Lactate is mainly fermented to butyrate by human intestinal microfloras but inter-individual variation is evident


Catherine Michel, UFDNH – INRA, Rue de la Géraudière, BP 71627, F-44316 Nantes cedex 03, France (e-mail:


Aim:  To assess the role of lactate as a precursor for butyrate biosynthesis in human colonic microflora.

Methods and Results:  Three human faecal microfloras were incubated in vitro with media supplemented with 30 mmol l−1 unenriched or 13C-enriched lactate. Lactate metabolism and short-chain fatty acid (SCFA) production were quantified. Lactate conversion to butyrate was investigated by gas chromatography-mass spectrometry and the pathways involved were identified by 13C nuclear magnetic resonance spectroscopy. All human faecal microfloras rapidly and completely fermented lactate, yielding approx. 19 mmol l−1 total SCFAs. However, the SCFA composition varied markedly between microfloras. Butyrate was the main end-product for two microfloras but not for the third (60 and 61%vs 27% of the net concentration of SCFA produced respectively). The latter was typified by its ability to produce propionate as a major product (37%), and valerate (3%). 13C-Labelling showed that butyrate was produced through the acetyl-CoA pathway and that the three microfloras possessed significant differences in their metabolic pathways for lactate consumption.

Conclusions:  In contrast to the ruminal microflora, the human intestinal microflora can utilize both d- and l-lactate as precursors for butyrate synthesis. Inter-individual variation is found.

Significance and Impact of the Study:  This study suggests that the butyrogenic capability of colonic prebiotics could be related to lactate availability. These findings will direct the development of selection strategies for the isolation of new butyrate-producing bacteria among the lactate-utilizing bacteria present in the human intestinal microfloras.


Butyrate is a short-chain fatty acid (SCFA) produced by the intestinal bacteria during fermentation of nondigestible carbohydrates (NDC). Based on its physiological properties, butyrate is perceived as a key factor in controlling colonic mucosal homeostasis (Scheppach et al. 2001). Therefore, favouring its production may be strategic for preventing colonic pathologies such as colorectal cancer or inflammatory bowel disease.

Controlling butyrate production in the colon is difficult, however, due to a limited knowledge of the intestinal bacteria that are responsible for butyrate synthesis (Pryde et al. 2002). Butyrate-producing bacteria (BPB) are assumed to belong mainly to the genera Clostridium, Eubacterium and Fusobacterium (e.g. Clostridium butyricum, Eubacterium limosum) (Holdeman et al. 1977; Salyers 1995). From both the enumeration of these species in human faeces (Finegold et al. 1977, 1983; Moore and Moore 1995) and from an estimation of total BPB by most-probable number after cultivation of human faecal microflora on glucose-containing medium (Macfarlane and Gibson 1994), BPB in the large bowel would account for approximately 1% of total cultivable gut bacteria. Such an estimate does not match with the butyrate concentrations observed in the large bowel, which accounted for up to 23% of total SCFA in sudden-death victims (Cummings et al. 1987) and in animals fed resistant starch (RS) or fructo-oligosaccharides (FOS) (Martin et al. 1998; Le Blay et al. 2003), two substrates known to enhance butyrate synthesis (Macfarlane and Englyst 1986; Michel et al. 1998). Supporting this contradiction, Schwiertz et al. (2000) failed to demonstrate any correlation between increases in butyrate production and changes in numbers of Eubacterium spp. in humans consuming RS.

Several hypotheses, which are not necessarily mutually exclusive, have been proposed to explain the apparent discrepancy between the low numbers of BPB reported and the actual production of butyrate. First, the BPB could be very metabolically active (Macfarlane and Gibson 1994; Sharp and Macfarlane 2000). Secondly, the main intestinal BPB could not yet have been isolated because of poor culturability (Salyers 1995; Sharp and Macfarlane 2000; Pryde et al. 2002). The latter hypothesis is supported by the fact that new culture-independent characterizations of the intestinal microflora based on molecular techniques suggest that <50% of the microbiota has been described (McCartney 2002). Moreover, this issue has been successfully addressed by Barcenilla et al. (2000) and Sharp and Macfarlane (2000) who, using a more sophisticated isolation strategy or a combination of molecular and conventional methods, identified 74 and one new BPB respectively, all belonging to Clostridium clusters. Thirdly, Salyers (1995) has suggested that butyrate synthesis may be influenced by some environmental conditions that were not reproduced when metabolic characterization of pure isolates was performed. In this respect, it is noteworthy that higher proportions of butyrate are synthesized by the same microfloras when carbon flux is increased through modulation of the dilution rate (in vitro; Sharp and Macfarlane 2000) or of the transit time (in vivo; El Oufir et al. 1999) or when the continuous culture of faeces is slightly acidified (Michel et al. 1998).

