Evidence that the Enterococcal Polysaccharide Antigen Gene (epa) Cluster Is Widespread in Enterococcus faecalis and Influences Resistance to Phagocytic Killing of E. faecalis

Abstract

In previous studies, we cloned a cluster of genes involved in polysaccharide biosynthesis (epa) from Enterococcus faecalis strain OG1RF and showed that this gene cluster mediated synthesis of a polysaccharide in Escherichia coli. Disruption of two open reading frames in the epa gene cluster of OG1RF generated two mutants, TX5179 and TX5180, which were attenuated in a mouse peritonitis model. In the current study, Western blotting was performed with serum from a patient with E. faecalis endocarditis and polysaccharide extracts from OG1RF and the mutants TX5179 and TX5180. OG1RF showed a smear in the high-molecular-weight region and discrete bands in the low-molecular-weight region, which were missing from the mutants; periodate treatment and carbohydrate staining confirmed the polysaccharide nature of this material. In a neutrophil killing assay using OG1RF-absorbed normal human serum, the mutants TX5179 and TX5180, respectively, were 50 and 2.4 times more susceptible to killing than wild-type OG1RF (P ≤ 0.01). With a fluorescence phagocytosis assay, 2.5 to 3 times more of the mutants were taken up by neutrophils than OG1RF (P ≤ 0.001). Finally, with restriction digestion and hybridization under high-stringency conditions, the epa gene cluster of OG1RF (which is also present in the sequenced E. faecalis strain V583) was detected in 12 of 12 other clonally distinct E. faecalis strains tested: a similar polysaccharide pattern was detected for the 12 strains on Western blots using an E. faecalis endocarditis patient serum, and sera from four other patients with E. faecalis endocarditis all reacted with polysaccharide extracts of OG1RF. These results indicate that the epa gene cluster is widespread among E. faecalis and confers some protection against human host defenses.

Enterococci are a leading cause of nosocomial infections in the United States and account for 5 to 15% of cases of infective endocarditis, with most isolates being Enterococcus faecalis (9). Better understanding of the pathogenicity of enterococci may help to develop more effective therapies or preventative modalities for E. faecalis infections.

Polysaccharides on bacterial surfaces may interact with the human host and play important roles in bacterial pathogenesis. There is evidence indicating that polysaccharides are involved in hindering leukocyte killing of enterococci. Arduino and colleagues (1, 12) reported that exposure of Enterococcus faecium TX0016 (also called TEX16 as well as DO), a strain resistant to phagocytosis and killing by leukocytes in the presence of normal human sera (NHS); for partial sequence, see http://www.hgsc.bcm.tmc.edu/microbial/efaecium/), to sodium periodate, but not to trypsin, pronase, or phospholipase C, eliminated its resistance to phagocytosis. Recently, members of our group showed that rabbit antiserum against formalin-killed E. faecium TX0016 promoted opsonization and killing, and this effect was dramatically reduced by adsorption of the antiserum with carbohydrate purified from TX0016, but not by incubation with surface protein extracts from TX0016 (12). In addition, Huebner and colleagues have shown that antibody to a polysaccharide component purified from an E. faecalis strain enhanced phagocytosis and killing of 6 of 16 E. faecalis strains and 2 of 7 vancomycin-resistant E. faecium strains (5, 6), suggesting some intra- and interspecies sharing of a common, or related, polysaccharide.

Members of our group previously identified a gene cluster (epa) of E. faecalis encoding homologues of many genes involved in polysaccharide biosynthesis (21-23) and showed that the gene cluster of E. faecalis strain OG1RF (10) mediated production of a polysaccharide in Escherichia coli. While we were not able to detect production of the polysaccharide-specific antigen by OG1RF or several other E. faecalis strains by using human-derived antibody eluted from the E. coli clone expressing the E. faecalis epa genes, we showed that epa genes were transcribed in OG1RF with at least three transcriptional start sites. Moreover, two mucoid E. faecalis clinical isolates (3) showed positive reactions with antibodies eluted from the polysaccharide extracts of the E. coli clone. In addition, two OG1RF mutants, TX5179 and TX5180, with disruptions in two of the genes (orfde4 and orfde6) in the epa cluster, showed a significant delay in killing in a mouse peritonitis model and a slightly higher 50% lethal dose than wild-type OG1RF (23).

