Fsr-Independent Production of Protease(s) May Explain the Lack of Attenuation of an Enterococcus faecalis fsr Mutant Versus a gelE-sprE Muta
http://www.100md.com
感染与免疫杂志 2005年第8期
Center for the Study of Emerging and Reemerging Pathogens
Division of Infectious Diseases, Department of Internal Medicine
Department of Microbiology and Molecular Genetics, The University of Texas Medical School at Houston, Houston, Texas 77030
ABSTRACT
An Enterococcus faecalis gelE insertion disruption mutant (TX5128), which produces neither gelatinase (GelE) nor the cotranscribed (in the wild type) serine protease (SprE), was found to be attenuated in a rat endocarditis model with a significant decrease in the endocarditis induction rate versus wild-type E. faecalis OG1RF (GelE+, SprE+). TX5266, which has a nonpolar deletion in fsrB and, like TX5128, is phenotypically GelE– under usual conditions, was also studied; fsrB is a homologue of agrB of staphylococci and participates in regulation of gelE-sprE expression. Unexpectedly, TX5266 approximated wild-type OG1RF in the endocarditis model and was significantly less attenuated than TX5128. This is in contrast to other models which have found fsr mutants to be as or more attenuated than TX5128. Further study found that the fsrB mutant produced very low levels of gelatinase activity after prolonged incubation in vitro versus no gelatinase activity with TX5128 and did not show the extensive chaining characteristic of TX5128. Reverse transcription-PCR confirmed that gelE was expressed in TX5266 at a very low level versus wild-type OG1RF and was not expressed at all in TX5128. Possible explanations for the increased induction of endocarditis by TX5266 versus TX5128 include the production of low levels of protease(s) or some other effect(s) of the inactivation of the E. faecalis fsr regulator. The equivalent ability of OG1RF and its fsr mutant to initiate endocarditis may explain why we did not find naturally occurring fsr mutants, which account for ca. 35% of E. faecalis isolates, unrepresented in endocarditis versus fecal isolates (J. C. Roberts, K. V. Singh, P. C. Okhuysen, and B. E. Murray, J. Clin. Microbiol. 42:2317-2320, 2004).
INTRODUCTION
We have previously described the fsr locus, a regulatory system of Enterococcus faecalis and a homologue of staphylococcal agr loci, and shown that it encodes a quorum-sensing system that positively regulates the expression of gelatinase and serine protease (encoded by the cotranscribed gelE and sprE genes) via the Fsr-dependent gelE promoter (20, 21). We have also reported absence of a 23.9-kb region, as originally described by Nakayama and colleagues (16), in ca. 35% of 215 E. faecalis isolates tested; this deletion results in loss of fsrA, fsrB, and part of fsrC, resulting in an fsr-gelE locus functionally equivalent to constructed fsr mutants (24). OG1RF mutants with an insertion disruption in the gelE or fsr genes have shown significant delays in mortality in a mouse peritonitis model (21, 27), and a nonpolar fsrB deletion mutant (TX5266) (20) showed even more attenuation than a gelE disruption mutant in a Caenorhabditis elegans infection model (26) and a rabbit endophthalmitis model (6, 7, 14, 26). The first reports implicating an important role for gelatinase in biofilm formation were by Mohamed et al. (J. A. Mohamed, K. V. Singh, W. Huang, F. Teng, B. E. Murray, Abstr. 43rd Intersci. Conf. Antimicrob. Agents Chemother., abstr. B-821, 2003) and Kristich et al. (11). We have also shown that Fsr mutants show decreased biofilm production (12, 31), albeit to a slightly but significantly lesser extent than the gelatinase and serine protease mutant (Gel–, Spr–) TX5128, and both TX5128 and Fsr mutants showed markedly decreased translocation across intestinal T84 cells (33). The fsr mutants are phenotypically nonproducers of GelE and SprE by standard assay (21), which was presumed to explain, at least in part, their attenuation in animal models and their effect on translocation and biofilm (12, 31). The importance of gelatinase for biofilm formation was recently confirmed using purified enzyme (9).
Neither an fsr nor a protease mutant has been studied in an endocarditis model. However, data from Gutschik et al. (8) using 10 nonisogenic "Streptococcus faecalis" (5 Gel+ and 5 Gel–) strains showed that rabbits inoculated with proteolytic strains had significantly shorter survival times and more emboli than those inoculated with nonproteolytic strains. In the present study, we compared a nonpolar fsrB deletion mutant (TX5266) (20), the gelE insertion disruption mutant (TX5128) (21), and wild-type E. faecalis strain OG1RF (13) for induction of endocarditis in a rat model and found that TX5128 was significantly more attenuated than the wild type or the fsr mutant. We also found low-level expression of gelE in the fsr mutant, which may explain its being less attenuated. Another explanation, supported by C. elegans and rabbit endopthalmitis models, is that Fsr regulates other genes besides gelE-sprE, as is true for the Agr systems in Staphylococcus aureus.
MATERIALS AND METHODS
Bacterial strains. Bacteria used (Table 1) include wild-type OG1RF (Gel+, Spr+) (13), the fsrB deletion mutant TX5266, and the gelE insertion disruption mutant TX5128 (10, 21); the latter two are Gel– using a standard plate assay, (20, 21) and by Northern blot analysis we had previously shown that gelE and sprE are cotranscribed in OG1RF and no sprE mRNA is detected in TX5128 (21). TX5283 (20) is the fsrB deletion mutant TX5266 containing the shuttle vector pTCV-lac with promoterless lacZ (19), while TX5286 is TX5266 carrying pTEX5270 (gelE promoter::lacZ fusion in pTCV-lac) (20). Brain heart infusion broth (BHI) (Difco Laboratories, Detroit, Mich.) was used to grow bacterial cultures. Antibiotics kanamycin (Kan) and erythromycin (Ery) were purchased from Sigma Chemical Co., St. Louis, Mo. The concentration of antibiotics used in BHI agar plates for the growth of TX5128 was 2,000 μg/ml Kan and for TX5283 and TX5286 was 10 μg/ml Ery.
Endocarditis model. For preparation for administration to rats, bacteria were grown overnight at 37°C with or without (for OG1RF) antibiotics in 10 ml of BHI broth (Difco Laboratories, Detroit, Mich.) with gentle shaking. Cells were pelleted for 10 min (10,000 rpm at 4 to 6°C), washed once with 0.9% saline, and resuspended in 10 ml of saline. Further dilutions were also prepared in saline and optical density was adjusted, and suspensions were plated to determine the actual inocula administered. Infective aortic valve endocarditis was established in male Sprague-Dawley rats using previously published methods (25) with some modifications and following interpretation criteria described by others (3, 4, 25, 32). In brief, the animals were anesthetized with isoflurane for placement of intravascular catheters. The right carotid artery was exposed and a sterile polyethylene catheter (PE10; inside and outside diameters, 0.28 and 0.61 mm, respectively; Braintree Scientific, Braintree, MA) was inserted through a small incision and advanced to 3.5 cm across the aortic valve into the left ventricle. Proper positioning was assured by feeling the resistance and vigorous pulsation of the line. Twenty minutes after catheterization, each rat was inoculated via the catheter with 0.5 ml of test bacteria in a range of 103 to 107 CFU in 0.9% saline. After inoculation, the catheter was heat sealed and left in place during the course of experiments (18, 28), and the skin was closed with sutures. Rats were euthanized by CO2 inhalation at 24 h, the aortic valves were examined, the vegetations were excised, weighed, and homogenized in 0.5 ml saline, and dilutions were plated on BHI agar and/or BHI agar plus 2,000 μg/ml Kan where appropriate to confirm retention of disruption mutations. Rats with sterile cultures of their undiluted vegetation homogenates were considered to have had no induction of endocarditis (4). The 50% infective dose (ID50) values were calculated by a previously published method (23). Randomly picked fsrB deletion mutant colonies recovered from infected vegetations were also tested by PCR to reconfirm the deletions. The activity of OG1RF and TX5266 was reconfirmed in the mouse peritonitis (26). Our protocol was preapproved by the Animal Welfare Committee, The University of Texas Health Science Center, Houston, Texas. Fisher's exact test was used to compare the rates of induction of endocarditis among test bacteria and to compare the percentages of numbers of cells per chain (2 cells/chain versus 3 cells/chain) among OG1RF, TX5266, and TX5128.