Implicit in Salyers's hypothesis is that using glucose to assess the capability of purified isolates to produce butyrate will only detect bacteria that produce butyrate directly from sugar utilization. Yet, carbohydrate fermentation requires bacterial trophic chains (Macfarlane et al. 1994) among which different organisms might catalyse (i) depolymerization, (ii) sugar utilization and production of intermediate metabolites such as hydrogen, lactate or ethanol, and (iii) conversion of these intermediates into end-products. Thus, if most of the colonic butyrate synthesis relies on bacteria from the last group, these bacteria would not have been identified as BPB.

That this is the case is supported by numerous observations arising from different models suggesting that butyrate synthesis could stem from lactate conversion by intestinal bacteria. Kanauchi et al. (1999) first proposed this hypothesis after observing that the co-culture of E. limosum and Bifidobacterium longum led to higher butyrate and lower lactate production as compared with that in pure cultures. This was also suggested by Kabel et al. (2002) who observed that during in vitro incubations of human faecal microfloras, lactate decreased in favour of butyrate. Furthermore, Megasphaera elsdenii, which has been identified as a predominant lactate utilizer in ruminants (Newbold et al. 1987) and which is a normal constituent of the human colonic microflora (Finegold et al. 1974, 1977), is capable of metabolizing lactate to butyrate (Counotte et al. 1981). This conversion is pH sensitive and increases when pH values are decreased to values commonly observed in the human colon with highly fermentable NDC (Cummings and Macfarlane 1991). Interestingly, M. elsdenii supplementation increased faecal butyrate concentrations in rats fed FOS (Hashizume et al. 2003) and its bacterial numbers rose in the presence of butyrogenic NDC as observed in faeces from humans fed FOS (Hidaka et al. 1986) or in pig caecal microfloras incubated with gluconic acid (Tsukahara et al. 2002).

It is apparent, therefore, that lactate could be a major precursor for butyrate synthesis by the human intestinal microflora. In order to test this hypothesis, in vitro incubations of human faecal microfloras supplemented with 13C-labelled lactate have been carried out to characterize lactate fermentation by human intestinal microbiota. The percentage of lactate that was converted into butyrate has been quantified and the metabolic pathways involved in this conversion have been identified.

Materials and methods

Human faecal microflora incubations

Faecal samples were collected from three separate healthy individuals (A, B and C), two males (51 and 30 years old) and one female (36 years old) with no history of antibiotic treatment over the preceding 3-month period and a normal nonvegetarian varied diet. All were methane excretors and none was a lactose malabsorber.

Faecal slurries were prepared from fresh faeces homogenized with anaerobic culture medium adapted from Gibson and Wang (1994) and composed of (g l−1): tryptone 2·5; yeast extract 0·5; peptone water 0·5; pectin 0·5; xylan 0·5; arabinogalactan 0·5; mucin 0·5; guar gum 0·5; starch 0·5; K2HPO4 2; NaHCO3 0·2; NaCl 4·5; MgSO4·7H2O 0·5; CaCl2·2H2O 0·45; cysteine 0·8; MnCl2·4H2O 0·2; haemin 0·05; FeSO4 0·005; CoCl2·6H2O 0·05; bile salts 0·05. Tween-80 (2 ml l−1) was also added in order to avoid bacterial clumping and to stimulate Gram-positive anaerobes growth. A solution of vitamins (5 ml l−1), which had been sterilized by filtration (Minisart® 0·20 μm, Sartorius, Palaiseau, France), was added to the cooled growth medium after autoclaving (121°C, 21 min). This consisted of (mg l−1): pyridoxine-HCl 20; panthothenate 20; p-aminobenzoic acid 10; nicotinic acid 10; thiamine 10; riboflavin 10; biotin 4·0; folic acid 4·0; menadione 2·0; vitamin B12 1·0; vitamin K1 0·015.

The faecal slurry was filtered through six layers of surgical gauze under CO2 flux to remove food residues. An aliquot (100 ml) of the filtrate was immediately used to inoculate a glass reactor (200 ml working volume) containing 100 ml of sterile culture medium supplemented or not with a test substrate. Anaerobic conditions were maintained by sparging the culture with O2-free N2/CO2 (80/20) gas. Temperature (37°C) was automatically regulated by a circulating water-bath and pH (5·8) maintained by the addition of 0·5 mol l−1 NaHCO3 or 0·5 mol l−1 HCl directed by a pH controller, as described in Michel et al. (1998). The cultures were carried out for 8 h under continuous stirring. Human faecal slurries from donors A, B and C were incubated with five different media:

  • ibasal medium
  • iibasal medium supplemented with 30 mmol l−1 of sodium d,l-lactate (60% syrup, Sigma, Saint-Quentin Fallavier, France)
  • iiibasal medium supplemented with 30 mmol l−1 of sodium l-lactate (Sigma)
  • ivbasal medium supplemented with 30 mmol l−1 of sodium l-lactate enriched at 20% with sodium l-[1-13C]lactate (99% enriched; Euriso-top, Saclay, France)
  • vbasal medium supplemented with 30 mmol l−1 of sodium l-lactate enriched at 10 or 20% with sodium l-[3-13C]lactate (99% enriched; Euriso-top, Saclay).