In this study, using polysaccharide preparations from E. faecalis and serum from a patient with E. faecalis endocarditis, we were able to demonstrate the production of immunoreactive polysaccharide by OG1RF and the absence of this polysaccharide in mutants TX5179 and TX5180. The epa gene cluster was demonstrated in 12 of 12 E. faecalis strains tested and appears to be involved in enabling E. faecalis to resist neutrophil-mediated phagocytosis and killing in the presence of complement.

MATERIALS AND METHODS

Bacterial strains, culture conditions, antisera, and NHS.

E. faecalis OG1RF has been previously described (10); other strains, previously shown to be distinct by pulsed-field gel electrophoresis (PFGE) and multilocus enzyme electrophoresis (MLEE), and the E. coli strain used in this paper are listed in Table 1. Cultures of E. coli cells were grown in Luria-Bertani broth or agar with the appropriate antibiotics at 37°C. Cultures of E. faecalis cells were grown in brain heart infusion (BHI) broth or agar (Difco Laboratories, Detroit, Mich.) at 37°C. Sera S0014, S0001, S0005, S0013, and S0032 were collected from patients with E. faecalis endocarditis. NHS were obtained as described previously (1), and the sera from eight normal healthy adult volunteers were pooled.

TABLE 1.

Strains used in this study^a

Strain (alternative designation)	Origin	PFGE pattern (7, 17, 18)	MLEE electrophoretic type (17)
E. faecalis
TX0043 (MC 25561)b	Mayo Clinic, Rochester, Minn.	MC-5	18
TX0045 (END6)b	Boston, Mass.	E-1	9
TX0048 (END27)b	Boston, Mass.	E3	18
TX0630 (HG6280)	Buenos Aires, Argentina	19	16
TX0635 (WH245)	West Haven, Conn.	8	5
TX0645 (Beirut)	Beirut, Lebanon	7	7
TX0668 (PA)	Philadelphia, Pa.	5	17
TX0771 (CE36)	Santiago, Chile	C-2	15
TX0860 (BE88)	Bangkok, Thailand	B-3	18
TX2486	Houston, Tex.	V-1b	—e
TX2783c,d	Logrono, Spain	—	—
TX4000 (JH2-2)	Laboratory strain	VII	17
TX4002 (OG1RF)	Laboratory strain	VIII	25
TX2708 (V583)d	St. Louis, Mo. (15)	—	—
TX5179	epa mutant (orfde4 disrupted) of OG1RF (23)
TX5180	epa mutant (orfde6 disrupted) of OG1RF (23)
E. coli
TX5159	DH5α containing the epa cluster of OG1RF (22, 23)

^aReference numbers are given in parentheses.

^bIsolated from patient with endocarditis.

^cIsolated from chicken product.

^dVancomycin resistant.

^e—, PFGE pattern or MLEE electrophoretic type has not been determined.

Polysaccharide extraction, carbohydrate labeling, Western blotting, and periodate treatment.

For polysaccharide extraction, 10 ml each of overnight or log-phase cultures of E. faecalis was collected, washed, and resuspended in phosphate-buffered saline (PBS) (0.14 M NaCl, 1.5 mM KH₂PO₄, 15 mM Na₂HPO₄ · 7H₂O, 2.7 mM KCl [pH 7.4]). Fifty microliters of lysozyme (40 mg/ml in PBS) was added, and the suspension was vortexed. Fifty microliters of proteinase K (20 mg/ml in dH₂O) was added, and the cell suspension was incubated at 50°C overnight (cells lysed after incubation). After phenol-chloroform extraction, a 15-μl sample was loaded onto sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels. Following electrophoresis, the material in the gel was transferred to Immobilon-P transfer membranes (Millipore Corporation, Bedford, Mass.) and processed either for carbohydrate labeling, as described in the manufacturer's protocol (ECL enhanced chemiluminescence glycoprotein detection system, Amersham Pharmacia Biotech UK Ltd., Little Chalfont, Buckinghamshire, United Kingdom), or for Western blotting by the method described previously (22). The stability of the antigens in the presence of periodate was determined by treatment of 10 μl of the extract with 5 μl of 0.6 M sodium periodate (pH 7.2) overnight at room temperature. Excess periodate was eliminated by the addition of 5 μl of 0.4 M sodium metabisulfite.

