Spent culture supernatant (SCS) of the probiotic Lactobacillus rhamnosus GG had been reported to exert antibacterial activity against Salmonella typhimurium. However, the chemical identity of the antimicrobial compound(s) responsible remained unknown. A survey of the antimicrobial compounds produced by L. rhamnosus GG was performed. Lactobacillus rhamnosus GG produced a low-molecular weight, heat-stable, non-proteinaceous bactericidal substance, active at acidic pH against a wide range of bacterial species. SCS of L. rhamnosus GG grown in MRS medium contained five compounds that could meet the above description, if present at the appropriate concentration. Based on different experimental approaches, it could be concluded that under the growth conditions tested, the strong antimicrobial activity of L. rhamnosus GG against Salmonella was mediated by lactic acid.
In different ecological niches such as the gastrointestinal tract, lactic acid bacteria (LAB) produce a variety of antimicrobial compounds (Ouwehand, 1998; Bongaerts & Severijnen, 2001; Servin, 2004). LAB which, when administered in adequate amounts, confer a health benefit on the host, are referred to as probiotic strains (FAO/WHO, 2001). One such strain is Lactobacillus rhamnosus GG (Sherwood & Gorbach, 1996) for which many health benefits have been postulated, including the defeat of intestinal pathogens. Lactobacillus rhamnosus GG spent culture supernatant (LGG-SCS) was reported to be antimicrobial against Salmonella enterica serovar Typhimurium (S. typhimurium) and other intestinal pathogens (Silvaet al.,1987; Hudaultet al.,1997; Lehto & Salminen, 1997). Silva (1987) reported that L. rhamnosus GG secretes an antimicrobial substance, distinct from lactic acid, with inhibitory activity against other bacteria in the pH range from 3 to 5. Despite many speculations, the purification and/or structural identification of the antimicrobial compound(s) has not been described. Therefore, this study aimed at unequivocally identifying these antimicrobial compound(s) produced by L. rhamnosus GG in vitro.
Materials and methods
Bacterial strains and growth conditions
Lactobacillus rhamnosus GG (ATCC 53103) (Sherwood & Gorbach, 1996) and Salmonella typhimurium SL1344 (Hoiseth & Stocker, 1981) were grown at 37°C. Lactobacillus rhamnosus GG was inoculated from glycerol stocks (–80°C), propagated twice in De Man–Rogosa–Sharpe medium (MRS; Difco) (De Manet al.,1960) and grown in nonshaking conditions. Salmonella typhimurium was grown in Luria–Bertani (LB) broth (Sambrooket al.,1989). Solid media contained 1.5% (w/v) agar.
Coculture of L. rhamnosus GG and S. typhimurium
The coculture was performed as described previously (Dragoet al.,1997), with modification of the coculture medium used. One liter of coculture-medium was composed of a mixture of the ingredients for 1 L of LB (Sambrooket al.,1989) and for 1 L of MRS medium (De Manet al.,1960) to which 1 L of distilled water was added. For glucose containing medium, filter sterilized glucose solution was added at a final concentration of 20 g L−1.
Preparation of SCS
MRS broth was inoculated with L. rhamnosus GG with a starting concentration of 2.35 × 107 CFU per mL. SCS was obtained from a 24 h culture by centrifugation for 30 min at 10 000 g at 4°C. To monitor the production of antimicrobial compounds during the growth of L. rhamnosus GG, samples were taken and used to prepare SCS at distinct time points. Growth of L. rhamnosus GG was monitored by measuring the optical density at 600 nm with an UV-visible light spectrophotometer (UV/VIS Lambda 2, Perkin Elmer). A pH ranging from 4.5 to 3.7 was observed for LGG-SCS over time. For standardization, the pH of tested LGG-SCS was adjusted to 4.5 with HCl or NaOH for all experiments, followed by filter sterilization (0.22 μm; Millipore).
Sensitivity of antimicrobial compounds to heat and proteolytic enzymes
The physical and enzymatic treatment of the SCS was performed as described previously (Bernet-Camardet al.,1997; Coconnieret al.,1997). Briefly, the LGG-SCS (24 h; pH 4.5) was heated at 110°C for 1 h. To test the sensitivity to proteases (all purchased from Sigma), the LGG-SCS was incubated at 37°C for 1 h with and without pronase E (200 μg mL−1), trypsin (200 μg mL−1), proteinase K (100 μg mL−1), or pepsin (200 μg mL−1). As an alternative method to eliminate proteinaceous compounds, the flow-through of a Microcon-SCX filter (Millipore, cut-off 3 kDa) was used.