Incubations were carried out in duplicate except for media (ii) and (iv). In all incubations, samples were harvested after 0, 2, 4 and 8 h for measurement of lactate and SCFA concentrations and identification of labelled compounds, and after 8 h for quantification of 13C-label into butyrate.

Analysis of SCFA and lactate

Samples for lactate analysis (500 μl) were immediately mixed with a 0·1 mol l−1 TRIS-ethanolamine solution (pH = 9·2) in 1 : 1 (v/v) proportion and stored at −70°C prior to analysis. d- and l-lactate concentrations were determined enzymatically (Kit Boehringer, Rhône-diagnostic, France) from thawed and centrifuged (10 000 g, 20 min, 4°C) samples. d- and l-lactate concentrations (mmol l−1) were summed (total lactate concentration) and relative proportions of l- and d-isomers were calculated as percentages.

Samples for SCFA analysis (750 μl) were immediately mixed with a HgCl2/H3PO4 solution (37 and 880 mmol l−1 respectively) in 1 : 10 (v/v) proportion, and stored at −20°C prior to analysis. After thawing and centrifugation (10 000 g, 20 min, 4°C), supernatants were analysed by gas chromatography (GC) using 95 mmol l−1 4-methylvaleric acid (added in 1 : 10, v/v proportion) as internal standard (Jouany 1982). SCFA production is expressed in two forms:


NMR spectrometry

Supernatants from centrifuged samples (10 000 g, 10 min, 4°C) were filtered (Minisart® 0·20 μm, Sartorius) then stored at −20°C until analysis. Thawed supernatant (3·0 ml) was acidified (pH = 2·0) by adding 6·38 mol l−1 potassium phosphate buffer (200 μl) and an aliquot (800 μl) was transferred to a 5-mm nuclear magnetic resonance (NMR) tube containing 200 μl of D2O (Euriso-top, Saclay). The 13C-NMR spectra were recorded using a spectrometer (Bruker Biospin, Wissembourg, France) fitted with a dual probe 13C/1H tuned at the recording frequency of 125·76 MHz, and using the ERETIC method (electronic reference to access in vivo concentrations) to provide a quantitative reference (Akoka et al. 1999). Chemical shifts for SCFA and lactic acid were assigned by comparison with literature values and confirmed by recording spectra of authentic reference compounds under identical conditions. Peak areas were measured using Perch software (Perch NMR SoftwareTM, University of Koupio, Koupio, Finland) then corrected using the ratio of the ERETIC area in the test sample to the ERETIC area in the reference sample. Labelling due to natural 13C-abundance, measured from spectra obtained with unenriched lactate samples, was deducted, taking into account the exact proportion of unlabelled lactate in the medium (80 and 90% for l-[1-13C]lactate and l-[3-13C]lactate respectively). Data were expressed as percentages of relative enrichment.

Gas chromatography-mass spectrometry

Samples for gas chromatography-mass spectrometry (GC-MS) analysis were centrifuged (10 000 g, 10 min, +4°C) then supernatants were filtered (Minisart® 0·20 μm, Sartorius) and stored at −70°C until analysis. Thawed samples were diluted (1 : 200) with highly purified water and organic acids extracted and derivatized with N-tert-butyldimethylsilyl-N-methyl-trifluoroacetamide (MTBSTFA; Fluka, Saint Quentin-Fallavier, France) according to Moreau et al. (2003) using 2-ethylbutyric acid (Sigma) as internal standard. The derivatization was achieved by incubating 500 μl of the pooled diethyl ether extracts with 50 μl MTBSTFA at room temperature for 45 min.

