Lactococcus lactis produces short-chain quinones that cross-feed Group B Streptococcus to activate respiration growth


*E-mail; Tel. (+33) 1 34 65 20 80; Fax (+33) 1 34 65 20 65.


Quinones are essential components of the respiration chain that shuttle electrons between oxidoreductases. We characterized the quinones synthesized by Lactococcus lactis, a fermenting bacterium that activates aerobic respiration when a haem source is provided. Two distinct subgroups were characterized: Menaquinones (MK) MK-8 to MK-10, considered as hallmarks of L. lactis, are produced throughout growth. MK-3 and demethylMK-3 [(D)MK-3] are newly identified and are present only late in growth. Production of (D)MK-3 was conditional on the carbon sugar and on the presence of carbon catabolite regulator gene ccpA. Electron flux driven by both (D)MK fractions was shared between the quinol oxidase and extracellular acceptors O2, iron and, with remarkable efficiency, copper. Purified (D)MK-3, but not MK-8–10, complemented a menB defect in L. lactis. We previously showed that a respiratory metabolism is activated in Group B Streptococcus (GBS) by exogenous haem and MK, and that this activity is implicated in virulence. Here we show that growing lactococci donate (D)MK to GBS to activate respiration and stimulate growth of this opportunist pathogen. We propose that conditions favouring (D)MK production in dense microbial ecosystems, as present in the intestinal tract, could favour implantation of (D)MK-scavengers like GBS within the complex.


Quinones are lipophilic, non-protein molecules present in cell membranes, which are essential for respiratory chain activity; they shuttle electrons between oxidoreductases, and concomitantly contribute to maintaining an electrochemical gradient (proton-motive force) (Taber, 1993; Gennis and Stewart, 1996). In bacteria, quinones also accept electrons from proteins other than dehydrogenases, like redox proteins, and participate in multiple cellular functions, such as disulphide bond formation of periplasmic proteins (Bader et al., 1999), defence against oxidative stress (Soballe and Poole, 2000; Maruyama et al., 2003; Wang and Maier, 2004), signalling of the environmental O2 status (Georgellis et al., 2001; Bock and Gross, 2002) and sporulation (Farrand and Taber, 1974). Quinones are implicated in extracellular electron transfer in Shewanella putrefaciens (Newman and Kolter, 2000), and in the circadian cycle of cyanobacterium Synechococcus elongatus (Ivleva et al., 2006).

Most prokaryotes produce three main types of quinones: menaquinones (MK) and MK precursor demethylmenaquinones DMK, both of which are naphthoquinones, and ubiquinones, which belong to the structural group of benzoquinones. In Gram-negative bacteria, (D)MKs are mainly used for anaerobic respiration, and ubiquinones, for aerobic respiration. Gram-positive bacteria usually produce one major DMK species, which is used for either type of respiration (Meganathan, 2001). MK production requires a DMK methyltransferase, the last enzyme in MK synthesis, which some bacteria, e.g. Enterococcus faecalis, are lacking. Allylic chain length of (D)MKs was presumed to be conserved for a given species, and has been used as a criterion for microbial typing (Collins and Jones, 1979). Chain elongation depends on two isoprenyl diphosphate synthases: IspA for up to three units, and IspB, which should give the chain length specificity (Liang et al., 2002). However, several quinone species can be present within a given bacterium: lactic acid bacterium Lactococcus lactis, reportedly produces mainly MK-9, but also MK-8 and MK-10, plus minor MKs (Morishita et al., 1999), while the closely related species E. faecalis produces DMK-9, DMK-8 and minor DMK-7 (Collins and Jones, 1979). Other than methylation of naphthoic group, which affects the midpoint potential of quinones, the impact of these chain length variations on function is not yet known.

Despite the recognized importance of quinones in metabolism, few studies address regulation of their production in bacteria. One common feature is that the amounts of quinone change with growth phase. The quinone pool increases when cells approach stationary phase in Bacillus subtilis, Escherichia coli and Staphylococcus aureus (Farrand and Taber, 1974). This increase might be correlated with extracellular factors such as acid pH, or end-products of glucose degradation, such as acetoin and acetate (Qin and Taber, 1996).

Lactococcus lactis is used industrially for its fermentative capacity. Lactococci switch to aerobic respiration when a haem source is present, resulting in marked increases in growth and stationary phase survival (Duwat et al., 2001). The haem requirement reflects the absence of haem biosynthesis genes on the L. lactis genome (Bolotin et al., 2001). All other respiratory chain components are present in L. lactis, including the complete set of genes needed for MK biosynthesis (Gaudu et al., 2002).

Here, we characterize the nature and function of MKs in L. lactis. Our work revealed the existence of short-chain quinone species (D)MK-3, which in addition to previously reported MK-8–10, are abundant in late exponential phase when lactococci are grown in glucose. (D)MKs are needed for L. lactis respiration and for reduction of O2, Fe and, with high efficiency, Cu. We demonstrate that (D)MK donation by lactococci is the basis for cooperative behaviour with the opportunist pathogen Group B Streptococcus (GBS), which is naturally MK-deficient. These findings point to a significant role for MKs in the bacterial composition of microaerobic ecosystems.