Determination of organic acid concentration
A commercial kit for the determination of d- and l-lactic acid was used (Roche). For acetic and formic acid, an Aminex HPX 87 H HPLC column (Bio-Rad Laboratories) was used. This column was also used for routine qualitative assessment of organic acids present in SCS. Control experiments were carried out using commercial lactic, acetic and formic acid (VWR International). The column was used with 6 mM HCl as carrier liquid at 0.6 mL min−1 flow rate, the column temperature was 35°C, and the elution of the compounds was monitored at 210 nm.
Tests for antimicrobial activity of SCS
Monitoring bacterial growth via Bioscreen assay
An overnight culture of S. typhimurium was washed in phosphate-buffered saline (PBS) and±105 CFU per mL were applied to growth medium to which LGG-SCS was added and with the final pH adjusted with HCl or NaOH as indicated in the text. Different concentrations of LGG-SCS were tested. 1 : 12 dilutions of SCS were routinely used as this is the highest concentration of LGG-SCS which still allows growth of Salmonella. Bacteria were grown, and the optical density at 600 nm was measured automatically each 30 min during at least 80 h in a BioscreenC instrument (Labsystems Oy). For each time point, the average optical density was calculated from three independent measurements. The generation time (g) was calculated as follows: g=[(t2−t1) log 2]/[log OD2−log OD1] with t, time; OD, optical density at 600 nm; 1 and 2 are successive time points in exponential growth phase.
Radial diffusion test
The radial diffusion test was performed as described previously (Coconnieret al.,1997). Briefly, 5 × 106 CFU per mL of S. typhimurium were added to 10 mL of Trypticase soy broth (TSB) agar [1% agarose, 0.02% (v/v) Tween 20] (42°C) and poured into a square Petri dish. The test material (15 μL) was applied in a well punched in the agar and incubated for 3 h at 37°C. Subsequently, the plates were overlaid with 10 mL of sterile TSB agar solution (1% agarose). Clear zones were measured after 18–24 h incubation at 37°C.
Viability assays were essentially performed as described previously (Coconnieret al.,1997). Briefly, colony count assays were performed by incubating ±108 CFU per mL with the different test solutions at 37°C. Initially and at predetermined intervals, aliquots were removed, serially diluted and plated on LB agar to determine colony counts. The results of the viability assay experiments are represented as mean values of the three independent replicates.
Microtiter plate assay
Fractions of liquid chromatographic runs were subjected to a modified growth inhibition assay. Specific amounts (10–200 μL) of fractions of different sources were applied to the wells of a 96-well plate and lyophilized. The dry material was resuspended in 100 μL of a Salmonella inoculated LB culture (final optical density at 600 nm of 0.150, pH 5.5). The 96-well plate was incubated at 37°C (non-shaking) and every 2 h the optical density at 600 nm was measured (VERSAmax, Molecular Devices).
Purification of antimicrobial substances from LGG-SCS
Chromatographic purification steps were performed as previously reported (Huttunenet al.,1995; Coconnieret al.,1997), with minor modifications. Consequently, interpretation of all data refers only to those materials that were soluble in the elution buffers used. Both freeze dried powder and methanol-acetone extract (Coconnieret al.,1997) of the LGG-SCS were dissolved in the elution buffer (50 mM NHAc, pH 4.8) resulting in 660 mg mL−1 final concentration of dry material. In the first purification step, 0.5 mL of the sample was loaded and fractionated on a gel filtration Biogel P2 column (1.5 × 30 cm; exclusion 100–1800 Da; Bio-Rad Laboratories). The active fractions of gel filtration were pooled and loaded on DEAE-Sephacell anion exchange column for further purification. Elution was carried out using a gradient elution program (Huttunenet al.,1995). Active fractions originating from DEAE separation were pooled and subjected to Rainin C8 RP-HPLC column (250 × 4.6 mm) for further purification (Huttunenet al.,1995). This resulted in the elution of one major peak with antimicrobial activity (elution time 5.6 min). A portion of the pooled active fractions (20 μL) was subjected to organic acid analysis using the Aminex HPX 87 H HPLC column (Bio-Rad Laboratories). To identify organic acids, standard solutions of lactic acid, acetic acid, formic acid, and pyroglutamic acid (PCA) (Sigma) were used.