Gelatinase and cell morphology. Screening for gelatinase production was done as described previously using 3% gelatin on Todd-Hewitt agar plates with incubation at 37°C for 24 h (5, 24). After some plates were accidentally left at ambient temperature for prolonged periods (3 weeks), we then systematically assessed gelatinase production after an initial incubation of 24 h at 37°C followed by a prolonged incubation of 16 to 20 days at room temperature. A previously published assay (15) for gelatinase activity using Azocoll (Azo dye-impregnated collagen, 0.25 g, <50 mesh; Calbiochem, Darmstadt, Germany) as substrate was also used to detect proteolytic activity of 24-h- and 12-day-old broth cultures of OG1RF, the fsrB deletion mutant TX5266, and the gelE insertion disruption mutant TX5128. We also determined the activity of the gelE promoter in 24-h-grown cultures of TX5286 (TX5266 carrying the gelE promoter-lacZ fusion) by comparing it with TX5283 (TX5266 carrying pTCV-lac with promoterless lacZ) and using 2-nitrophenyl--D-galactopyranoside (Sigma-Aldrich, St. Louis, MO) as a substrate described in previously published methods (19, 20). For the -galactosidase assay, the experiment was performed in quadruplicate.
For cell morphology examination, bacteria were grown under the same conditions as those described above for experimental endocarditis sections. Cells were washed once with saline and resuspended in saline, and a 1:10 dilution was prepared resulting in 1 x 108 CFU/ml of bacterial suspension. Bacterial cells from 1 x 108 stock suspension were Gram stained (Difco Laboratories) and were viewed by phase-contrast microscopy (magnification, 1,250x) with an Olympus BX60 (Leeds Instruments, Inc., Dallas, TX). The number of cells per chain was also counted under light microscopy in 16 randomly chosen microscopic fields, and data were analyzed.
RNA extraction and reverse transcription-PCR (RT-PCR). RNA was extracted from cultures grown overnight in BHI broth medium and using RNeasy kits (QIAGEN Sciences, MD) from OG1RF, TX5128, and TX5266. Extracted RNA samples were treated two to three times with 20 U of RQ1 RNase-Free DNase (Promega, Madison, WI) to remove possible DNA contamination from the samples. RT-PCR was performed by using SuperScript One-Step kits (Invitrogen, Carlsbad, CA) following the manufacturer's recommendations. Primers used for RT-PCR are listed in Table 2. Intragenic primers used for mRNA gelE expression were designed from a OG1-10 gelE sequence (29) while the internal control primers used in the study for unrelated genes were based on intragenic regions of a 16S rRNA sequence we have previously used in Lactococcus lactis (2) and E. faecalis studies and gls24 (30) of E. faecalis. Various concentrations (200 ng, 20 ng, and 5 ng/reaction) of total RNA/reaction were used in RT-PCR. Processed RNA from OG1RF, TX5266, and TX5128 was also used in PCRs with Platinum Taq DNA polymerase (Invitrogen) to detect DNA contamination. The RT-PCR conditions consisted of 1 cycle at 49°C for 30 min for cDNA synthesis and then 1 cycle at 94°C for 2 min for predenaturation, followed by 29 cycles consisting of denaturation at 94°C for 30 s, annealing at 55°C for 45 s, and extension at 72°C for 1 min, followed by 1 cycle of extension at 72°C for 10 min. The total reaction volume was 50 μl, and 10 μl was loaded on the agarose gel for viewing the amplified products.
RESULTS AND DISCUSSION
Experimental endocarditis. The development of infective endocarditis using wild-type OG1RF, the fsrB deletion mutant (TX5266), and the gelE insertion mutant (TX5128) was evaluated for various challenge inocula. The ID50 of the percent of vegetations infected and P values are shown in Table 3, and the geometric means of the inocula in half-log increments and the moving average of the percentage of infected rats are shown in Fig. 1. Wild-type OG1RF at inocula of 103 and 104 CFU showed 0% and 12% endocarditis induction rate, while at inocula of 105, 106, and 107 CFU OG1RF showed endocarditis induction rate of 61%, 78%, and 100%, respectively. The ID50 of OG1RF was determined to be 1.6 x 105 CFU.
The ID50 of the gelE-sprE mutant (TX5128) was determined to be 1.8 x 106 CFU, approximately 12 times that of wild-type OG1RF. Because it would require inordinately large numbers of animals to statistically compare ID50s, we instead compared the percentage of rats infected with different inocula. At inocula of 105 and 106 CFU, TX5128 infected significantly fewer vegetations than wild-type OG1RF (3 out of 19 and 6 out of 17 [16% and 35%] with TX5128 versus 11 out of 18 and 11 out of 14 [61% and 78%] with OG1RF; P = 0.0069 and 0.0292, respectively), while at the highest inoculum of 107 CFU, 100% of vegetations were infected for both organisms (Table 3). The attenuation of TX5128 in comparison to the wild-type strain OG1RF is consistent with it being less virulent in other published animal models (6, 14, 21, 26, 27). It should also be noted that, in the present study, we tried to use equivalent numbers of CFU of test bacteria to infect animals; indeed, the geometric means of the CFU of the inocula (Fig. 1) of TX5128 were actually slightly greater than those of OG1RF and TX5266 for two of the three inocula showing marked differences. Moreover, our observations here, as well as previously published reports, of chaining (possibly because of less maturation of the muramidase-1 autolysin) of the mutant TX5128 (22, 31) implies that we delivered an even higher number of actual TX5128 cells (Fig. 3) in the inoculum than was deduced from CFU counts.
The nonpolar fsrB deletion mutant TX5266 has been shown to be highly attenuated versus wild-type OG1RF in mouse peritonitis, C. elegans (7, 26), and rabbit endophthalmitis (6, 14) models. In C. elegans and the endophthalmitis model, TX5266 was even more attenuated than the double protease mutant TX5128 (6, 26). However, in the present study, it was similar to OG1RF and was significantly less attenuated at a 106 CFU inoculum than the double protease mutant TX5128 (P = 0.0137) (Table 3). TX5266 showed 0 out of 4, 4 out of 9, and 6 out of 6 (0%, 44%, and 100%) infected vegetations at inocula of 103, 105, and >106 CFU, respectively (Fig. 1), and the ID50 was determined to be 3.0 x 105, very close to that of OG1RF. TX5266 was also noted to have much less chaining than TX5128 but slightly more chaining than OG1RF (Fig. 3). To test the possibility that a secondary mutation had occurred that restored virulence, the TX5266 culture used in these experiments was retested in the mouse peritonitis model and found to be as highly attenuated as previously reported (26 and data not shown). The current literature also shows that mutations affecting agr of S. aureus have been tested in a variety of animal models, including endocarditis (4), septic arthritis (1), and mouse peritoneal sepsis (J. Yu, C. Bellinger-Kawahara, P. Winterberg, K. Francis, Abstr. 102nd Gen. Meet. Am. Soc. Microbiol., abstr. B-322, 2002), and it has been shown that agr mutants were attenuated in endocarditis (4) and septic arthritis models (1) versus no virulence attenuation in a low-inoculum (versus our studies with OG1RF) mouse peritoneal sepsis model (17).
Investigation of gelE and gelatinase expression. Although fsr mutants, including TX5266, do not produce detectable gelatinase in the standard plate assay (20, 21), we serendipitously noted in the current study, having left some plates at room temperature for weeks, that there was a small amount of gelatinase activity observed with TX5266 (but not TX5128) in the plate assay. Using Azocoll (three independent determinations), very low levels of proteolytic activity were also observed with TX5266 with 12-day- but not 1-day-old cultures versus no activity in TX5128, while the wild-type strain OG1RF showed high proteolytic activity (data not shown) at both time points.