Lactococcus lactis produces quinones that vary as a function of age, but not metabolic growth mode

Several genes putatively involved in MK biosynthesis (yhdB, menE, menB, menD and menF) are clustered on the L. lactis genome (Bolotin et al., 2001; Wegmann et al., 2007). We inactivated menB, coding for a 1,4-dihydroxy-2-naphthoate synthase (DHNA synthase), which catalyses cyclization of o-succinylbenzoyl-CoA into 1,4-dihydroxy-2-naphthoic acid. Quinone profiles of the wild type (WT) and menB mutant were examined by adsorption chromatography (thin-layer chromatography; TLC) (Fig. 1A). Acetone extraction of early exponential phase WT cells (OD600 = 0.6) grown on glucose in static fermentation yielded one band with a retention factor (Rf) of 0.6 in our conditions, compared with Rf = 0.55 for the MK-4 standard. This band (called fraction-a) was absent in the menB mutant, indicating that it corresponded to a L. lactis quinone.

Figure 1.

Quinone production varies as a function of growth phase and carbon source. Quinones extracted from WT and menB strains were analysed by TLC.
A. Profiles of extracts from exponential (E, OD600 = 0.5) and stationary (S, 24 h) phase cells. Elution from preparative silica gel cartridge with 10% diethyl ether in heptane. Fs, static fermentation; Fa, aerated fermentation; R, respiration-permissive (aeration plus haem). Rf of the control quinone, MK-4 (ctl), was 0.55.
B. Separation of MK species by two-step extraction. MKs were eluted from silica gel cartridge with 3% diethyl ether in heptane. Stationary phase static cultures (grown in GM) were harvested, and MKs extracted directly by acetone (acet*), or by the two-step protocol: HIP mixture followed by acetone extraction (acet). Samples were adjusted to equal amounts of wet cells.

The WT strain was cultured on glucose under conditions of static fermentation, aerobic fermentation or respiration (aerobically with haem). Extracts of early exponential (OD600 = 0.6) or stationary phase (> 16 h) cultures were analysed for quinones (Fig. 1A). Fraction-a was produced in all tested conditions. Two new nearly comigrating bands appeared in stationary phase extracts, with Rf values (0.52 and 0.51) close to that of MK-4. As WT, but not menB extracts, produced these species, they were assumed to be MKs and are referred to collectively as fraction-b. A time-course experiment indicated that fraction-b appeared once cell density attained an OD600 = 1.0 (data not shown). Quinone profiles were similar for extracts derived from static, aeration or respiration (i.e. plus haem and O2) cultures. These analyses show that L. lactis produces several quinones and that fraction-b production depends on growth phase but not on metabolic growth mode. Significantly, respiration metabolism is not a major inducing factor for quinone production.

Distinct compositions of L. lactis quinones

We developed a two-step procedure to differentially extract each quinone fraction from cells: a first treatment with heptane–isopropanol (HIP) mixture extracted only fraction-b. Subsequent acetone treatment extracted the remaining fraction-a (Fig. 1B). As expected, no quinones were extracted by HIP from cells collected in the early growth phase (OD600 = 0.6; data not shown).

Reverse-phase high-performance liquid chromatography (RP-HPLC) coupled with UV spectrum analyses were used to characterize the quinone fractions. Analysis showed that each fraction was heterogeneous. Fraction-a comprised MK-9 as the major species (∼70% of the fraction), as well as MK-8 (∼20%), MK-10 (∼3%) and other minor compounds (data not shown). This fraction corresponded to MKs previously identified in lactococci (Collins and Jones, 1979; Morishita et al., 1999). Fraction-b (about 1/3 of fraction-a, based on A245) displayed two main peaks on RP-HPLC (Fig. 2A). A compound with a retention time (RT) of 5.25 min was identified as MK-3 when compared with RTs and UV spectra of MK-4 and MK-7 controls (derived from B. subtilis; Collins, 1985; Fig. 2B). A second compound was eluted slightly before MK-3 (RT = 4.65 min). This compound was identified as DMK-3, by its elution and UV spectrum, plotted using as controls DMKs extracted from E. faecalis, which produces only DMKs (Collins and Jones, 1979; Fig. 2). Treatment of samples from each peak with reducing agent KBH4 resulted in a characteristic shift in UV spectrum, confirming that samples corresponded to quinones (data not shown).

Figure 2.