Results and discussion
Production of antimicrobial compounds by L. rhamnosus GG parallels its growth curve
A relatively constant antimicrobial activity against Salmonella typhimurium SL1344 appeared after 7.5 h in culture of Lactobacillus rhamnosus GG (clear zone size in radial diffusion test corresponding to 0 mm at 0, 2.5 and 5 h, 6 mm at 7.5 h, 7.5 mm at 12 h, 8.5 mm at 24, 34, and 48 h). Concentrations of lactic acid, mainly the l-enantiomer, increased during time and were approximately 70 mM at 7.5 h, 136 mM at 12 h, and 215 mM at 24, 34, and 48 h. The pH determined during the time course ranged between 6.0 (0 h) and 3.7 (24–48 h). For subsequent experiments, the SCS prepared from 24 h old MRS L. rhamnosus GG cultures (LGG-SCS) was used.
The antimicrobial activity of LGG-SCS was examined as a function of contact time (Fig. 1). Salmonella viability decreased rapidly after 1 h of contact with LGG-SCS (±2 logs) and dramatically after 3 h (8 logs) (Fig. 1). In contrast, the controls PBS and MRS (pH 4.5) did not show bactericidal effects.
Lactobacillus rhamnosus GG produces a pH dependent, heat-stable, nonproteinaceous, low molecular weight antimicrobial compound
When tested at a pH of 6.6, the antimicrobial activity of LGG-SCS was no longer present as is reflected by the faster generation of Salmonella compared with the one observed at pH 5 (Table 1). To rule out the possibility that the antimicrobial activity of LGG-SCS was only due to the low pH, the growth in sterile MRS at pH 5 was tested. In this case, a shorter generation time and lag phase of Salmonella growth was observed as compared with LGG-SCS at pH 5.0. This can most probably be related to the presence of undissociated organic acids in the LGG-SCS; the pH influences the ratio of dissociated to undissociated acid, as explained by the Henderson–Hasselbach equation. Although both forms can inhibit bacterial growth, the undissociated form of organic acids was reported to be more inhibitory, per mole, than its corresponding dissociated form (Eklund, 1983; Presseret al.,1997).
|Condition||Final pH||Lag (h)||Generation time (g) (h)|
|LB + SCS||5||34||2.62|
|LB + SCS||6.6||4||0.79|
|LB + MRS||5||22||1.35|
|LB + MRS||6.6||4||0.79|
|Condition||Final pH||Lag (h)||Generation time (g) (h)|
|LB + SCS||5||34||2.62|
|LB + SCS||6.6||4||0.79|
|LB + MRS||5||22||1.35|
|LB + MRS||6.6||4||0.79|
LB, Luria–Bertani; MRS, De Man–Rogosa–Sharpe; LGG, Lactobacillus rhamnosus GG; SCS, spent culture supernatant.
The characteristics of the antimicrobial activity of LGG-SCS (pH 4.5) were examined, using S. typhimurium as indicator strain (Fig. 2). As a control, sterile MRS at the same pH 4.5 was used, showing±20% inhibitory activity. This inhibitory activity can be partially attributed to acetic acid as sterile MRS contains 60 mM sodium acetate as a basic ingredient. The antimicrobial activity of LGG-SCS (set at 100%) was compared with that of sterile MRS supplemented with lactic acid. While dl-lactic acid at a concentration of 250 mM and at pH 4.5, had 1.5-fold more antimicrobial activity as compared with LGG-SCS, dl-lactic acid sodium salt (250 mM and at pH 4.5) showed no effect on Salmonella growth. The growth inhibitory effect of LGG-SCS could not significantly be alleviated by the addition of proteases, by passing over a Microcon-SCX filter, or by heat treatment. Therefore, the antimicrobial activity of L. rhamnosus GG against S. typhimurium is not due the production of a bacteriocin (Stevenset al.,1991). This is in line with the broad spectrum of antimicrobial activity of LGG-SCS, including other lactobacilli, Pediococcus pentosaceus and even L. rhamnosus GG itself (data not shown). The latter would be very unusual in view of bacteriocin production, as generally, the producer strain is immune against its own bacteriocin (Baba & Schneewind, 1998). Consequently, as expected, ammonium sulphate precipitation and subsequent chloroform/methanol extractions could not indicate the presence of bacteriocins in LGG-SCS (data not shown). Note that, although rarely, some bacteriocins are inactivated after chloroform/methanol extraction.