Since the low-level gelatinase activity could be related to basal or stress-induced expression from the gelE promoter or, alternatively, from activation of an unidentified proteolytic activity, we next examined our gelE promoter::lacZ fusion construct (20) using 24-h-grown cultures of TX5286 (TX5266 carrying the gelE promoter-lacZ fusion in pTCV-lac). A slightly increased level of -galactosidase activity of the gelE promoter was observed (quadruplicate determination) versus TX5283 (TX5266 carrying the shuttle vector pTCV-lac) (data not shown).
We also applied RT-PCR to RNA obtained from 24-h-grown cultures. Results (Fig. 2) showed that, with the fsrB deletion mutant TX5266, no transcript was seen with 5 ng total RNA; using RNA concentrations of 20 and 200 ng, gelE transcript was observed but was much less than that obtained with wild-type OG1RF (Fig. 2A, lanes 1, 4, and 7 of OG1RF versus lanes 2, 5, and 8 of TX5266). No RT-PCR-amplified bands were detected with RNA from TX5128 (Fig. 2A, lanes 3, 6, and 9), consistent with the disruption of gelE (20) and the lack of detectable gelatinase activity by TX5128 on Todd-Hewitt agar-gelatin plates even after prolonged incubation of 12 days. Control genes gls24 and 16S rRNA showed similar amounts of amplified bands in OG1RF, TX5266, and TX5128 (Fig. 2B and C). We did not see RT-PCR-amplified bands with total RNA of TX5266 from a 20-day-old culture, suggesting that the low level of gelatinase activity observed on plates may be due to the accumulative effect of gelatinase produced, while the mRNA may have been degraded after such prolonged incubation. We also found evidence of expression by RT-PCR using 4-h cultures with amounts of mRNA again much reduced for TX5266 versus OG1RF. We previously were unable to detect gelE mRNA in fsrA, fsrB, and fsrC mutants from cells in exponential phase (4 h) by RT-PCR (21). The difference in the current study may be that we used a different set of primers or that the methodology for RT-PCR is more sensitive than what was used 6 years ago.
Chaining. The distribution of the number of cells per chain of OG1RF, TX5266, and TX5128 is shown in Fig. 3, which confirms our previous results (31) with TX5128 and also shows that TX5128 displayed much more chaining than TX5266 (P = <0.0001); TX5266, in turn, showed slightly more chaining than OG1RF (P = 0.0212). Complementation of TX5128 (Gel–, Spr–) and a gelE deletion mutant (TX5264) with gelE in trans was previously shown to restore the diplococcus morphology of these strains (31), indicating that gelatinase is sufficient for restoration. Residual, low-level gelatinase with or without serine protease production may be the reason that the fsrB deletion mutant had mostly diplococcal chains, with only a small number of longer chains relative to the gelE insertion mutant (TX5128) which showed almost all the cells in longer chains with a small number present as diplococci (Fig. 3); alternatively, this difference could be due to another effect of the fsr disruption.
One limitation of this work is that we have not, because of the cost and labor requirements of this model, distinguished between effects of gelatinase and serine protease nor tested complemented mutants. In addition, whether TX5128 is attenuated because of effects resulting from loss of proteolytic activity on E. faecalis cells themselves (e.g., increased chaining) or because of a direct effect of protease(s) on host tissues or cells is not known. Moreover, this model only assessed the ability of bacteria to cause endocarditis, not whether the protease(s) had effects on disease manifestations, as was suggested by the study of Gutschik et al. (8), in which Gel+ strains had more evidence of systemic disease and embolic phenomena than nonisogenic Gel– strains. Since Fsr activation would be expected to occur in vegetations due to high organism density, which should then be followed by increased protease production, we envision that there could be subsequent proteolytic digestion of vegetations. Of note, in preliminary studies in rabbits, we noted a different consistency of vegetations, with soft, friable ones with OG1RF versus hard, nonfriable ones with TX5128 (unpublished observations). We would also like to point out that, in the current model, bacteria were inoculated after 20 min of catheterization and vegetations were studied 24 h after bacterial challenge; thus, bacteria were colonizing early vegetation lesions, not mature vegetations, leaving open the possibility that the outcome may be different if more mature vegetations (e.g., 48 h after catheterization) were to be studied.
It is of interest that, in our recent survey of 215 E. faecalis, we found that, although the gelE gene is present in over 90% of isolates, less than 65% had a functional fsr locus and produced gelatinase by routine assay; i.e., many isolates phenotypically mimic constructed fsr mutants. Since gelatinase production and fsr function have been shown to contribute to virulence in several models, as well as to biofilm production (11, 12) and translocation across T84 cells (33), it was intriguing to find that the fsr mutant studied here was not diminished in its ability to initiate endocarditis. This may explain a failure to find differences in the percentage of gelatinase-positive isolates or of isolates with an intact fsr locus between endocarditis and fecal isolates (5, 24). Our finding of low-level expression of gelatinase by the fsr mutant suggests that basal expression of protease activity could be what allows both Fsr+ and Fsr– E. faecalis to initiate endocarditis equally well. In contrast, in the peritonitis model, which involves inoculation of a large number of bacteria (108) into a very small space, a major difference was observed between wild-type OG1RF and both TX5128 and TX5266, particularly in terms of the rapidity of death. It is likely that this high number of organisms of OG1RF administered locally activates the fsr-gel locus via quorum sensing and results in local protease production. On the other hand, the inoculation of 106 to 107 organisms intravenously, plus their subsequent dispersion and dilution in the bloodstream, as with an endocarditis model, would result in such a low density of circulating organisms that there would not be activation of fsr, at least until vegetations are formed. Thus, following intravenous inoculation, we predict that fsr activation would neither be seen nor be expected during the bloodstream phase, nor would there be an increase in gelatinase production over basal levels, even with organisms fully capable of doing so. That basal levels of gelatinase are sufficient to allow induction of endocarditis is supported by the near equal ability of TX5266 to OG1RF to cause endocarditis; an alternative hypothesis is that some other effect of the loss of fsr function restores this ability.
In conclusion, a Gel–, Spr– mutant, TX5128, showed a significantly decreased rate of induction of endocarditis in comparison to wild-type E. faecalis OG1RF as well as in comparison to the very slow protease producer TX5266 (an fsrB deletion mutant), which in turn was not significantly attenuated versus OG1RF. This lesser degree of attenuation of TX5266 is similar, qualitatively, to our previous report showing less reduction in biofilm production with fsr mutants than with TX5128 (12). These results are in contrast to results from the rabbit endophthalmitis and the C. elegans model in which TX5266 was more attenuated than TX5128. The endocarditis model and biofilm results may be explainable by fsr-independent gelE expression. On the other hand, models showing greater attenuation of fsr mutants suggest that the fsr genes affect more than just protease production, as is the case for the homologous agr system in staphylococci; this possibility will be addressed in future studies.
ACKNOWLEDGMENTS
This work was supported in part by NIH grant R37 AI 47923 from the Division of Microbiology and Infectious Diseases, NIAID, to Barbara E. Murray.
We express our sincere thanks to Gary M. Dunny, Pat Schlievert, and John McCormick, University of Minnesota Medical School, Minneapolis, Minn., for helping us learn the cardiac catheterization techniques for the experimental endocarditis model.
Present address: Cordoba 1614, Rosario (2000), Argentina.
REFERENCES
1. Abdelnour, A., S. Arvidson, T. Bremell, C. Ryden, and A. Tarkowski. 1993. The accessory gene regulator (agr) controls Staphylococcus aureus virulence in a murine arthritis model. Infect. Immun. 61:3879-3885.
2. Bolotin, A., P. Wincker, S. Mauger, O. Jaillon, K. Malarme, J. Weissenbach, S. D. Ehrlich, and A. Sorokin. 2001. The complete genome sequence of the lactic acid bacterium Lactococcus lactis ssp. lactis IL-1403. Genome Res. 11:731-753.
3. Cheung, A. L., C. Wolz, M. R. Yeaman, and A. S. Bayer. 1995. Insertional inactivation of a chromosomal locus that modulates expression of potential virulence determinants in Staphylococcus aureus. J. Bacteriol. 177:3220-3226.
