Lack of Correlation of Gelatinase Production and Biofilm Formation in a Large Collection of Enterococcus faecalis Isolates

Enterococci, although part of the normal flora of the human gastrointestinal tract, have been recognized as an important cause of nosocomial infection for over two decades (8, 11) and are commonly implicated in urinary tract infections, bacteremia, intra-abdominal and surgical wound infections, catheter-related infections, and endocarditis (11). Enterococci have also been associated with the production of biofilms in root canals (3), on different biomaterials (1, 7, 9, 20), and on various indwelling medical devices, such as ureteral stents (6), intravascular catheters (17), and silicone gastrostomy devices (2). The genetic determinants involved in biofilm formation by Enterococcus faecalis have just begun to be understood (4, 5, 7, 9, 13, 19, 20). Several studies have evaluated esp (9, 19, 20), including one recent study which reported that the presence of esp was significantly associated with the ability to form biofilm on abiotic surfaces (20); that study also found that insertional inactivation of esp of two strains, but not of another, markedly impaired biofilm production. We have reported a lack of association between esp presence and the presence or absence of biofilm among a large collection of clinical isolates from various sources; however, we found that greater amounts of biofilm formation were highly significantly associated with the presence of esp (9). We also found that endocarditis isolates more often made biofilm than nonendocarditis isolates and also made significantly more biofilm (9). We further demonstrated that disruption of epa (enterococcal polysaccharide antigen), atn (autolysin), the fsr locus (E. faecalis regulator), and gelE (gelatinase) genes of E. faecalis OG1RF (an esp-negative strain) markedly reduced biofilm formation in a microtiter plate assay as well as by phase-contrast microscopy (9); this is consistent with other reports about the importance of gelatinase for E. faecalis and the production of biofilm (4, 7, 9, 10, 13). The marked impact of gelE disruption on biofilm production led us to speculate that biofilm production was likely to be greater among naturally occurring gelatinase-producing isolates than among non-gelatinase-producing isolates. To test this hypothesis, we examined the 163 isolates previously analyzed for esp and biofilm (using a microtiter plate assay after 24 h of growth and crystal violet staining [9]) for the production of gelatinase (15, 16) and for the presence of gelE and fsr genes, and we correlated these results with biofilm production.

Among the 163 isolates tested, 71 were GelE⁻ and 92 (56%) were GelE⁺ (all 92 were fsrB⁺ gelE⁺) (Table 1) (16). Eighty-four of the 92 GelE⁺ isolates were biofilm producers (optical density at 570 nm [OD₅₇₀] > 0.5), while 67 of the 71 GelE⁻ isolates produced biofilm (P = 0.55, Fisher's exact test); the median biofilm ODs for GelE⁺ and GelE⁻ isolates were also essentially equal (1.457 versus 1.488) (P = 0.43, Mann-Whitney test). For the subgroup of esp-lacking isolates, the median biofilm OD of GelE⁺ isolates was higher, although not significantly so, than that of GelE⁻ isolates (median OD, 1.155 for the 19 esp-lacking/GelE⁻ isolates versus 1.424 for the 70 esp-lacking/GelE⁺ isolates), suggesting that GelE may contribute to biofilm in the esp-lacking background.

TABLE 1.

Correlation of esp and gelatinase production with biofilm formation by E. faecalis isolates^a

Strain characteristics	OD of isolates with:				Average
	Strong (OD570, >2) biofilm formation	Medium (OD570, 1-2) biofilm formation	Weak (0.5 < OD570 < 1) biofilm formation	No (OD570 ≤ 0.5) biofilm formation
esp+/GelE+	2.456 (2.206-2.717) (10)	1.328 (1.162-1.617) (10)	0.864 (0.810-0.918)b (2)	0.000 (0)	1.633 (1.178-2.456) (22)
esp+/GelE−	2.658 (2.286-3.102) (15)	1.494 (1.267-1.696) (32)	0.823 (0.622-0.890) (5)	0.000 (0)	1.587 (1.267-2.214) (52)
No esp/GelE+	2.692 (2.274-3.060) (8)	1.503 (1.346-1.680) (41)	0.830 (0.795-0.892) (13)	0.279 (0.242-0.383) (8)	1.424 (0.872-1.706) (70)
No esp/GelE−	2.462 (2.143-3.055)b (3)c	1.438 (1.154-1.609) (9)c	0.769 (0.586-0.817)b (3)	0.398 (0.316-0.444) (4)	1.155 (0.586-1.746) (19)
Total	36	92	23	12	163

^aThe median ODs and interquartile ranges are shown; the number of isolates in each group is shown in parentheses. P was <0.001 for median ODs of esp-positive versus esp-negative isolates (median ODs, 1.62 [IQR, 1.25-2.29] versus 1.37 [IQR, 0.85-1.71]), as previously reported (9). Also, 0 of 74 esp-positive isolates generated no biofilm (ODs that were 0.5) versus 12 of 89 esp-negative isolates (P = 0.0014). P was 0.425 for the 92 GelE⁺ versus the 71 GelE⁻ strains. P was <0.02 for 22 esp⁺/GelE⁺ isolates versus the 19 esp-lacking/GelE⁻ isolates.