On the other hand, the antimicrobial activity of LGG-SCS was lost upon dialysis (Fig. 2). Conclusively, and in line with previous results with Escherichia coli as indicator strain (Silvaet al.,1987), under the conditions tested, the antimicrobial activity of LGG-SCS is due to (a) heat-stable, nonproteinaceous low molecular weight compound(s).
Purification of antimicrobial substance(s) of L. rhamnosus GG
Throughout purification, the organic acid composition of the antimicrobial fractions was monitored, as preliminary characterization suggested that undissociated organic acids might mediate the antimicrobial activity of LGG-SCS. As a control, the organic acid profile of sterile MRS medium was analysed, revealing besides several non-identified minor peaks, peaks of two unidentified acids (5.7 and 7.6 min), of acetic acid (AA) (14.6 min), and of pyroglutamic acid (PCA) (17.6 min) (Fig. 3a). Compared with this control chromatogram (Fig. 3a), the one of LGG-SCS (Fig. 3b) contained two additional peaks, one corresponding to lactic acid (LA) (11.9 min) and one to formic acid (FA) (13.3 min) (with lactic acid being the major compound).
The methanol-acetone extract of LGG-SCS contained lactic, formic and acetic acid, PCA and traces of several unidentified acids (Fig. 3c). The same major acid compounds were found in the active fractions separated by the Biogel-2 column. The major peak with antimicrobial activity, eluted from a RP-HPLC column (data not shown), showed to contain after organic acid analysis an unidentified compound also present in sterile MRS (5.7 min), lactic acid (11.9 min), acetic acid (14.6 min), PCA (17.9 min) and a compound having the same retention time as formic acid (13.3 min) (Fig. 3d). Note that the enrichment during purification obviates estimating the actual concentrations of organic acids present in LGG-SCS.
PCA seemed a good potential candidate, as already previously suggested to be responsible for the antimicrobial activity in LGG-SCS (Yanget al.,1997; Lehto & Salminen, 1997). However, this study showed that PCA is already present in sterile MRS medium, and that its amount does not significantly change during growth of L. rhamnosus GG (Fig. 3a and b). The fact that glutamic acid is converted to PCA when heated above 180°C (Hartmannet al.,1981) can explain why it is also detected in sterile MRS. As in this study production of PCA by Lactobacillus could not be confirmed, it is not the compound conferring the antimicrobial activity in LGG-SCS.
Antimicrobial activity of organic acids identified in the enriched antimicrobial fraction of LGG-SCS
The only compounds present in the active fraction of purified LGG-SCS and absent in sterile MRS are lactic acid and formic acid. Lactobacillus rhamnosus GG is classified as a facultative heterofermentative Lactobacillus (Kandler & Weiss, 1986): glucose is mainly converted into lactic acid while acetic acid and formic acid are produced in much smaller amounts [i.e. at ±20-fold lower concentration than lactic acid (12 mM), data not shown]. While MRS supplemented with such low concentrations of these acids (72 mM acetate final concentration and 12 mM formic acid, respectively), at pH 4.5, did not effect S. typhimurium viability (Fig. 4), 3 h of contact with MRS supplemented with lactic acid (Fig. 4) or with LGG-SCS (Fig. 1), drastically decreased Salmonella viability (8 logs).
Interestingly, when using PBS instead of MRS, the bactericidal effect of lactic acid on Salmonella disappeared (Fig. 4). This clearly demonstrates that the specific composition of the medium or solvent influences the results of comparative viability tests. This medium dependency can be attributed to the equilibrium of dissociated to nondissociated forms of organic acids, which is influenced by the pH and composition of the medium (Thomaset al.,2002). Therefore, care should be taken when conclusions about the role of lactic acid as antimicrobial compound are drawn based on experiments that compare the antimicrobial activity of SCS with the inhibitory effect of solutions containing externally added lactic acid at concentrations higher than present in the SCS (Silvaet al.,1987; Bernet-Camardet al.,1997; Coconnieret al.,1997).