4. Cheung, A. L., M. R. Yeaman, P. M. Sullam, M. D. Witt, and A. S. Bayer. 1994. Role of the sar locus of Staphylococcus aureus in induction of endocarditis in rabbits. Infect. Immun. 62:1719-1725.
5. Coque, T. M., J. E. Patterson, J. M. Steckelberg, and B. E. Murray. 1995. Incidence of hemolysin, gelatinase, and aggregation substance among enterococci isolated from patients with endocarditis and other infections and from feces of hospitalized and community-based persons. J. Infect. Dis. 171:1223-1229.
6. Engelbert, M., E. Mylonakis, F. M. Ausubel, S. B. Calderwood, and M. S. Gilmore. 2004. Contribution of gelatinase, serine protease, and fsr to the pathogenesis of Enterococcus faecalis endophthalmitis. Infect. Immun. 72:3628-3633.
7. Garsin, D. A., C. D. Sifri, E. Mylonakis, X. Qin, K. V. Singh, B. E. Murray, S. B. Calderwood, and F. M. Ausubel. 2001. A simple model host for identifying gram-positive virulence factors. Proc. Natl. Acad. Sci. USA 98:10892-10897.
8. Gutschik, E., S. Moller, and N. Christensen. 1979. Experimental endocarditis in rabbits. 3. Significance of the proteolytic capacity of the infecting strains of Streptococcus faecalis. Acta Pathol. Microbiol. Scand. B 87:353-362.
9. Hancock, L. E., and M. Perego. 2004. The Enterococcus faecalis fsr two-component system controls biofilm development through production of gelatinase. J. Bacteriol. 186:5629-5639.
10. Kawalec, M., J. Potempa, J. L. Moon, J. Travis, and B. E. Murray. 2005. Molecular diversity of a putative virulence factor: purification and characterization of isoforms of an extracellular serine glutamyl endopeptidase of Enterococcus faecalis with different enzymatic activities. J. Bacteriol. 187:266-275.
11. Kristich, C. J., Y. H. Li, D. G. Cvitkovitch, and G. M. Dunny. 2004. Esp-independent biofilm formation by Enterococcus faecalis. J. Bacteriol. 186:154-163.
12. Mohamed, J. A., W. Huang, S. R. Nallapareddy, F. Teng, and B. E. Murray. 2004. Influence of origin of isolates, especially endocarditis isolates, and various genes on biofilm formation by Enterococcus faecalis. Infect. Immun. 72:3658-3663.
13. Murray, B. E., K. V. Singh, R. P. Ross, J. D. Heath, G. M. Dunny, and G. M. Weinstock. 1993. Generation of restriction map of Enterococcus faecalis OG1 and investigation of growth requirements and regions encoding biosynthetic function. J. Bacteriol. 175:5216-5223.
14. Mylonakis, E., M. Engelbert, X. Qin, C. D. Sifri, B. E. Murray, F. M. Ausubel, M. S. Gilmore, and S. B. Calderwood. 2002. The Enterococcus faecalis fsrB gene, a key component of the fsr quorum-sensing system, is associated with virulence in the rabbit endophthalmitis model. Infect. Immun. 70:4678-4681.
15. Nakayama, J., Y. Cao, T. Horii, S. Sakuda, A. D. Akkermans, W. M. de Vos, and H. Nagasawa. 2001. Gelatinase biosynthesis-activating pheromone: a peptide lactone that mediates a quorum sensing in Enterococcus faecalis. Mol. Microbiol. 41:145-154.
16. Nakayama, J., R. Kariyama, and H. Kumon. 2002. Description of a 23.9-kilobase chromosomal deletion containing a region encoding fsr genes which mainly determines the gelatinase-negative phenotype of clinical isolates of Enterococcus faecalis in urine. Appl. Environ. Microbiol. 68:3152-3155.
17. Novick, R. P. 2003. Autoinduction and signal transduction in the regulation of staphylococcal virulence. Mol. Microbiol. 48:1429-1449.
18. Patel, R., M. S. Rouse, K. E. Piper, and J. M. Steckelberg. 2001. Linezolid therapy of vancomycin-resistant Enterococcus faecium experimental endocarditis. Antimicrob. Agents Chemother. 45:621-623.
19. Poyart, C., and P. Trieu-Cuot. 1997. A broad-host-range mobilizable shuttle vector for the construction of transcriptional fusions to beta-galactosidase in gram-positive bacteria. FEMS Microbiol. Lett. 156:193-198.
20. Qin, X., K. V. Singh, G. M. Weinstock, and B. E. Murray. 2001. Characterization of fsr, a regulator controlling expression of gelatinase and serine protease in Enterococcus faecalis OG1RF. J. Bacteriol. 183:3372-3382.
21. Qin, X., K. V. Singh, G. M. Weinstock, and B. E. Murray. 2000. Effects of Enterococcus faecalis fsr genes on production of gelatinase and a serine protease and virulence. Infect. Immun. 68:2579-2586.
22. Qin, X., K. V. Singh, Y. Xu, G. M. Weinstock, and B. E. Murray. 1998. Effect of disruption of a gene encoding an autolysin of Enterococcus faecalis OG1RF. Antimicrob. Agents Chemother. 42:2883-2888.
23. Reed, L. J., and H. Muench. 1938. A simple method of estimating fifty percent end points. Am. J. Hygiene. 27:493-497.
24. Roberts, J. C., K. V. Singh, P. C. Okhuysen, and B. E. Murray. 2004. Molecular epidemiology of the fsr locus and of gelatinase production among different subsets of Enterococcus faecalis isolates. J. Clin. Microbiol. 42:2317-2320.
25. Santoro, J., and M. E. Levison. 1978. Rat model of experimental endocarditis. Infect. Immun. 19:915-918.
26. Sifri, C. D., E. Mylonakis, K. V. Singh, X. Qin, D. A. Garsin, B. E. Murray, F. M. Ausubel, and S. B. Calderwood. 2002. Virulence effect of Enterococcus faecalis protease genes and the quorum-sensing locus fsr in Caenorhabditis elegans and mice. Infect. Immun. 70:5647-5650.
27. Singh, K. V., X. Qin, G. M. Weinstock, and B. E. Murray. 1998. Generation and testing of mutants of Enterococcus faecalis in a mouse peritonitis model. J. Infect. Dis. 178:1416-1420.
28. Souli, M., C. Thauvin-Eliopoulos, and G. M. Eliopoulos. 2000. In vivo activities of evernimicin (SCH 27899) against vancomycin-susceptible and vancomycin-resistant enterococci in experimental endocarditis. Antimicrob. Agents Chemother. 44:2733-2739.
29. Su, Y. A., M. C. Sulavik, P. He, K. K. Makinen, P. L. Makinen, S. Fiedler, R. Wirth, and D. B. Clewell. 1991. Nucleotide sequence of the gelatinase gene (gelE) from Enterococcus faecalis subsp. liquefaciens. Infect. Immun. 59:415-420.
30. Teng, F., E. C. Nannini, and B. E. Murray. 2005. Importance of gls24 in virulence and stress response of Enterococcus faecalis and use of the Gls24 protein as a possible immunotherapy target. J. Infect. Dis. 191:472-480.
31. Waters, C. M., M. H. Antiporta, B. E. Murray, and G. M. Dunny. 2003. Role of the Enterococcus faecalis GelE protease in determination of cellular chain length, supernatant pheromone levels, and degradation of fibrin and misfolded surface proteins. J. Bacteriol. 185:3613-3623.
32. Weiss, W. J., T. Murphy, E. Lenoy, and M. Young. 2004. In vivo efficacy and pharmacokinetics of AC98-6446, a novel cyclic glycopeptide, in experimental infection models. Antimicrob. Agents Chemother. 48:1708-1712.