DNA amplification and Southern blotting.

Internal fragments of orfde4 to orfde10 on pTEX5159 (harboring E. faecalis epa gene cluster) (22) were amplified by the standard method (23) with the oligonucleotides listed in Table 2. Genomic DNA was extracted from E. faecalis as described previously (23) and digested with HindIII, and Southern blotting was performed by standard methods (23) with the mixed PCR fragments described above as a probe.

TABLE 2.

Oligonucleotides used in this study

Oligonucleotide	Sequence (5′→3′)	Position
GW301	ATGGATCCTATGAGCATGCAAGAAAT	1, orfde4, upper strand
GW302	CATTGAATAACACTTGATTATGACC	220, orfde4, lower strand
GW337	GAGTCAAAGATTAGCGGTAGTC	24, orfde5, upper strand
GW338	TCAAGGAGCAACAATAATTCAC	269, orfde5, lower strand
GW339	ACAAGGTCCAGGGCAAGG	51, orfde5 to -6, upper strand
GW371	TAAGAGGGATTGGTAACTTG	244, orfde5 to -6, lower strand
GW303	ATGGATCCTATGAAAGGAATTATTTT	1, orfde6, upper strand
GW304	CATCTGGGCTTTCTTGTACCGC	232, orfde6, lower strand
GW365	GAAGGCTACATTTTATCAGAAC	295, orfde7, upper strand
GW366	CGCTTCAAATTCTTTTAAGG	524, orfde7, lower strand
GW316	CGACAACTCATTAAACGACC	258, orfde8, upper strand
GW317	GAACCCACGCTTTGACTAAC	504, orfde8, lower strand
GW318	CAAAAGCTACTTTGCCCTGCC	15, orfde9, upper strand
GW319	ACGTAAACAAGTGTCGCCCC	337, orfde9, lower strand
GW320	GAGCGGGAAATCGGAAAGTG	494, orfde10, upper strand
GW321	CAATCAGAATTGCTGAGCCGAC	682, orfde10, lower strand

^aThe oligonucleotides are described in reference 20.

PMN killing assay.

An opsonophagocytic killing assay was conducted to quantify the killing activity of polymorphonuclear leukocytes (PMNs) against OG1RF and the mutants. The assay was performed as previously described (13) with some modifications. PMNs were isolated from blood of healthy adult volunteers by the method described previously (14) and suspended in Hanks balanced salt solution without Ca²⁺ and Mg²⁺ (HBBS). The concentration of PMNs was determined by trypan blue staining and counting in a hemocytometer, and the final concentration was adjusted to 2 × 10⁷/ml. The PMN suspension was kept on ice and used within 1 h. NHS was diluted in HBSS (1:4), absorbed with 10⁹CFU of OG1RF/ml for 2 h at 4°C, aliquoted, and kept at −80°C. The PMN killing assay was performed in a 96-well plate with 100 μl of PMN suspension, 20 μl of bacteria (concentration adjusted spectrophotometrically to 10⁸/ml and confirmed by viable counts), 40 μl of absorbed NHS (final dilution, 1:20), 2 μl of 100 × Ca²⁺ and Mg²⁺ (to activate PMNs), and 38 μl of HBSS. The 96-well plates were incubated on a shaker at 37°C. Samples were taken at 0, 30, and 60 min; diluted 1:10 in H₂O; incubated at room temperature for 10 min (to lyse PMNs); and then diluted in 0.5% NaOAc and plated onto BHI agar plates. Each assay was performed in duplicate, and the experiment was performed twice. The percentage of survival at each time point was compared.

Phagocytosis assay.

The assay was performed as described previously (1) with minor modifications: the absorbed NHS (final concentration, 1:20) was used in opsonization, and 20 consecutive individual PMNs per sample were examined to determine the number of ingested organisms per cell. The experiment was performed twice, and the significance of the difference between wild-type OG1RF and the mutants and between the two mutants was determined by a t test (for two-group comparisons).