Identification of short-chain lactococcal quinones by RP-HPLC and UV spectra. Extractions were performed on a saturated L. lactis WT static culture.
A. Analysis of HIP-extracts. MKs were first separated on silica gel cartridge, and then eluted on RP-HPLC with acetonitrile as the mobile phase. Insets show UV spectra of peaks (correspondence indicated by arrow). Insets at right correspond to reference UV spectra for MK-4, and DMKs (from E. faecalis: DMK-3, dotted line; DMK-9, solid line).
B. Plots of isoprenyl unit lengths for short-chain lactococcal (D)MKs. The plot depicts log10(net RT in min) versus number of isoprenyl units, using RP-HPLC results in (A). MK references are MK-4 (+), MK-7 (X; extracted from B. subtilis). DMK references are DMK-3, -7, -8, -9 (Δ; extracted from E. faecalis). The L. lactis peak at RT = 5.247 corresponds to MK-3 (▪); the peak at 4.65 corresponds to DMK-3 (□). Straight lines describing the MKs and DMKs are parallel. Slopes SMK and SDMK of the respective lines are indicated.

Production of (D)MK-3 depends on the carbon source and the presence of catabolite control regulator CcpA

Production of (D)MK-3 only in late growth, when glucose was depleted, led us to test whether changing the carbon source to galactose impacted on (D)MK profiles. Whereas L. lactis produced both quinone subgroups when grown to stationary phase in glucose, cells grown on galactose failed to produce (D)MK-3. Amounts of MK-8–10 were similar for cells grown with either sugar. (D)MK-3 production therefore appears to be regulated by central carbon metabolism.

Carbon sugar had a clear impact on (D)MK profiles. This led us to ask whether deficiency of carbon catabolite control protein, CcpA, a major regulator leading to glucose repression in Gram-positive bacteria, also affected quinone profiles (Luesink et al., 1998; Gaudu et al., 2003; Zomer et al., 2007) (Fig. 3). Interestingly, (D)MK-3 species were present in WT, but absent in ccpA extracts. Nevertheless, production of MK-8–10 was comparable in both strains. These results point to a role for sugar metabolism in (D)MK-3 production by lactococci.

Figure 3.

Inactivation of ccpA abolishes (D)MK-3 production. WT and ccpA strains were grown in static conditions to stationary phase in GM medium. Quinones were extracted by the two-step procedure as in Fig. 1.

Reduction of extracellular O2 and iron by (D)MKs

Quinones mediate reduction of respiratory chain quinol oxidase, but also transfer electrons to exogenous compounds like O2 (Huycke et al., 2001; Korshunov and Imlay, 2006) or metal ions (Saffarini et al., 2002). The capacity of L. lactis to reduce O2 and Fe3+ was investigated in the absence or presence of haem, in early or late exponential phase lactococci (Table 1, Fig. 4). E. faecalis strain OG1RF, which produces large amounts of superoxide, was used as positive control for superoxide production studies; DMKs produced by E. faecalis are required for superoxide production, but in respiration conditions are rerouted to CydAB at the expense of O2 (Huycke et al., 2001).

Table 1.  Extracellular superoxide production in L. lactis is decreased by menB inactivation or by respiration activity.
ConditionsStrainsSpecific activitya
Early expoLate expo
  • a. 

    Superoxide production was determined by an SOD-sensitive cytochrome c reduction assay (see Experimental procedures).

  • Cell suspensions (OD600 = 0.2) were supplemented with 20 μM cytochrome c and reaction started with 1% glucose addition. Results of at least four independent experiments are shown (average, and standard deviation in parentheses). E. faecalis is used as control.

  • Expo, exponential phase; ND, not determined.

AerationL. lactis (pIL252)0.85 ( ± 0.1)1.00 ( ± 0.15)
L. lactis menB0.07 ( ± 0.04)0.05 ( ± 0.08)
E. faecalis OG1RF5.3 ( ± 0.5)ND
Aeration plus haem 10 μML. lactis (pIL252)0.15 ( ± 0.1)0.5 ( ± 0.15)
L. lactis menB0.02 ( ± 0.04)0.23 ( ± 0.1)
E. faecalis OG1RF1.5 ( ± 0.3)ND
Figure 4.

Quinone-mediated metal reduction by lactococci. Late exponential WT and menB cultures grown in GM under aeration or respiration conditions were assayed for metal reduction capacity (see Experimental procedures).
A. Fe3+ reduction. Readings at A562 correspond to Fe2-Ferrozine complex formed after 16 h incubation.
B. Cu2+ reduction. Readings at A480 correspond to Cu+1-Bathocuproine complex formed after 15 min incubation. Determinations from cells harvested in late exponential phase are presented; results of early phase cells were essentially the same. Averages (plus standard deviation) are shown for at least four independent experiments. Boxed conditions correspond to those activating L. lactis respiration. SOD: superoxide dismutase.

In early aerobic exponential growth, L. lactis produced extracellular superoxide when only MK-8–10 were detected. A >10-fold drop in superoxide levels in the menB mutant confirmed that MKs were implicated in this reaction. Thus, MK-8–10 produced by L. lactis mediate O2 reduction. Activation of WT respiration activity by haem addition lowered superoxide production by about fivefold, indicating that electron flux via MK is rerouted to the respiration chain. In late aerobic growth, when (D)MK-3 species appear, superoxide production was slightly increased and remained menB-dependent. Unexpectedly, respiration conditions decreased superoxide production only twofold as compared with fermentation. This possibly suggests that quinones are in excess and can mediate both activities. However, superoxide levels were elevated even in the menB mutant in the presence of haem, suggesting the existence of an alternative O2 reduction pathway. The above results show that lactococcal (D)MKs can reduce O2 to superoxide, and that this reaction competes to some extent with CydAB respiration activity.