33. Zeng, J., F. Teng, G. M. Weinstock, and B. E. Murray. 2004. Translocation of Enterococcus faecalis strains across a monolayer of polarized human enterocyte-like T84 cells. J. Clin. Microbiol. 42:1149-1154.(Kavindra V. Singh, Sreedh)
Division of Infectious Diseases, Department of Internal Medicine
Department of Microbiology and Molecular Genetics, The University of Texas Medical School at Houston, Houston, Texas 77030
ABSTRACT
An Enterococcus faecalis gelE insertion disruption mutant (TX5128), which produces neither gelatinase (GelE) nor the cotranscribed (in the wild type) serine protease (SprE), was found to be attenuated in a rat endocarditis model with a significant decrease in the endocarditis induction rate versus wild-type E. faecalis OG1RF (GelE+, SprE+). TX5266, which has a nonpolar deletion in fsrB and, like TX5128, is phenotypically GelE– under usual conditions, was also studied; fsrB is a homologue of agrB of staphylococci and participates in regulation of gelE-sprE expression. Unexpectedly, TX5266 approximated wild-type OG1RF in the endocarditis model and was significantly less attenuated than TX5128. This is in contrast to other models which have found fsr mutants to be as or more attenuated than TX5128. Further study found that the fsrB mutant produced very low levels of gelatinase activity after prolonged incubation in vitro versus no gelatinase activity with TX5128 and did not show the extensive chaining characteristic of TX5128. Reverse transcription-PCR confirmed that gelE was expressed in TX5266 at a very low level versus wild-type OG1RF and was not expressed at all in TX5128. Possible explanations for the increased induction of endocarditis by TX5266 versus TX5128 include the production of low levels of protease(s) or some other effect(s) of the inactivation of the E. faecalis fsr regulator. The equivalent ability of OG1RF and its fsr mutant to initiate endocarditis may explain why we did not find naturally occurring fsr mutants, which account for ca. 35% of E. faecalis isolates, unrepresented in endocarditis versus fecal isolates (J. C. Roberts, K. V. Singh, P. C. Okhuysen, and B. E. Murray, J. Clin. Microbiol. 42:2317-2320, 2004).
INTRODUCTION
We have previously described the fsr locus, a regulatory system of Enterococcus faecalis and a homologue of staphylococcal agr loci, and shown that it encodes a quorum-sensing system that positively regulates the expression of gelatinase and serine protease (encoded by the cotranscribed gelE and sprE genes) via the Fsr-dependent gelE promoter (20, 21). We have also reported absence of a 23.9-kb region, as originally described by Nakayama and colleagues (16), in ca. 35% of 215 E. faecalis isolates tested; this deletion results in loss of fsrA, fsrB, and part of fsrC, resulting in an fsr-gelE locus functionally equivalent to constructed fsr mutants (24). OG1RF mutants with an insertion disruption in the gelE or fsr genes have shown significant delays in mortality in a mouse peritonitis model (21, 27), and a nonpolar fsrB deletion mutant (TX5266) (20) showed even more attenuation than a gelE disruption mutant in a Caenorhabditis elegans infection model (26) and a rabbit endophthalmitis model (6, 7, 14, 26). The first reports implicating an important role for gelatinase in biofilm formation were by Mohamed et al. (J. A. Mohamed, K. V. Singh, W. Huang, F. Teng, B. E. Murray, Abstr. 43rd Intersci. Conf. Antimicrob. Agents Chemother., abstr. B-821, 2003) and Kristich et al. (11). We have also shown that Fsr mutants show decreased biofilm production (12, 31), albeit to a slightly but significantly lesser extent than the gelatinase and serine protease mutant (Gel–, Spr–) TX5128, and both TX5128 and Fsr mutants showed markedly decreased translocation across intestinal T84 cells (33). The fsr mutants are phenotypically nonproducers of GelE and SprE by standard assay (21), which was presumed to explain, at least in part, their attenuation in animal models and their effect on translocation and biofilm (12, 31). The importance of gelatinase for biofilm formation was recently confirmed using purified enzyme (9).
Neither an fsr nor a protease mutant has been studied in an endocarditis model. However, data from Gutschik et al. (8) using 10 nonisogenic "Streptococcus faecalis" (5 Gel+ and 5 Gel–) strains showed that rabbits inoculated with proteolytic strains had significantly shorter survival times and more emboli than those inoculated with nonproteolytic strains. In the present study, we compared a nonpolar fsrB deletion mutant (TX5266) (20), the gelE insertion disruption mutant (TX5128) (21), and wild-type E. faecalis strain OG1RF (13) for induction of endocarditis in a rat model and found that TX5128 was significantly more attenuated than the wild type or the fsr mutant. We also found low-level expression of gelE in the fsr mutant, which may explain its being less attenuated. Another explanation, supported by C. elegans and rabbit endopthalmitis models, is that Fsr regulates other genes besides gelE-sprE, as is true for the Agr systems in Staphylococcus aureus.
MATERIALS AND METHODS
Bacterial strains. Bacteria used (Table 1) include wild-type OG1RF (Gel+, Spr+) (13), the fsrB deletion mutant TX5266, and the gelE insertion disruption mutant TX5128 (10, 21); the latter two are Gel– using a standard plate assay, (20, 21) and by Northern blot analysis we had previously shown that gelE and sprE are cotranscribed in OG1RF and no sprE mRNA is detected in TX5128 (21). TX5283 (20) is the fsrB deletion mutant TX5266 containing the shuttle vector pTCV-lac with promoterless lacZ (19), while TX5286 is TX5266 carrying pTEX5270 (gelE promoter::lacZ fusion in pTCV-lac) (20). Brain heart infusion broth (BHI) (Difco Laboratories, Detroit, Mich.) was used to grow bacterial cultures. Antibiotics kanamycin (Kan) and erythromycin (Ery) were purchased from Sigma Chemical Co., St. Louis, Mo. The concentration of antibiotics used in BHI agar plates for the growth of TX5128 was 2,000 μg/ml Kan and for TX5283 and TX5286 was 10 μg/ml Ery.
Endocarditis model. For preparation for administration to rats, bacteria were grown overnight at 37°C with or without (for OG1RF) antibiotics in 10 ml of BHI broth (Difco Laboratories, Detroit, Mich.) with gentle shaking. Cells were pelleted for 10 min (10,000 rpm at 4 to 6°C), washed once with 0.9% saline, and resuspended in 10 ml of saline. Further dilutions were also prepared in saline and optical density was adjusted, and suspensions were plated to determine the actual inocula administered. Infective aortic valve endocarditis was established in male Sprague-Dawley rats using previously published methods (25) with some modifications and following interpretation criteria described by others (3, 4, 25, 32). In brief, the animals were anesthetized with isoflurane for placement of intravascular catheters. The right carotid artery was exposed and a sterile polyethylene catheter (PE10; inside and outside diameters, 0.28 and 0.61 mm, respectively; Braintree Scientific, Braintree, MA) was inserted through a small incision and advanced to 3.5 cm across the aortic valve into the left ventricle. Proper positioning was assured by feeling the resistance and vigorous pulsation of the line. Twenty minutes after catheterization, each rat was inoculated via the catheter with 0.5 ml of test bacteria in a range of 103 to 107 CFU in 0.9% saline. After inoculation, the catheter was heat sealed and left in place during the course of experiments (18, 28), and the skin was closed with sutures. Rats were euthanized by CO2 inhalation at 24 h, the aortic valves were examined, the vegetations were excised, weighed, and homogenized in 0.5 ml saline, and dilutions were plated on BHI agar and/or BHI agar plus 2,000 μg/ml Kan where appropriate to confirm retention of disruption mutations. Rats with sterile cultures of their undiluted vegetation homogenates were considered to have had no induction of endocarditis (4). The 50% infective dose (ID50) values were calculated by a previously published method (23). Randomly picked fsrB deletion mutant colonies recovered from infected vegetations were also tested by PCR to reconfirm the deletions. The activity of OG1RF and TX5266 was reconfirmed in the mouse peritonitis (26). Our protocol was preapproved by the Animal Welfare Committee, The University of Texas Health Science Center, Houston, Texas. Fisher's exact test was used to compare the rates of induction of endocarditis among test bacteria and to compare the percentages of numbers of cells per chain (2 cells/chain versus 3 cells/chain) among OG1RF, TX5266, and TX5128.
