthritogenic L. fermentum cell wall.

TABLE 2.

Chemical composition of cell wall preparations used^a

Component	% Composition of arthritogenic strains			% Composition of nonarthritogenic L. fermentum ATCC 14931
	L. casei B ATCC 25303	L. casei C ATCC 25302	L. casei B ATCC 11578
Carbohydrates
Rhamnose	5.2	<0.1	15.2	<0.1
Glucose	5.0	12.0	9.9	26.9
Galactose	ND	ND	ND	9.7
Arabinose	ND	<0.1	ND	ND
Fucose	ND	<0.1	ND	<0.1
N-Acetylmuramic acid	1.3	4.0	5.3	6.0
N-Acetylglucosamine	4.3	5.4	8.2	8.1
N-Acetylgalactosamine	3.0	5.5	6.2	ND
Amino acids
Alanine	4.3	2.9	1.7	4.5
Glutamic acid/glutamine	6.0	12.2	8.6	10.4
Lysine	3.7	7.0	4.8	1.9
Aspartic acid/asparagine	8.6	15.9	11.1	16.1
Ornithine	0.4	0.3	0.7	3.9
Valine	15.5	3.0	4.4	ND
Glycine	3.1	2.0	2.4	2.1
Serine	2.9	1.5	1.2	2.4
Leucine	8.2	2.9	2.6	9.2
Isoleucine	12.8	4.0	6.4	9.5
Threonine	5.0	1.9	3.8	3.8
Proline	1.6	1.1	1.6	0.7
Tyrosine	2.9	3.3	1.9	3.0
α-Aminoisobutyric acid	2.1	1.8	ND	ND
Phenylalanine	1.2	0.9	1.4	0.9
Totalb	96.7	87.3	97.4	118.8

^aValues are expressed as a percentage of cell wall (dry weight). Values of components of PG are in bold. ND, not detected. 

^bDiaminopimelic acid, cystine, arginine, histidine, and cysteine were not detected. 

Effect of lysozyme on Lactobacillus cell wall.

We also tested if the arthritogenicity of the cell wall correlates with resistance to lysozyme. A 24-h incubation with the enzyme did not decrease the optical density (OD) of 10S cell wall preparations inducing chronic arthritis (Table 1). The cell wall of the nonarthritogenic L. fermentum was lysozyme sensitive, with a 44% decrease in OD.

Tissue distribution of cell wall.

To study tissue distribution of arthritogenic and nonarthritogenic cell walls, mononuclear cells from liver, spleen, and mesenteric lymph nodes were analyzed for the muramic acid content. This was performed on days 14, 28, and 63 with organs from the rats injected i.p. with a cell wall preparation of the arthritogenic L. casei B ATCC 11578 or of the nonarthritogenic L. fermentum. The results obtained indicate that the cell wall of the arthritogenic strain is deposited in all three organs studied considerably more effectively than that of the nonarthritogenic L. fermentum (Fig. 2). For instance, on day 14, the differences were at least sixfold, and the same trend was also seen on days 28 and 63. Particularly striking is the high content of N-acetylmuramic acid in the liver after injection of the arthritogenic cell wall, even on day 63.

FIG. 2.

Muramic acid content in mononuclear cells from rat liver, spleen, and mesenteric lymph nodes, analyzed by GC-MS. On day zero, rats were injected i.p. with cell wall from L. casei B ATCC 11578 or L. fermentum ATCC 14931. Each column represents a mean value ± standard error of the mean for three rats. Asterisks indicate significant differences (P < 0.05) when the rats injected with cell wall of L. casei B ATCC 11578 or L. fermentum ATCC 14931 are compared, done separately for each day of analysis.

DISCUSSION

The purpose of this work was to provide insights to the question of what determines arthritogenicity of the bacterial cell wall. For this, we used three strains of L. casei, which are closely related as confirmed by 16S rDNA and cellular fatty acid profile analyses, and one strain of L. fermentum. Here, we report that all the three L. casei strains tested were arthritogenic, whereas L. fermentum was not. Furthermore, we found that cell walls inducing chronic arthritis are resistant to lysozyme degradation in vitro, which is consistent with the previous observations (28). Moreover, we show that arthritogenic Lactobacillus is characterized by a prolonged persistence in the rat tissues. Additionally, all arthritogenic lactobacilli share the same PG type, different from that in the nonarthritogenic one.