Iron reduction capacities by WT and menB cells were examined early (data not shown) and late (Fig. 4A) in growth. Fe3+ reduction in both early and late exponential phase was mediated by superoxide, as addition of superoxide dismutase (SOD) abolished the reduction. Accordingly, results for Fe3+ were, with one exception, similar to those observed for superoxides. As with O2, Fe3+ reduction in early exponential cells was diminished in respiration, likely as a result of competition with the respiration chain. Also, in late exponential phase, when (D)MK amounts were higher, significant Fe3+ reduction still occurred in respiration, suggesting that excess quinone levels mediate both activities. However, unlike results on O2, virtually no Fe3+ reduction occurred in the menB mutant, showing that (D)MKs are specifically required. The above results show that lactococcal (D)MKs are responsible for reduction of O2 and Fe3+, and that respiratory chain activity partly competes with these reactions. We suggest that higher quinone production late in growth augments the reducing capacity of L. lactis.

Copper is a direct exogenous electron acceptor of (D)MKs

Cu2+ reduction by early and late exponential phase cultures was similar (Fig. 4B shows results with late exponential phase cells). However, reaction kinetics of Cu2+ reduction in L. lactis was unexpectedly much faster than with iron (i.e. minutes versus hours respectively). In contrast to Fe3+ reduction, SOD addition did not affect the reaction, indicating that superoxide was not involved. As expected, the menB mutant displayed very weak reducing activity, confirming that (D)MKs were actors of Cu2+ reduction. Finally, respiration-grown cells in both early and late exponential phases failed to reduce copper. These results indicate that Cu2+ may be a preferred substrate for (D)MKs compared with O2 or Fe3+.

menB inactivation increases L. lactis resistance to oxidative stress-inducing metals

The reducing properties of quinones have implicated these molecules in defence against oxidative stress (Do et al., 1996; Soballe and Poole, 2000). We compared the capacity of the WT and menB mutant strains to tolerate different oxidative stress conditions, by treating cells with H2O2, tertiobutylhydroperoxide and menadione, or by a shift from anaerobic to aerobic conditions. No differences in WT and menB mutant strain growth and survival were observed under the different tested oxidative challenges (data not shown).

Oxidative stress can be exacerbated by redox metals like iron or copper, which catalyse the Fenton reaction (H2O2 + Metalred→ HO°, HO-, Metalox), and provoke cellular damage in L. lactis (Rezaiki et al., 2004). As (D)MKs reduced Fe3+ and Cu2+, WT and menB growth sensitivity to these metals was tested in aerobic conditions. In the case of Fe3+ (0.6 mM), both strains grew equally well in the presence or absence of the metal (data not shown). However, CuCl2 addition (0.6 mM) markedly slowed growth of WT L. lactis, while the menB mutant was unaffected (Fig. 5). (D)MKs might therefore exacerbate oxidative stress when copper is available.

Figure 5.

Copper resistance is increased by menB inactivation. WT (circles) and menB (squares) strains were grown aerobically in GM medium (closed symbols). Parallel cultures contained 0.6 mM CuCl2 (open symbols). Results shown are representative of three experiments.

Purified (D)MK-3, but not MK-8–10, restores L. lactis menB mutant respiration capacity in trans

Differences in physical characteristics and expression of MK-8–10 and (D)MK-3 could impact on their roles in cell physiology. We examined the capacity of each species to restore menB mutant respiration. As expected, the menB mutant is strictly fermentative when grown in the presence of haem. Addition of MK-4 to haem-containing medium complemented the respiration defect as reflected by the gain of biomass and decreased acidification. Interestingly, the HIP-extracted fraction, as well as RP-HPLC-purified MK-3 and DMK-3 fractions, each restored respiration efficiency when added at only 2 μM (data not shown). In contrast, the MK-8–10 fraction (or RP-HPLC-purified MK-9) failed to restore menB respiration even when added at ∼20 μM (data not shown). These experiments could indicate that (D)MK-3 is a better substrate for respiratory chain activity than MK-8–10. Alternatively, exogenously added (D)MKs with a longer isoprenoid chain may be unable to insert into the membrane. As cells producing only MK-8–10 (e.g. early log- or galactose-grown cells, or the ccpA mutant) can respire, we favour the hypothesis that (D)MK-3, but not MK-8–10, is more mobile for complementation and can readily insert into L. lactis membranes.