Gelatinase and cell morphology. Screening for gelatinase production was done as described previously using 3% gelatin on Todd-Hewitt agar plates with incubation at 37°C for 24 h (5, 24). After some plates were accidentally left at ambient temperature for prolonged periods (3 weeks), we then systematically assessed gelatinase production after an initial incubation of 24 h at 37°C followed by a prolonged incubation of 16 to 20 days at room temperature. A previously published assay (15) for gelatinase activity using Azocoll (Azo dye-impregnated collagen, 0.25 g, <50 mesh; Calbiochem, Darmstadt, Germany) as substrate was also used to detect proteolytic activity of 24-h- and 12-day-old broth cultures of OG1RF, the fsrB deletion mutant TX5266, and the gelE insertion disruption mutant TX5128. We also determined the activity of the gelE promoter in 24-h-grown cultures of TX5286 (TX5266 carrying the gelE promoter-lacZ fusion) by comparing it with TX5283 (TX5266 carrying pTCV-lac with promoterless lacZ) and using 2-nitrophenyl--D-galactopyranoside (Sigma-Aldrich, St. Louis, MO) as a substrate described in previously published methods (19, 20). For the -galactosidase assay, the experiment was performed in quadruplicate.
For cell morphology examination, bacteria were grown under the same conditions as those described above for experimental endocarditis sections. Cells were washed once with saline and resuspended in saline, and a 1:10 dilution was prepared resulting in 1 x 108 CFU/ml of bacterial suspension. Bacterial cells from 1 x 108 stock suspension were Gram stained (Difco Laboratories) and were viewed by phase-contrast microscopy (magnification, 1,250x) with an Olympus BX60 (Leeds Instruments, Inc., Dallas, TX). The number of cells per chain was also counted under light microscopy in 16 randomly chosen microscopic fields, and data were analyzed.
RNA extraction and reverse transcription-PCR (RT-PCR). RNA was extracted from cultures grown overnight in BHI broth medium and using RNeasy kits (QIAGEN Sciences, MD) from OG1RF, TX5128, and TX5266. Extracted RNA samples were treated two to three times with 20 U of RQ1 RNase-Free DNase (Promega, Madison, WI) to remove possible DNA contamination from the samples. RT-PCR was performed by using SuperScript One-Step kits (Invitrogen, Carlsbad, CA) following the manufacturer's recommendations. Primers used for RT-PCR are listed in Table 2. Intragenic primers used for mRNA gelE expression were designed from a OG1-10 gelE sequence (29) while the internal control primers used in the study for unrelated genes were based on intragenic regions of a 16S rRNA sequence we have previously used in Lactococcus lactis (2) and E. faecalis studies and gls24 (30) of E. faecalis. Various concentrations (200 ng, 20 ng, and 5 ng/reaction) of total RNA/reaction were used in RT-PCR. Processed RNA from OG1RF, TX5266, and TX5128 was also used in PCRs with Platinum Taq DNA polymerase (Invitrogen) to detect DNA contamination. The RT-PCR conditions consisted of 1 cycle at 49°C for 30 min for cDNA synthesis and then 1 cycle at 94°C for 2 min for predenaturation, followed by 29 cycles consisting of denaturation at 94°C for 30 s, annealing at 55°C for 45 s, and extension at 72°C for 1 min, followed by 1 cycle of extension at 72°C for 10 min. The total reaction volume was 50 μl, and 10 μl was loaded on the agarose gel for viewing the amplified products.
RESULTS AND DISCUSSION
Experimental endocarditis. The development of infective endocarditis using wild-type OG1RF, the fsrB deletion mutant (TX5266), and the gelE insertion mutant (TX5128) was evaluated for various challenge inocula. The ID50 of the percent of vegetations infected and P values are shown in Table 3, and the geometric means of the inocula in half-log increments and the moving average of the percentage of infected rats are shown in Fig. 1. Wild-type OG1RF at inocula of 103 and 104 CFU showed 0% and 12% endocarditis induction rate, while at inocula of 105, 106, and 107 CFU OG1RF showed endocarditis induction rate of 61%, 78%, and 100%, respectively. The ID50 of OG1RF was determined to be 1.6 x 105 CFU.
The ID50 of the gelE-sprE mutant (TX5128) was determined to be 1.8 x 106 CFU, approximately 12 times that of wild-type OG1RF. Because it would require inordinately large numbers of animals to statistically compare ID50s, we instead compared the percentage of rats infected with different inocula. At inocula of 105 and 106 CFU, TX5128 infected significantly fewer vegetations than wild-type OG1RF (3 out of 19 and 6 out of 17 [16% and 35%] with TX5128 versus 11 out of 18 and 11 out of 14 [61% and 78%] with OG1RF; P = 0.0069 and 0.0292, respectively), while at the highest inoculum of 107 CFU, 100% of vegetations were infected for both organisms (Table 3). The attenuation of TX5128 in comparison to the wild-type strain OG1RF is consistent with it being less virulent in other published animal models (6, 14, 21, 26, 27). It should also be noted that, in the present study, we tried to use equivalent numbers of CFU of test bacteria to infect animals; indeed, the geometric means of the CFU of the inocula (Fig. 1) of TX5128 were actually slightly greater than those of OG1RF and TX5266 for two of the three inocula showing marked differences. Moreover, our observations here, as well as previously published reports, of chaining (possibly because of less maturation of the muramidase-1 autolysin) of the mutant TX5128 (22, 31) implies that we delivered an even higher number of actual TX5128 cells (Fig. 3) in the inoculum than was deduced from CFU counts.
The nonpolar fsrB deletion mutant TX5266 has been shown to be highly attenuated versus wild-type OG1RF in mouse peritonitis, C. elegans (7, 26), and rabbit endophthalmitis (6, 14) models. In C. elegans and the endophthalmitis model, TX5266 was even more attenuated than the double protease mutant TX5128 (6, 26). However, in the present study, it was similar to OG1RF and was significantly less attenuated at a 106 CFU inoculum than the double protease mutant TX5128 (P = 0.0137) (Table 3). TX5266 showed 0 out of 4, 4 out of 9, and 6 out of 6 (0%, 44%, and 100%) infected vegetations at inocula of 103, 105, and >106 CFU, respectively (Fig. 1), and the ID50 was determined to be 3.0 x 105, very close to that of OG1RF. TX5266 was also noted to have much less chaining than TX5128 but slightly more chaining than OG1RF (Fig. 3). To test the possibility that a secondary mutation had occurred that restored virulence, the TX5266 culture used in these experiments was retested in the mouse peritonitis model and found to be as highly attenuated as previously reported (26 and data not shown). The current literature also shows that mutations affecting agr of S. aureus have been tested in a variety of animal models, including endocarditis (4), septic arthritis (1), and mouse peritoneal sepsis (J. Yu, C. Bellinger-Kawahara, P. Winterberg, K. Francis, Abstr. 102nd Gen. Meet. Am. Soc. Microbiol., abstr. B-322, 2002), and it has been shown that agr mutants were attenuated in endocarditis (4) and septic arthritis models (1) versus no virulence attenuation in a low-inoculum (versus our studies with OG1RF) mouse peritoneal sepsis model (17).
Investigation of gelE and gelatinase expression. Although fsr mutants, including TX5266, do not produce detectable gelatinase in the standard plate assay (20, 21), we serendipitously noted in the current study, having left some plates at room temperature for weeks, that there was a small amount of gelatinase activity observed with TX5266 (but not TX5128) in the plate assay. Using Azocoll (three independent determinations), very low levels of proteolytic activity were also observed with TX5266 with 12-day- but not 1-day-old cultures versus no activity in TX5128, while the wild-type strain OG1RF showed high proteolytic activity (data not shown) at both time points.