RESULTS

Detection of polysaccharide antigens in OG1RF and loss of the antigens in epa mutants.

In our previous study (22), an E. coli clone (TX5159) containing the enterococcal gene cluster epa showed a ladder-like pattern on a Western blot with sera from patients with enterococcal infection as the primary antibodies. Specific antibodies against the polysaccharide (eluted from the ladder-like polysaccharide of TX5159) did not react with whole-cell preparations from E. faecalis OG1RF or from several other E. faecalis clinical isolates, suggesting that E. faecalis might not express this form of polysaccharide in vitro or that the method was not sufficiently sensitive. In the present study, we used an approach to detect polysaccharide antigens by treatment of E. faecalis OG1RF cell lysates with proteinase K and then assayed the crude polysaccharide extracts on Western blots with a serum sample from a patient with enterococcal infection (Fig. 1). A smear in the high-molecular-weight region and a discrete banding pattern in the lower region were detected on the Western blot; after periodate treatment, the smear and the discrete bands disappeared (data not shown). This material was positively labeled by the carbohydrate labeling reagent, further indicating these immunoreactive moieties were polysaccharides. The polysaccharide extracts of the two epa mutants were examined on the Western blot along with OG1RF (Fig. 1). Another OG1RF mutant, TX10293 (16), was also used as a control; this mutant has a two-component system disrupted by the same vector, pTEX4577 (11), with which the epa mutants were also constructed, and showed no change in the Western blot. However, the major antigen or antigens present in polysaccharide extracts of OG1RF and TX10293 were not seen in preparations from the two epa mutants, suggesting that the epa genes (including orfde4, orfde6, and/or their downstream genes) are involved in synthesis of the immunoreactive polysaccharides. In addition, the major carbohydrate-staining region of OG1RF was not present in extracts of two epa mutants, confirming that the reactive material present in wild-type OG1RF extracts and absent in the mutants is carbohydrate in nature. Four other sera from patients with E. faecalis endocarditis also showed a positive reaction with OG1RF extracts. NHS, on the other hand, showed a negative reaction with OG1RF extracts, indicating that antibodies against the polysaccharides were not present, at least not in sufficient quantities to detect the polysaccharide in this assay. Both log- and stationary-phase cultures of OG1RF, TX5179, and TX5180 were assayed, and similar results were obtained, suggesting that the polysaccharides are expressed in both growth phases.

FIG. 1.

Western blot of polysaccharides in E. faecalis. Serum 00014 from a patient with endocarditis was used. Lanes: 1, TX5180; 2, TX5179; 3, TX10293; 4, OG1RF.

PMN killing assays.

Based on the report by Gaglani and colleagues (4), who showed that absorption to remove specific antibodies from NHS significantly reduced neutrophil killing of E. faecalis, we compared unabsorbed and absorbed NHS for their opsonic activity in PMN killing assay. Different serum dilutions were tested, including 10, 5, 2.5, 1, and 0.5%. It was found that with a 5% dilution, the absorbed and unabsorbed sera showed a significant difference in killing; higher serum dilutions eliminated opsonic activity of the NHS (both unabsorbed and absorbed; data not shown). For this reason, 5% absorbed serum was used in subsequent experiments.

In order to determine if the epa gene cluster was important for protecting OG1RF from PMN killing in the presence of absorbed NHS, OG1RF, TX5179, and TX5180 were compared in the PMN killing assay (Fig. 2). At 30 min, OG1RF and TX5180 had about 90% survival, while approximately 3% of TX5179 cells survived. At 60 min, the survival rates for OG1RF, TX5180, and TX5179 were approximately 45, 19, and 0.9% (P < 0.01 for OG1RF versus TX5180, OG1RF versus TX5179, and TX5179 versus TX5180) (Fig. 2). Heat-inactivated absorbed serum was inactive in this assay, consistent with previous results that complement is necessary for the process (2, 5). This experiment was performed twice with reproducible results.

FIG. 2.