Derivatized samples (1 μl) were injected into an HP5890 series II gas chromatograph connected to a 5971A quadrupole mass spectrometer (Hewlett-Packard, Palo Alto, CA, USA). Chromatographic separation was performed in splitless mode on an OV-1 capillary column (30 m × 0·25 mm i.d., 0·25 μm film thickness; Interchim, Montluçon, France) according to Moreau et al. (2003). The mass spectrometer was used under selected ion monitoring acquisition mode at m/z 145, 146, 147 and 148 for unlabelled butyrate (M), monolabelled butyrate (M + 1), bilabelled butyrate (M + 2), and trilabelled butyrate (M + 3) respectively. 13C-Enrichments were calculated from GC-MS peak areas with the following ratio: (M + n) to [(M) + (M + 1) + (M + 2) + (M + 3)] with n = 1 to 3 according to the number of atoms of 13C incorporated. Only values that exceeded ratios from natural 13C-abundance by more than 1% were considered as significant. The concentration of each labelled butyrate was calculated using the enrichment values obtained by GC-MS and the total butyrate concentration determined by GC analysis. Concentrations of labelled butyrate were corrected for natural 13C-abundance, estimated from incubations with nonenriched l-lactate, and taking into account the proportion of unenriched l-lactate in the incubation medium. Finally, as NMR analysis showed that butyrate was produced from lactate by the acetoacetyl-CoA pathway, the percentage of lactate that was converted into butyrate was calculated as follows: [monolabelled butyrate] + 2 × [bilabelled butyrate] × 100/[initial labelled lactate].

Statistical analysis

When appropriate, mean values and standard deviations were calculated. Two-way analysis of variance (anova) was performed using the StatView® 5·0 syst 9 package (Abacus Concepts, Berkeley, CA, USA) in order to assess the effect of substrate, the effect of microflora and the effect of their interactions. When a significant effect was found, the mean values were compared using Fisher's test. Statistical significance was accepted at the P < 0·05 level.


The statistical analysis of the utilization of lactate and the production of SCFA revealed significant differences between the ways in which each of the three microfloras fermented the available substrate (e.g. P between floras < 0·001 for propionate, butyrate and valerate total concentrations at 8 h, P between floras <0 ·01 for residual lactate at 2 h). This was not due to the nature of the substrate supplied (Pmicroflora × substrate > 0·1). Hence, it was apparent that the differences were intrinsic to each individual microfloral isolate: data are therefore presented specifically in relation to the individual microfloras.

Lactate utilization by human faecal microfloras

All lactate sources were entirely consumed between 4 and 8 h by the three microfloras investigated (Fig. 1). For a given microflora, no significant differences were observed between lactate sources with respect to residual lactate concentration measured at each experimental time. This shows that the intensity of lactate utilization was affected neither by the isomer nor by the labelling of the lactate provided.

Figure 1.

Kinetics of residual lactic acid concentrations (mmol l−1, mean ± SD). during incubations of three human faecal floras in the presence of d,l-lactate ( inline image), l-lactate (○), [1-13C]lactate (bsl00072) or [3-13C]lactate (bsl00000). For each incubation time and each microflora, there was no significant difference between lactate sources (see P values)

For microfloras A and C, lactate consumption was completed by the fourth hour of incubation, as shown by the average concentrations of residual lactate calculated for all lactate sources at this time (0·6 ± 0·8 and 0·8 ± 0·5 mmol l−1 respectively). In contrast, microflora B only attained such low residual concentrations of lactate (0·3 ± 2·6 mmol l−1) after 8 h of culture.

As expected, the l-isomer proportions were statistically different at 0 h between d,l-lactate- and l-lactate-supplemented media (P = 0·025; 51·6 ± 2·7 and 84·8 ± 18·1% respectively), with no difference between all l-lactate sources (P = 0·885). This difference was no longer detectable at 2 h (P = 0·395; 56·6 ± 4·1 and 64·5 ± 7·1% for d,l- and all l-lactate respectively), suggesting that racemization occurred in the media containing l-lactate. Both isomers were equally consumed by the microfloras, as shown by the consistency of l-isomer proportions from d,l-lactate between 0 and 2 h.

SCFA production from lactate by human faecal microfloras

Whatever the lactate source, lactate supplementation resulted in significantly (P < 0·0001) increased total concentrations of the sum of SCFA as compared with the basal medium after 8 h of incubation (Table 1). On average, this increase reached 18·6 ± 6·0 mmol l−1 and resulted in significantly higher net concentrations of butyrate (+10·0 ± 3·7 mmol l−1; P < 0·0001) and propionate (+6·4 ± 1·6 mmol l−1; P < 0·0001), whereas the net concentrations of acetate and minor SCFA were not affected (+1·7 ± 4·9 mmol l−1, P = 0·403 and +0·6 ± 1·4 mmol l−1, P = 0·295 respectively). Overall, this production constituted an average yield of 70·2 ± 10·3% when expressed as carbon balance, with no differences between lactate sources (P = 0·781).