Since the low-level gelatinase activity could be related to basal or stress-induced expression from the gelE promoter or, alternatively, from activation of an unidentified proteolytic activity, we next examined our gelE promoter::lacZ fusion construct (20) using 24-h-grown cultures of TX5286 (TX5266 carrying the gelE promoter-lacZ fusion in pTCV-lac). A slightly increased level of -galactosidase activity of the gelE promoter was observed (quadruplicate determination) versus TX5283 (TX5266 carrying the shuttle vector pTCV-lac) (data not shown).
We also applied RT-PCR to RNA obtained from 24-h-grown cultures. Results (Fig. 2) showed that, with the fsrB deletion mutant TX5266, no transcript was seen with 5 ng total RNA; using RNA concentrations of 20 and 200 ng, gelE transcript was observed but was much less than that obtained with wild-type OG1RF (Fig. 2A, lanes 1, 4, and 7 of OG1RF versus lanes 2, 5, and 8 of TX5266). No RT-PCR-amplified bands were detected with RNA from TX5128 (Fig. 2A, lanes 3, 6, and 9), consistent with the disruption of gelE (20) and the lack of detectable gelatinase activity by TX5128 on Todd-Hewitt agar-gelatin plates even after prolonged incubation of 12 days. Control genes gls24 and 16S rRNA showed similar amounts of amplified bands in OG1RF, TX5266, and TX5128 (Fig. 2B and C). We did not see RT-PCR-amplified bands with total RNA of TX5266 from a 20-day-old culture, suggesting that the low level of gelatinase activity observed on plates may be due to the accumulative effect of gelatinase produced, while the mRNA may have been degraded after such prolonged incubation. We also found evidence of expression by RT-PCR using 4-h cultures with amounts of mRNA again much reduced for TX5266 versus OG1RF. We previously were unable to detect gelE mRNA in fsrA, fsrB, and fsrC mutants from cells in exponential phase (4 h) by RT-PCR (21). The difference in the current study may be that we used a different set of primers or that the methodology for RT-PCR is more sensitive than what was used 6 years ago.
Chaining. The distribution of the number of cells per chain of OG1RF, TX5266, and TX5128 is shown in Fig. 3, which confirms our previous results (31) with TX5128 and also shows that TX5128 displayed much more chaining than TX5266 (P = <0.0001); TX5266, in turn, showed slightly more chaining than OG1RF (P = 0.0212). Complementation of TX5128 (Gel–, Spr–) and a gelE deletion mutant (TX5264) with gelE in trans was previously shown to restore the diplococcus morphology of these strains (31), indicating that gelatinase is sufficient for restoration. Residual, low-level gelatinase with or without serine protease production may be the reason that the fsrB deletion mutant had mostly diplococcal chains, with only a small number of longer chains relative to the gelE insertion mutant (TX5128) which showed almost all the cells in longer chains with a small number present as diplococci (Fig. 3); alternatively, this difference could be due to another effect of the fsr disruption.
One limitation of this work is that we have not, because of the cost and labor requirements of this model, distinguished between effects of gelatinase and serine protease nor tested complemented mutants. In addition, whether TX5128 is attenuated because of effects resulting from loss of proteolytic activity on E. faecalis cells themselves (e.g., increased chaining) or because of a direct effect of protease(s) on host tissues or cells is not known. Moreover, this model only assessed the ability of bacteria to cause endocarditis, not whether the protease(s) had effects on disease manifestations, as was suggested by the study of Gutschik et al. (8), in which Gel+ strains had more evidence of systemic disease and embolic phenomena than nonisogenic Gel– strains. Since Fsr activation would be expected to occur in vegetations due to high organism density, which should then be followed by increased protease production, we envision that there could be subsequent proteolytic digestion of vegetations. Of note, in preliminary studies in rabbits, we noted a different consistency of vegetations, with soft, friable ones with OG1RF versus hard, nonfriable ones with TX5128 (unpublished observations). We would also like to point out that, in the current model, bacteria were inoculated after 20 min of catheterization and vegetations were studied 24 h after bacterial challenge; thus, bacteria were colonizing early vegetation lesions, not mature vegetations, leaving open the possibility that the outcome may be different if more mature vegetations (e.g., 48 h after catheterization) were to be studied.
It is of interest that, in our recent survey of 215 E. faecalis, we found that, although the gelE gene is present in over 90% of isolates, less than 65% had a functional fsr locus and produced gelatinase by routine assay; i.e., many isolates phenotypically mimic constructed fsr mutants. Since gelatinase production and fsr function have been shown to contribute to virulence in several models, as well as to biofilm production (11, 12) and translocation across T84 cells (33), it was intriguing to find that the fsr mutant studied here was not diminished in its ability to initiate endocarditis. This may explain a failure to find differences in the percentage of gelatinase-positive isolates or of isolates with an intact fsr locus between endocarditis and fecal isolates (5, 24). Our finding of low-level expression of gelatinase by the fsr mutant suggests that basal expression of protease activity could be what allows both Fsr+ and Fsr– E. faecalis to initiate endocarditis equally well. In contrast, in the peritonitis model, which involves inoculation of a large number of bacteria (108) into a very small space, a major difference was observed between wild-type OG1RF and both TX5128 and TX5266, particularly in terms of the rapidity of death. It is likely that this high number of organisms of OG1RF administered locally activates the fsr-gel locus via quorum sensing and results in local protease production. On the other hand, the inoculation of 106 to 107 organisms intravenously, plus their subsequent dispersion and dilution in the bloodstream, as with an endocarditis model, would result in such a low density of circulating organisms that there would not be activation of fsr, at least until vegetations are formed. Thus, following intravenous inoculation, we predict that fsr activation would neither be seen nor be expected during the bloodstream phase, nor would there be an increase in gelatinase production over basal levels, even with organisms fully capable of doing so. That basal levels of gelatinase are sufficient to allow induction of endocarditis is supported by the near equal ability of TX5266 to OG1RF to cause endocarditis; an alternative hypothesis is that some other effect of the loss of fsr function restores this ability.
In conclusion, a Gel–, Spr– mutant, TX5128, showed a significantly decreased rate of induction of endocarditis in comparison to wild-type E. faecalis OG1RF as well as in comparison to the very slow protease producer TX5266 (an fsrB deletion mutant), which in turn was not significantly attenuated versus OG1RF. This lesser degree of attenuation of TX5266 is similar, qualitatively, to our previous report showing less reduction in biofilm production with fsr mutants than with TX5128 (12). These results are in contrast to results from the rabbit endophthalmitis and the C. elegans model in which TX5266 was more attenuated than TX5128. The endocarditis model and biofilm results may be explainable by fsr-independent gelE expression. On the other hand, models showing greater attenuation of fsr mutants suggest that the fsr genes affect more than just protease production, as is the case for the homologous agr system in staphylococci; this possibility will be addressed in future studies.
ACKNOWLEDGMENTS
This work was supported in part by NIH grant R37 AI 47923 from the Division of Microbiology and Infectious Diseases, NIAID, to Barbara E. Murray.
We express our sincere thanks to Gary M. Dunny, Pat Schlievert, and John McCormick, University of Minnesota Medical School, Minneapolis, Minn., for helping us learn the cardiac catheterization techniques for the experimental endocarditis model.
Present address: Cordoba 1614, Rosario (2000), Argentina.
REFERENCES
1. Abdelnour, A., S. Arvidson, T. Bremell, C. Ryden, and A. Tarkowski. 1993. The accessory gene regulator (agr) controls Staphylococcus aureus virulence in a murine arthritis model. Infect. Immun. 61:3879-3885.
2. Bolotin, A., P. Wincker, S. Mauger, O. Jaillon, K. Malarme, J. Weissenbach, S. D. Ehrlich, and A. Sorokin. 2001. The complete genome sequence of the lactic acid bacterium Lactococcus lactis ssp. lactis IL-1403. Genome Res. 11:731-753.
3. Cheung, A. L., C. Wolz, M. R. Yeaman, and A. S. Bayer. 1995. Insertional inactivation of a chromosomal locus that modulates expression of potential virulence determinants in Staphylococcus aureus. J. Bacteriol. 177:3220-3226.