PMN-mediated killing of E. faecalis. Percentage of E. faecalis cells surviving in the presence of PMNs and absorbed NHS as the complement source. P < 0.01 for TX5179 versus OG1RF, TX5180 versus OG1RF, or TX5179 versus OG1RF at 60 min. Error bars represent standard deviations.

PMN phagocytosis assay.

In the phagocytosis assay, ingestion of OG1RF, TX5179, and TX5180 by PMNs was quantified by counting bacteria inside PMN after a 30-min incubation, and the results are shown in Fig. 3. The average counts of bacteria inside PMNs for OG1RF, TX5179, and TX5180 were approximately 2.4, 7.5, and 6.0 (P < 0.001 for TX5179 versus OG1RF and TX5180 versus OG1RF), suggesting that the protective effect involving the epa gene cluster is at least in part due to resistance to phagocytosis. There was no significant difference between the two mutants in the phagocytosis assay.

FIG. 3.

PMN-mediated phagocytosis of E. faecalis. The experiment was performed twice, and the results were consistent with each other. Results from one of the experiments are shown. Twenty PMNs were observed for each strain, and the average counts of the bacteria inside each PMN were plotted. P < 0.001 for TX5179 versus OG1RF and for TX5180 versus OG1RF. Error bars represent standard deviations.

Distribution of the epa gene cluster in E. faecalis.

In order to determine if the epa gene cluster is commonly distributed in E. faecalis, 12 distinct E. faecalis strains (Table 1) were evaluated by high-stringency hybridization with mixed PCR products of orfde4 to orfde10 of the epa cluster of OG1RF as a probe. Eleven of the E. faecalis strains showed three bands of 6.8, 3.5, and 1.6 kb, as seen for OG1RF (22) (Fig. 4). One of the strains, TX0048, showed two smaller bands of similar sizes and a slightly larger third band (Fig. 4). Polysaccharides were extracted from these strains and assayed on the Western blots by using E. faecalis patient serum S0014. All of the strains showed a smear in the high-molecular-weight region and discrete bands in the low-molecular-weight region, a polysaccharide pattern similar to that of OG1RF (data not shown).

FIG. 4.

Southern blot of E. faecalis with the PCR products of orfde4 to orfde10 of the epa gene cluster of OG1RF. Lanes: 1, TX0043; 2, TX0045; 3, TX0048; 4, TX0630; 5, TX0635; 6, TX0645; 7, TX0668; 8, TX0771; 9, TX0860; 10, TX2486; 11, TX2783; 12, TX4000.

DISCUSSION

In our previous studies (22, 23), sequence analysis of the epa gene cluster revealed similarity to genes for rhamnose biosynthesis, glycosyl transferases, and ATP-binding cassette transporters, which are involved in polysaccharide synthesis in other bacteria. The E. coli clone (TX5159) containing the gene cluster was shown to produce a polysaccharide and to react with five sera from patients with E. faecalis endocarditis; however, the specific antibody eluted from that polysaccharide did not react with any of several strains of E. faecalis, including OG1RF. We felt that it was possible that the specific antigen was regulated and not produced in an amount detectable in vitro or that the polysaccharide antigen produced with the epa genes in E. coli was different in some way from that produced in E. faecalis. Disruption in two of the epa genes (orfde4 and orfde6) in OG1RF generated two mutants, TX5179 and TX5180, both of which were attenuated in a mouse peritonitis model. Based on identification of potential promoters, transposon insertion, and the complementation assay, orfde4 appears to be cotranscribed with orfde5, but not with orfde6 and its downstream genes, and based on promoter and reverse transcription-PCR analysis, orfde6 to orfde10 appear to form another transcriptional unit.

In the present study, total polysaccharide content was extracted from E. faecalis and was tested with serum from a patient with E. faecalis endocarditis. Wild-type OG1RF showed a smear in the high-molecular-weight region and discrete bands in the low-molecular-weight region on Western blotting with patient serum, which were missing in the TX5179 and TX5180 mutants, suggesting that both gene operons (orfde4 and orfde5 and orfde6 to orfde10) are important for synthesis of the immunoreactive polysaccharides. The material present in polysaccharide extracts of OG1RF and absent in extracts of the epa mutants was confirmed to be polysaccharide by periodate treatment and by carbohydrate labeling. On Western blotting, the reaction of the polysaccharide extracts of OG1RF with NHS was undetectable, while five sera from patients with E. faecalis endocarditis showed a positive reaction with polysaccharide extracts of OG1RF, suggesting that the polysaccharides related to epa genes frequently elicit an antibody response during infections.