Table 1.  Total concentration of short-chain fatty acid (SCFA, mmol l−1) produced by three human faecal microfloras after 8 h of incubation with basal medium or basal medium supplemented with various sources of lactate
 Sum of SCFAAcetatePropionateButyrateMinor SCFA
Basal medium (i)
 All floras42·06·120·93·46·82·17·92·66·51·0
  Flora A46·60·322·13·68·71·19·81·35·91·0
  Flora B38·910·619·96·24·61·98·43·26·00·6
  Flora C40·41·720·71·26·90·55·40·57·40·6
d,l-lactate medium (ii)
 All floras61·15·924·22·212·52·117·54·96·81·8
  Flora A65·5 24·6 14·2 21·0 5·7 
  Flora B55·7 21·9 10·0 17·4 6·3 
  Flora C57·7 25·9 11·7 10·7 9·4 
l-lactate medium (iii)
 All floras62·54·823·34·113·51·918·55·67·22·1
  Flora A65·66·222·96·015·00·321·60·46·00·8
  Flora B60·21·721·22·111·01·222·02·56·00·9
  Flora C60·22·725·92·013·70·610·51·210·10·0
l-[1-13C]lactate medium (iv)
 All floras61·80·621·53·714·22·718·77·67·42·9
  Flora A61·8 18·5 15·3 21·4 6·6 
  Flora B61·3 20·5 11·2 24·7 5·0 
  Flora C62·4 25·6 16·1 10·1 10·6 
l-[3-13C]lactate medium (v)
 All floras59·05·120·84·113·42·217·86·26·92·4
  Flora A59·311·118·27·214·61·521·03·25·50·8
  Flora B58·90·420·30·411·01·822·13·85·41·2
  Flora C58·72·223·90·214·51·810·40·39·90·4
All lactate sources
 All floras61·14·722·53·713·42·018·15·57·12·1
  Flora A63·56·421·65·114·70·821·31·35·90·7
  Flora B59·22·120·91·210·91·121·73·15·70·9
  Flora C59·72·325·21·414·11·710·40·610·00·4
P between microfloras 0·244 0·239 0·0001<0·0001 <0·0001
P microflora × substrate 0·970 0·946 0·842 0·101 0·139
P between all substrates 0·0004 0·637<0·0001<0·0001 0·321
P between lactate sources 0·749 0·598 0·191 0·525 0·777

The net concentration of SCFA produced from lactate fermentation was not significantly different for all microfloras (16·9 ± 6·4, 20·3 ± 7·8 and 19·3 ± 3·0 mmol l−1 for microfloras A, B and C, respectively; P = 0·575) but the nature of the major end-product(s) of lactate fermentation varied. This is illustrated in Fig. 2, where net concentration changes over time for each SCFA produced are averaged from all lactate sources (as similar amounts of SCFA were produced at each experimental time within a given microflora, whatever the lactate source). Microfloras A and B consistently produced significantly more butyrate than microflora C (P < 0·0001), which produced significantly more propionate at the beginning of incubation (P = 0·0002 at 2 h) and more valerate from the second hour as compared with microfloras A and B (P = 0·020; P < 0·0001; P < 0·0001 at 2, 4 and 8 h respectively). Consequently, for microfloras A, B and C, respectively, butyrate accounted for 60·2 ± 10·5, 61·4 ± 15·9 and 26·6 ± 3·2% of the total absolute net SCFA production after 8 h of incubation, while valerate accounted for 0·9 ± 0·5, 1·9 ± 1·1 and 13·2 ± 1·7% at 8 h.

Figure 2.

Kinetics of absolute net production of individual short-chain fatty acid during incubations of three human faecal floras (flora A: bsl00000, flora B: •, flora C: inline image) in the presence of different lactate sources (mean values for all lactate sources ± SD). For each experimental time, values harbouring different letters were statistically different (P < 0·05)

Identification of labelled compounds produced

Relative enrichment percentages of labelled compounds detected in fermentations conducted in media supplemented with l-[3-13C]lactate are presented in Table 2. The diversity of labelled compounds detected differed between microfloras. Residual [3-13C]lactate was detected with all microfloras at 0 and 2 h but also at 4 h with microflora B. The lower enrichment value observed in the lattermost case suggests that lactate production occurred concomitantly with lactate utilization. [2-13C]Acetate, [2-13C]butyrate and [4-13C]butyrate were produced by all microfloras. Enrichment values were comparable for acetate with all microfloras but higher butyrate enrichment values were obtained with microfloras A and B than with microflora C. In addition, flora A synthesized [3-13C]propionate. Microflora C showed a much more complex metabolism, with label being recovered in [2-13C]propionate and [3-13C]propionate, both exhibiting comparable specific 13C-incorporation. With microflora C, [2-13C]valerate, [4-13C]valerate and [5-13C]valerate were also detected from the fourth hour of incubation.