Lactococcal (D)MKs cross-feed GBS, which stimulates growth by activating the respiration chain

The opportunist pathogen GBS is a main cause of invasive infection (sepsis, pneumonia and meningitis) in neonates and mortality in immuno-compromised, or elderly adults (Spellerberg, 2000). We recently reported that GBS undergoes a respiratory metabolism when haem and MK-4 are supplied (Yamamoto et al., 2005). We asked whether L. lactis could be a source of quinones for GBS respiration. We first checked whether purified L. lactis (D)MKs can activate the GBS respiration chain. The (D)MK-3, but not the MK-8–10 fraction, activated GBS respiration in the presence of haem (data not shown), as observed above for the L. lactis menB mutant.

The physiological relevance of these results was addressed in co-culture conditions, where GBS was put in contact with a L. lactis strain that could act as quinone donor; haem was provided exogenously (Fig. 6). A plate test was performed in which a broad streak of GBS culture was spread on solid medium containing haem. Drops of WT or menB L. lactis cultures were deposited on the edge of the GBS band. Vitamin K2 was deposited as a positive control. A zone of strong GBS growth stimulation was seen around the WT L. lactis strain, but not around the menB mutant. No such zone was seen when a GBS cydA derivative was used, or if plates lacked haem.

Figure 6.

Cross-feeding in vivo of L. lactis (D)MKs activates GBS respiration. GBS strain NEM316 WT or the cydA mutant is streaked as a broad band on solid medium containing 10 μM haem. L. lactis WT or menB cultures (3 μl) were deposited on the border of the streak. Note that WT L. lactis, but not the menB mutant can undergo respiration on plates because of the presence of haem. Vitamin K2 was deposited as positive control. L. lactis WT culture provides (D)MKs to stimulate respiration growth of GBS WT, but not of the GBS cydA mutant.

Group B Streptococcus respiration leads to increases in pH and acetoin production (Yamamoto et al., 2005). These features were evaluated in liquid overnight co-cultures of GBS with different L. lactis strains: the quinone donor was L. lactis strain cydA::ISS1, which is respiration-deficient (Rezaiki et al., 2004); the L. lactis menB strain was used as negative control (Fig. 7). The pair WT GBS × L. lactis cydA showed a clearly higher culture pH and acetoin content than the respiration-negative controls, indicating that GBS underwent respiration growth (L. lactis alone generates a culture pH of 4.5, and accumulates small amounts of acetoin in these conditions; data not shown). These results show that GBS benefits from lactococcal quinones, which activate its respiratory chain and thus stimulate growth.

Figure 7.

Growth characteristics and metabolic by-products in mono- versus mixed L. lactis–GBS co-culture. L. lactis (L. lac) and GBS strains were inoculated in a 1:1 ratio in GM supplemented with 1 μM haem, and were cultured under aeration for overnight at 37°C. Acetoin (grey bars) and pH (white bars) production was determined from culture supernatants. Note that pH and acetoin production are higher in conditions where GBS respires.


Bacterial quinones are essential for respiration, and also participate in numerous redox reactions. Although quinones have been long thought to be species-specific, and have even been used for bacterial taxonomy (Hess et al., 1979; Collins and Jones, 1981), a closer look at quinone synthesis indicates that quinone profiles may significantly change with the growth condition. In addition to MK-8–10, we showed that L. lactis produces (D)MK-3 species late in growth, which can represent around 25% of quinones produced by lactococci. These species seem to remain cell-associated, as no MKs could be detected in a cell-free culture supernatant (data not shown). Interestingly, galactose [which alleviates CcpA repression (Luesink et al., 1998)], or ccpA inactivation, abolishes (D)MK-3 production without significantly affecting MK-8–10 production. This may indicate that pathways leading to production of these quinones are differentially regulated. We were initially surprised that (D)MK-3 levels were increased in late growth phase, when CcpA is expectedly less active. However, this could be explained if regulation of (D)MK-3 synthesis requires CcpA independently of its role in glucose repression. Recent transcriptomic analysis of a ccpA mutant (Zomer et al., 2007) did not reveal modified expression of genes expectedly involved in isoprenyl chain elongation, such as ispA, ispB or fni (a 2-isopentenyl diphosphate isomerase), possibly indicating that regulation is indirect (Ludwig et al., 2002). In conclusion, qualitative differences in quinones associated with changes in growth conditions suggest that they may be differentially expressed to satisfy their diverse uses in response to the environment.

One role of lactococcal quinones is in respiratory chain activity. (D)MKs and CydAB are produced throughout growth and, during early growth, the respiration chain likely uses MK-8–10. However, later in growth, CydAB might accept electrons from different (D)MK species. The ability of (D)MK-3 to restore respiration of the menB mutant indicates that MKs with different length side-chains can promote L. lactis respiration, as was proposed in yeast (Okada et al., 1998). Increased amounts of (D)MKs late in lactococcal growth may allow cells to respire and simultaneously mediate other reduction activities. The finding that ferric reduction persists even in respiration conditions (Fig. 4) supports this view.