By high-stringency hybridization, genes highly related to orfde4 to orfde10 of OG1RF were shown to be present in 12 of 12 other E. faecalis strains tested. These strains are geographically diverse, have different PFGE and MLEE patterns (7, 17, 18), indicating they are all distinct strains. Previous sequence comparison of OG1RF and database V583 epa gene clusters indicated that the epa clusters (including orfde4 to orfde10) of both strains have similar gene composition and organization (20), adding another strain with a similar epa gene cluster to the list. These distinct strains of E. faecalis also showed similar polysaccharide patterns on Western blots with the sera from E. faecalis endocarditis patients, suggesting that the polysaccharide is common in E. faecalis and that the antibody cross-reacting with the polysaccharide is present in the patient serum. In other studies of enterococcal polysaccharides, Huebner and colleagues (19) isolated a capsular polysaccharide antigen from an E. faecalis strain, and the antiserum raised against this polysaccharide was opsonic for 6 of 16 E. faecalis strains in PMN killing assays. The lack of opsonic capability for 10 of these 16 E. faecalis strains suggests that these 10 strains did not produce the polysaccharide against which antibodies were raised, further suggesting that their polysaccharide may be different from the one we detected, which appeared to be present in 14 of 14 E. faecalis strains assessed directly (hybridization, sequence analysis, or Western blots) and in 5 assessed indirectly (antibody present in serum from infected patients).

It has been shown previously that both antibodies and complement are involved in phagocytic killing of enterococci (2, 4, 5, 12). In this study, we confirmed that absorption to remove specific antibody and/or heat inactivation significantly reduced the bactericidal activity of PMNs in the presence of NHS. After the sera were absorbed with E. faecalis OG1RF, this strain was more resistant to PMN killing in the presence of this absorbed serum, similar to the results of Gaglani (4). Absorbed NHS was then used to examine OG1RF and the two epa mutants, TX5179 and TX5180, in PMN killing assays. With the absorbed NHS as a complement source, the mutants were more susceptible than wild-type OG1RF to PMN killing, suggesting that the epa genes are involved in protecting against PMN killing, possibly by resistance to phagocytosis and/or enhancement of survival in phagosomes. We also examined phagocytosis of OG1RF and the mutants by using fluorescence labeling; OG1RF was found to be more resistant to phagocytosis, suggesting that the epa genes are involved in this process. While the difference in phagocytosis between the wild-type OG1RF and the mutants is not as large as the difference in killing, suggesting that the epa genes may be related to survival in the phagosome, the assays are not directly comparable, and further studies will be needed to evaluate this possibility. Susceptibility to neutrophil-mediated killing may be at least a partial explanation for the observation that the two epa mutants showed less virulence in the mouse peritonitis model. How the polysaccharides protect E. faecalis from PMN killing is unclear, although prevention of C3 deposition by capsular polysaccharide has been shown in type III group B streptococci (8).

Although the two epa mutants TX5179 and TX5180 showed some similarity in phenotypes (as described above), they also appear different at some level. In this study, TX5179 was more susceptible to PMN killing than TX5180, although no significant difference was detected in the phagocytosis assay, suggesting TX5180 may survive better in the phagosome. This raises the possibility that orfde4 and/or its downstream genes in the operon may function in more than one pathway or that some function of orfde6 to orfde10 may be partially complemented by other genes, which are yet to be identified.

In conclusion, our results showed that disruption of epa genes results in loss of an immunoreactive polysaccharide and that the commonly distributed E. faecalis epa genes contribute to resistant to PMN phagocytic killing of E. faecalis. These results, the widespread occurrence of the epa genes, and our previous results showing attenuation of mutants in a mouse peritonitis model suggest that the common Epa polysaccharide warrants further study for a possible role in human infection.