In this study, the cell wall of L. casei C ATCC 25302, which has previously been reported to induce only a mild arthritis (28), clearly proved to induce chronic arthritis. This seeming discrepancy can, however, be understood on the basis of differences in the preparation of the cell wall. The heat treatment of bacteria and a differential centrifugation in our experiments yielded cell wall fragments of a different size than those used by Lehman et al. (28). The same was experienced by Fox et al. (11) when studying streptococcal cell wall arthritis and by Tuomanen et al. (47) in studies of meningeal inflammation induced by pneumococcal cell wall products. For instance, Fox et al. observed the most severe joint inflammation by using PG fragments of intermediate size (50 × 10⁶ Da), as compared to large (500 × 10⁶ Da) and small (5.3 × 10⁶ Da) fragments.

All three arthritogenic L. casei strains used by us have the same variation of PG structure, known as A4α (15, 21, 34), whereas the nonarthritogenic L. fermentum has another variation (A4β) (34). The arthritogenic variation α in the subgroup A4 has lysine as the third amino acid of the PG stem peptide in contrast to ornithine in the nonarthritogenic variation β. This finding is of interest in the light of a study by Zhang et al. of Eubacterium cell wall arthritis (48); an arthritogenic strain of Eubacterium aerofaciens had a PG structure of variation A4α with lysine in the third position of the stem peptide, whereas a nonarthritogenic strain of E. aerofaciens, 100% identical by 16S rDNA analysis with the former one, did not possess this structure.

Different variants of PG are degraded differentially; degradation of PG is sterically hindered by covalently linked polysaccharides. Evidence for this exists from experiments both in vivo and in vitro. Mutanolysine, an enzyme with substrate specificity analogous to that of lysozyme (23), as well as muramidase (2), N-acetylmuramyl-l-alanine amidase (17–19), and N- and O-acetylation (3, 42) have been applied for this purpose. Lysozyme and N-acetylmuramyl-l-alanine amidase degrade bacterial cell walls in vivo to residues with varying inflammatory capacity (20). Lysozyme, which degrades the sugar backbone of PG by hydrolyzing the bond between N-acetylglucosamine and N-acetylmuramic acid, is thought to be the most important enzyme for the inactivation of PG. Considerably less is known about the lactamide bond between N-acetylmuramic acid and the peptide side chain attacking N-acetylmuramyl-l-alanine amidase, for which the substrate specificity in vitro seems to be determined by the first three amino acids of the PG stem peptide (20). It is also known that variation of the third amino acid changes the biochemical activity of PG fragments; those with lysine in position three have been demonstrated to be highly inflammatory in the rabbit subarachnoid space (2, 46, 47). Quite recently, branched stem peptides isolated from Streptococcus pneumoniae PG were observed to carry high tumor necrosis factor-stimulating activity; the third amino acid of the stem peptide was lysine (33). These results support our view that a specific structural part of PG is responsible for the arthritogenicity. Therefore, the exact structure of PG arthritogenic component remains to be further elucidated.

Susceptibility to biodegradation may also explain the different persistence in tissues of different cell walls, as described here for L. casei and L. fermentum cell walls (Fig. 2). Several other studies have demonstrated that the occurrence of arthritis correlates with the amount of cell walls deposited in the tissues and persisting in the joints (1, 6, 9, 10, 13, 29, 41, 44). On the other hand, bacterial strains capable of inducing chronic arthritis in the rat are resistant to lysozyme degradation (30, 38, 39, 42). However, it is apparent that resistance to lysozyme degradation alone is not critical for the arthritis-inducing capacity; this is shown by a study on Eubacterium limosum cell wall, which is nonarthritogenic and which has been reported to be lysozyme resistant (38).

Altogether, the present results together with those presented earlier underline the possibility that the chemical structure of the bacterial cell wall is important in determining arthritogenicity or nonarthritogenicity. To confirm this hypothesis, further arthritogenic and nonarthritogenic bacterial strains should be studied, particularly regarding their PG structure, susceptibility to biodegradation, and tissue persistence.