Quinones are mediators of metal reduction. L. lactis does not produce siderophores, but it does encode potential ferrous transporters. The capacity of quinones to reduce Fe3+ may be a means for L. lactis to assimilate this metal. Strikingly, Cu2+ reduction by (D)MKs occurred in minutes, compared with hours for Fe3+ reduction, and did not involve a superoxide intermediate. Furthermore, (D)MKs appear to be the unique site of Cu2+ reduction, as no reduction occurred in the menB mutant. This Cu2+ reduction property might mimic a respiration-like activity. In agreement with this suggestion, L. lactis growth was reportedly stimulated by low copper concentrations (Kaneko et al., 1990). Copper in human tissues is reportedly at concentrations of 1–4 μg ml−1 (Lech and Sadlik, 2007). These levels are only around 10-fold lower than those used in growth and metal reduction assays (40 and 15 μg ml−1). Lactococcus lactis interactions with copper are thus potentially relevant under in vivo conditions.

A third role for quinones could be to generate reactive oxygen species to eliminate competing cells in the surrounding environment. In E. faecalis, extracellular superoxide was proposed to favour its development in the host by destroying gut epithelial cells, probably mediated via Fenton reaction (Huycke et al., 2002) based on iron or copper. As respiration activity markedly decreases superoxide production (Table 1), sites where bacterial respiration occurs may be less hostile for neighbouring cells.

The commensal bacterium and opportunist pathogen GBS can grow via respiration only if haem and quinones are supplied externally. We recently provided evidence for a role of respiration in GBS virulence and persistence in blood, indicating that this bacterium finds these compounds in the host (Yamamoto et al., 2005). While it seemed obvious that haem can be obtained from red blood cell degradation (Rouault, 2004), the source of (D)MKs, which are not synthesized by animals, was not clearly identified. (D)MKs present in human fluids are likely provided by ingested foods and by bacteria that transit or reside in the intestines. Bacterial cross-feeding has previously been reported for other vitamins and cofactors, such as folate and nicotimanide or menaquinone (Kerr and Tritz, 1973; Myers and Myers, 2004; Graber and Breznak, 2005). In this work, co-culture experiments showed that L. lactis can supply (D)MKs for GBS respiration. Purified (D)MK-3, but not (D)MK-8–10 activated GBS respiration. However, we cannot rule out that the latter species can activate GBS respiration in co-culture conditions; (D)MKs might be delivered upon bacterial lysis or by direct cell–cell interaction.

Although L. lactis is not part of the normal flora, it can persist for 3–4 days after ingestion in the human gastrointestinal tract (Klijn et al., 1995). In view of the close contact between bacteria under colonization conditions, we suggest that food bacteria like lactococci, or more generally, commensal bacteria that synthesize (D)MKs (e.g. E. faecalis, Enterobacteriaceae and Bacteroidaceae) and/or haem, may cooperate with intestinal commensal neighbours like GBS to activate their incomplete respiratory chains, and thus stimulate their growth. A very recent study of E. coli persistence in a mouse intestinal model revealed that cydAB inactivation causes a significant colonization defect in a competitive assay, thus supporting the view that respiration is important for colonization and that oxygen, albeit at low concentrations, is normally available (Jones et al., 2007). In our initial report showing GBS respiration capacity, we provided evidence that respiration is important for GBS infection, but proposed that the anaerobic gut environment would preclude respiration in that compartment (Yamamoto et al., 2005). Based on our present results and the evidence for oxygen in the gut, we revise this model, and suggest that respiration may favour GBS colonization.

Cross-feeding of quinones may be modulated by diet or changing flora. Moreover, non-digestible sugars that accumulate in the lower intestine could also affect cross-feeding, as we have shown that sugars (i.e. glucose and galactose) modulate production of short-chain (D)MKs. The availability of (D)MKs for in trans activation may be a critical point in considering the impact of diet on bacterial interactions in the gut ecosystem.

Experimental procedures

Strains and culture conditions

Strains and plasmids used in this work are listed in Table 2. Plasmid pIL252 confers erythromycin (Em) resistance, and was established in L. lactis WT strain MG1363 in experiments where strain selection was needed. L. lactis strains were cultured at 30°C in M17 medium (Difco) containing 1% glucose (GM) or 1% galactose, supplemented as necessary with 2 μg ml−1 Em. GBS, Streptococcus agalactiae strain NEM316 and derivatives were grown in GM at the exception of inoculum in M17 supplemented with 0.2% glucose and kanamycin 1 mg ml−1 as required, at 37°C. E. faecalis strain OGRF1 was grown in GM at 37°C without antibiotic. E. coli strains were cultured in Luria–Bertani broth to which ampicillin 100 μg ml−1 was added as needed. A 20 mM haem (Sigma) stock solution was sterile-filtered and used where indicated at 10 μM or 1 μM final concentration in L. lactis or GBS respectively. A 20 mM stock solution of MK-4 (chain comprising four isoprenyl residues; Sigma) was prepared in ethanol and used where indicated at 10 μM final concentration. For growth, an overnight static culture was diluted in fresh medium, to obtain an initial OD600 of 0.025, and cultures were then grown in static conditions, or in aeration without or with added haem, as described (Duwat et al., 2001). For liquid co-culture experiments, L. lactis and GBS strains were grown in a 1:1 ratio (OD600 = 0.025 for each strain) in fresh GM containing 1 μM haem under aeration at 37°C.