Table 2. 13C-Enrichment (%) of labelled compounds detected by 13C NMR spectrometry in extracts of human faecal microfloras incubated with l-lactate containing 10%l-[3-13C]lactate
CompoundChemical shift (δ) (ppm)Flora AFlora BFlora C
0 h2 h4 h8 h0 h2 h4 h8 h0 h2 h4 h8 h

When microfloras A and B were incubated with l-[1-13C]lactate, no 13C-compounds were detected except for the substrate provided. Conversely, microflora C produced [2-13C]acetate (2·3%13C-enriched at 4 h), [1-13C]propionate (8·9%13C-enriched at 4 h), [2-13C]propionate (2·7%13C-enriched at 4 h), [3-13C]propionate (2·7%13C-enriched at 4 h), and [3-13C]valerate (8·7%13C-enriched at 4 h).

Percentage of lactate converted into butyrate

Butyrate in samples collected at 8 h of incubation from media supplemented with l-[3-13C]lactate, showed 13C-enrichment higher than natural abundance only in mono- and bilabelled molecules. Their proportions, as related to the total labelled butyrate, were respectively 85·9 ± 2·7 and 14·1 ± 2·7% on average for the three microfloras. When added together, the label recovered in butyrate constituted 53·6 ± 8·8, 55·0 ± 12·1 and 22·9 ± 9·4% of the initial 13C provided as lactate for microfloras A, B and C respectively.

Conversely, in samples collected at 8 h of incubation from all three microfloras incubated in media supplemented with l-[1-13C]lactate, the 13C enrichment ratios of butyrate did not exceed natural abundance by more than 1% and were therefore considered negligible.


Acetate is generally assumed to be the major precursor for butyrate synthesis (Diez-Gonzalez et al. 1999; Duncan et al. 2002). Clearly, however, acetate does not constitute the only substrate for butyrate synthesis. NDC, such as pectin or alginate, are poor butyrogenic substrates although they favour a large production of acetate when incubated with whole human faecal bacteria (Michel et al. 1996). Furthermore, only 50% of the BPB isolated by Barcenilla et al. (2000) were net acetate utilizers. The hypothesis assessed in this study was that lactate could constitute an alternative precursor for butyrate synthesis by human intestinal microflora. In summary, our main results are:

  • both lactate enantiomers are equally utilized by human intestinal microflora;
  • these are both mainly converted to butyrate by certain microfloras;
  • marked inter-individual differences with respect to the metabolic orientation of lactate conversion occur, even in the relatively small sample of three individual microfloras.

Our results clearly show that the human faecal microflora rapidly and entirely utilized lactate and that both d- and l-lactate were similarly fermented by human faecal bacteria. This indicates that the human intestinal microflora differs from ruminal microflora, as the d-lactate fermentation rate by ruminal microflora typically represents only 25–50% of that of d,l-lactate (Counotte et al. 1983). Furthermore, more propionate was produced by the ruminal microflora of cows from d-lactate than from l-lactate (Counotte and Prins 1981). Such a difference could logically stem from the large contribution of protozoa in ruminal lactate fermentation: notably these micro-organisms are reported to utilize l-lactate more rapidly than d-lactate (Newbold et al. 1987).

Moreover, our data clearly demonstrate that l-lactate racemization is at least as rapid as l-lactate consumption as the isomeric ratio remains essentially unchanged during the first 2 h of the incubation. This finding is in agreement with the d- and l-lactate inter-conversion previously observed by Hove and Mortensen (1995).

On average, butyrate was the major net product of lactate conversion by human faecal microfloras. Other SCFA produced were, in decreasing order, propionate, acetate and valerate. These results were corroborated both by the quantification of 13C-enrichment of butyrate by GC-MS, which showed that up to 55% of the [3-13C]lactate was converted into butyrate, and by NMR analysis of 13C-labelled compounds, where [2-13C]butyrate and [4-13C]butyrate were consistently detected.