Lactic acid as antimicrobial substance produced by L. rhamnosus GG
The residual inhibitory effect of SCS present after removal of lactic acid was evaluated. After elimination of the lactic acid by dialysis of LGG-SCS against a buffer at pH 4.5, its antimicrobial activity disappeared completely (Fig. 2). However, when this dialysis was performed against a lactic acid solution (20 g L−1) at pH 4.5, the antimicrobial activity remained unaltered, pointing towards lactic acid as major antimicrobial compound.
Additional evidence supporting this role of lactic acid came from a second experiment in which the viability of Salmonella was monitored in coculture with L. rhamnosus GG under different conditions (Fig. 5). As compared with the reference condition at pH 7 (no L. rhamnosus GG present, Fig. 5, condition 1), adding L. rhamnosus GG severely inhibited Salmonella growth (Fig. 5, conditions 2 and 3). Dialysis of the coculture decreased the inhibitory effect of L. rhamnosus GG (Fig. 5, condition 9). The residual antimicrobial activity was of the same magnitude as that observed in the condition without L. rhamnosus GG, at low pH (Fig. 5, condition 4). Comparing the conditions without L. rhamnosus GG at pH 7 (Fig. 5, condition 1) and pH 5 (Fig. 5, condition 4) indeed showed that low pH can only be partially responsible for the observed Salmonella growth inhibition in coculture.
Omission of sugar in the medium does not significantly affect growth of L. rhamnosus GG, while the production of lactic acid is drastically reduced. This medium was used in conditions 4–8 (Fig. 5). In the absence of sugar, and at pH 7 (Fig. 5, conditions 7 and 8), Salmonella viability was restored to a level observed in the reference [pH 7, no L. rhamnosus GG present (Fig. 5, condition 1)]. Similarly, no significant difference in viability could be observed between the reference condition at pH 5 (Fig. 5, condition 4), and the cocultures at the same pH 5.0 lacking a sugar source (Fig. 5, conditions 5 and 6). Although not very plausible, antimicrobial compounds, if present in these conditions, are clearly independent of the presence of a fermentable sugar and their effect exerted on Salmonella just equals that of low pH. The increased antimicrobial effect observed in conditions 2 and 3 is thus clearly dependent on the presence of fermentable sugars and thus must be related to a fermentation product of L. rhamnosus GG, able to be dialysed, identified as lactic acid.
So far, in literature the role of lactic acid in the antimicrobial activity of L. rhamnosus GG is controversial with hypotheses being formulated that range from no role at all (Silvaet al.,1987; Fayol-Messaoudiet al.,2005) to lactic acid being the major player (Makras et al., 2005). In this study, however, clear indications for the latter hypothesis were found. The exact mode of action underlying this observed antimicrobial effect of lactic acid is not yet completely clarified (Ricke, 2003). Besides exerting its activity through lowering the pH and through its undissociated form (Rubinet al.,1982; Presseret al.,1997), lactic acid is also known to function as a permeabilizer of the Gram-negative bacterial outer membrane (Alakomiet al.,2000) allowing other compounds to act synergistically with lactic acid (Niku-Paavolaet al.,1999). By their chelating properties, organic acids such as lactic acid can capture elements essential for growth, such as iron, being another potential mechanism for inhibition (Presseret al.,1997). Lactic acid also specifically influences the expression of the Salmonella key virulence gene hilA (Durantet al.,2000).
Conclusively, HPLC analysis identified five components in the antimicrobial fraction of LGG-SCS: an unidentified compound with retention time of 5.7, acetic acid, PCA, formic acid and lactic acid. As the first three compounds were also present in sterile MRS medium and formic acid, at its actual concentration in LGG-SCS was not bactericidal against Salmonella, lactic acid was identified as the main antimicrobial compound produced by L. rhamnosus GG. Despite the fact that differences in L. rhamnosus GG physiology between in vivo and in vitro conditions could complicate generalization of these results, lactic acid remains an intriguing compound for further studies.
S. De Keersmaecker and K. Marchal were Research Associates of the Belgian Fund for Scientific Research (FWO-Vlaanderen) when this study was conducted. I. Nagy and T. Verhoeven were supported by the IWT through project STWW-00162. Additionally, this work was partially supported by GBOU-SQUAD-20160 of the IWT.