Table 2.  Strains and plasmids used in this study
 Main characteristicsaReference
  • a. 

    Emr, Ampr, Kanr, resistance to Em, ampicillin and kanamycin respectively.

E. coli
 Top10F- mcrA Δ(mrr-hsdRMS-mcrBC) Φ80ΔlacΔM15 ΔlacX74 recA1 araD139Δ (ara-leu)7697 galU galK rpsL (StrR), endA1 nupGInvitrogen
L. lactis
 MG1363Plasmid-free derivative of NCDO712Gasson (1983)
 menBEmr, MG1363, menB::pRV300This study
 ccpAEmr, MG1363, ccpA::pRV300Gaudu et al. (2003)
E. faecalis
 OG1RFNatural isolateHuycke et al. (2001)
S. agalactiae
 NEM316Serotype III isolated from neonatal blood cultureGlaser et al. (2002)
 NEMJ01Kanr, NEM316, cydA::aphA-3Yamamoto et al. (2005)
 pIL252Emr, L. lactis low copy vector, derivative of pAMβ1Simon and Chopin (1988)
 pCRII-TOPOAmpr Kanr, cloning vectorInvitrogen
 pRV300Ampr, Emr, cloning vectorLeloup et al. (1997)

Construction of the menB mutant

The MG1363 menB gene was inactivated by single cross-over insertion. A 537 bp internal menB DNA fragment was amplified by polymerase chain reaction (PCR) using non-degenerate primers designed from the IL1403-published sequence (Bolotin et al., 2001). To assure hybridization (IL1403 and MG1363 genomes share 80% identity), the primers corresponded to highly conserved menB regions. The forward primer was 5′-GGAATTGCTAAAATTACCATCAATCGTCC-3′; the reverse primer was 5′-TGGAACAACAGTATTAACCATTCCCATGTC-3′. The amplified menB fragment was first cloned into the pCRII-TOPO vector (Invitrogen) and established in E. coli, then re-cloned into the EcoRI site of pRV300, which is a suicide vector for L. lactis that confers Em resistance (Leloup et al., 1997). The insert was confirmed by DNA sequencing, and the resultant plasmid was then transformed into L. lactis MG1363 (Holo and Nes, 1989). Transformants were plated on solid medium containing Em and MK-4 (vitamin K2) in an anaerobic jar (BioMerieux). menB inactivation in Em-resistant clones was confirmed by Southern blot hybridization (ECL Kit), using the menB PCR-amplified fragment as probe.

Demethylmenaquinone extraction and identification

Cells grown in GM were harvested by centrifugation at 3400 g for 30 min at 4°C and washed twice in NaCl 0.9%. Total extraction was performed with cold acetone (Collins, 1985) as follows: 0.5 g of wet cell pellet was suspended in 2 ml of acetone and treated in an ultrasonic ice-cooled water bath for 15 min. Ten milliliters of acetone and 0.5 g of Na2SO4 were added, and the mixture was shaken end-over-end for 30 min at 4°C in the dark. The acetone phase was recovered after centrifugation at 3400 g for 15 min at 4°C. Two more re-extractions were done on cells with 10 ml of acetone, and the pooled acetone fraction was evaporated under vacuum (temperature < 40°C). Remaining water was eliminated by freeze-drying.

We devised a two-step protocol for differential quinone extraction: cell pellets were resuspended in water (1:1, w/v), to which we added two volumes of HIP (2:1, v/v) mixture. After 30 min shaking in the dark at 4°C, phase separation was done by centrifugation at 3400 g for 5 min at 4°C. The heptane upper phase was collected, and the cell pellet-hydroalcoholic phase extracted again, once with 2 vols of HIP and once with heptane in the same conditions. Cells were sedimented by centrifugation at 3400 g for 15 min at 4°C. The three pooled heptane phases were evaporated under nitrogen with gentle warming (< 40°C). The hydroalcoholic phase was discarded and the cell pellet was then extracted with acetone as described above. All extracts were recovered in ∼1 ml of heptane, and then submitted to partial purification on a preparative 100 mg of silica gel column [SPE DSC-Si silica tube, Supelco Bellefonte, CA; (Morishita et al., 1999)]. (D)MKs were eluted with 0.75–2 ml of 3% diethyl ether in heptane. Samples were conserved for further analysis at −20°C in glass tubes with Teflon-ringed caps after drying under nitrogen.