Our finding that lactate is a precursor for butyrate synthesis is in marked contrast to the findings of those few studies available that are dedicated to lactate metabolism by the intestinal microflora of monogastrics. These have identified propionate and, to a lesser extent, acetate as the major end-products of lactate metabolism (Stevani et al. 1991; Hove and Mortensen 1995; Ushida et al. 2002). However, the present work differs fundamentally from these previous studies. First, the intestinal microfloras investigated by Stevani et al. (1991) and by Ushida et al. (2002) were isolated from pigs. Secondly, our study differed with respect to the composition of the incubation media. All these previous workers used very simple buffers whereas we used a complex medium – as did Gibson and Wang (1994)– in order to sustain the growth of the maximal number of intestinal bacteria. In a preliminary experiment, we also failed to observe lactate conversion into butyrate when microfloras B and C were incubated with lactate in a simple mineral solution supplemented with urea (data not shown). This suggests that some of the lactate-utilizing bacteria, particularly those capable of producing butyrate, may possess complex nutritional requirements, which would not be satisfied in slurries made in essentially mineral solutions. A third probable cause of discrepancy is the pH value at which the incubations were performed. Working pH values were initially set at 6·5, 6·8 and 7·8 by Stevani et al. (1991), Ushida et al. (2002) and Hove and Mortensen (1995) respectively. These values are higher than those measured in the proximal colonic contents (5·4 to 5·9) collected from sudden-death victims (Cummings and Macfarlane 1991) and from rats fed with FOS (5·6), a NDC favouring both lactate and butyrate production in the colon (Le Blay et al. 1999). For this reason, we set the pH value of our incubations at 5·8 and we applied continuous pH control. Acidity has been shown to act as a key regulatory factor in lactate utilization by ruminal microflora (Mackie and Gilchrist 1979; Counotte and Prins 1981) with more lactate being fermented into butyrate at pH 5·8 to 6·0 as compared with pH 6·9 (Counotte et al. 1983). This effect is supposed to result from both the tolerance of lactate-utilizing bacteria to acidity (Mackie and Gilchrist 1979) and from the particularity of M. elsdenii, which produces a variable amount of butyrate from lactate according to the environmental pH (Counotte et al. 1981). A further probable cause of the observed discrepancies is that individual microfloras show marked differences in SCFA production profiles, even when cultivated under carefully controlled conditions. Our results clearly show that all human microfloras do not metabolize lactate in the same way, notably with respect to the relative production of butyrate and propionate. The mix of the three human microfloras investigated by Hove and Mortensen (1995) may have been of a propionate-producing type.

13C NMR spectroscopy was used to elucidate the route by which lactate was converted to butyrate. Essentially, the 13C-labelled compounds detected after incubation of the three microfloras in the presence of [3-13C]lactate were the monolabelled compounds expected from the main metabolic pathways of pyruvate conversion (Bernalier et al. 1999) and valerate synthesis (Ladd 1959) in anaerobic bacteria (Fig. 3). Similarly, both 13C monolabelled [1-13C]propionate and [3-13C]valerate were detected, as predicted from fermentation of [1-13C]lactate according to these pathways (Fig. 3). [2-13C]Acetate was also detected with microflora C, showing that reductive acetogenesis was also active in this microflora. This pathway can theoretically produce [1-13C]acetate as well (Bernalier et al. 1999), but this end-product was not detected, probably because of the lower signal provided by carboxyl groups in our experimental conditions, due to partial saturation induced by the 13C-T1 of these groups (Franconi 2000). Unexpectedly, [2-13C]propionate and [3-13C]propionate were also detected when microflora C was incubated with [1-13C]lactate. The presence of these products suggests that inter-conversion between labelled acetate, resulting from acetogenesis, and pyruvate occurred. This may have been reversibly catalysed by pyruvate-formate lyase or pyruvate ferredoxin oxidoreductase (Gottschalk 1979). Overall, these observations suggest that no new specific pathway was involved in lactate fermentation by human faecal microfloras and that inter-conversion between lactate and pyruvate was highly active.

Figure 3.

Schematic chart of the labelled compounds theoretically produced from [13C]lactate by microbial fermentation pathways. Plain symbols: direct metabolites; empty symbols: subsequent metabolites; framed symbols: labelled compounds actually detected in this study. (a) Labelled compounds from [1-13C]lactate (bsl00079) and (b) labelled compounds from [3-13C]lactate (•)

The 13C-NMR studies also clearly demonstrated inter-individual differences between microfloras with respect to the relative importance of the different metabolic pathways involved in lactate fermentation. Microflora A metabolized lactate through both acetyl-CoA and acrylate pathways. Microflora B seemed to use solely the acetyl-CoA pathway. In microflora C, acetyl-CoA-, succinate-, valerate-syntheses and acetogenesis were all active. Considering that 13C enrichment values were identical for [2-13C]propionate and [3-13C]propionate, the acrylate pathway did not seem active in this microflora. This effectively corroborates the differences evident from SCFA analysis, which show that microflora C produced a larger range of SCFA and less butyrate from lactate as compared with the microfloras A and B. The advantage of 13C NMR is that it indicates throughput of metabolite in a particular pool, rather than necessarily the final accumulation. In this respect, GC and NMR are complementary. Two marked differences were observed between accumulation and throughput using these two analytical approaches.