4. Cheung, A. L., M. R. Yeaman, P. M. Sullam, M. D. Witt, and A. S. Bayer. 1994. Role of the sar locus of Staphylococcus aureus in induction of endocarditis in rabbits. Infect. Immun. 62:1719-1725.
5. Coque, T. M., J. E. Patterson, J. M. Steckelberg, and B. E. Murray. 1995. Incidence of hemolysin, gelatinase, and aggregation substance among enterococci isolated from patients with endocarditis and other infections and from feces of hospitalized and community-based persons. J. Infect. Dis. 171:1223-1229.
6. Engelbert, M., E. Mylonakis, F. M. Ausubel, S. B. Calderwood, and M. S. Gilmore. 2004. Contribution of gelatinase, serine protease, and fsr to the pathogenesis of Enterococcus faecalis endophthalmitis. Infect. Immun. 72:3628-3633.
7. Garsin, D. A., C. D. Sifri, E. Mylonakis, X. Qin, K. V. Singh, B. E. Murray, S. B. Calderwood, and F. M. Ausubel. 2001. A simple model host for identifying gram-positive virulence factors. Proc. Natl. Acad. Sci. USA 98:10892-10897.
8. Gutschik, E., S. Moller, and N. Christensen. 1979. Experimental endocarditis in rabbits. 3. Significance of the proteolytic capacity of the infecting strains of Streptococcus faecalis. Acta Pathol. Microbiol. Scand. B 87:353-362.
9. Hancock, L. E., and M. Perego. 2004. The Enterococcus faecalis fsr two-component system controls biofilm development through production of gelatinase. J. Bacteriol. 186:5629-5639.
10. Kawalec, M., J. Potempa, J. L. Moon, J. Travis, and B. E. Murray. 2005. Molecular diversity of a putative virulence factor: purification and characterization of isoforms of an extracellular serine glutamyl endopeptidase of Enterococcus faecalis with different enzymatic activities. J. Bacteriol. 187:266-275.
11. Kristich, C. J., Y. H. Li, D. G. Cvitkovitch, and G. M. Dunny. 2004. Esp-independent biofilm formation by Enterococcus faecalis. J. Bacteriol. 186:154-163.
12. Mohamed, J. A., W. Huang, S. R. Nallapareddy, F. Teng, and B. E. Murray. 2004. Influence of origin of isolates, especially endocarditis isolates, and various genes on biofilm formation by Enterococcus faecalis. Infect. Immun. 72:3658-3663.
13. Murray, B. E., K. V. Singh, R. P. Ross, J. D. Heath, G. M. Dunny, and G. M. Weinstock. 1993. Generation of restriction map of Enterococcus faecalis OG1 and investigation of growth requirements and regions encoding biosynthetic function. J. Bacteriol. 175:5216-5223.
14. Mylonakis, E., M. Engelbert, X. Qin, C. D. Sifri, B. E. Murray, F. M. Ausubel, M. S. Gilmore, and S. B. Calderwood. 2002. The Enterococcus faecalis fsrB gene, a key component of the fsr quorum-sensing system, is associated with virulence in the rabbit endophthalmitis model. Infect. Immun. 70:4678-4681.
15. Nakayama, J., Y. Cao, T. Horii, S. Sakuda, A. D. Akkermans, W. M. de Vos, and H. Nagasawa. 2001. Gelatinase biosynthesis-activating pheromone: a peptide lactone that mediates a quorum sensing in Enterococcus faecalis. Mol. Microbiol. 41:145-154.
16. Nakayama, J., R. Kariyama, and H. Kumon. 2002. Description of a 23.9-kilobase chromosomal deletion containing a region encoding fsr genes which mainly determines the gelatinase-negative phenotype of clinical isolates of Enterococcus faecalis in urine. Appl. Environ. Microbiol. 68:3152-3155.
17. Novick, R. P. 2003. Autoinduction and signal transduction in the regulation of staphylococcal virulence. Mol. Microbiol. 48:1429-1449.
18. Patel, R., M. S. Rouse, K. E. Piper, and J. M. Steckelberg. 2001. Linezolid therapy of vancomycin-resistant Enterococcus faecium experimental endocarditis. Antimicrob. Agents Chemother. 45:621-623.
19. Poyart, C., and P. Trieu-Cuot. 1997. A broad-host-range mobilizable shuttle vector for the construction of transcriptional fusions to beta-galactosidase in gram-positive bacteria. FEMS Microbiol. Lett. 156:193-198.
20. Qin, X., K. V. Singh, G. M. Weinstock, and B. E. Murray. 2001. Characterization of fsr, a regulator controlling expression of gelatinase and serine protease in Enterococcus faecalis OG1RF. J. Bacteriol. 183:3372-3382.
21. Qin, X., K. V. Singh, G. M. Weinstock, and B. E. Murray. 2000. Effects of Enterococcus faecalis fsr genes on production of gelatinase and a serine protease and virulence. Infect. Immun. 68:2579-2586.
22. Qin, X., K. V. Singh, Y. Xu, G. M. Weinstock, and B. E. Murray. 1998. Effect of disruption of a gene encoding an autolysin of Enterococcus faecalis OG1RF. Antimicrob. Agents Chemother. 42:2883-2888.
23. Reed, L. J., and H. Muench. 1938. A simple method of estimating fifty percent end points. Am. J. Hygiene. 27:493-497.
24. Roberts, J. C., K. V. Singh, P. C. Okhuysen, and B. E. Murray. 2004. Molecular epidemiology of the fsr locus and of gelatinase production among different subsets of Enterococcus faecalis isolates. J. Clin. Microbiol. 42:2317-2320.
25. Santoro, J., and M. E. Levison. 1978. Rat model of experimental endocarditis. Infect. Immun. 19:915-918.
26. Sifri, C. D., E. Mylonakis, K. V. Singh, X. Qin, D. A. Garsin, B. E. Murray, F. M. Ausubel, and S. B. Calderwood. 2002. Virulence effect of Enterococcus faecalis protease genes and the quorum-sensing locus fsr in Caenorhabditis elegans and mice. Infect. Immun. 70:5647-5650.
27. Singh, K. V., X. Qin, G. M. Weinstock, and B. E. Murray. 1998. Generation and testing of mutants of Enterococcus faecalis in a mouse peritonitis model. J. Infect. Dis. 178:1416-1420.
28. Souli, M., C. Thauvin-Eliopoulos, and G. M. Eliopoulos. 2000. In vivo activities of evernimicin (SCH 27899) against vancomycin-susceptible and vancomycin-resistant enterococci in experimental endocarditis. Antimicrob. Agents Chemother. 44:2733-2739.
29. Su, Y. A., M. C. Sulavik, P. He, K. K. Makinen, P. L. Makinen, S. Fiedler, R. Wirth, and D. B. Clewell. 1991. Nucleotide sequence of the gelatinase gene (gelE) from Enterococcus faecalis subsp. liquefaciens. Infect. Immun. 59:415-420.
30. Teng, F., E. C. Nannini, and B. E. Murray. 2005. Importance of gls24 in virulence and stress response of Enterococcus faecalis and use of the Gls24 protein as a possible immunotherapy target. J. Infect. Dis. 191:472-480.
31. Waters, C. M., M. H. Antiporta, B. E. Murray, and G. M. Dunny. 2003. Role of the Enterococcus faecalis GelE protease in determination of cellular chain length, supernatant pheromone levels, and degradation of fibrin and misfolded surface proteins. J. Bacteriol. 185:3613-3623.
32. Weiss, W. J., T. Murphy, E. Lenoy, and M. Young. 2004. In vivo efficacy and pharmacokinetics of AC98-6446, a novel cyclic glycopeptide, in experimental infection models. Antimicrob. Agents Chemother. 48:1708-1712.
33. Zeng, J., F. Teng, G. M. Weinstock, and B. E. Murray. 2004. Translocation of Enterococcus faecalis strains across a monolayer of polarized human enterocyte-like T84 cells. J. Clin. Microbiol. 42:1149-1154.(Kavindra V. Singh, Sreedh)