^bMinimum and maximum ODs are shown when n is ≤4.

^cTwo strong and two medium biofilm producers lacked esp, gelE, and fsrB.

It is of interest (Table 1) that the majority of isolates (122) had either only gelatinase production or only esp, with fewer strains having (22) or lacking (19) both. We also noted that the biofilm ODs were highest for the 22 esp⁺/GelE⁺ isolates (median OD, 1.633; interquartile range [IQR], 1.178 to 2.456) and lowest for the 19 esp-lacking/GelE⁻ isolates (median OD, 1.155; IQR, 0.586 to 1.746) (P < 0.02, Mann-Whitney test), consistent with our previous observation that esp-positive isolates produce more biofilm than esp-negatives ones (9). However, even among the esp-lacking/GelE⁻ isolates, three produced strong biofilm and nine produced medium biofilm (Table 1), indicating that neither esp nor gelatinase is essential for biofilm production.

It has been previously noted that almost all E. faecalis isolates have the gelE gene, but in two studies, only 60% (16) and 62% (15) produced gelatinase by standard assay. The lack of gelatinase production by these gelE⁺ isolates is usually related to the absence of a 23.9-kb region including most of the fsr locus (12, 16), the positive regulator of gelE expression (15). These isolates thus resemble previously described in vitro fsr mutants (14, 15), except that these fsr mutants still have the 23.9-kb region. We have also recently shown that there is very-low-level production of gelatinase, not detected by standard assay, by in vitro fsr mutants, which is due to basal expression from the gelE promoter (18). Even low levels of gelatinase might, in theory, influence biofilm production; indeed, our OG1RF-derived in vitro fsr mutants showed slightly but significantly more biofilm production than a gelE disruption mutant (9). Among the 71 GelE⁻ isolates in the present study, 52 were gelE⁺ but lacked fsrB, while 8 possessed both fsrB and gelE; none of those with fsrB lacked gelE. We also found 11 GelE⁻ isolates that lacked both gelE and fsrB; 9 of these 11 (2 of which also lack esp) produced strong biofilm, and 2 (both lacking esp) produced medium biofilm. These results indicate that gelE (and thus, by inference, low-level gelatinase) is not required for biofilm production among clinical isolates, even among esp-lacking strains.

In summary, despite a marked effect of gelatinase and of in vitro gelE and fsr interruption on biofilm production (4, 7, 9), the present study found no difference in in vitro biofilm production between GelE⁺ and GelE⁻ isolates derived from clinical or fecal sources; this indicates that, as with esp, neither gelE, gelatinase, nor fsr is required for biofilm formation. However, esp⁺/GelE⁺ isolates produced significantly more biofilm than esp-lacking/GelE⁻ ones, and in the esp-lacking subset, there was a trend towards more biofilm production by GelE⁺ than by GelE⁻ isolates, suggesting that gelatinase may contribute to biofilm formation for some naturally occurring strains, as it does for strain OG1RF. The lack of correlation between gelatinase production and biofilm in the overall collection of E. faecalis isolates, despite the marked in vitro effect, in an esp-lacking background, of gelE or fsr disruption or complementation, and the finding of strong biofilm production even among some isolates lacking both gelE and esp indicate that other factors are also important for the marked strain differences in biofilm production among E. faecalis isolates found in nature. Since most GelE⁻ clinical isolates lack a 23.9-kb region encompassing much of the fsr locus, it is possible that the loss of this region compensates in some unknown way for the loss of Fsr function and of GelE production. Other possible explanations for the failure to find reduced biofilm among GelE⁻ versus GelE⁺ isolates include the possibilities that basal expression of gelE in the naturally occurring fsr mutants may be sufficient to allow biofilm formation and/or that other compensatory changes (e.g., mutations or gene acquisition [for example, esp]) have occurred in GelE⁻ strains that endow these organisms with the ability to produce biofilm.