First, although the net concentration of propionate observed in microflora B was equivalent to that in microfloras A and C (approx. 6 mmol l−1; Table 1 and Fig. 2), no labelled propionate was detected in microflora B by 13C NMR (Table 2). This probably resulted from the combination of the lower total propionate concentrations observed with microflora B as compared with those observed with microfloras A and C (Table 1), and from the diluting out of the 13C-label incorporation from the provided 13C-lactate, due to the concomitant lactate production which was observed with this microflora.

Secondly, 13C NMR showed an enrichment of the acetate pool in all three microfloras, even though the net increase in acetate was low (0·1 to 4·5 mmol l−1). This suggests that lactate-derived acetate was submitted to a rapid turnover with further incorporation principally into butyrate (microfloras A and B) or into butyrate and valerate (microflora C).

These results unequivocally demonstrate differences between the three microfloras in the nature of the main metabolic pathways involved in lactate utilization. Such inter-individual variations have previously been reported for in vitro fermentation studies (McBurney and Thompson 1989; Weaver et al. 1989, 1992; Kabel et al. 2002) but have not been satisfactorily explained. Most probably, these arise from differences in the bacterial composition of the intestinal microflora, even though the relationship between composition and metabolic activity is difficult to establish (Macfarlane and Gibson 1994; Macfarlane and Macfarlane 2003). Taking this into consideration, and that no reliable method currently exists to quantify accurately lactate-utilizing bacteria that belong to different bacterial genera, the bacterial composition of the microfloras investigated in this study was not determined. It is, however, probable that they differ in the number and/or the nature of lactate-utilizing bacteria present, as previous specific enumerations of M. elsdenii, Veillonella parvula and propionibacteria from human faeces showed extreme variation in both the number and the detection frequency between subjects (Finegold et al. 1974, 1977).

Another important influence of differences in the bacterial composition of the microflora would stem from differences in the type of hydrogenotrophs present (i.e. acetogens, methanogens, sulphate-reducing or nitrate-reducing bacteria). Regeneration of redox balance is of crucial importance in the fermentation process. To achieve this, intestinal bacteria can produce H2 either by pyruvate : ferredoxin oxidoreductase linked to hydrogenase or by NADH : ferredoxin oxidoreductase and hydrogenase. The latter reaction is assumed to be the main route for H2 production by intestinal microflora but, in contrast to the former, it requires a low partial pressure of H2 to be thermodynamically attainable. Thus, when the partial pressure of H2 is high, the redox balance is regenerated through production of reduced metabolites such as succinate, ethanol or butyrate (Macfarlane and Macfarlane 2003). As in the present study all the donors were methanoexcretors, it can be assumed that all the three microfloras investigated harboured methanogens. However, acetogenesis was only demonstrated for microflora C. Although methanogens are theoretically more competitive than acetogens, a moderately acidic pH, such as used in our study, may give a competitive advantage to acetogens (Bernalier et al. 1999). In this respect, microflora C was probably more efficient than other microfloras for maintaining a low partial pressure of H2 and therefore for regenerating the redox balance without the need for the production of reduced metabolites. This could explain why microflora C produced less butyrate than microfloras A and B.

Interestingly, similar differences between microfloras with regard to butyrate production were also observed when these microfloras were incubated with FOS as substrate. Thus, microflora C, the low lactate-derived butyrate producer, also produced a smaller proportion of butyrate from FOS (data not shown). Such a similarity suggests that lactate conversion into butyrate is of key importance in butyrate production from NDC.

Overall, this study shows that in conditions mimicking those occurring in the human colonic lumen when highly fermentable NDC are present, lactate can be a major precursor of butyrate synthesis in some individuals. In this respect, the human intestinal microflora clearly differs from the ruminal microflora. These findings will direct the development of selection strategies for the isolation of new BPB among the lactate-utilizing bacteria present in the human intestinal microfloras. Besides, this work underlines the necessity to determine the exact role of colonic butyrate with respect to colonic diseases considering, on the one hand, the inter-individual differences in butyrogenic capabilities demonstrated here and, on the other, the recent interest in prebiotic oligosaccharides (e.g. FOS). These food components are not digestible but are very rapidly fermented once in the large intestine where they transiently induce accumulation of lactate and therefore could greatly affect butyrate production. A greater understanding of the complex interaction between the bacterial composition of the colonic microflora and the various possible routes of lactate metabolism is required in order to model effectively the butyrogenic capacities of different fermentable substrates in different physiological conditions.


C. Bourriaud received financial support from the Regional Council of the Pays de la Loire. The authors are grateful to J. Doré, H. Dumon, M. Champ, M.C. Alexandre-Gouabau and Serge Akoka for helpful discussions and critical review of the typescript; to S. Goupry, P. Maugère and G. Chaigneau for technical assistance; to C. Wrigglesworth (Scientific English, Nantes) for linguistic assistance.