Thin-layer chromatography was performed on 10 × 5 cm silica gel plates (60 F540, Merck, Darmstat, RDA), using heptane–diethyl ether (85:15, v/v) as the mobile phase. Sample amounts in heptane were all adjusted to correspond to equal cell wet weights (100 mg). MK-4 was used as migration standard. Numeric photography of plates was done with UV light. RP-HPLC was performed with a Waters LC System on a Nucleosil 100 Å C18 5 μ column (150 × 4.6, Colochrom, Gagny, France). The mobile phase was acetonitrile for HIP extracts and acetonitrile–tetrahydrofuran (70:30, v/v) for acetone extracts, at 1 ml min−1. A diode array detector (Waters 2996) was used between 210 and 380 nm. Samples were dissolved in acetonitrile. MK size was determined by plotting log10 of net retention time versus isoprenyl unit number, using MK-4 and MK-7 from B. subtilis and DMKs from E. faecalis as standards (Collins, 1985). Quinone reduction was done with KBH4 in the presence of acetate buffer, as described (Kröger and Dadák, 1969) except that acetonitrile was used instead of ethanol.

O2 and metal reduction determination

Strains were grown in GM and harvested in exponential (OD600 between 0.5 and 0.7) or late exponential phase (OD600 = 2 in fermentation or OD600 = 3 in respiration). Cells were pelleted, washed once in PBS (pH 7.4), and resuspended at a final cell density of OD600 = 0.2 in PBS. Superoxide determinations were performed at room temperature by comparing reduction of 20 μM cytochrome c (Sigma) in identical samples, except that to one sample we added 45 units of SOD (Sigma). Activity was determined as a superoxide-dependent increase of absorption at A550; specific activity was defined as nanomoles cytochrome c reduced per min per cell density OD600 = 1, using a value of 21 per mM per cm as absorption coefficient (Imlay and Fridovich, 1991). E. faecalis was used as a superoxide-producing control strain (Huycke et al., 2001). Reduction of Fe3+ to Fe2+, and of Cu2+ to Cu+ was performed as described (Coves and Fontecave, 1993; Rapisarda et al., 1999). Cells were resuspended in 1 ml of PBS containing 0.1 mM Fe-citrate and 2 mM ferrozine (Sigma), or 0.2 mM CuCl2 and 2 mM bathocuproine sulphonic acid (Sigma), for respective reduction assays. For Fe3+ reduction, samples were incubated 16 h at 30°C, and Fe2+–ferrozine complex formation in supernatants was read at A562562 = 27.9 mM−1 cm−1), whereas for Cu2+ reduction, samples were incubated 15 min at 30°C and absorption of complex was read at A480480 = 12.9 mM−1 cm−1). Reactions were started by adding glucose (1% final concentration). Determinations were done at least four times.

Oxidative stress sensitivity

Effects of the menB mutation on oxidative stress sensitivity were tested by comparing aerobic growth of the WT and mutant strains, without or with addition of the following oxidizing agents at OD600 = ∼0.6: H2O2 (1 and 4 mM), menadione bisulfite (5 μM), tertiobutylhydroperoxyde (10 mM). Growth was followed for 8 h. We also performed plate inhibition tests, in which exponential phase cells (OD600 = ∼0.5) or 24 h stationary phase cells were suspended in 5 ml of soft agar (0.6% agar) to a final density of OD600 = 0.025, and plated on GM solid medium. Disks (0.9 cm) containing 30 μl of H2O2 (1 M), or menadione (100 mM) were then deposited on plates. After 24 h incubation at 30°C, diameters of inhibition zones were measured, and compared between mutant and WT strains.

Metal-sensitive growth tests

Fe3+ and Cu2+ sensitivities were determined by comparing OD600 of L. lactis strains grown aerobically in GM supplemented or not with 0.6 mM Fe3+-citrate or CuCl2.

Use of L. lactis (D)MKs by GBS for respiration growth

Group B Streptococcus overnight static cultures were prepared in GM medium, diluted 1:50 in PBS, and ∼50 μl was spread as a band on GM plates containing 10 μM haem. M17 medium shows some variability; GM medium was supplemented with 1% yeast extract, and 3% each glycerol and sucrose, which stimulated MK production in B. subtilis (Sato et al. 2001). After drying, a 5 μl drop of L. lactis WT or menB culture was deposited at the edge of the GBS culture band. After drying, plates were incubated at 30°C overnight and photographed.

Acetoin and pH determinations in co-culture medium

After overnight growth of an L. lactis–GBS co-culture (1:1) in GM supplemented with 1 μM haemin, pH was taken, and acetoin production was measured from cell-free supernatants by gas chromatography as described (Yamamoto et al., 2005).


We are grateful to L. Cauchy, R. Boudjelloul and A. Maes for technical assistance. We thank our colleagues from the BLO laboratory and C. Garrigues, E. Johansen, M. Pedersen and H. Møllgaard (Chr Hansen A/S, Denmark), C. Poyart (Hôpital Cochin), P. Trieu-Cuot (Institut Pasteur) and J.-M. Faurie (Danone) for stimulating discussions. This work is supported in part by a research Grant Chr-Hansen A/S, Denmark, and partial thesis funding (L.R.).